Reduced Stability of Mitogen-activated Protein Kinase Kinase-2 mRNA and Phosphorylation of Poly(A)-binding Protein (PABP) in Cells Overexpressing PABP*

The poly(A)-binding protein (PABP) is an important regulator of mRNA translation and stability. The cellular level of PABP is controlled by regulating its mRNA translation by a feedback mechanism. The important aspect of this mechanism is that PABP binds to an adenosine-rich cis-element at the 5′-untranslated region of its own mRNA and inhibits its translation. To assess the importance of controlling the PABP level, we studied the effect of PABP overexpression on the transcription profile using the microarray technique. In PABP-overexpressing cells, 19 mRNAs showed a reduction in cellular levels due to reduced mRNA stability, and one showed an increase due to increased mRNA stability. Among these mRNAs, the MKK-2 mRNA encodes the protein kinase activator of ERK1/2 kinase, which is involved in the phosphorylation of eukaryotic initiation factor (eIF) 4E. As a result, mRNA translation may be regulated by the cellular level of MKK-2. In this study, we show that the abundance of the MKK-2 polypeptide is reduced in PABP-overexpressing cells. In these cells, the levels of phosphorylated PABP, eIF4E, and ERK2 are also reduced. Treatment of HeLa cells with the MKK-2 inhibitor U0126 reduced PABP phosphorylation, suggesting that the phosphorylation of PABP is mediated by the MKK-2/ERK signaling pathway. Thus, a novel signaling pathway involving MKK-2 and ERK1/2 may down-regulate the activity of PABP and eIF4E by controlling their phosphorylation and compensates for the effect of excess cellular PABP.

Regulation of the rate of protein synthesis is important for the control of cellular growth and differentiation. Cells respond to changes in growth conditions by fine-tuning the rate of mRNA translation. Initiation of mRNA translation is the rate-limiting step and is often regulated by controlling the interaction between the 5Ј-cap of the mRNA and several initiation factors, including eukaryotic initiation factor (eIF) 2 4E, eIF4G, and eIF4A (1). A recent study has shown that the poly(A)-binding protein (PABP) also behaves like a genuine initiation factor because depletion of PABP from a cell-free extract prevents initiation of mRNA translation (2). The 3Ј-poly(A) tail-bound PABP interacts with eIF4G and circularizes the translating mRNA. This process is believed to enhance mRNA translation by promoting recirculation of terminating ribosomal subunits from the 3Ј-end of the mRNA for another round of initiation (3)(4)(5)(6). Although this model is widely accepted, it has not been proven to occur in vivo. In addition, if PABP is essential for mRNA translation, it is possible that PABP participates in mRNA translation by a yet unknown mechanism, as the presence of 3Ј-poly(A) (and thus circularization of mRNA) is not essential for translation initiation of capped mRNAs (7,8).
PABP binds to the 3Ј-poly(A) track of eukaryotic mRNA. It also interacts with polypeptides involved in regulating the translation and stability of mRNA (9 -12). Among the four highly conserved RNA-binding domains of PABP, the first two show specificity toward poly(A), whereas the third and fourth RNA-binding domains can also bind to poly(G) and poly(U) (13)(14)(15). The C-terminal region of PABP does not bind RNA, but can interact with other polypeptides and promotes oligomerization of PABP on poly(A) (16). The C-terminal region contains a highly conserved 74-amino acid-long PABC (for PABP C-terminal) domain that binds to the eukaryotic release factor 3 and two regulators of mRNA translation, Paip1 and Paip2 (17,18). The ability of PABP to interact with eIF4G, Paip1, and Paip2 also resides within its RNA-binding domain (19,20).
Because of the important role of PABP in the initiation step of protein synthesis, a change in the cellular PABP level may affect not only the rate of protein synthesis, but also the protein expression profile of the cells. Therefore, control of PABP expression may be critical for cellular physiology. Accumulated evidences show that PABP expression is regulated primarily at the post-transcriptional level (21)(22)(23)(24)(25)(26)(27). However, in some instances, transcription of the PABP gene can also be regulated (28). A number of studies have shown that regulation of cellular PABP levels is indeed important for embryonic development and growth control. For example, ectopic expression of PABP prevents maturation-specific deadenylation and translational inactivation of maternal mRNAs in Xenopus oocytes (29). PABP overexpression also leads to defects in cell divisions in Schizosaccharomyces pombe (30). Interestingly, in the early stages of cancer, an increase in cellular PABP levels has been observed (31), suggesting a link between growth control and PABP levels.
In this work, we studied the effect of PABP overexpression on cellular mRNA levels in cultured HeLa cells. We report here that 19 mRNAs showed 2-3-fold reduced cellular levels in PABP-overexpressing cells and that the level of only one mRNA (encoding the TIMP-1 polypeptide) was increased by ϳ6-fold. We show that changes in the stability of these mRNAs in PABP-overexpressing cells were responsible for the observed effects on mRNA levels. One of these mRNAs that showed reduced levels in PABP-overexpressing cells encodes MKK-2. Because MKK-2 may regulate the phosphorylation of eIF4E via the ERK1/2 signaling pathway (32)(33)(34), we studied how the stability of MKK-2 mRNA is affected by the cellular PABP level. We identified a novel MKK-2 mRNA stability control element that binds PABP. We show that, as a cellular response to excess PABP, the phosphorylation of PABP along with eIF4E was reduced. In addition, the immediate downstream target of MKK-2, ERK2 phosphorylation was reduced in PABP-overexpressing cells. Using a specific inhibitor of MKK-2, we show that the phosphorylation of PABP is regulated by the MKK-2/ERK1/2 kinase signaling pathway. We propose that down-regulation of MKK-2 expression to control the activity of PABP and eIF4E is a signaling mechanism of mammalian cells to override the adverse effect of excess PABP.

EXPERIMENTAL PROCEDURES
Plasmids-A PABP cDNA clone (pHu73 plasmid) (35) was digested with BamHI and SacII, which removed the first 263 nucleotides of the PABP cDNA containing the adenosine-rich autoregulatory region and released the cDNA insert from the vector (24). The PABP cDNA fragment was cloned at the AvrII/BamHI sites of the pCMV-SPORT-␤-gal vector (Invitrogen) from which the ␤-galactosidase coding region was removed. This vector was named pCMV⌬PABP. A second PABP expression vector was created with the FLAG epitope tag (pCMV⌬PABP-FLAG) at the N-terminal end of PABP using a synthetic double-stranded oligodeoxynucleotide. ␤-Galactosidase reporter constructs containing different regions of the 3Ј-untranslated region (UTR) of MKK-2 mRNA were generated by ligating double-stranded oligodeoxynucleotides with EcoRI/NheI sticky ends to the EcoRI/BamHI fragment of the pCMV-SPORT-␤-gal vector.
To synthesize N-terminally His 6 -tagged PABP, the cDNA was amplified by PCR using pQE primers. The forward primer (5Ј-aaaggatccaaccccagtgcccc-3Ј) contained a BamHI restriction site, and the reverse primer (5Ј-ctaaagcttaaacagttggaacacc-3Ј) contained a HindIII restriction site. PCR was performed using a PfuUltra Hotstart DNA polymerase kit (Stratagene) with an initial denaturing step at 95°C for 5 min, followed by 33 cycles of denaturation at 95°C for 1 min, annealing at 55°C for 1 min, and extension at 72°C for 2 min. The PCR product was digested with the appropriate restriction enzymes (Fermentas GmbH), purified on a 1% agarose gel using a QIAquick gel extraction kit (Qiagen Inc.), and cloned into the pQE80L prokaryotic expression vector (Qiagen Inc.).
Transfection of Cells-Approximately 3 ϫ 10 5 subconfluent HeLa cells grown on a 35-mm dish in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum were used for transfection. Plasmid DNA (2-3 g) was incubated with 10 g of Lipofectamine in 100 l of Opti-MEMI (Invitrogen) for 30 min at room temperature before addition to the cells. Cells were incubated at 37°C for 5 h with the DNA/ liposome mixture in 1 ml of Opti-MEMI. Following incubation, 1 ml of growth medium containing 20% fetal bovine serum was added to the culture. After 12 h of incubation, the medium was replaced with fresh complete Dulbecco's modified Eagle's medium containing 10% fetal bovine serum.
Analysis of RNA-RNAs from mock-, pCMV⌬PABP-, or pCMV⌬PABP-FLAG-transfected cells were isolated using a high pure RNA isolation kit (Roche Applied Science) according to the manufacturer's directions. Human cDNA microarrays (University Network Microarray Centre, Toronto, Ontario, Canada) were used to examine the changes in the gene expression profile during PABP overexpression in HeLa cells. An equal amount (15 g) of RNA from either mock-or PABP-transfected cells was mixed with an oligo(dT) 18 primer, a master mixture for reverse transcription (250 mM Tris-HCl (pH 8.3), 375 mM KCI, 15 mM MgCl 2 , 5 mM dithiothreitol (DTT), aminoallyl-dUTP, and dNTP), and 20 units of reverse transcriptase (Invitrogen). Reverse transcription proceeded for 2 h, followed by hydrolysis of the remaining RNA with NaOH, and the cDNA was purified using a PCR purification kit (Qiagen Inc). The probe was precipitated using isopropyl alcohol and resuspended in 5 l of water. Probes were labeled using Alexa Fluor CyDye during a 1-h incubation period, and the remaining CyDye was removed using the PCR purification kit, followed again by alcohol precipitation. The probes were mixed with hybridization solution (4% each calf thymus DNA and yeast tRNA in DIG Easy Hyb (Roche Applied Science)) and incubated in a prewarmed and prehumidified chamber at 37°C for 18 h in the dark. Microarray slides were washed with decreasing amounts of salt and SDS (first wash, 2ϫ SSC and 0.1% SDS at 42°C for 5 min; second wash, 1ϫ SSC and 0.1% SDS at 25°C for 10 min; and third wash, 0.1ϫ SSC at 25°C for 1 min.). The slides were dried, scanned using an Axon scanner, and analyzed using R Package software (University of California, Berkeley, CA). Each RNA sample was analyzed on three microarray slides, and three independent transfection experiments were performed.
The level of a specific mRNA in the sample was determined by comparative real-time reverse transcription (RT)-PCR using Rotor-Gene 3000 (Corbett Research Australia) (36). An aliquot of total RNA (1 g) was reverse-transcribed at 50°C for 1 h in a total reaction volume of 25 l using SuperScript II reverse transcriptase (Invitrogen) and 150 ng of random primers. After the reaction, 5 l of the cDNA sample was amplified by PCR in a total reaction volume of 25 l using the Platinum SYBR Green qPCR SuperMix-UDG kit (Invitrogen) and 100 ng of the forward (sense) and reverse (antisense) primers specific to individual mRNAs (Table 1). Amplification was performed using an initial denaturation step at 95°C for 4 min, followed by 40 cycles of denaturation at 95°C for 20 s, annealing at 60°C for 20 s, and extension at 72°C for 20 s. The specificity of the PCR product was examined after the final cycle by generating a melting curve with a heating rate of 1°C/s between 72 and 99°C. The data were analyzed using Rotor-Gene 3000 software and the 2 Ϫ⌬⌬Ct method. The relative expression values of all mRNAs were normalized to the ␤-actin mRNA level.
Measurement of the Effect of PABP on Cell Survival and Cell Divisions-To measure cell death, paraformaldehyde-fixed cells were stained with 4Ј,6diamidino-2-phenylindole, and the nuclei were viewed under a fluorescence microscope. The cells were considered dead if the nuclei appeared fragmented and pyknotic, as described previously (37). Approximately 200 cells were examined for each experiment. The cell-doubling time was determined by counting the number of cells in the culture after removing the cells from the plate by trypsinization at 12-h intervals. The cloning efficiency of trypsinized cells was determined by plating ϳ100 cells on a 5-cm 2 dish and counting the number of colonies after 4 -6 days in culture.
Measurement of Protein Levels-Forty-eight hours after transfection, cells were washed three times with phosphate-buffered saline and lysed in protein sample loading buffer (6% glycerol, 2% SDS, 100 mM DTT, and 0.02% bromphenol blue in 60 mM Tris-HCl (pH 6.6)) for 5 min. The polypeptides were separated by SDS-10% PAGE and electrophoretically transferred to a nitrocellulose membrane. After treating the transfer membrane for 1 h with blocking buffer (2% nonfat dry milk and 0.2% Tween 20 in phosphate-buffered saline), the membrane was incubated with a diluted primary antibody for 2 h at room temperature in blocking buffer and then incubated with an alkaline phosphatase-conjugated secondary antibody for 1 h. The bound antibody was detected with 4-nitro blue tetrazolium chloride and 5-bromo-4-chloro-3-indolyl phosphate as described previously (38). In some experiments, a horseradish peroxidase-conjugated secondary antibody was used for detecting antigens by chemiluminescence using a Western Lightning kit (PerkinElmer Life Sciences) as described by the manufacturer.
In Vitro RNA Synthesis-Different regions of MKK-2 mRNA were amplified by PCR using primers containing a T7 promoter ( Table 2). PCR was performed using the PfuUltra Hotstart DNA polymerase kit as described above. The gel-purified PCR product was used as a template for RNA synthesis. The accuracy of the amplicons was confirmed by sequencing.
Transcription was usually performed at 37°C for 3 h in a final volume of 25 l containing 1 g of DNA template, 2.5 mM each NTP, and 20 units of T7 RNA polymerase (Promega) in 20 mM MgCl 2 , 1 mM spermidine, 0.01% Triton X-100, 20 mM DTT, and 40 mM Tris-HCl (pH 8.1). Uniformly radiolabeled RNA was synthesized under similar conditions in a final reaction volume of 25 l containing 150 Ci of [␣-32 P]CTP (MP Biomedicals), and the final concentration of unlabeled CTP was reduced to 25 mol. The contaminating nucleotides, incompletely transcribed products, and DNA template were removed by fractioning reaction mixtures on an 8% polyacrylamide gel under denaturing conditions (51). The amount of RNA and its specific radioactivity were determined using a spectrophotometer (1 A 260 nm unit ϭ 40 g/ml RNA) and scintillation counter, respectively.
Preparation of Cytoplasmic Extracts-HeLa cells were grown on 100-mm dishes to 30 -40% confluency and transfected with 25 g of DNA as described above. Forty-eight hours after transfection, cells were harvested to prepare the cytoplasmic protein extract as described previously (39). In short, cells were grown to 90% confluency in Dulbecco's modified Eagle's medium, harvested by scraping in chilled phosphatebuffered saline (137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , and 2 mM KH 2 PO 4 (pH 7.4)), and collected by centrifugation at 5000 ϫ g. The cells were lysed in hypotonic buffer (10 mM HEPES-KOH (pH 7.5), 3 mM MgCl 2 , 14 mM KCl, 5% glycerol, 1 mM DTT, 0.02% Igepal CA-630, 0.5 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, and 2 g/l aprotinin) by repeated passages through a 28-gauge needle. The cell lysate was centrifuged at 12,000 ϫ g for 2 min at 4°C to remove the nuclei and cell fragments, and the supernatant was stored at Ϫ85°C in small aliquots. The protein concentration was estimated by the method of Bradford (40).
UV Cross-linking Assay-For UV light-induced cross-linking assays, 60 g of cytoplasmic extract or 1-10 ng of purified PABP was incubated with Ϸ2-3 ng (2 ϫ 10 5 cpm) of radiolabeled RNA at 22°C for 10 min in a total reaction volume of 30 l in binding buffer (10 mM HEPES-KOH (pH 7.5), 3 mM MgCl 2 , 140 mM KCl, 5% glycerol, 1 mM DTT, 0.02% Igepal CA-630, 10 g of E. coli tRNA, and 0.001% bromphenol blue). Heparin was added to a final concentration of 10 g/l, and the sample was irradiated by UV light (254 nm, 4000 microwatts/cm 2 ) at 4°C for 5 min. The sample was treated with RNase T1 (25 units)/RNase A (1 g) at 37°C for 5 min. Finally, the sample was boiled in protein sample loading buffer for 5 min and analyzed by 10% SDS-PAGE. The gels were vacuum-dried and autoradiographed.
RNA Electrophoretic Mobility Shift Assay-20 g of cytoplasmic extract was incubated with Ϸ1-2 ng (1 ϫ 10 5 cpm) of radiolabeled RNA for 10 min at 22°C in 18 l of binding buffer. Subsequently, heparin was added to a final concentration of 10 g/l, and incubation was resumed for another 5 min. The sample was treated with RNase T1 (25 units) for   Detection of Phosphoproteins-The phosphorylation of ERK1/2 and eIF4E was measured by Western blotting using the respective phosphospecific antibodies (Cell Signaling Technology, Inc.). Proteins were detected either with an alkaline phosphatase-conjugated secondary antibody using the chromogenic substrate 5-bromo-4-chloro-3-indolyl phosphate or with a horseradish peroxidase-conjugated secondary antibody using a chemiluminescence detection system (PerkinElmer Life Sciences). To detect phosphorylated PABP, the cellular phosphoproteins were labeled with 32 PO 4 3Ϫ . Approximately 3 ϫ 10 5 HeLa cells grown on a 35-mm dish in 10% fetal bovine serum were transfected with the pCMV⌬PABP vector. Following 24 h of transfection, 2 mCi/ml [ 32 P]orthophosphate (9000 Ci/mmol; ICN Biomedicals) was added to the medium and incubated at 37°C in 5% CO 2 for an additional 24 h. The cytoplasmic extracts from mock-and PABP-transfected cells were prepared in hypotonic buffer as described (39). The cytoplasmic extract was adjusted to 1 M NaCl and 1 g/ml tRNA, and 200 g of proteins from each sample was incubated with a 50-l packed volume of poly(A)-agarose (Sigma) for 60 min at 4°C as described previously (13). Beads were washed five times each with 0.5 ml of hypotonic buffer supplemented with 1 M NaCl and 1 mg/ml tRNA. The bound PABP was eluted with 100 l of SDS gel loading buffer and analyzed by SDS-PAGE and autoradiography to detect phosphorylated PABP. The level of total cellular PABP in the eluted fraction was analyzed by Western blotting. Phosphorylated PABP was also detected by Western blotting after separating the cellular proteins by two-dimensional gel electrophoresis using immobilized isoelectric focusing pH 3-10 dry strips (Amersham Biosciences) as described previously (41). Sample preparations and electrophoresis conditions were according to the procedures provided by Amersham Biosciences.
Statistical Analysis-All values in the figures are expressed as the means Ϯ S.E. Student's t test was used to determine whether there were significant differences between sample means. The p values were derived from three to four independent experiments. Differences were considered significant when at p Ͻ 0.05.

RESULTS
Effect of PABP Overexpression on the Gene Expression Profile-In mammalian cells, PABP expression is regulated at the mRNA translational level through two cis-regulatory elements: the terminal oligopyrimidine tract and the autoregulatory sequence (ARS) (23)(24)(25)(26). Although how PABP mRNA translation is controlled has been well studied, the consequence of unregulated PABP expression on cellular physiology is not clear. Therefore, to examine the effect of PABP overexpression on mammalian cells in culture, we transiently transfected cells with a CMV promoter-driven PABP expression construct and examined the changes in several mRNA levels by the microarray technique. To achieve unregulated expression of ectopic PABP, we used a construct lacking both the terminal oligopyrimidine tract and the ARS from the 5Ј-UTR. In addition, we overexpressed both native PABP and FLAG-tagged PABP for our studies. Using coexpression of green fluorescent protein (GFP) to monitor the transfection efficiency by fluorescence microscopy, we found that 80 -90% of the HeLa cells were routinely transfected under the experimental conditions. The results of RT-PCR analyses measuring the PABP and PABP-FLAG mRNA levels showed 2-3-fold higher levels of PABP mRNA in cells transfected with the pCMV⌬PABP-FLAG expression construct compared with the mock-transfected cells (Fig. 1, A and C). Significant levels of PABP-FLAG mRNA were detected using a FLAG-specific primer in the RT-PCR (Fig. 1A). Western blot analyses of the polypeptide levels in cells showed that PABP polypeptide levels were increased in cells transfected with the native PABP expression vector by ϳ3-3.5-fold compared with the mock-transfected cells (Fig. 1, B and D). Similarly, 2-2.5-fold more PABP-FLAG than endogenous PABP was present in cells transfected with the pCMV⌬PABP-FLAG expression vector (Fig. 1B). The mocktransfected and PABP-overexpressing cells showed very few dead or apoptotic cells by 4Ј,6-diamidino-2-phenylindole staining (data not shown). However, PABP overexpression resulted in a small (25-30%)  increase in cell-doubling time and reduced (25-35%) cloning efficiency (Table 3). PABP plays an important role in modulating mRNA stability; we postulated that PABP overexpression could lead to changes in the level of many transcripts. The results of microarray analyses using pCMV⌬PABP-and pCMV⌬PABP-FLAG-transfected cells showed that, under both conditions, of 1800 different mRNAs tested, the levels of 20 mRNAs were altered reproducibly in all experiments (Table 4). Furthermore, both PABP and PABP-FLAG overexpression produced similar results. Among the 20 mRNAs, the levels of 19 mRNAs were reduced by Ͼ2-fold. The mRNA encoding TIMP-1 was the only mRNA that showed a significant (6-fold) and highly reproducible increase in its cellular level in PABP-and PABP-FLAG-overexpressing cells.

MKK-2 gene Oligonucleotide
To validate the microarray results, the levels of the 20 mRNAs shown in Table 4 were measured by comparative real-time RT-PCR using mRNA-specific primers ( Table 1). The results from three separate transfection experiments showed changes in mRNA levels similar to those observed by microarrays (Fig. 2). The results show that, in pCMV⌬PABP-FLAG-transfected cells, the levels of the 19 mRNAs were ϳ50% of those observed in the mock-transfected cells. The cellular level of TIMP-1 mRNA was increased by 6 -7-fold in PABP-FLAGexpressing cells (Fig. 2). In contrast, the level of ␤-actin mRNA showed very little difference between mock-transfected and PABP-FLAG-expressing cells.
Among the 20 mRNAs listed in Table 4, the MKK-2 mRNA encodes a protein kinase that may be involved in regulating mRNA translation by signaling phosphorylation of initiation factors, including eIF4E (2, 34). We therefore further examined whether reduced mRNA levels result in reduced levels of the MKK-2 polypeptide and found that the cellular level of MKK-2 in PABP-transfected cells was indeed ϳ3-fold lower than in the mock-transfected cells (Fig. 3).
Altered Stability of mRNA by PABP Overexpression-A change in the steady-state level of mRNA may result from a change at the level of transcription or at the level of mRNA stability. Because PABP is known to function in controlling mRNA stability, we first examined whether the stability of the 19 mRNAs discussed above was reduced and whether   the TIMP-1 mRNA was increased in PABP-overexpressing cells. To measure the stability of mRNA, transfected cells were treated with 5 g/ml actinomycin D for 0, 2, 4, 8, 12, and 16 h, and the levels of different mRNAs were measured by real-time PCR (Fig. 4). The results show that the half-lives of 17 of the 19 mRNAs showing reduced levels in PABPoverexpressing cells were also reduced by ϳ2-3-fold. The reduction in the stability of these mRNAs correlated well with their reduced steadystate levels in PABP-transfected cells. Therefore, the transcription of these genes was probably not affected by increased cellular levels of PABP. The stability of ribosomal protein L27 and coagulation factor 2 was not significantly decreased by PABP overexpression, suggesting that the transcription and/or transport of these mRNAs was probably affected. Interestingly, the stability of TIMP-1 mRNA, the cellular level of which was increased in PABP-transfected cells, was increased in these cells. However, the increase in the stability of this mRNA was only 2-fold, which cannot fully account for the 6-fold increase in its steadystate level. It is possible that transcriptional control of the TIMP-1 gene was also responsible for the observed increase in TIMP-1 mRNA in PABP-transfected cells. As a control, we measured the half-life of the ␤-actin mRNA, the level of which was not affected by PABP overexpression. We show here that the half-life of the ␤-actin mRNA was similar in both mock-and PABP-transfected cells. PABP Binds to a cis-Element of MKK-2 mRNA-PABP is a phosphoprotein, and its ability to bind poly(A) and eIF4G can be modulated by phosphorylation (41,42). Therefore, it is possible that a reduced MKK-2 level could down-regulate PABP phosphorylation and reduce PABP activity in cells overexpressing PABP. To determine whether the stability of MKK-2 mRNA is regulated by the interaction between PABP and a destabilizing cis-element of the MKK-2 mRNA, we first tested the ability of purified PABP to bind different regions of the MKK-2 mRNA by UV cross-linking of 32 P-labeled RNA to purified PABP. RNAs corresponding to different regions of MKK-2 mRNA were transcribed in vitro and allowed to interact with purified PABP. The binding of PABP to the RNA was examined by the presence of labeled RNA fragments covalently linked to PABP following SDS-PAGE. The results of these studies are summarized in Fig. 5A. We first tested the 5Ј-UTR, 3Ј-UTR, and coding region of MKK-2 mRNA for the presence of a PABP-binding site. The results show that the PABP binding ability was limited to the 3Ј-UTR of MKK-2 mRNA. Both the 5Ј-UTR and different segments of the coding region were unable to bind PABP. We further delineated the PABP-binding element by testing several smaller RNAs representing different regions of the 3Ј-UTR of MKK-2 mRNA. The results show that a 62-nucleotide-long region of the 3Ј-UTR (M3U-62-1, where M3U is MKK-2 3Ј-UTR) was sufficient for binding to PABP (Fig. 5, A and C). The ability of M3U-62-1 RNA to bind PABP was similar to that observed with the entire 3Ј-UTR-containing RNA. Furthermore, we compared the ability of M3U-62-1 RNA to bind to PABP with that of the oligo(A)-rich ARS RNA, derived from the 5Ј-UTR of PABP mRNA, which was previously shown to bind PABP (25). Both M3U-62-1 RNA and the ARS RNA showed almost equal ability to bind to PABP. In control experiments in which the RNA and proteins were not exposed  to UV, binding between PABP and the RNA was not observed (Fig. 5C, ϪUV lane). In addition, RNAs similar in size (M3U-63 and M3U-62-2) (Fig. 5A) to M3U-62-1 RNA did not bind PABP (Fig. 5, A and C), suggesting that the observed interaction between M3U-62-1 RNA and PABP was sequence-specific.
As PABP functions in cells by interacting with several polypeptides, we also tested whether additional polypeptides bind to the 3Ј-UTR of MKK-2 mRNA. Cell extracts from both mock-and PABP-transfected cells were incubated with 32 P-labeled RNA representing different regions of the MKK-2 mRNA and analyzed for RNA-protein binding following UV treatment. The results show that only a 72-kDa polypeptide from both mock-and PABP-transfected cell extracts was able to bind to the 62-nucleotide-long M3U-62-1 RNA (Fig. 5D). The binding of twice more 72-kDa polypeptide was observed with the PABP-trans-  Either different regions of MKK-2 mRNA were subcloned into a T7 promoter-containing vector, or double-stranded synthetic oligodeoxynucleotides containing a T7 promoter were created by annealing the sense and antisense strands as described under "Experimental Procedures." 32 P-Labeled RNAs corresponding to different regions of the MKK-2 mRNA were synthesized by in vitro runoff transcription using the T7 RNA polymerase. Approximately 3 ng of RNA (Ϸ2 ϫ 10 5 cpm) was incubated with 10 ng of purified PABP and cross-linked by UV light. Following RNase A/T1 digestions of the RNA, the samples were analyzed by SDS-PAGE, followed by autoradiography as described under "Experimental Procedures." Nucleotide numbering is from the 5Ј-cap site of the mRNA. Each cross-linking study was repeated three times with reproducible results. A summary of the UV cross-linking experiments is shown in A. B shows the nucleotide sequences of the shortest region of the MKK-2 mRNA that binds PABP (M3U-62-1) and another region of similar length that does not bind PABP (M3U-63) and the reporter constructs used in additional studies. ␤-Gal, ␤-galactosidase. C and D show the autoradiograms of UV-cross-linked samples using various RNAs and either purified PABP (C) or 60 g of cytoplasmic extract from mocktransfected (M-CE) or PABP-overexpressing (P-CE) cells (D). The ARS RNA (C) represents the adenosine-rich autoregulatory region of PABP mRNA (24). The 72-kDa cross-linked polypeptide shown in D was scanned, and the relative intensity of this band in samples from the mock-transfected and PABP-overexpressing cell extracts was determined using NIH Image analysis software. The average results of three independent experiments are shown in E. ***, p Ͻ 0.001. fected cell extract than with the mock-transfected cell extract (Fig. 5E). Because there was 2-3-fold more PABP in the PABP-transfected cells, it is likely that the 72-kDa polypeptide represents cellular PABP. The lack of binding of additional proteins to the M3U-62-1 RNA was somewhat surprising. It is possible that some proteins may bind to this RNA transiently and are not easily detectable. Furthermore, detection of RNAprotein interaction by UV cross-linking relies on direct interaction between RNA and the polypeptide. Therefore, polypeptides that are present in the ribonucleoprotein (RNP) complex through protein-protein interaction may not be detected by the UV cross-linking approach. As native RNP complexes (containing proteins that are present through interaction with an RNA-binding protein) can be detected by RNA electrophoretic mobility shift assay, we analyzed the RNP complexes of M3U-62-1 RNA using this approach. The results show that a single RNP complex was formed between the M3U-62-1 RNA and purified PABP (Fig. 6A). Incubation of M3U-62 RNA with cell extracts from mock-and PABP-transfected cells also showed formation of a complex similar to that observed with purified PABP, but a second slower migrating RNP complex was also observed with both extracts. However, more RNP complex was formed when extracts from PABP-transfected cells were used. Furthermore, the M3U-63 RNA was unable to form a distinct RNP complex. We also tested whether formation of these complexes is influenced by the PABP level in the cell extract. The addition of purified PABP to the mock-transfected cell extract resulted in a shift toward the formation of the slower migrating complex (Fig. 6B). However, the levels of RNP complexes did not increase linearly with the addition of exogenous PABP to the cell extract, suggesting that some cellular components may be limiting. These results nevertheless indicate that PABP could recruit additional polypeptides to form a multimeric RNP complex with a cis-element located between nucleotides 1463 and 1524 of the 3Ј-UTR of MKK-2 mRNA.
The Destabilizing cis-Element of MKK-2 mRNA-To determine whether the PABP-binding region of the MKK-2 mRNA can work as a destabilizing sequence in PABP-overexpressing cells, we used chimeric ␤-galactosidase reporter expression constructs containing different regions of the MKK-2 mRNA at the 3Ј-UTR of the ␤-galactosidase gene. Three constructs were tested in transient transfection assays in HeLa cells. The results show that similar levels of ␤-galactosidase were expressed in both mock-and PABP-transfected cells when the parental pCMV-SPORT-␤-gal construct was used (Fig. 7, A and B). When the M3U-63 region of MKK-2 mRNA was present at the 3Ј-UTR of the ␤-galactosidase mRNA, similar levels of ␤-galactosidase was expressed FIGURE 6. Analysis of RNP complex formation by electrophoretic mobility shift assay. A, in vitro synthesized, 32 P-labeled RNA (2 ng, Ϸ2 ϫ 10 5 cpm) was incubated either with 10 ng of purified PABP or with 20 g of cytoplasmic extract from mock-transfected (M-CE) or PABP-transfected (P-CE) cells at 22°C for 10 min. The samples were treated with 25 units of RNase T1 at 37°C for 5 min and subjected to electrophoresis on a 5% nondenaturing polyacrylamide gel. The gel was dried and processed for autoradiography. B, RNA electrophoretic mobility shift assay was performed with M3U-62-1 RNA, mocktransfected cell extract, and the indicated amounts of purified PABP as described for A. in both mock-and PABP-transfected cells. There was also no difference in the level of ␤-galactosidase expression between the parental and chimeric constructs containing the PABP-binding region (M3U-62-1) of the MKK-2 mRNA in mock-transfected cells, but the level of ␤-galactosidase expression from the M3U-62-1-containing chimeric construct was significantly reduced in the PABP-transfected cells. This difference was not due to a difference in the transfection efficiency, as the levels of cotransfected GFP and the loading control ␤-actin were similar between mock-and PABP-transfected cells. The average results of three separate transfection experiments are presented in Fig. 7B after adjusting for the difference in loading and transfection efficiency. The results show an ϳ3-4-fold difference in the ␤-galactosidase levels between mock-and PABP-transfected cells when the PABP-binding sequence of the MKK-2 mRNA was present in the reporter mRNA.
To further asses the underlying mechanism of reduced ␤-galactosidase expression, we examined the level of the reporter mRNA in cells transfected with different reporter constructs by real-time RT-PCR. The results show that the level of ␤-galactosidase mRNA derived with all constructs was similar to that in cells that were not cotransfected with the PABP expression vector (Fig. 8A). In contrast, the ␤-galactosidase mRNA level in PABP-transfected cells was only one-third of that in the mock-transfected cells when the M3U-62-1 region (containing the PABP-binding sequence of the MKK-2 mRNA) was present in the ␤-galactosidase reporter mRNA. We further analyzed the molecular basis of the reduction in ␤-galactosidase mRNA levels in PABP-overexpressing cells. As the stability of MKK-2 mRNA was reduced in PABPtransfected cells, we decided to examine whether the stability of the ␤-galactosidase reporter mRNA was also reduced by the PABP-binding sequence of the MKK-2 mRNA. The results of mRNA stability measurements (Fig. 8B) show that the half-lives of the ␤-actin and ectopic GFP mRNAs were ϳ9 and 13 h, respectively, in both mock-and PABPtransfected cells. The stability of the ␤-galactosidase mRNA with the M3U-63 sequence was also similar in both mock-and PABP-transfected cells. However, the presence of the M3U-62-1 region in the ␤-galactosidase mRNA brought about a sharp decline in the half-life of the reporter mRNA in PABP-transfected cells. The results show that the ␤-galactosidase reporter mRNA decayed with a half-life of ϳ17-18 h in mock-transfected cells, regardless of the presence of any MKK-2 mRNA sequence. The reporter mRNA half-life was only 8 -9 h in PABP-transfected cells when the PABP-binding region of the MKK-2 mRNA was present at the 3Ј-UTR of the ␤-galactosidase reporter mRNA. Collectively, these results suggest that a destabilizing cis-element present within the 62-nucleotide-long region of the MKK-2 mRNA is involved in reducing mRNA stability in response to an increase in cellular PABP levels.
PABP Overexpression Results in Reduced Levels of Phosphorylated PABP-As MKK-2 is involved in activating the ERK1/2 signaling pathway, reduction of its cellular level could affect phosphorylation of PABP and eIF4E (36,37). It is known that ERK1/2 mediates phosphorylation of eIF4E via the MNK1/2 pathway (38). PABP is also a phosphoprotein and functions in conjunction with eIF4E and eIF4G to stimulate 5Ј-cap-and 3Ј-poly(A)-dependent mRNA translation. However, how PABP is phosphorylated is not known. Studies have shown that hyperphosphorylated PABP interacts more efficiently with eIF4G than its hypophosphorylated form (41,42); thus, the phosphorylation of the factors involved in initiation of mRNA translation may be a compensatory mechanism of PABP-overexpressing cells. To investigate this possibility, we examined the levels of phosphorylated ERK1/2 and eIF4E by Western blotting using specific antibodies that recognize only the phosphorylated forms of these polypeptides. The results show that, although the levels of total cellular ERK1/2 and eIF4E were similar in mock-and PABP-transfected cells, the levels of the respective phosphoproteins were reduced in PABP-overexpressing cells by ϳ3-fold (Fig. 9, A and B). As a phosphospecific PABP antibody is not available, we examined the level of phosphorylated PABP after metabolic labeling with 32 PO 4 3Ϫ and subsequent purification of PABP by affinity chromatography. The results show that there was also ϳ3-fold less phosphorylated PABP in cells overexpressing it (Fig. 9, A and B). Because there was 2-3-fold more PABP in these cells than in the mock-transfected cells, these results suggest a significant change in PABP phosphorylation.
We further tested whether the MKK-2/ERK pathway is involved in the phosphorylation of PABP using an inhibitor of the MKK-2/ERK1/2 signaling pathway (43). HeLa cells were treated with different concentrations of U0126, and the phosphorylation of ERK1/2 was examined. The results show that the phosphorylation of ERK2, which is the substrate of MKK-2, was inhibited by U0126, whereas the total cellular levels of ERK1/2 and the ␤-actin control were not affected (Fig. 10, A  and B). As very little ERK1 was detectable in HeLa cells, the effect of U0126 on the phosphorylation of ERK1 was also undetectable. We then examined the effect of 20 M U0126 on PABP phosphorylation by Western blotting after separating the cellular proteins by two-dimensional gel electrophoresis. The results show that PABP from the solvent (Me 2 SO)-treated cells migrated with three distinct pI values, whereas PABP from the drug-treated cells exhibited only one pI value (Fig. 10C). The higher pI value of PABP presumably represents the hypophosphorylated state of this polypeptide. As such, these results suggest that PABP is phosphorylated by the MKK-2/ERK pathway. Quantification of PABP on the Western blots showed that there was no significant change in the total cellular level of PABP following the drug treatment (Fig. 10,  D and E). U0126 did not affect the migration of ␤-actin on two-dimension gels (Fig. 10B).

DISCUSSION
We have shown that overexpression of PABP in HeLa cells results in changes in the levels of several mRNAs. These mRNAs encode proteins involved in mRNA translation (ribosomal protein L27 and ribosomal protein L39-like protein), glycolysis (glyceraldehyde-3-phosphate dehydrogenase and ADH6), and heavy metal storage (ferritin L chain and metallothionein 1F); membrane receptors/transporters (colorectal mutant cancer protein and glucosyltransferase-5); enzymes involved in oxidation/reduction (PRDX2 and ABP1); and known or potential candidates that control cellular growth and differentiation (macrophage-specific colony-stimulating factor (MCSF), peroxisomal proliferatoractivated receptor-␦, MKK-2, ZNF136, and BAG1). Among the mRNAs affected by PABP overexpression, only the TIMP-1 mRNA level was increased significantly. TIMP-1 is a member of a family of multifunctional metalloprotease inhibitors. Its biological functions include regulation of cell proliferation and apoptosis (44,45). Increased TIMP-1 levels have been associated with poor prognosis of a variety of human cancers (46).
Changes in the levels of all 20 mRNAs reported here might not be due to a direct effect of PABP on the transcriptional and/or post-transcriptional control of the expression of these genes. In some cases, PABP may indirectly affect the cellular mRNA level by regulating one or more key signaling pathways. Although the levels of several mRNAs involved in cell signaling (including MCSF, MKK-2 and TIMP-1) were affected, overexpression of PABP did not result in an increase in apoptosis. However, the PABP-transfected cells showed an ϳ25-35% decrease in cloning efficiency and increase in cell-doubling time.
We investigated the molecular mechanism of changes in the levels of the 20 mRNAs in PABP-overexpressing cells. Most of the mRNAs that showed reduced levels in PABP-overexpressing cells showed a 2-3-fold Western blotting was performed with the appropriate antibodies to detect either the total cellular levels of PABP, MKK-2, ERK1/2, and eIF4E or the cellular levels of phosphorylated (p) ERK1/2 and eIF4E. The level of phosphorylated PABP was measured by metabolic labeling of cells with 32 PO 4 3Ϫ and isolating the PABP by affinity chromatography using poly(A)-Sepharose as described under "Experimental Procedures." Total cell lysates from mock-transfected (M) and PABP-transfected (P) cells were used for the analysis. The numbers below each panel in A show the relative levels of the polypeptides. The averages of four separate measurements are shown in B. **, p Ͻ 0.01; ***, p Ͻ 0.001; NS (not significant), p Ͼ 0.05. reduction in their stability, which was generally proportional to the observed changes in mRNA levels. For some mRNAs, including the MKK-2 mRNA, we did not see an exact correspondence between the reduction in mRNA stability and mRNA levels. However, the differences were small and within the range of experimental errors inherent to the procedures of measuring mRNA stability by actinomycin D treatment. The non-agreement between the changes in mRNA levels and stability was particularly evident for the TIMP-1 mRNA. Its stability was increased by ϳ2-fold in PABP-transfected cells, which was not directly proportional to the ϳ6-fold increase in TIMP-1 mRNA levels observed in these cells. This discrepancy was greater than what was observed for the other mRNAs. Therefore, it is likely that transcription and/or transport of TIMP-1 mRNA is also affected by the cellular PABP level. Although PABP is a cytoplasmic protein, it was shown previously (47) and confirmed by us (data not shown) that, in 10 -15% of transfected cells, PABP is also present in the cell nucleus. It has been suggested that a fraction of the cellular PABP could shuttle between the nucleus and the cytoplasm and is involved in mRNA transport (46,47). Therefore, the presence of excess PABP may enhance post-transcriptional processing and/or transport of TIMP-1 mRNA. In addition, the role of PABP in the transcription of the TIMP-1 gene cannot be ruled out. PABP may exhibit a preferential effect on some mRNAs if they contain an internal PABP-binding sequence.
As MKK-2 is the upstream regulator of mitogen-activated protein kinases and ERK1/2 (32), which are known to regulate protein synthesis by phosphorylating initiation factors (1,2,32,49,50), we further investigated how MKK-2 levels are reduced by PABP overexpression. We have shown here that MKK-2 mRNA has a destabilizing cis-element at its 3Ј-UTR that binds to PABP. We have identified a 62-nucleotide-long region of the MKK-2 mRNA 3Ј-UTR that binds PABP and that can confer reduced stability to a heterologous ␤-galactosidase reporter mRNA in PABP-overexpressing cells, but not in control mock-transfected cells. The ability of PABP to bind to this cis-element was also demonstrated using a HeLa cell extract. Interaction between the 62-nucleotide-long cis-element and PABP was sequence-specific and not just an electrostatic interaction between PABP and the negatively charged RNA. First, this interaction was seen in the presence of the polyanionic agent heparin, which is known to reduce nonspecific interactions between RNA and proteins. Second, other RNAs of similar sizes derived from different regions of the 3Ј-UTR of MKK-2 mRNA did not bind PABP in our binding assays. The nucleotide sequence of the MKK-2 destabilizing cis-element is rich in guanines and cytosines (72%) and may form a stable stem-loop structure as judged by the RNA fold program (data not shown).
The 3Ј-UTR of MKK-2 mRNA does not have any previously known mRNA destabilizing element, including the AU-rich element and major protein-coding region determinant of instability (mCRD) regions present in some unstable mRNAs (51,52). These cis-elements are known to bind PABP (50,53). Among the 20 mRNAs affected by the cellular PABP level, only two mRNAs (MCSF and ribosomal protein L39) have AU-rich sequences in their 3Ј-UTRs. The MCSF mRNA is similar to the granulocyte/MCSF mRNA, the stability of which is regulated by the AU-rich cis-element (46). The majority of the mRNAs (Fig. 2 and Table 4), including the MKK-2 mRNA, have GC-rich 3Ј-UTRs. Because PABP can also bind to poly(G) in vitro (37) and is known to bind a GC-rich region of vasopressin mRNA (12), PABP may bind to the GC-rich region at the 3Ј-UTR of MKK-2 mRNA when its cellular level exceeds a threshold level. The binding of PABP to the GC-rich region of MKK-2 mRNA region may promote the assembly of an mRNA decay complex (48) and/or decrease the residency time of PABP on the 3Ј-poly(A) tail to allow degradation from the 3Ј-end (13).
Previous studies have shown that the cellular PABP level can be controlled by regulating mRNA translation through a negative feedback mechanism involving the binding of PABP to an adenosine-rich element at the 5Ј-UTR of its own mRNA (23,24,26). In this study, we have provided evidence that a second control mechanism may work by down-regulating both PABP and eIF4E activity via the MKK-2/ERK1/2 signaling pathway. Thus, the polypeptides that work together appear to be regulated by the same signaling pathway. This compensatory process of down-regulating the activity of translation initiation factors may protect cells from any adverse effect of excess PABP on the regulation of cell divisions. However, this model does not fully explain the 25-30% increase in the cell-doubling time of PABP-transfected cells. One possibility is that a subpopulation of cells with a very high level of PABP (especially nuclear) were not able to completely compensate for the adverse effect of excess PABP and grew poorly.