Metastasis is the leading cause of death in cancer patients, and understanding the molecular mechanism of cancer metastasis is therefore one of the urgent problems. Phosphatase of regenerating liver-3 (PRL-3) was identified as the only gene that is consistently overexpressed in liver metastases derived from colorectal cancer (CRC), whereas it is undetectable in normal colorectal epithelia and shows intermediate expression in advanced primary cancer.1 Increasing evidence suggests that PRL-3 plays multiple roles in cancer migration and metastasis: PRL-3 expression is elevated in most metastatic lesions from various human malignancies including those of the colorectum,2, 3 stomach,4 mammary gland5 and ovary.6 In the in vivo cancer-metastasis models, overexpression of PRL-3 in Chinese hamster ovary (CHO) cells promoted the incidence of metastasis,7, 8 and knockdown of PRL-3 expression in CRC-derived DLD-1 cells by RNA interference effectively abrogated the activities of cancer cell motility in vitro and hepatic colonization in vivo.3
The PRL family members (PRL-1, PRL-2 and PRL-3) comprise a distinct subclass of protein tyrosine phosphatases (PTPs),9 whereas the C(X)5R motif within the P-loop of the PRL-3 protein is typical for dual-specific phosphatases: It is close in structure to the Rho-type small GTP-binding protein (GTPase) Cdc42, and to phosphatase tensin homologues deleted on chromosome 10 (PTEN).10 The catalytic domain of the PRL-3 protein is critical for its regulation of PRL-3-mediated cancer cell migration and invasion: site-directed mutagenesis experiments revealed that catalytic domain-mutants of PRL-3 (C104A) within the P-loop diminished PRL-3-induced invasion and Rho activation in CRC-derived SW480 cells.11 Interestingly, recent studies have demonstrated the possible role of PRL-3 phosphatase activity in the promotion of the epithelial-mesenchymal transition (EMT) in cultured cells: In DLD-1 cells, PRL-3 overexpression promoted Akt activation and upregulated Snail expression levels, consequently leading upregulation of the mesenchymal marker fibronectin and downregulation of adhesion molecules such as E-cadherin, γ-catenin and integrin β3.12 Interestingly, the catalytically inactive PRL-3 mutant (C104S) was impaired in the above PRL-3-triggered EMT.12
At present, integrin α1 has been identified as a PRL-3 PTP-targeting protein, which may subsequently downregulate the tyrosine phosporyation of integrin β1.13 However, the mechanistic basis for this PRL-3 function has remained unknown. As another approach in defining the PRL-3 signal pathway, we herein used protein expression profiling to search for novel PRL-3-mediated effector proteins. Because proteomics is advantageous for analyzing the complete complement of proteins expressed by a biological system in response to various stimuli and/or under different physiological or pathological conditions, we attempted to find the phosphorylated proteins that are differentially expressed in SW480 cells overexpressing wild-type and catalytically inactive mutant PRL-3 to uncover novel molecular targets that would be dephosphorylated during PRL-3-mediated promotion of cancer cell motility and progression. Also, a novel PRL-3 inhibitor14 was used to assess the biological significance of PRL-3 phosphatase activity.
Material and methods
Human CRC cells SW480 (American Type Culture Collection, Manassas, VA) were maintained in RPMI-1640 supplemented with 10% fetal bovine serum (Sigma, St. Louis, MO) and antibiotics (Sigma) in a humidified atmosphere containing 5% CO2. Cells were treated with the PRL-3 inhibitor (Sigma), which was dissolved in dimethylsulfoxide (DMSO). We confirmed that the final concentration of DMSO did not influence the cell viability and the status of proteins expressed (data not shown).
Coding sequence for human PRL-3 (Accession No. NCBI database NM_032611) was amplified by reverse transcription-polymerase chain reaction, and the cDNA was subcloned into the mammalian expression vector pEGFP-C1 (Clontech, Palo Alto, CA) to generate pEGFP-PRL-3 vector, and PRL-3 (C104S) phosphatase ‘dead’ mutant expression vector pEGFP-C104S.7 Each plasmid was transfected into SW480 cells using FuGENE 6 (Roche, Indianapolis, IN) according to the manufacturer's instruction. Similarly, the mammalian expression vector pFLAG-CMV2 (Invitrogen, Carlsbad, CA) was also utilized for generation of pFLAG-PRL-3 and pFLAG-C104S that expressing FLAG-tagged recombinant proteins.
Cell migration analysis and wound-healing assay
For the cell migration analysis, cells (3.0 × 104 cells/well) were seeded on cell culture inserts and allowed to migrate to the underside of the insert for 48 hr at 37°C. The insert membrane was fixed with formamide and stained with 1% hematoxylin, and the cells that migrated to the lower side of the membrane were counted under microscope. For the wound-healing assay, cells grown to 90% confluence in 6-well plate were wounded using a sterile 1000 μl pipette tip, washed to remove the floating cells and returned to the incubator before imaging.
The antibodies used in this study were as follows: anti-PRL-3 (Sigma); anti-GFP (MBL, Nagoya, Japan); anti-FLAG (Sigma); anti-p130Cas (Santa Cruz, Santa Cruz, CA); anti-FAK (Santa Cruz); anti-E-cadherin (Santa Cruz); anti-paxillin (BD Transduction, Lexington, KY); anti-vimentin (Thermo Shandon, Pittsburgh, PA) and anti-β-actin (Sigma). Anti-PRL-3 antibody is shown not to cross-react with human PRL-1 and PRL-2 by manufacturer. For the phosphorylated keratin 8 (KRT8) levels, anti-phospho-KRT8-specific antibodies at S73 (Thermo Fisher, Fremont, CA) and S431 (Thermo Fisher) were used as well as anti-KRT8 antibody (Santa Cruz).
Cells were lysed in a buffer containing 150 mM NaCl, 10 mM Tris-HCl (pH 7.4), 0.5 mM ethylene diamine tetra acetic acid, 1% Triton-X, 1% Protease Inhibitor Cocktail (Sigma) and 1% Phosphatase Inhibitor Cocktail (Sigma). Immunoprecipitation Matrix (Santa Cruz) was incubated with 5 μg of each antibody against GFP or KRT8 overnight, and the lysates were transferred to the complex, followed by overnight incubation. Total cell lysates and immunoprecipitates were resolved by sodium dodecylsulfate (SDS)-polyacrylamide gel electrophoresis (PAGE), transferred to a polyvinylidene difluoride membrane (Millipore, Bedford, MA), and analyzed by immunoblot using horseradish peroxidase (HRP)-conjugated anti-mouse/rabbit IgG (GE Healthcare, Piscataway, NJ) and ExactaCruz Western Blot Reagent (Santa Cruz) as secondary antibodies, respectively. The specific bindings were visualized using ImmunoStar Reagents (Wako, Osaka, Japan).
Immunofluorescence and immunohistochemistry
For immunofluorescence, cells were grown on glass coverslips and then fixed with precooled methanol (−20°C) for 10 min. After washing with phosphate-buffered saline, cells were double-stained with anti-PRL-3 and anti-KRT8. Staining patterns obtained with antibodies against PRL-3 and KRT8 were visualized with Cy2-conjugated antibody against rabbit IgG (GE Healthcare) and Cy3-conjugated antibody against mouse IgG (GE Healthcare), respectively. The nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI). Cells were observed under Olympus BX50 microscope (Olympus, Tokyo, Japan). Also, anti-KRT8 antibodies at S73 and S431 were used for immunofluorescence.
Immunohistochemistry was performed using the streptavidin-biotin-peroxidase method with LSAB kit (Dako, Glostrup, Denmark). Deparaffinized and rehydrated 4-μm sections were autoclaved to retrieve antigenicity. After blocking endogenous peroxidase and nonspecific binding sites, each primary antibody was applied to sections. Sections were incubated with biotinylated goat anti-mouse/rabbit IgG and streptavidin conjugated to HRP. Chromogenic fixation was carried out by immersing the sections in a solution of 3,3′-diaminobenzidine tetrahydrochloride. Sections were counterstained with Mayer's hematoxylin. The degree of protein expressions was evaluated according to the number of stained cells and the staining intensity in individual cells.4
SW480 cells (1 × 106) transfected with pFLAG-PRL-3 or pFLAG-C104S were lysed in sample buffer containing 7 M urea, 2 M triourea, 4% CHAPS, 2 mM tributyl phosphine and 0.2% Bio-Lyte 3/10 ampholytes (Bio-Rad, Hercules, CA). The crude cell homogenate was sonicated and centrifuged at 10,000g for 10 min. Eleven centimeter immobilized pH gradient strips with pH range 3.5–10 were hydrated overnight in sample buffer containing 200 μg of total protein. After isoelectric focusing, using Ampholine (GE Healthcare), proteins were separated in the second dimension by 9–18% gradient SDS-PAGE. After 2D separation, the gels were stained with SYPRO Ruby (Invitrogen) or Pro-Q Diamond phospho-protein gel stain (Invitrogen). For each sample, 2 biological and 2 technical replicate gels were run. Images were scanned with Versadoc 3000 image system (Bio-Rad), and then analyzed with the ImageMaster 2D ver. 5.0 (GE Healthcare).
Matrix-assisted laser desorption/ionization (MALDI)-time of flight (TOF) and mass spectrometry (MS)
MS was carried out in Pro Phoenix (Hiroshima, Japan). Proteins from spots that were consistently increased at least 1.5-fold in the pFLAG-C104S transfected SW480 cells were subjected to MS. Gel pieces were incubated in 100 mM NH4CO3, dehydrated with acetonitrile and then dried in centrifugal concentrator. This process was repeated twice. In-gel digestion of the extracted proteins was carried out with 10 μg/mL trypsin in 50 mM NH4CO3/5 mM CaCl2 for 16 hr. The digested peptides were extracted with a mixture of 5% formic acid/50% acetonitrile and applied onto a stainless steel MALDI plate (Waters, Milford, MA). MS of the resulting peptides was recorded on the MALDI-TOF spectrometer (Waters) in reflectron mode. Resulting peptides were matched with their corresponding proteins with MASCOT searching the NCBI database.
PRL-3 phosphatase activity is essential for the promotion of PRL-3-mediated cell motility
The importance of PRL-3 phosphatase activity has been shown in the acceleration of cancer cell motility, invasion and metastasis7, 8, 11; we therefore confirmed these PRL-3 characteristics using the phosphatase ‘dead’ mutant PRL-3 expression vector pEGFP-C104S and the PRL-3 inhibitor in SW480 cells. As reported previously,11 PRL-3 overexpression reduced paxillin expression levels, whereas no significant change was found in the other PRL-3-related adhesion molecules, including p130cas and FAK (Fig. 1a). Recombinant PRL-3 protein fused to EGFP was detected mainly in the cytoplasm with faint staining at the plasma membrane (Fig. 1b). Although recombinant PRL-3 proteins enhanced cancer cell motility in the migration and wound-healing assays, the substitution of C104 with serine effectively diminished the PRL-3-mediated cancer cell migration activity in vitro (Figs. 1c and 1d).
Next, we examined the effects of the PRL-3 inhibitor (Fig. 2a) on the regulation of cell-cell adhesion molecules and cell migration. This PRL-3 inhibitor slightly decreased paxillin levels (Fig. 2b) but altered the spindle morphology of SW480 cells. The spindle shape was impaired but an epithelial-like morphology was acquired in the presence of the inhibitor (Fig. 2c). The cell proliferation and viability were not affected by the presence of the inhibitor (data not shown). Inhibition of cell motility by this PRL-3 inhibitor was confirmed (Figs. 2d and 2e). To confirm the importance of PRL-3 phosphatase activity in the regulation of cell migration in the other cell line, we performed the wound-healing assay in DLD-1 cells using the above gene expression system and the PRL-3 inhibitor (Supporting Fig. S1).
PRL-3 direct interaction with KRT8 modulates dephosphorylation of KRT8
To investigate the candidate molecules that can be dephosphorylated by PRL-3 phosphatase activity, the differentially expressed phosphorylated protein profiles between wild-type and C104S mutant PRL-3 recombinant proteins overexpressing SW480 cells were obtained by proteomic analysis. Protein extracts were prepared 48 hr after pFLAG-PRL-3 and pFLAG-C104S transfection and subjected to 2D-PAGE. After staining with SYPRO Ruby to confirm equivalent transduction of each ectopic expression vector (pFLAG-PRL-3 transfectant, 552 spots; and pFLAG-C104S transfectant, 524 spots; Fig. 3a), phosphorylated proteins were visualized by Pro-Q Diamond (pFLAG-PRL-3 transfectant, 292 spots; and pFLAG-C104S transfectant, 329 spots; Fig. 3b). MALDI-TOF and MS analysis were carried out on the several highest-ranking phospho-protein spots that were increased in SW480 cells transfected with pFLAG-C104S compared with the pFLAG-PRL-3 transfectant. Finally, the intermediate filament protein KRT8 (Accession No. NCBI database NM_ 002273; average difference value, 1.788) was identified.
Next, the possible modulation of KRT8 phosphorylation by PRL-3 was confirmed in SW480 cells. Treatment with the PRL-3 inhibitor increased the expression levels of phosphorylated KRT8 at S73 and S431 not only in parental cells in a dose-dependent manner (Fig. 4a) but also in the pEGFP-PRL-3 transfectant (Fig. 4b). Loss of phosphorylated KRT8 expression at S73 and S431 were also found in the pEGFP-PRL-3 transfectant but not in the pEGFP-C104S mutant transfectant (Fig. 4c). Conversely, induction of phospho-Tyr-positive KRT8 expression was not detected (data not shown), supporting a possible physiological modulation of KRT8 phosphorylation at both S73 and S431 by PRL-3. To investigate the potential physiological relevance between PRL-3 and KRT8, immunoprecipitation was conducted. Direct interaction was detected between endogenous KRT8 and recombinant PRL-3 C104S mutant protein; however, the interaction was faint with wild-type PRL-3 (Fig. 5a). The possible interaction between PRL-3 and KRT8 was suggested in SW480 cells but we further ascertained direct interaction between PRL-3 and KRT8 in DLD-1 cells. Similar to the results in SW480 cells, a stable physiological interaction with KRT8 was detected in the DLD-1 cells transfected with pEGFP-C104S (Supporting Fig. S2a). Furthermore, increased levels of phosphorylated KRT8 at S73 and S431 were detected when the DLD-1 cells were treated with PRL-3 inhibitor (Supporting Fig. S2b). Endogenous PRL-3 and KRT8 colocalized at the cellular lamellipodias and ruffles, with a meintenance of spindle morphology at the edge of the cells. Nevertheless, when the cells were treated with PRL-3 inhibitor (50 μM), colocalization of PRL-3 and KRT8 proteins disappeared and intermediate bundles composed of KRT8 distributed diffusely in the cytoplasm (Fig. 5b).
Elevated PRL-3 expression and the phospho-KRT8 status during the progression of CRC
Finally, the results in the current study were confirmed in human tissues of primary CRC and liver metastasis. In Figure 6, representative results of immunohistochemical analysis in the same patient with CRC and liver metastasis are shown. PRL-3 expression levels were extremely low at the superficial (apical side) part of these CRCs and normal mucosae; however, elevation of PRL-3 levels was significant at the invasive front (serosal site) and liver metastases, along with remarkable reduction of phosphorylated KRT8 (S73 and S431) expression. However, there was no significant change in the levels of total nonphosphorylated KRT8 expression.
In this study, we attempted to identify the target molecules dephosphorylated by PRL-3 that is potentially important for PRL-3-modulated cancer metastasis. Using proteome analysis of phosphorylated proteins differentially expressed in SW480 cells transfected with pFLAG-PRL-3 and pFLAG-C104S vectors, KRT8 was found to be a target molecule dephosphorylated by PRL-3. This is the first report that recognized a PRL-3-interacting protein by 2D-PAGE analysis, followed by MALDI-TOF and MS analyses. However, in our system, we could not detect the other 2 substrates, integrin α1 and ezrin, both of which had been reported as the PRL-3 substrates.13, 15 This may be probably caused by the methodological difference. Integrin α1 was detected by screening of human placenta brain cDNA library using yeast 2-hybrid system with an intention to identify a direct PRL-3-interacting protein.13 In addition, Forte et al.15 performed mono-dimensional-PAGE for the identification of a specific cellular substrate of PRL-3 using wild-type and catalytically ‘dead’ mutant PRL-3 expression vectors. Although this approach is basically similar to our system, we believe that our 2D-PAGE system with a high resolution has an advantage to determine the specific protein spots differentially expressed in SW480 cells transfected with pFLAG-PRL-3 and pFLAG-C104S vectors. Moreover, binding of the C104S mutant PRL-3 to substrate KRT8 was much stronger than that of the wild-type PRL-3. This may possibly be caused by the transient nature of the binding between the wild-type phosphatase and its substrate. Because the Cys to Ser mutant at the PTP active site is known to impair the dephosphorylation activity without disturbing the binding ability to its substrate,16 the C104S mutant protein may acted as a substrate-trapping mutant, forming stable complex with KRT8.
PRL-3 modulation of KRT8 phosphorylation was confirmed by treatment with the PRL-3 inhibitor; however, upregulation of phosphorylated KRT8 levels was detected at S73 and S431 but not at tyrosine residues. Although these results seem to be contradictory to the PRL-3 function as a PTP, recent research has shown that ezrin, an ezrin-radixin-moesin family member, can be directly dephosphorylated by PRL-3 protein at its threonine residue T567,15 suggesting the dual phosphatase function of PRL-3 for the modulation of protein dephosphorylation at both tyrosine and serine/threonine residues. According to our unpublished data (Mizuuchi et al.) phosphorylation of tyrosine residues of KRT8 was undetectable with an antibody against phospho-tyrosine in the pEGFP-PRL-3 transfected SW480 cells (data not shown). Keratins are intermediate filaments that form the cytoskeletal framework in the cytoplasm of various eukaryotic cells, and they serve as markers of the epithelial tissue origin of tumors.17 Interestingly, the intracellular organization of intermediate filament networks is under the control of protein kinases and phosphatases: The site-specific phosphorylation of keratins induces the disassembly of these filaments, and the balance between their phosphorylation and dephosphorylation controls the continuous exchange of intermediate filament subunits between a soluble pool and polymerized filaments. Although we did not investigate KRT8's role in the enhancement of cancer cell motility and metastatic potential, our data implicate KRT8 as a physiologic target molecule for PRL-3, which may dephosphorylate S73 and S431 of the KRT8 protein but accelerate the switch of reorganization and stabilization of keratin filaments. Hence, we hypothesize that activation of phosphorylation and dephosphorylation of KRT8 is required for promoting KRT8-mediated cancer cell migration. Dissociation and reorganization of keratins is expected to accelerate cell motility and metastatic potential in CRC cells (Fig. 7); however, identification of the detailed mechanism(s) of KRT8 in the PRL-3 signaling pathway awaits further study.
Cell–cell adhesiveness is generally reduced in human cancers, and reduced intercellular adhesion is indispensable for cancer invasion and metastasis.18 Loss or decrease of adhesion molecules is a common event that promotes cancer progression and metastasis, and reduction of E-cadherin has been documented in many human malignancies.19 The PRL-3 effects on PTEN expression and phosphatidylinositol 3-OH kinase signaling to promote EMT have been reported: Induction of PRL-3 expression in HeLa and CHO cells reduced paxillin, and PRL-3 upregulated mesenchymal markers fibronectin and Snail and downregulated epithelial markers E-cadherin, γ-catenin and integrin β3.12 We observed that the PRL-3 inhibitor weakly suppressed vimentin expression levels with decreased levels of E-cadherin (Mizuuchi et al., unpublished data) supporting the idea that PRL-3 can be a good biomarker for determining cancer cell differentiation. Additional investigations on PRL-3 may not only provide information useful in determining the grade of malignancy of each CRC case, but may also point to a novel therapeutic target for intractable CRC metastasis.