Epigallocatechin-gallate Suppresses Tumorigenesis by Directly Targeting Pin1 Free

See perspective on p. 1343

Epigallocatechin-3-gallate (EGCG), the most potent active anticancer tea polyphenol, inhibits multiple signal transduction pathways, including activator protein 1 (AP-1), NF-κB, nuclear factor of activated T-cells (NFAT), and β-catenin (1–5). The suppression of these signals is believed to contribute to the anticancer activities of EGCG because these pathways are critical in carcinogenesis. The human peptidyl prolyl cis/trans isomerase (Pin1) is composed of 2 distinct functional domains, the protein–protein interaction N-terminal WW domain (1–39 aa) and the isomerization catalysis PPIase domain (45–168 aa), which are connected by a flexible linker region (40–44 aa; ref. 6). Pin1 isomerizes the peptide bond of specific phosphorylated serine or threonine residues preceding proline in several proteins involved in various events including mitosis, transcription, differentiation, and the DNA damage response (7,8). The Ser/Thr-Pro motifs appear to be exclusive phosphorylation sites for many key protein kinases involved in control of cell growth and transformation (9,10). Pin1 is most likely required for the full activation of multiple signal transduction pathways including AP-1, NF-κB, NFAT, and β-catenin (7, 11,12). Furthermore, Pin1 plays a key role in oncogenic signaling pathways (13,14) and is highly abundant in various tumors (11, 15) including breast (12), colorectal (16), prostate (17), and thyroid (18). Notably, inhibition of Pin1 in cancer cells triggers apoptosis or suppresses the transformed phenotype (19,20). EGCG was reported to inhibit the activation of HER-2/neu and its downstream signaling pathway in human head and neck and breast carcinoma cells (21). Pin1 is also a downstream effector of oncogenic Neu/Ras signaling (20, 22). Therefore, we hypothesized that EGCG might directly target Pin1 to suppress critical oncogenic signaling pathways and neoplastic cell transformation.

The crystals of the complex belong to space group P3121 with 1 molecule in the asymmetric unit. The crystals of the Pin1/EGCG complex were isomorphous to the crystals of Pin1 complexed with the L-peptide (PDB code: 2itk; ref. 23). Therefore, the 2itk coordinates were used as the starting model for the molecular replacement. Refinement was conducted with REFMAC5 (24) coupled with automated building and updating of the solvent structure using ARP/wARP (25). The signal coordinates and experimental data were deposited to PDB with FDB code 3OOB. Manual adjustment of the models and interpretation of extra electron densities were carried out with the programs O (26) and COOT (27) using 2Fo-Fc and Fo-Fc electron density maps. The 2Fo-Fc maps for the 2 EGCG molecules were calculated without contribution of coordinates of EGCG (omit) with superimposed refined molecules of EGCG (Supplementary Fig. S2). The structure model was refined to a final R-factor/free R-factor of 0.19/0.23. The final structure contains residues 6 to 163 of Pin1 (but without the interdomain linker region of 38–51 aa due to the lack of electron density), 2 EGCG molecules, 3 PEG 400 molecules, 1 sulfate ion, and 113 water molecules. The structure has good stereochemistry and no outliers were observed in the disallowed regions of the Ramachandran plot. Refinement statistics are summarized (Supplementary Table S1).

To investigate the interaction of Pin1 and c-Jun, a glutathione-S-transferase (GST) pull-down assay was conducted as described (12, 28). The pcDNA3.1/c-Jun plasmid was a kind gift from Dr. Michael J. Birrer (National Cancer Institute, Bethesda, MD). For expression of c-Jun, the c-Jun plasmid (pcDNA3.1/c-Jun) was transfected into HEK293T cells. For the GST/Pin1 pull-down assay, 4 μg GST and GST/Pin1 full-length (FL) or GST/Pin1 WW domain fusion proteins was collected on glutathione-Sepharose beads (Amersham Biosciences) and treated with EGCG (0, 10, 20, 30, or 40 μmol/L). The bound proteins were denatured in sample buffer, separated, and analyzed by immunoblotting with anti-phospho-c-Jun (Ser73).

Athymic nude mice [Cr:NIH(S), NIH Swiss nude, 5 weeks old] purchased from Jackson Laboratory were pretreated by intraperitoneal injection with vehicle (0.9% NaCl) or 50 mg/kg body weight of EGCG in vehicle (3× week for 2 weeks). At the end of 2 weeks, mice were injected subcutaneously in the right flank with 1 × 10⁶ Neu/Pin1 wild-type (WT) or Neu/Pin1 knockout (KO) mouse embryonic fibroblasts (MEF) in a volume of 0.5 mL. Treatment with vehicle (n = 15 per group) or EGCG (n = 15 per group) continued (3× week for 32 days) until most of the vehicle-treated mice had tumors measuring 1 cm³, which is the endpoint allowed by University of Minnesota Institutional Animal Care and Use Committee. In an additional study, mice were pretreated by intraperitoneal injection with vehicle (10% ethanol/PBS) or 75 mg/kg of EGCG in vehicle (3× week for 1 week). Mice were then injected subcutaneously in the right flank with 1 × 10⁶ HCT116 cells. Treatment with vehicle (n = 10) or EGCG (n = 10) continued (3× week for 27 days). For both studies, tumors were measured twice a week by caliper and volume calculated by the formula: length × width × height × 0.52.

The mU6Pro-Pin1 plasmid was constructed using the mU6Pro vector. The inserted sequence was: 5′-TTTGGTCAGGAGAGGAAGACTTTTTCAAGA GAAAAGTC TTCCTCTCCTGACTTTTT-3′. The siRNA sequence targeted against mRNA Pin1 was GUCAGGAGAGGAAGACUUU.

The coding sequences of Pin1 were obtained after PCR amplification (30 cycles, annealing at 65°C) of the cDNA derived from reverse transcription of the total RNA of Neu/PIN1 WT MEFs. The amplified PCR products were ligated into the pcDNA4.0-HisMaxC plasmid (Invitrogen) to produce an expression vector for 6× His-and Xpress-tagged Pin1. The mutant Pin1 R17A was constructed using the Site-Directed Mutagenesis Kit II (Stratagene). Transformants of Pin1 were obtained by transfecting 1 μg of the pcDNA4.0-WT, mutant-Pin1-HisMaxC, or control pcDNA4.0-HisMaxC vector into Neu/Pin1 KO MEFs using jetPEI (Qbiogen, Inc.) and then maintained by addition of Zeocin into the culture medium.

Escherichia coli BL21 (DE3) bacteria served as the host strain for the N-terminal His6-tagged Pin1 R14A mutant protein. Transformed BL21 bacteria were grown in Luria Bertani (LB) medium with kanamycin at 37°C. Bacteria were induced with 0.25 mmol/L isopropyl β-d-thiogalactoside overnight at 15°C. PIN1 A mutant proteins were purified by Ni-affinity chromatography, digested with thrombin, and further purified by gel filtration chromatography (Superdex 75).

Expression, purification and crystallization of the Pin1 protein was performed as described previously (23). The complex of Pin1 and EGCG was obtained by soaking Pin1 crystals in 40% (v/v) PEG 400, 100 mmol/L HEPES-Na⁺ (pH 7.5), and 200 mmol/L EGCG for 15–20 min at room temperature. The soaking solution served as a cryoprotectant, and during X-ray exposure, crystals of the complex were picked up by a fiber loop and frozen in a stream of liquid nitrogen. Diffraction data were collected to 1.9-Å resolution at the 23 IDD NET-CAT beamline (Advanced Photons Source, Argonne National Laboratory) using a MAR-300 CCD detector. The Pin1/EGCG structure (PDB ID 3OOB) was solved by molecular replacement method, and were integrated and scaled with the HKL2000 suite.

After suspending Sepharose 4B freeze-dried powder (0.3 g) with 1 mmol/L HCl, it was mixed with EGCG (1.5 mg) in coupling solution (0.1 mol/L NaHCO₃, pH 8.3, and 0.5 mol/L NaCl) and rotated at 4°C overnight. Then, EGCG coupled to the CNBr-activated Sepharose 4B matrix was transferred to 0.1 mol/L Tris-HCl buffer (pH 8.0) and again rotated end over end at 4°C overnight. The mixture was washed 3 times with 0.1 mol/L acetate buffer (pH 4.0) containing 0.5 mol/L NaCl followed by a wash with 0.1 mol/L Tris-HCl (pH 8.0) containing 0.5 mol/L NaCl. The resulting EGCG-Sepharose 4B beads were used to assess binding of EGCG with soluble cellular proteins from Neu/Pin1 WT, Neu/Pin1 KO, Neu/Pin1 KO + Pin1, Neu/Pin1 KO + Pin1R17A or Neu/Pin1 KO + mock cell lysates in reaction buffer [50 mmol/L Tris, pH 7.5, 5 mmol/L EDTA, 150 mmol/L NaCl, 1 mmol/L DTT, 0.01% NP-40, 2 μg/mL bovine serum albumin, 0.02 mmol/L phenylmethylsulfonylfluoride (PMSF), 1× proteinase inhibitor]. The beads were washed 5 times with buffer (50 mmol/L Tris, pH 7.5, 5 mmol/L EDTA, 150 mmol/L NaCl, 1 mmol/L DTT, 0.01% NP-40, 0.02 mmol/L PMSF) and proteins bound to the beads were analyzed by immunoblotting with the appropriate antibodies or by autoradiography (29).

The effect of EGCG on PPIase activity of purified Pin1 was determined using the proline isomerase assay based on the isomer-specific cleavage of the phosphorylated peptide WFY(pS)PR-pNA (Bachem) by trypsin as described (30) with some modifications. In brief, the isomerase activity of Pin1 was estimated by absorbance of p-nitroaniline at 390 nm measured every 10 seconds for a total of 100 seconds using a Beckman DU800 spectrophotometer and a temperature-controlled cuvette holder. Prior to kinetic measurements, sample buffer containing Pin1 and EGCG was incubated in the spectrophotometer at 10°C for 10 minutes. Then, trypsin (final concentration of 0.1 mg/mL from stock solution 50 mg/mL) in 1 mmol/L HCl and 2 mmol/L CaCl₂ was added, and the reaction was immediately initiated by addition of peptide dissolved in 0.47 mol/L LiCl/TFE (trifluoroethanol, anhydrous) at a final concentration of 20 mg/mL. The total volume of reaction was 1 mL, including 35 mmol/L HEPES (pH 7.8), 44 nmol/L Pin1, 20 μmol/L peptide, 0.1 mg/mL trypsin, and various concentrations of EGCG (final concentrations 0–30 μmol/L). Determination of inhibition constant (K_i) for EGCG was conducted using the Dixon approach (31) with N-α-benzoyl-dl-arginine-4-nitroanilida hydrochloride (BAPNA) as a substrate to build a calibration curve of dependence on relative absorbance of BAPNA concentration and to determine reaction velocities. From the graphics plot, the kinetic parameter (V_max) was determined (Supplementary Fig. S4A) as described previously (31). Then, the K_i was determined from the plot of the reciprocal inhibition rates as a function of concentrations of EGCG from the negative of the x-axis value at the point of intersection of the plot with the plot at V = V_max (Supplementary Fig. S4B–E). For ex vivo experiments, Pin1WT, Pin1KO, Neu/Pin1 WT, Neu/Pin1 KO, Neu/Pin1 KO + Pin1, Neu/Pin1 KO + mock, or Neu/Pin1KO + Pin1R17A MEF lysates were used. Cells were grown in 10-cm dishes for 24 hours at a density of 2 × 10⁵/mL and then treated with various concentrations of EGCG (0, 10, 15, 20, or 25 μmol/L) for 24 hours. Cells were disrupted in lysis buffer (35 mmol/L HEPES, pH 7.4, containing proteinase inhibitors) and equivalent concentrations of soluble cellular proteins were used in the assays. The K_i was determined as for the in vitro experiments. All K_i values obtained from the graphical approach were confirmed by calculations made using the program GraphPad Prism (GraphPad Prism 5.0 for Windows; GraphPad Software: www.graphpad.com).

To detect the FK506-binding protein (FKBP) and cyclophilin A, a GST pull-down assay was conducted as described for the ex vivo experiment using 100 ng of FKBP or cyclophilin A (Sigma), respectively. Proteins bound to the beads were analyzed by immunoblotting with the appropriate antibodies (FKBP12 antibody, BD Biosciences; cyclophilin A antibody, Cell Signaling Technology). The levels of FKBP and cyclophilin A in Neu/Pin1 WT and Neu/Pin1 KO or Pin1WT and Pin1KO MEFs were also determined by immunoblotting with the appropriate antibodies.

The purified mutant human Pin1R14A protein was subjected to gel filtration using a Superdex 75 column in buffer containing 10 mmol/L HEPES (pH 7.5) and 100 mmol/L NaCl. The EGCG solution was prepared in the same buffer and centrifuged at 14,000 rpm for 10 minutes to remove any insoluble particles. Isothermal titration calorimetry (ITC) experiments were carried out at 25°C using a NanoITC instrument that was coupled with NanoAnalyze v.2.0.4 calculation software (www.tainstruments.com). The titration was carried out with 25 injections of 10 μL EGCG (500 μmol/L; spaced 300 seconds apart) into the sample cell containing a solution of 1,000 μL of 5 μmol/L Pin1. The thermogram was plotted by the NanoAnalyze software v.2.0.4 (www.tainstruments.com) using the best-fit parameters obtained by a nonlinear least-squares fit to the independent binding model after subtracting the heats of dilution from the raw binding heats. Dilution heats were determined by a blank titration and the association constant (K_a) was obtained and the dissociation constant (K_d) was calculated from the relationship: K_d = 1/K_a.

The binding of Pin1 and EGCG was confirmed in an affinity chromatography pull-down experiment and Western blotting with anti-Pin1 (Fig. 1A). We carried out ITC and calculated a dissociation constant of 21.6 μmol/L (Supplementary Fig. S1). To obtain direct information about the interaction between Pin1 and EGCG, the X-ray crystal structure of the Pin1/EGCG complex was determined and refined to 1.9 Å resolution (Supplementary Table S1). The analysis of the difference electron density maps revealed that 2 EGCG molecules (Supplementary Fig. S2A) were bound to Pin1–one EGCG molecule bound the WW domain and the other to the PPIase domain (Fig. 1B and Supplementary Fig. S2B and C). An additional affinity chromatography experiment using Sepharose CNBR-EGCG beads supported these findings (Fig. 1C).

In the WW domain, EGCG forms contacts with the pSer/Thr–Pro recognition loop of Met15–Tyr23 through hydrogen bonding with the side chain and the amide group of Arg17 (Fig. 1D). In this structure, the PEG 400 molecule (from crystallization conditions) occupies the hydrophobic pocket formed by the interface between both domains. The PEG 400 molecule might be thought to compete with EGCG for binding to the WW domain and prevent EGCG from occupying the appropriate position. To eliminate this possibility, we mutated arginine 17 to alanine (R17A) and showed that this mutant failed to bind EGCG ex vivo (Fig. 1E). This result supports the idea that EGCG binds with WW domain in the manner observed in the solved crystal structure. We also found that a second molecule of EGCG bound to the PPIase domain of Pin1 (Fig. 1B and C). Although most of the atoms of the EGCG molecule are exposed to the surrounding solution, EGCG still creates several strong contacts with Pin1. Its gallate moiety forms bonds with the main chain carbonyl group of Asp112, the side chain of Ser114, and with the amide groups of Trp73 and Ser114 (Fig. 1F). In addition, from the crystallization conditions, the complex has a sulfate ion in the phosphate-binding cluster, which mimics the phosphorylated substrate moiety and another PEG 400 molecule in the active site (Fig. 1F).

First, we showed that Pin1 is required for tumor cell proliferation in Pin1WT MEFs and Pin1-deficient (Pin1KO) MEFs (Fig. 2A). Cell growth is decreased time dependently in Pin1WT MEFs compared with Pin1KO MEFs (Fig. 2B). To study the effect of EGCG on cell proliferation, we treated Pin1WT and Pin1KO MEFs with various concentrations of EGCG. Proliferation was assessed using the MTS assay as described in Supplementary Methods. Proliferation of Pin1KO MEFs was not affected by EGCG (Fig. 2C, left) compared with Pin1WT (Fig. 2C, right). These data suggest that Pin1WT MEFs are more sensitive to the antiproliferation effects of EGCG than Pin1KO MEFs. Neu/Pin1 WT MEFs overexpress both the Neu oncogene and Pin1; and Neu/Pin1 KO MEFs express the Neu oncogene but not Pin1 (Fig. 2D). Pin1 is also very important for cell growth in these cell lines (Fig. 2E). Treating these cells with EGCG revealed that EGCG inhibits proliferation of the Neu/Pin1 WT (Fig. 2F, right) but has no effect on Neu/Pin1 KO MEFs (Fig. 2F, left). To show that the resistance of Pin1-deficient MEFs to EGCG treatment is specifically due to the lack of Pin1 or the lack of Pin1 affinity for EGCG, Neu/Pin1 KO cells were transfected with the mammalian expression pcDNA4/HisMax-Pin1 plasmid, R17A-mutant-Pin1-HisMaxC plasmid, or with pcDNA4.0-HisMaxC vector as a control. These cells were then used to study the effect of EGCG on proliferation. Results indicated that only cells reexpressing Pin1 (Fig. 2H) regained the sensitivity to the antiproliferation effects of EGCG, whereas cells transfected with vector (Fig. 2G) or the R17A mutant (Fig. 2I), which failed to bind EGCG, remained resistant to EGCG. To study the functional significance of the binding of EGCG to Pin1, Pin1 PPIase activity measurements were made using an α-trypsin–coupled assay as described in Supplementary Information. For the ex vivo assays, lysates were prepared from Pin1WT and Pin1KO MEFs and Neu/Pin1 WT and Neu/Pin1 KO MEFs and treated for 24 hours with increasing concentrations of EGCG. The results indicated that EGCG suppressed PPIase activity in vitro (Fig. 3A), in Pin1WT (Fig. 3B), in Neu/Pin1 WT (Fig. 3C), and in Neu/Pin1 KO + WT (Fig. 3D) cells in a dose-dependent manner but had no effect on Pin1KO (Supplementary Fig. S3A) or Neu/Pin1 KO (Supplementary Fig. S3B). For an ex vivo recovery experiment, Neu/Pin1 KO MEFs were transfected with a mammalian expression plasmid, pcDNA4/HisMax-Pin1, or a scrambled mammalian expression plasmid, pcDNA4/HisMax as a mock control. Also, Neu/Pin1 KO MEFs were transfected with the R17A-mutant-Pin1-HisMaxC plasmid to determine whether the failure of the Pin1 mutant to bind EGCG will affect PPIase activity. The cells with reexpression of Pin1 showed a restoration of sensitivity to inhibitory effect of EGCG on PPIase activity (Fig. 3D), whereas the Neu/Pin1 KO + pcDNA4/HisMax or KO + R17A-mutant-Pin1-HisMaxC showed no response to EGCG (Supplementary Fig. S3C and D). The inhibition constant (K_i) for EGCG in vitro was determined to be 20 μmol/L [K_cat/K_m = 19.8 × 10⁶ (mol/L)⁻¹s⁻¹] and for the ex vivo experiments, 45, 48 and 50 μmol/L for PinWT, Neu/Pin1 WT, and Neu/Pin1 KO + Pin1 cells, respectively (Supplementary Fig. S4B–E).

To investigate the specificity of EGCG for Pin1 compared with other isomerases, we determined the binding affinity of EGCG with the FKBP and human cyclophilin A. Western blot analysis of the application of FKBP and cyclophilin A to Sepharose 4B or a Sepharose 4B-EGCG column using a specific antibody was conducted. Results indicated that both FKBP and cyclophilin A are expressed at similar levels in Neu/Pin1 WT, Neu/Pin1 KO, Pin1WT, and Pin1KO cells (Supplementary Fig. S5A). However, EGCG did not bind to either protein (Supplementary Fig. S5B).

The formation of the Pin1/c-Jun complex is an important regulator of c-Jun stabilization, and Pin1 has been shown to bind phosphorylated c-Jun and potentiate its transcriptional activity toward cyclin D1 (12). We therefore used a GST/Pin1 pull-down assay to determine the effect of increasing concentrations of EGCG on the formation of the Pin1 and c-Jun complex in cell lysates prepared from HEK293T cells that were transfected with c-Jun. Pin1 was phosphorylated on Ser73, as indicated by Western blot analysis (Fig. 4A and B). No binding was observed between GST alone and phosphorylated c-Jun [pc-Jun (Ser73)], but binding did occur between phosphorylated c-Jun (Ser73) and the GST/Pin1 FL (Fig. 4A) or the GST/Pin1 WW domain (Fig. 4B). Importantly, the binding of phosphorylated c-Jun (Ser73) with either FL Pin1 or the WW domain of Pin1 was dramatically decreased by EGCG (Fig. 4A and B). The phosphorylation of c-Jun at Ser63 or Ser73 was examined in Neu/Pin1 WT and Neu/Pin1 KO MEFs (Fig. 4C). EGCG treatment resulted in the dose-dependent decrease in the phosphorylation of c-Jun at both sites in Neu/Pin1 WT but had no effect on Neu/Pin1 KO MEFs (Fig. 4C). Also, we found that the level of cyclin D1 was decreased in EGCG-treated cells that overexpress Pin1 but was unchanged in Pin1-deficient cells treated with EGCG (Fig. 4C). The expression level of Pin1 in Neu/Pin1 WT did not change.

EGCG has been reported to inhibit AP-1 or NF-κB activity (2–4) and Pin1 is required for the full activation of signal transduction pathways including AP-1 and NF-κB (7, 12, 32). Therefore, we hypothesized that the inhibitory effect of EGCG might effectively suppress AP-1 or NF-κB activity through Pin1. We examined the effect of EGCG on AP-1 and NF-κB promoter activity in Pin1WT and Pin1KO MEFs. Results indicated that EGCG significantly suppressed AP-1 (Fig. 4D) and NF-κB (Fig. 4E) reporter activity in Pin1WT cells but had little effect on these activities in Pin1KO cells.

Pin1 has been shown to play an important role in cell-cycle regulation (8, 33) and to trigger apoptosis with stress (11, 12, 34). We compared the effect of EGCG on cell cycle and apoptosis in Pin1WT and Pin1KO MEFs. We found that EGCG induces accumulation of Pin1WT cells in S-phase and apoptosis (Fig. 5A and B) and also regulates several pro- or antiapoptotic marker genes (Fig. 5C). We observed that the number of cells in S-phase was substantially increased in EGCG-treated Neu/Pin1 WT cells but not in Neu/Pin1 KO cells (Fig. 5A). EGCG induced more apoptosis in Neu/Pin1 WT than in Neu/Pin1 KO cells. In addition, we found that the level of cyclin B1 and cyclin D1 was decreased in EGCG-treated cells overexpressing Pin1 but was not changed in Pin1-deficient cells treated with EGCG (Figs. 4C and 5C).

Next, we evaluated the effect of EGCG on colony formation mediated through Pin1 by conducting an anchorage-independent soft agar cell transformation assay using Neu/Pin1 WT and Neu/Pin1 KO MEFs. Neu/Pin1 WT MEFs were treated with increasing concentrations (0–30 μmol/L) of EGCG and results indicated that colony numbers (Fig. 6A) were significantly decreased in cells treated with EGCG compared with untreated cells. To compare the inhibitory efficiency of EGCG with that of the well-known Pin1 inhibitors, Juglone (35) and PiB (30), we conducted an anchorage-independent soft agar cell transformation assay with Neu/Pin1 WT and Neu/Pin1 KO MEFs using the same conditions and the same concentrations (0-30 μmol/L) of each of the inhibitors (Fig. 6A). Results indicated that EGCG was similar to Juglone in effectiveness but more effective than PiB to inhibit colony formation of Neu/Pin1 WT MEFs. To confirm the essential role of Pin1 in colony formation, we conducted an anchorage-independent soft agar cell transformation assay using Neu/Pin1 KO MEFs reexpressing Pin1 WT or the R17A mutant and found that Neu/Pin1 KO MEFs reexpressing Pin1WT regained sensitivity to EGCG in contrast to the mutant (Fig. 6B).

We finally determined whether EGCG could suppress tumor growth in a xenograft study in vivo using mice injected with Neu/Pin1 WT or KO cells (1 × 10⁶). Mice (n = 15 mice per group) were either treated with vehicle or EGCG (50 mg/kg body weight) 3× week by intraperitoneal injection. We found that tumor formation was significantly delayed by EGCG treatment. At day 14, 60% of wild-type vehicle-treated mice had visible tumors, whereas none of the EGCG-treated wild-type mice had tumors. At day 17, 100% of vehicle-treated wild-type mice had tumors and only 40% of EGCG-treated mice had tumors. Notably, only 1 mouse injected with Neu/Pin1 KO cells in each of the vehicle-treated and EGCG-treated groups developed a small visible tumor (day 28; Fig. 6C). Furthermore, EGCG also suppressed tumor formation in nude mice injected with HCT116 colorectal cancer cells, which also express high levels of Pin1 (Supplementary Fig. S6). The significant effect of EGCG and knockout of Pin1 on tumor development in vivo confirmed that Pin1 is an oncoprotein and the antitumor activity of EGCG is associated with Pin1 activity.

Effective inhibitors of human Pin1 activation could provide a novel clinical strategy for cancer prevention or therapy. Both the WW and PPIase domain activities of Pin1 are important for function of Pin1 (6, 36–40). Thus, a highly effective inhibitor could be targeted either against the catalytic center of the PPIase domain or against the phospho-peptide–binding site of the WW domain. The only reported crystal structure of Pin1 with a ligand bound to the WW domain is the Pin1 complex with the doubly phosphorylated CTD peptide (PDB code: 1f8a; ref. 41). All other crystal structures with peptides or inhibitors of Pin1 were screened against the PPIase domain [(PDB code: 1pin; ref. 6); (PDB codes: 2q5a, 2itk; ref. 23); (PDB codes: 3ik8, 3ikd, 3ikg; ref. 42); (PDB codes: 3kac, 3kad, 3kaf, 3kag, 3kah, 3kai, 3kce; ref. 43); and (PDB codes: 3i6c, 3jyj; ref. 44)]. Here, we present the first crystal structure of the complex of Pin1 with a natural inhibitor where the inhibitor is targeted against not only the WW domain of Pin1 but also against the catalytic PPlase domain of Pin1. The binding of EGCG in both the WW domain and PPIase domain might explain why EGCG is an effective inhibitor of Pin1 and Pin1regulated genes or signaling pathways. The crystallographic findings were supported by our studies in vitro, ex vivo, and in vivo. Previous studies showed that inhibition of Pin1 in cancer cells triggers apoptosis or suppresses the transformed phenotype (19,20). We showed that the Pin1/EGCG association plays a regulatory role in cell proliferation (Fig. 2) and malignant cell transformation (Fig. 6A and B). In a xenograft athymic nude mouse model, our results indicated that EGCG significantly suppresses tumor growth of Neu/Pin1 WT MEFs (Fig. 6C) as well as of HCT116 cells, which express high levels of Pin1 (Supplementary Fig. S6). Pin1 has many reported protein targets and an important Pin1 substrate is the pSer63/73-Pro motif in c-Jun (12). Consistent with our data, the binding of phosphorylated c-Jun (Ser73) with either FL Pin1 or the WW domain of Pin1 was dramatically decreased by EGCG (Fig. 4A and B). The binding of Pin1 to phosphorylated c-Jun is associated with an increased expression of cyclin D1 (12). We found that EGCG treatment corresponded with a decreased abundance of cyclin D1 in cells overexpressing Pin1, but cyclin D1 was unchanged in Pin1-deficient cells (Fig. 4C). EGCG was reported to inhibit tumor promoter–induced cell transformation mediated by AP-1 (2) and NF-κB activation (3). Pin1 was also shown to regulate NF-κB DNA binding and reporter activity after Pin1 recognition of its phosphorylated p65/RelA subunit (45). We found that EGCG effectively inhibited AP-1 or NF-κB reporter activity in Pin1WT MEFs but not in Pin1-deficient MEFs (Fig. 4D and E). We also showed that EGCG might effectively suppress AP-1 or NF-κB activity through Pin1 because EGCG decreased the phosphorylation of the Pin1 substrate, c-Jun at Ser63 and Ser73, in cells expressing Pin1 but not in Pin1-deficient MEFs (Fig. 4C). These data also support the hypothesis that EGCG directly targets Pin1 and this association can inhibit Pin1 tumor promoter activity.

Apoptosis was markedly induced with EGCG treatment in wild-type compared with KO MEFs (Fig. 5B); and EGCG-induced changes in levels of various proapoptotic- or antiapoptotic-associated proteins was more apparent in Pin1-overexpressing cells (Fig. 5C). The comparison of the inhibitory effect of EGCG with other known Pin1 inhibitors, Juglone (35) and PiB (30), revealed that EGCG was comparable to these inhibitors in its ability to suppress colony formation of Neu/Pin1 WT MEFs (Fig. 6A). On the other hand, Pin1-deficient cells were unresponsive to all 3 inhibitors. Juglone and PiB were shown to be specific for the parvulin family of peptidyl-prolyl cis/trans isomerases and do not affect the PPIase activity of other isomerases such as FK506-binding proteins or cyclophilins (30, 35). To determine the ability of EGCG to bind with FKBP12 or cyclophilin A, we prepared an in vitro pull-down assay with EGCG that revealed no binding between EGCG and these isomerases (Supplementary Fig. S5B). In addition, our isomerase activity assays showed that Pin1-expressing and -deficient cells have a differential sensitivity to EGCG despite the same abundance of these isomerases (Supplementary Fig. S5A). Overall, these results indicate a specificity of EGCG in interacting with Pin1.

The crystal structure of the Pin1/EGCG complex revealed that EGCG is bound to the Met15–Tyr23 loop in the WW domain of Pin1 through interaction with the backbone and side chain of Arg17. The importance of this loop and especially Arg17 for the binding activity and function of Pin1 was confirmed by mutational analysis (41), which revealed that the side chain of Arg17 contributes substantially to the ligand-binding energy through the recognition of a phosphorylated serine/threonine residue of target proteins. Replacing Arg17 with Ala resulted in a 6.3-fold decrease in binding activity (41). Notably, our results indicated that mutating Arg17 to Ala resulted in an inability of Pin1 to bind EGCG ex vivo (Fig. 1E). These findings might explain why EGCG binding with WW domain can inhibit its phospho-group binding ability. The other consequence of the EGCG association with Arg17 might be its effect on Ser16 phosphorylation. The Pin1 WW domain is phosphorylated on Ser16 both in vitro and in vivo, and the phosphorylation regulates the ability of the WW domain to mediate Pin1 substrate interaction and cellular localization (39). Pin1 and WW domain mutants, which cannot be phosphorylated at Ser16, were shown to act as dominant-negative mutants to induce mitotic block and apoptosis and increase multinucleated cells (39). Even though the distances from EGCG to the hydroxyl group of Ser16 in the solved structure are too large to form hydrogen bonds, EGCG binding with Arg17 could prevent phosphate group access to this residue.

The suppressive effect of EGCG on Pin1 isomerase activity in vitro and ex vivo provides evidence that EGCG inhibits Pin1 isomerase function even after reexpression of Pin1 in Pin1KO cells and confirms our results showing that EGCG also binds with the PPIase domain of Pin1. Cys113 was shown to be important for the isomerase activity of Pin1 because its thiol group is involved in catalysis (6). The flexibility of a main chain of residues spatially neighboring Cys113, including Trp73 and Ser114, might be critical for rotational freedom of thiol group of Cys113 (46), which is necessary for the catalytic reaction (6). In the solved structure, EGCG forms hydrogen bonds with the amide groups of the backbone of Trp73 and Ser114 and the carbonyl group of the backbone of Asp112, making them more rigid. Thus, EGCG can affect the unrestricted movement of Cys113.

Finally, the changing of the surface charges around the side chain of Cys113 upon the negatively charged group binding might alter the reactivity of its thiol group (46). The apparent pK_a of Cys113 increases (i.e., becomes more basic), making the residue less nucleophilic, which will inhibit the catalytic reaction (46). EGCG might possibly have an effect on changing the surface charge around Cys113 to control the isomerase activity of Pin1. Overall, our crystallographic findings and accompanying biochemical data showed that EGCG binds with both domains of Pin1 and directly inhibits the tumor-promoting activity of Pin1 and, thus, might have practical implications for cancer prevention or therapy.

No potential conflicts of interest were disclosed.

Z. Dong designed research; D.V. Urusova, J.-H. Shim, D.J. Kim, S.K. Jung, A.M. Bode, and A. Carper carried out research experiments; D.V. Urusova, J.-H. Shim, T.A. Zykova, D.J. Kim, S.K. Jung, A.M. Bode, and A. Carper analyzed data; and D.V. Urusova, D.J. Kim, and A.M. Bode wrote the manuscript.

The authors thank Dr. Kun Ping Lu (Harvard Medical School, Boston, MA) for providing the GST/Pin1, Pin1-deficient (Pin1^−/−), Neu/Pin1 KO, Pin1 expressing (Pin1^+/+), and Neu/Pin1 WT cell lines and Dr. Zhiwei He for providing the siPin1 construct and assisting in the creation of the knockdown and overexpression cells. They also thank Dr. Jessie Zhang (Salk Institute, University of California, Berkeley, CA) for the gift of the Pin1 R14A mutant plasmid. They also acknowledge the Argonne Photon Source and Argonne National Laboratory for providing access to the X-ray facilities for data collection and Drs. Valentina Terechko (University of Chicago, Chicago, IL) for her assistance during the data collection and Svetlana Ermakova (Pacific Institute of Bioorganic Chemistry, Vladivostokm, Russia), Sergei Pletnev (National Cancer Institute, Argonne, IL), and Svetlana Antonyuk (University of Liverpool, Liverpool, United Kingdom) for their advice and assistance.

This work was supported by The Hormel Foundation and NIH grants R37 CA081064 and CA120388.

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