Phosphorylation of Serine 147 of tis21/BTG2/pc3 by p-Erk1/2 Induces Pin-1 Binding in Cytoplasm and Cell Death*

Treatment of U937 cells with epidermal growth factor (EGF) induces phosphorylation of tis21 and subsequent interaction of tis21 with Pin-1, resulting in the increased cell death with mitochondrial depolarization. Ser147 and Ser149 residues of tis21 were strongly phosphorylated by p-Erk1/2 and p-p38MAPK, respectively, but not by JNK. To investigate the significance of phosphorylation of the Ser147 residue, Pin-1, one of the mitotic regulators that binds to the Ser(P)/Thr(P)-Pro region, was employed. Wild type tis21 phosphorylated by p-Erk1/2 clearly increased its binding to Pin-1, but not the P148A mutant, indicating that Pin-1 was bound to the Ser(P)147-Pro148 region of tis21. Transfection of tis21 significantly enhanced EGF-induced Pin-1 diffusion to cytoplasm, compared with that in the vector-transfected cells. Knockdown of tis21 expression by using shRNAi significantly inhibited EGF-induced Pin-1 diffusion, and analysis by flow cytometry after JC-1 stain and confocal microscope revealed that EGF aggravated tis21-induced mitochondrial depolarization and cell death. Furthermore, tis21 was bound to cyclin B1 and Cdc2 and inhibited its activity in vivo and in vitro. In summary, treatment of U937 cells with EGF activates Erk1/2, which in turn phosphorylates Ser147 of tis21 and induces tis21 and Pin-1 binding and mitochondrial depolarization. These data suggest, for the first time, a mechanism of how EGF can be antiproliferative in human tumor cells: binding of tis21/BTG2/pc3 to Pin-1 or cyclin B1-Cdc2 complex and induction of mitochondrial depolarization.

tis21 (TPA-inducible sequences 21), antiproliferative gene, has been known as one of the early growth response genes (1) and isolated from SW3T3 cells treated with 12-O-tetradecanoylphorbol-13-acetate (TPA) 1 (2). tis21 mRNA is very rapidly and transiently induced by growth factors and tumor promoter and superinduced by inhibition of protein biosynthesis (3). pc3 found in rat PC12 cells (4) and BTG2 (B-cell translocation gene 2) in humans (5) are tis21-homologous genes. BTG2 has been cloned from chromosomal segment 1q32 and contains a wild type p53 response element in the 5Ј-flanking region (Ϫ74 to Ϫ122) (6), thus confirming the observation that BTG2 expression is dependent on wild type p53 function (5).
According to the published sequences of tis21, there is only one serine residue ( 147 SPSK 150 ) that can potentially be phosphorylated by proline-directed kinases such as Erk1/2, p38 MAPK , JNK, and Cdc2 (cell division cycle 2 kinase) (14), suggesting a possibility of TIS21 as one of the PIN-1-interacting proteins. Pin-1 (protein-interacting NIMA), NIMA-interacting protein, is a nuclear peptidyl-prolyl cis/trans-isomerase essential for regulation of mitosis (15), which contains group IV WW domain and binds to Ser(P)/Thr(P)-Pro in mitotic phosphoproteins (16). It has been suggested to be required for both normal mitotic progression (17) and reentry into the cell cycle from quiescence (18). Recently, increasing interest has been focused on the role of Pin-1 and its binding proteins in cell cycle regulation and on the value of Pin-1 as a potential therapeutic target in breast cancer (19), prostate carcinoma (20), oral squamous cell carcinoma (21,22), and hepatocellular carcinoma (23). Indeed, Pin-1 has been shown to play an important role in oncogenesis (24) through several different mechanisms, such as the expression of cyclin D1 by cooperating with Ras signaling (25,26), direct stabilization of cyclin D1 protein (27), and inhibition of ␤-catenin interaction with the tumor suppressor adenomatous polyposis coli (28). Furthermore, Pin-1 is an E2F target gene essential for the Neu/Ras-induced transformation of mammary epithelial cells (26) as well as a critical regulator of the tumor suppressor p53 and p21 WAF1 in response to DNA damage (29,30). These findings strongly imply that several key regulators of G 1 /S and G 2 /M phases, such as p53, Cdc25C, Cdc2, and cyclin B1, might be the potential proteins to interact with Pin-1. However, a number of studies on the cellular consequence of Pin-1 binding to these proteins are quite limited until now.
Considering the fact that half of human tumors are either p53-deleted or mutant (31), it is not too far-fetched to suggest p53-independent regulation of the cell division cycle by the interaction of tis21/BTG2/pc3 with Pin-1 in human tumors. Contrary to the definition that tis21/BTG2/pc3 is a primary response gene induced by TPA or various growth factors (1-3), neither TPA nor high cell densities are able to increase BTG2 expression in the tumor cells to the level observed in normal renal tissues (32). It should be noted that all of these counterpart normal tissues constitutively express tis21 mRNAs. The above described findings strongly support tumor suppressive function of tis21 in carcinogenic process and indicate that down-regulation of BTG2 is a pivotal event involved in cancer development. However, precise mechanisms of tumor suppression by tis21/BTG2/pc3 in primary tumors have not yet been known.
We demonstrate in the present study that tis21 protein interacts with Pin-1 after phosphorylation of its Ser 147 residue by p-Erk1/2, which was induced by EGF and inhibition of cyclin B1-associated Cdc2 activity. Moreover, treatment of U937 cells with EGF clearly enhanced tis21-mediated Pin-1 diffusion from nucleus to cytoplasm, confirmed by knockdown of tis21 expression with shRNAi (short hairpin RNA interference) application. Concurrently, mitochondrial depolarization and increased cell death occurred. Taken together, we have found a mechanism of how EGF can be anti-proliferative through Pin-1 binding to tis21/BTG2/pc3, resulting in increased mitochondrial depolarization.

Construction of tis21 Mutants
tis21 cDNA was earlier subcloned in our laboratory (26), and tis21(open reading frame (ORF)) was amplified by PCR with the sense primer, 5Ј-ccgaattcaggatccatgagccacgggaagagaacc-3Ј, and the antisense primer, 5Ј-ggctcgaggatccctagctggagacggccatcac-3Ј. The PCR products were inserted into pGEX-4T-3 (Amersham Biosciences) after digestion with EcoRI and XhoI. To prepare S147A, P148A and S149A tis21 mutants, several mutagenic primers were designed (Table I), and PCR was performed twice. The first PCR was tried with mutagenic primer as a forward primer and the above described antisense primer, which is complementary to 3Ј region of tis21(ORF). The second PCR was done with the first PCR product (61, 58, and 55 bp, respectively) as a reverse primer and the above described sense primer of tis21(ORF). Concurrently, one more primer set was prepared to construct C-terminal deletion mutant of the tis21 gene (tis21⌬C), which contained amino acids 1-129 of the full sequence, using 5Ј-ccgaattcaggatccatgagccacgggaagagaacc-3Ј and 5Ј-gggctcgagagccaccggggc-3Ј as forward and reverse primers, respectively. The PCR products amplified with tis21 cDNA as a template were inserted into pGEX-4T-3 after digestion with EcoRI and XhoI. Mutant clones obtained were confirmed by automatic DNA sequence analyzer (ABI 377).

Preparation of Recombinant tis21 Proteins
Expression of glutathione S-transferase (GST) fusion proteins was induced overnight in DH5␣ cells with isopropyl ␤-D-thiogalactoside (500 M) at 30°C, and the cells were resuspended in extraction buffer (phosphate-buffered saline, pH 7.4, 5 mM 2-mercaptoethanol, 1 g/ml leupeptin, and 0.5% Triton X-100) and disrupted by sonication. After removal of debris by centrifugation (12,000 ϫ g, 30 min, 4°C), the GST fusion proteins in the supernatant were purified with glutathione-Sepharose 4B beads (Amersham Biosciences) according to the manufacturer's instruction. The purified proteins were concentrated using Centricon 10 (Millipore Corp.). Various tis21 proteins were visualized in SDS-PAGE after further purification from the GST recombinant constructs by thrombin digestion for 4 h at 20°C (Fig. 1A). ⌬C indicates C-terminal deleted tis21 protein that has 1-129 amino acids instead of 158 (Fig. 1B).

Production of Anti-tis21 Polyclonal Antibody
Rabbits (New Zealand White) were immunized with the synthetic peptide, keyhole limpet hemocyanin-coupled N terminus of tis21 (MSH-GKRTDMLPEC; 13-mer), whose sequence did not share any homology with other BTG family genes and appeared to be hydrophilic. The synthetic peptide was emulsified with an equal volume of complete Freund's adjuvant for the first injection, and with incomplete Freund's adjuvant for three subsequent booster injections. The immune sera were purified by immobilized protein A column (Pierce) and concentrated using a Centricon 10 concentrator.

Immunoblot Analyses
For protein expression assays, 10 -50 g of cell lysates were subjected to electrophoresis on a 12% SDS-PAGE. The separated proteins were transferred to a polyvinylidene difluoride membrane (Bio-Rad). Commercial sources of antibodies used were as follows: anti-p-Erk1/2 and anti-p-p38 MAPK from Cell Signaling Technology; anti-Cdc2 and anti-␣-tubulin from Oncogene. The membranes were then incubated with horseradish peroxidase conjugated with goat anti-mouse or antirabbit antibodies, and the proteins were visualized by means of the enhanced chemiluminescence kit (Amersham Biosciences).

Protein Phosphorylation Assay
To investigate whether tis21 can be phosphorylated and which kinase potentially phosphorylates tis21 protein, in vivo and in vitro IP kinase assays were performed using the above IP products. MAPKs and cyclin B1 immunocomplexes were washed twice with mitogen-activated protein kinase buffer (25 mM Hepes, pH 7.4, 25 mM MgCl 2 , 5 mM dithiothreitol, 25 mM glycerophosphate, 1 mM sodium orthovanadate, 1 mM sodium fluoride, 10 g/ml leupeptin) and Cdc2 kinase buffer (50 mM Tris-Cl, pH 7.5, 10 mM MgCl 2 , 5 mM dithiothreitol, 1 mM sodium orthovanadate, 1 mM sodium fluoride, 10 g/ml leupeptin), respectively. Recombinant JNK (constitutively active forms; Upstate Biotechnology, Inc., Lake Placid, NY) was also used for in vitro kinase assay. The assays were initiated by adding substrate proteins and 5 M [␥-32 P]ATP (30 Ci/mmol) in a final volume of 30 l. After incubation for 30 min at 30°C, the reactions were terminated by adding 6ϫ SDS sample buffer and analyzed by autoradiography.

In Vitro Assay for tis21-Cdc2 and tis21-Pin-1 Interactions
GST Pull-down Assay-To discern mitotic phase proteins interacting with WT TIS21, a GST pull-down assay was performed. Recombinant GST-WT TIS21, GST-TIS21⌬C, and GST proteins (25 g) were pre- 5Ј-ggc agg agc gcc ccc tcg aag-3Ј (a 439 gc 3 g 439 cc) P148A mutant 65.8 5Ј-c agg agc agc gcc tcg aag aac-3Ј (c 442 cc 3 g 442 cc) S149A mutant 66.6 5Ј-gg agc agc ccc gcg aag aac tat g-3Ј (t 445 cg 3 g 445 cg) bound to glutathione-agarose beads and then incubated with lysates of the nocodazole-treated U937 cells (500 g) in the Triton X-100-containing lysis buffer (300 l) at 4°C for 12 h. The agarose beads were vigorously washed with the lysis buffer, and the bound proteins were evaluated by immunoblot analyses using antibodies against cyclin B1 and Cdc2.
Immunoprecipitated Pin-1 Pull-down Assay-Pin-1 immunoprecipitates obtained from U937 cell lysates (500 g of protein) were incubated at 4°C for 3 h with either thrombin-cleaved WT tis21 or tis21 P148A mutant proteins (20 g), which had previously been phosphorylated overnight with or without p-Erk1/2 immunocomplex. Pin-1-agarose beads were then washed three times with the lysis buffer, and the bound proteins were evaluated with immunoblot analyses, using polyclonal anti-tis21 antibody.

Inhibition of Cyclin B1-associated Kinase Activity by tis21 in Vivo and in Vitro
To evaluate the effect of tis21 on Cdc2 activity, endogenous Cdc2 immunocomplexes were prepared from U937 cells transfected either with WT tis21, tis21⌬C, or pcDNA3 vector with anti-cyclin B1 antibody. Immunocomplexes were washed twice with kinase buffer, and the assay was initiated by adding histone H1 (5 g) and [␥-32 P]ATP (5 M; 30 Ci/mmol) in a final volume of 30 l. The reaction was carried out for 30 min at 30°C and then terminated by adding the sample buffer. Phosphorylation of the substrate protein was examined by autoradiography after SDS-PAGE. To further confirm in vitro the inhibition of Cdc2 activity by tis21, GST-tis21 protein (1, 5, and 25 g) was added to the reaction mixture, and Cdc2 kinase activity was then measured by the method described above.
FIG. 1. Purification of wild type and mutant tis21 proteins. A, GST-fused wild type and mutant tis21 proteins were digested with thrombin (2 units for 0.1-1.0 mg of protein), revealing 17 kDa of tis21 proteins except for the ⌬C mutant. B, GST-tis21 proteins, wild type, and ⌬C mutant, which has the C-terminal 1-129 residues, were eluted with reduced glutathione (10 mM). GST-tis21⌬C corresponds to 40 kDa. All of the recombinant proteins were resolved on 12-15% SDS-PAGE and stained with Coomassie Blue dye.
FIG. 2. Phosphorylation of tis21 by p-Erk1/2 and p-p38 MAPK . A, U937 cells treated with EGF (10 ng/ml) for 15 min were subjected to immunoblot analysis to determine the stimulation of Erk1/2. ␣-Tubulin was used as a loading control. B, IP-Western blot analysis. The above cell lysates (750 g) were incubated with anti-p-Erk1/2 antibody overnight at 4°C, and the immunocomplex was then harvested with protein A-agarose beads. Input, ppt, and sup, the cell lysates (50 g), IP complex, and supernatant of the IP complex, respectively. The blot revealed p-Erk1/2 present in each fraction. C, autoradiography. To investigate whether tis21 could be phosphorylated by p-Erk1/2, the p-Erk1/2 IP complex was used as the enzyme source for a kinase assay. As shown in the upper panel, only the WT tis21 (10 g) and not the tis21 S147A mutant (10 g) could be phosphorylated by [␥-32 P]ATP (5 M, 30 Ci/mmol) when incubated for 30 min at 30°C. Myelin basic protein (MBP; 2 g) was also employed as a positive control of the activity of p-Erk1/2. D, to stimulate p38 MAPK , U937 cells were pretreated with sorbitol (100, 300, and 500 M) for 15 min, and the cell lysates (10 g) were evaluated by immunoblot analysis with anti p-p38 MAPK antibody. E, IP-Western blot. The lysates (750 g) of cells treated with 300 M sorbitol were incubated with anti p-p38 MAPK antibody overnight at 4°C and then immunoprecipitated by adding protein A-agarose beads. Input, ppt, and sup, whole cell lysates (10 g), IP complex, and supernatant of the reaction mixture, respectively. The blot was hybridized with anti-p-p38 MAPK antibody. F, autoradiograph. To investigate whether tis21 could be phosphorylated by p-p38 MAPK , the p-p38 MAPK IP complex was used as the enzyme source for a kinase assay. As shown in the upper panel, the tis21 S149A mutant (10 g) could not be phosphorylated by p-p38 MAPK when incubated for 30 min at 30°C. Phosphorylation of MBP (2 g) was also employed as a positive control of the activity of p-p38 MAPK immunoprecipitates.

Pin-1 Immunocytochemistry
U937 cells were stimulated with EGF (10 ng/ml) 24 h after transfection with either pcDNA3-tis21 or the vector alone and then spun down after 20, 40, 60 min and 24 h of EGF treatment. To investigate the intracellular location of Pin-1, the cells were fixed in 4% paraformaldehyde for 20 min at ice, permeabilized with 0.1% Triton X-100 in PBS, and then attached on slide glass (MUTO Pure Chemical Co.) for 20 min at room temperature. After blocking with 3% bovine serum albumin, anti-Pin-1 antibody (1:200) was applied at 4°C overnight, and the cells were washed 3 times with 0.1% Triton X-100 in PBS before treating them with donkey anti-rabbit IgG conjugated with Texas Red (1:200; Jackson ImmunoResearch Laboratories) for 1 h at room temperature. To observe nuclei, Hoechst 33258 dye (1:100,000; Molecular Probes, Inc., Eugene, OR) was also added to the above, and the cells were then washed three times with 0.1% Triton X-100 in PBS for 10 min. Cells were observed by confocal microscope (Olympus, Japan).

Preparation of tis21 Short Hairpin RNA
To knock down tis21 expression, shRNA against tis21 mRNA was prepared using the MessageMuter shRNAi production kit (Epicenter, WI) according to the manufacturer's instructions. Oligomers targeting the 5Ј end (named as LIK106) and middle (LIK107) of tis21 mRNA sequences including 1-21 (aagg 1 atg agc cac ggg aag aga acc 21 ctt gct tc ggt tct ctt ccc gtg gct cat tat agt ga) and 192-212 (aagg 192 cta tcg ctg tat ccg tat caa 212 ctt gct tc ttg ata cgg ata cag cga tag tat agt ga) bases of tis21(ORF), were synthesized; The boldface sequences can make hairpin loop region, and their downstream italic bases correspond to antisense of 1-21 and 192-212 of tis21 mRNA sequences. After annealing T7 promoter oligonucleotide with either of the LIK106 or LIK107 oligomers, the recessed ends were filled in by Klenow exo Ϫ polymerase and dNTP. Then in vitro transcription was performed at 42°C for 3 h, and the transcripts were harvested through the treatments with RNase-free DNase I, phenol/chloroform extraction, ethanol precipitation, and MicroSpin TM column purification (Amersham Biosciences).

Measurement of Mitochondrial Membrane Potential Differences
In order to investigate the effect of EGF to enhance tis21-induced cell death, changes of mitochondrial membrane potential difference were measured by employing JC-1 (5,5Ј,6,6Ј-tetrachloro-1,1Ј,3,3Ј-tetraethylbenzimidazolylcarbocyanine iodide), based on the living cell staining in a membrane potential-dependent fashion: green fluorescence ( em ϭ 527 nm) of JC-1 monomer and the red fluorescence ( em ϭ 590 nm) of JC-1 aggregates (33). U937 cells transfected with WT tis21 or its vector alone were incubated with EGF for 1 h and washed twice with PBS. The cell suspension was incubated with JC-1 (5 g/ml) dye for 15 min at 37°C, and green (FL-1) and red (FL-2) fluorescence was measured by fluorescence-activated cell sorting (FACS) after washing the cells with PBS twice.

FIG. 4. In vivo and in vitro inhibition of cyclin B1-Cdc2 activity by tis21.
A, cyclin B1-associated Cdc2 was prepared using anti-cyclin B1 antibody from U937 cells that had previously been transfected with either pCDNA3, WT tis21, or tis21⌬C plasmids. Reaction mixtures containing the cyclin B1 IP complex, [␥-32 P]ATP, and histone H1 (5 g) as a substrate were incubated for 30 min at 30°C. Autoradiography was performed to determine in vivo activity of Cdc2. Coomassie Blue stain revealed that histone H1 was added equally in each reaction. B, in vitro inhibition of Cdc2 activity by WT tis21 protein. A cyclin B1-associated Cdc2 kinase assay was performed by the same method described above after the addition of either GST alone (25 g) or 1, 5, or 25 g of GST-WT tis21 proteins. Note significant inhibition of the kinase activity by WT tis21.
were investigated as potential phosphorylation sites by these enzymes. Before carrying out the above investigation, however, highly active kinase preparations were needed. Therefore, U937 cells were stimulated with EGF (10 ng/ml) for 15 min to strongly activate Erk1/2 ( Fig. 2A). With the p-Erk1/2 immunoprecipitate (Fig. 2B), an in vitro kinase assay was performed, and only WT tis21 and not the S147A mutant could be detected by autoradiography (Fig. 2C), indicating that Ser 147 residue in tis21 protein might be the site of phosphorylation by p-Erk1/2. Phosphorylation of myelin basic protein (Sigma) was also employed to prove activity of the added immunoprecipitates in each reaction mixture. In order to activate p38 MAPK , U937 cells were stimulated with 0 -500 M sorbitol (Fig. 2D), and p-p38 MAPK was immunoprecipitated from the cells after stimulation with 300 M sorbitol (Fig. 2E). When the WT tis21, S149A, and ⌬C mutant proteins were added to the in vitro kinase assay mixture, p-p38 MAPK strongly phosphorylated WT tis21, but not S149A and ⌬C mutant proteins, indicating that the Ser 149 residue in tis21 is the most likely potential site for p-p38 MAPK (Fig. 2F). To test whether JNK also phosphorylates WT tis21 protein, an in vitro JNK kinase assay was performed; however, JNK could not phosphorylate any tis21 proteins, as opposed to the strong phosphorylation of c-Jun-GST, which served as a positive control for the constitutively active JNK (data not shown).
Pin-1 and Cyclin B1 Interact with tis21 Protein-Based on the observation that the Ser 147 residue in WT tis21 protein can be phosphorylated by p-Erk1/2 (Fig. 2, A-C) and the Ser(P)-Pro residue can be recognized by Pin-1 protein (35), we hypothesized that the Ser 147 residue of 147 SPSK 150 in the tis21 molecule might interact with Pin-1. To verify this probability, we performed Pin-1 immunoprecipitate pull-down assay. Only the WT tis21 protein that had previously been phosphorylated with p-Erk1/2 was bound to Pin-1 when the Pin-1-bound proteins were evaluated by immunoblot analysis with anti-tis21 antibody (Fig. 3A, lane 3). To prove the specificity of the binding between Pin-1 and tis21, the tis21 P148A mutant was also employed and phosphorylated with p-Erk1/2; however, there was no tis21 binding to Pin-1 immunoprecipitate (Fig. 3A, lane  4), thus indicating that Ser(P) 147 -Pro 148 residues of WT tis21 represent a binding site for the Pin-1 molecule. When tis21-Cdc2 interaction was also investigated by a GST pull-down assay (Fig. 3B), not only Cdc2 but also cyclin B1 was found to bind to GST-tis21 proteins, both the WT tis21 and tsi21⌬C-(1- 129) proteins, but not to GST alone. This indicates that the Ser 147 residue is very specific for tis21-Pin-1 interaction but not for the tis21-cyclin B1 and tis21-Cdc2 interaction; the C terminus of tis21 was redundant for binding of tis21 protein to the cyclin B1-Cdc2 complex.
Inhibition of Cdc2 Activity by tis21 in Vivo and in Vitro-To evaluate biological consequence of the interaction of tis21 with Cdc2, we measured in vivo and in vitro regulations of the Cdc2 activity by tis21. Thus, the active form of Cdc2 was prepared by cyclin B1 immunoprecipitation in U937 cells. When the cell lysates of the vector, WT tis21, and tis21⌬C transfection were immunoprecipitated with anti-cyclin B1 antibody and subjected to kinase assay with [␥-32 P]ATP and histone H1 (5 g) as a substrate, histone H1 phosphorylation by Cdc2 was found to be significantly inhibited only in the WT tis21 overexpresser and not in the vector-and tis21⌬Ctransfected cells (Fig. 4A). Inhibition of Cdc2 activity by tis21 was also confirmed by in vitro analysis. When GST or GST-tis21 proteins (1-25 g) were added to the reaction mixture, cyclin B1-associated kinase activity was concentration-dependently inhibited by GST-tis21 but not by GST itself (Fig. 4B). Taken together, the C terminus of tis21 might be essential to inhibit Cdc2 kinase activity, thereby regulating G 2 /M phase progression.
tis21 Induces Pin-1 Diffusion into Cytoplasm after EGF Treatment-To elucidate biological significance of Pin-1 binding to tis21, the vector-or tis21-transfected U937 cells were stimulated with EGF, and changes of subcellular localization of Pin-1 were immunocytochemically examined. As shown in Fig.  5A, Pin-1 was mainly located in nuclei of the vector-transfected cells; however, EGF treatment for 40 min significantly increased Pin-1 in the cytoplasm of both vector-and tis21-transfected cells. When the translocation was scrutinized in greater detail, Pin-1 translocation into the cytoplasm occurred more in the tis21-transfected cells than in the vector-transfected ones (Fig. 5B, arrowheads). In 24 h, EGF significantly induced fragmentation and condensation of nuclei in the tis21-transfected cells, compared with the vector-transfected cells (Fig. 5B,  stars), which was confirmed by cotransfection of tis21 and its shRNAi (LIK107) constructs (Fig. 5B, third panel). When the changes were evaluated under a confocal microscope, the Pin-1 translocation was significantly increased in the tis21 expresser relative to the vector-and the shRNAi-transfected cells (*, p Ͻ 0.02 versus tis21; Fig. 5C). Significant knockdown of tis21 expression by shRNAi (LIK107) was confirmed by RT-PCR analysis (Fig. 5D). These findings strongly suggest that treatment of U937 cells with EGF rapidly induces phosphorylation of the Ser 147 residue in tis21 protein by activated Erk1/2 and consequently enhances Pin-1 binding and its diffusion into cytoplasm with accompanying nuclear fragmentation of U937 cells.
EGF Aggravates tis21-induced G 2 /M Delay and Cell Death in U937 Cells-Based on our recent study (36), we investigated whether EGF treatment could modulate the effect of tis21 on G 2 /M delay of U937 cells. Therefore, the cells were synchronized with nocodazole treatment, and the cell cycle was monitored by FACS analysis. As shown in Fig. 6A, EGF significantly delayed G 2 /M exit in the tis21-transfected U937 cells 1 h after the release from nocodazole treatment. Moreover, the number of cells in interphase, mitosis, and apoptosis was counted according to nuclear changes under a fluorescence microscope (Fig. 6, B and C), tis21 was found to significantly induce cell death (*, p ϭ 0.004 versus vector), and EGF treatment enhanced the tis21 effect (#, p ϭ 0.055 versus TIS21).
EGF Enhances tis21-induced Mitochondrial Depolarization of U937 Cells-To confirm the effect of EGF on tis21-induced cell death, changes of mitochondrial membrane potential were examined by FACS analysis with JC-1 stain of the living cells. As shown in Fig. 7A, a scattergram revealed the presence of R2 cells with increased FL-1 intensity, compared with that of R1 cells. When the cells in the R1 and R2 region were analyzed, EGF treatment was found to markedly increase tis21-transfected cells in the R2 region, whereas much more of the population of vector transfected cells was in the R1 region (Fig. 7B). These data clearly indicate that EGF increases mitochondrial depolarization, an early event of cell death, when cells express the tis21 gene, thus resulting in cell death with G 2 /M arrest by inhibiting Cdc2 activity. DISCUSSION In the present study, we presented a possible mechanism of tumor-suppressive activity of tis21, through its direct binding to Pin-1 and Cdc2 (Fig. 3), both of which are very important regulators of G 2 /M progression. Binding of Pin-1 to tis21 absolutely required phosphorylation of Ser 147 by activated Erk1/2 (Fig. 2, A-C) and downstream Pro 148 residue in tis21 protein (Fig. 3A). The C-terminal residues (positions 130 -158) of tis21 were required for inhibition of Cdc2 activity (Fig. 4A), although they were not necessary for interaction with cyclin B1 and Cdc2 (Fig. 3B). Treatment of U937 cells with EGF clearly induced cytoplasmic Pin-1 localization in 40 min and increased cell death in 24 h (Fig. 5A). Interestingly, Pin-1 translocation was more prominent in the tis21 overexpresser than the vectortransfected cells (Fig. 5B). We previously reported that constitutive expression of tis21/BTG1/pc3 in U937 cells induces G 2 /M arrest, cell death, and inhibition of cyclin B1 binding to Cdc2 (36). Based on the above observations, we can suggest that EGF-induced tis21 phosphorylation at Ser 147 residue recruited Pin-1 to C terminus of tis21 protein, which in turn inhibited Cdc2 activity through the inhibition of cyclin B1-Cdc2 complex formation. These changes might result in increased cell death through mitochondrial depolarization (Fig. 8).
Numerous mitotic Cdc2 substrates have been identified, all of which are phosphorylated at Ser/Thr-Pro sites. Not only Cdc2 but also MAPKs have been included in the proline-directed kinases (34). Among the proline-directed kinases, MAPK family has an invariable consensus sequence of Ser/Thr-Pro, nevertheless, each has its own unique substrate specificity; it might be due to tertiary structural differences between substrates or different requirements for flanking amino acid sequence. We demonstrated in the present study that Ser 147 and Ser 149 residues could be phosphorylated by p-Erk1/2 and p-p38 MAPK , respectively (Fig. 2), and suggest p-Erk1/2 as the "in vitro noble tis21 Ser 147 kinase." It is, therefore, highly likely that there might be cross-talk between the cyclin-cyclin-dependent kinase complex and p-Erk1/2 signaling pathway, which can affect the phosphorylation pattern of tis21. The above possibility is in support with earlier reports that activation of Erk1/2 pathway inhibits Cdc2 activation, leading to blockade of M phase entry in Xenopus extracts (37), and that the active form of Cdc2 interacts with MEK1, upstream regulator of Erk1/2, and inhibits binding to Erk1/2 (38).
It has earlier been reported that the cyclin A-cyclin-dependent kinase 2 complex phosphorylates the Ser 147 residue of pc3 (85% homologous with tis21), suggesting that pc3/tis21 is relevant to S phase regulation of cell cycle (8). Furthermore, it should be noted that BTG2 (89% homologous with tis21) and pc3 proteins also have the 147 SPSK 150 sequence. We demonstrated in the present study that C terminus of tis21 was critical for tis21-Pin-1 interaction and inhibition of Cdc2 activity (Figs. 3 and 4). As a consequence to tis21-Cdc2 interaction, in vivo and in vitro kinase activity of the cyclin B1-Cdc2 com-plex was significantly decreased (Fig. 4), suggesting a mechanism of G 2 /M arrest by tis21. This further supports our recent report on the significant reduction of in vivo Cdc2 binding to cyclin B1 in the tis21 expressers (36). Furthermore, tis21 also regulates G 1 arrest through the decreased cyclin E biosynthesis in the pRB null 293 cells (9). Therefore, tis21 might be regarded as "a pan-cell cycle modulator" in both G 1 /S and G 2 /M phases, when the cells are pRB and p53 null, respectively.
Pin-1 is an 18-kDa protein and therefore small enough to freely diffuse through the nuclear pore, although it has no specific nuclear localization sequence. The protein remains normally in nuclear speckle; however, phosphorylation at Ser 16 in its WW domain is shown to change its intracellular localization FIG. 8. tis21, a potential cell cycle regulator in the p53 null U937 cells. The Ser 147 residue in tis21 can be phosphorylated by p-Erk1/2 activated by EGF treatment and be a target of Pin-1 binding. tis21 and Pin-1 interaction may induce Pin-1 diffusion in cytoplasm and then inhibit various mitotic regulators, which are essential for mitosis. At the same time, tis21 can bind with either cyclin B1, Cdc2, or its complex, thereby inhibiting Cdc2 activity. tis21-Pin-1 interaction and tis21-Cdc2 binding modulate mitotic regulators, thereby resulting in failure of G 2 /M phase exit and cell death via loss of mitochondrial membrane potential difference. FIG. 7. tis21 induces mitochondrial depolarization and EGF enhances the effect. U937 cells, transfected with either vector or tis21 cDNA construct, were incubated with 5 g/ml of JC-1 dye for 15 min at 37°C and then vigorously washed with PBS. FACS analysis was performed to evaluate changes of mitochondrial depolarization using fluorescence changes. A, scattergram showing two groups (R1 and R2) of cells with increased green fluorescence (FL1) in the R2 region, which reveals the early sign of apoptosis. B, when we analyzed components of the cells in the R1 and R2, EGF treatment for 1 h markedly increased mitochondrial depolarization of the tis21transfected U937 cells (compare orange color in R1 and R2), indicating that EGF enhanced tis21-induced loss of mitochondrial membrane potential difference. However, red fluorescence in FL-2 showed that there was not significant change among cells (data not shown). Control indicates endogenous fluorescence without the addition of JC-1 dye to the cells.
in HeLa cells (35). In the present study, we observed that EGF-induced Ser 147 phosphorylation of tis21 enhanced cytoplasmic localization of Pin-1 and nuclear condensation of U937 cells (Fig. 5). Pin-1 translocation observed in 40 min after EGF treatment (Fig. 5A) was in accordance with the report that Pin-1 dispersal from nucleus was found 30 min after forskolin treatment in HeLa cells (35); however, Pin-1 shuttling into the nucleus observed in HeLa 4 h after forskolin treatment was not observed until 24 h in the tis21-transfected U937 cells (Fig.  5A). Pin-1 overexpression is a prevalent and specific event in human cancers, such as prostate, lung, ovary, cervix, brain, melanoma, and breast cancers. Moreover, Pin-1 expression has been considered as an excellent prognostic marker in prostate cancer, and inhibition of Pin-1 can suppress transformed phenotypes and inhibit tumor cell growth (39). At this point, we can therefore suggest that tis21 might function as a tumor suppressor through sequestering Pin-1 in cytoplasm, thereby preventing it as a mitotic regulator in the various tissues with p53 defect.
Based on the reports that the PDZ domain of rPICK1 (protein that interacts with protein kinase C) interacts with tis21 and protein kinase C (40) and that PICK1 protein can selectively localize to mitochondria upon serum stimulation in NIH3T3 cells (41), we investigated mitochondrial depolarization in the tis21-transfected cells. As shown in Fig. 7B, EGF treatment significantly increased mitochondrial depolarization in U937 cells with tis21 expression, suggesting that EGF-induced cell death may start with tis21 phosphorylation and Pin-1 binding, which subsequently induces loss of mitochondrial membrane potential differences. As far as we are aware, our present data are the only report yet about the role of tis21 in mitochondria. Therefore, we strongly suggest that EGFinduced Pin-1 binding to tis21 and mitochondrial depolarization might explain a mechanism of EGF-induced growth inhibition of the various human cancer cells, such as A431, HN6 (42), breast cancer (43,44), and U937 cells (45), which express the amplification of EGF receptor. EGF-induced apoptosis has been known to reveal lower activation of caspases than those in the paclitaxel or 5-fluorodeoxyuridine-treated cells with overexpressed EGF receptor (46), and EGF induces anoikis; however, activation of procaspase was minimal (47). In support of the above findings, we also failed to find a caspase activation, Bax expression and other apoptotic signals such as poly(ADPribose) polymerase degradation (data not shown).
It remains to be elucidated, however, how Pin-1 interaction with phosphorylated tis21 affects the function of tis21 at cellular or tissue levels. The effect of tis21 on the induction of cell death and cytoplasmic location of Pin-1 was proved by using tis21 shRNAi transfection in U937 cells (Fig. 5B). However, the biological consequence of Pin-1 binding to tis21 has to be further investigated by focusing on the regulation of mitochondrial depolarization.