DLD-1 cells were cultured in RPMI medium supplemented with 10% fetal bovine serum (FBS). HEK293T cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% FBS. RKO cells were cultured in Eagle’s minimum essential medium (MEM) containing 10% FBS. Plasmids and siRNAs were transfected into cells using Lipofectamine 2000 and RNAiMAX (Invitrogen), respectively.

Myc-tagged PTPRD (corresponding to NM_002839.2) was generated by PCR from the cDNA of HEK293T cells and cloned into pcDNA3.1(+) (Invitrogen). For the preparation of GST-fusion proteins, cDNA fragments were subcloned into pGEX-5X (GE Healthcare). GST-fusion proteins were synthesized in Escherichia coli and isolated by adsorption to glutathione-conjugated Sepharose (GSH-Sepharose; Pharmacia). Catalytic-inactive (C1553S) and substrate-trapping (D1521A) mutants of PTPRD were generated using PCR mutagenesis and cloned into pcDNA3.1(+) and pGEX-5X, respectively.

Mouse monoclonal antibodies to β-catenin, Plakoglobin and E-cadherin, were obtained from BD Biosciences. The rabbit polyclonal antibody to PTPRD was generated by immunizing rabbits with a GST-fusion protein containing amino acids 1096–1127 of PTPRD. The antibodies were purified by affinity chromatography using columns to which the antigens used for immunization had been linked.

Total RNA was extracted using TRIsure (Bioline) and reverse-transcribed into cDNA using the ReverTra Ace qPCR RT kit (Toyobo). Real-time PCR was performed using LightCycler480 SYBR Green I Master and a LightCycler480 Instrument (Roche). The results were normalized with the detected value for GAPDH. Primers used in RT-PCR were as follows: GAPDH forward, 5′-GCA CCG TCA AGG CTG AGA AC-3′; GAPDH reverse, 5′-TGG TGA AGA CGC CAG TGG A-3′; PTPRD forward, 5′-GCT GCT GCT CCT CAC TTT CT-3′; PTPRD reverse, 5′-CGG GTG TTC GTG TAA ACC TT-3′.

Immunoblotting, immunoprecipitation and GST pull-down assays were performed as described previously (17).

Cells were fixed with 10% formalin/PBS for 15 min and stained overnight with each antibody at 4˚C. Primary antibodies were diluted as follows: anti-β-catenin (1:500), anti-Plakoglobin (1:500) and anti-E-cadherin (1:250). Staining patterns obtained with antibodies were visualized with Alexa Fluor 488-conjugated secondary antibodies (Molecular Probes). Cells were photographed with a Carl Zeiss LSM510 laser scanning microscope (Carl Zeiss).

Cell migration assays were performed as described previously (18) with minor modifications. Briefly, the underside of the filter membrane was coated with 4 μg/ml type-I collagen (Koken, Japan), 10 μg/ml fibronectin (Sigma-Aldrich) or 10 μg/ml laminin (Sigma-Aldrich), respectively, and was allowed to air-dry. Transfected DLD-1 cells (1×10⁴ cells per well) were added to the upper compartment of the Transwell and allowed to migrate to the underside for 6 h. For RKO cells, 5×10³ cells were allowed to migrate for 4 h.

For scattering assays, transfected DLD-1 cells (1.2×10⁵) were seeded on coverslips (18 mm in diameter; Matsunami) and cultured for 48 h. Cells were fixed with 10% formalin/PBS for 15 min and stained with TO-PRO 3 (Molecular Probes). For quantification of scattering, staining patterns were photographed and analyzed by Image-J software. The scattering index is defined as the ratio of the area of scattering nuclei divided by that of total nuclei.

Stealth siRNA duplexes against PTPRD were purchased from Invitrogen. The sequences of siRNAs targeting the human PTPRD were: PTPRD-1, 5′-ACA TCA TTC AGT TGT AGC ACA TTT C-3′; PTPRD-3, 5′-ATG ACT TGT ATG TCA CTT GAA AGG G-3′. Validated Stealth negative control siRNA duplex with low GC content (Invitrogen) was used as a control. DNA oligonucleotides encoding shRNA were subcloned into pSuper-retro-puro (OligoEngine). The sequence of shRNA targeting the human CD44 was 5′-CTG GCG CAG ATC GAT TTG AAT-3′.

Microarray datasets and clinical annotations were obtained from the Gene Expression Omnibus (GSE5206 and GSE17537). Signal values of each dataset were log-transformed and normalized before analysis. The probe set IDs were converted to Ensembl gene ID. In cases where one gene ID matched multiple probe set IDs, probe sets with low signal intensity were excluded, and the probe set which shows the most variance across the samples was mapped to the gene ID. Wnt target genes were obtained from The Wnt Homepage (http://www.stanford.edu/~rnusse/wntwindow.html) and converted to Ensembl gene ID. Hierarchical clustering and visualization of the results were performed using GenePattern software, and survival analysis was performed using R software.

To investigate the role of PTPRD in cell migration, knockdown experiments were performed using two distinct small interference RNAs (siRNAs). The expression of PTPRD was significantly reduced in DLD-1 cells transfected with the specific siRNAs, but not the control siRNA (Fig. 1A). When siRNA-transfected cells were subjected to migration assays using collagen-coated Boyden chambers, the migratory activity of these cells was significantly increased compared to that of the control siRNAtransfected cells (Fig. 1B). We next performed cell scattering assays and found that the knockdown of PTPRD induced cell scattering without growth factor stimulation (Fig. 1C and D). Consistent with these results, ectopic expression of PTPRD resulted in the inhibition of cell migration (Fig. 1E). Notably, the ectopic expression of a catalytic inactive mutant of PTPRD (PTPRD-CS) also suppressed migration to the same extent as the wild-type, implying that phosphatase activity is dispensable for this suppression.

It is widely accepted that the disruption of cell-cell adhesion is a crucial step in cancer invasion and metastasis (19). The extracellular domain of PTPRD undergoes homophilic binding, indicating that PTPRD plays a role in cell-cell adhesion (20). We therefore assessed the formation of cell-cell contacts in cells transfected with PTPRD-specific siRNAs by immunostaining for the adhesion molecules, β-catenin, Plakoglobin and E-cadherin. These adhesion molecules clearly accumulated at the cell-cell contacts in cells transfected with control siRNA, whereas they showed diffused localization in cells transfected with PTPRD-specific siRNAs (Fig. 2A, arrowheads). These results suggest that PTPRD is required for appropriate cell-cell adhesion.

We next investigated whether PTPRD directly interacts with these adhesion molecules. It is known that the membraneproximal D1 domain is catalytically active, while distal D2 domain and substrate-trapping DA mutant are inactive (1,2). We constructed GST-fusion proteins containing one of the two PTP domains (D1 and D2) or a substrate trapping mutant of D1 (DA), respectively. When cell lysates were subjected to pull-down assays with these GST-fusion proteins, β-catenin was found to interact with D2, but not D1, DA or GST alone (Fig. 2B). To confirm this result, in vivo immunoprecipitation experiments were performed with HEK293T and DLD-1 cells using anti-PTPRD antibody. Both β-catenin and E-cadherin co-precipitated with PTPRD, whereas N-cadherin did not (Fig. 2C). In these experiments, co-precipitations of β-catenin and E-cadherin were inhibited by pre-incubation of the antibody with the antigen used for immunization. These results suggest that PTPRD directly regulates the function of the β-catenin/E-cadherin complex.

DLD-1 cells harbor biallelic nonsense mutations of APC and express truncated mutant APC. Truncated mutant APC is capable of enhancing cell migration through activation of both β-catenin/TCF and Rac/Cdc42 signaling (18,21–24). Consistent with previous reports (22–24), ectopic expression of ΔN-TCF4, a dominant-negative form of TCF4, suppressed cell migration (Fig. 3A). The migratory activity of cells transfected with ΔN-TCF4 along with wild-type PTPRD was unexpectedly significantly higher than that of cells transfected with ΔN-TCF4 alone. By contrast, PTPRD-CS, a catalytic inactive mutant, did not show this effect. Thus, it appears that PTPRD stimulates cell migration in a phosphatase activity-dependent manner in the absence of Wnt signaling. To rule out the possibility that PTPRD overcomes or inhibits the dominant-negative effect of ΔN-TCF4, luciferase assays were performed using a TCF-responsive reporter. PTPRD failed to increase its reporter activity when co-expressed with ΔN-TCF4 (Fig. 3B). We next performed migration assays using RKO cells, which contain intact APC and β-catenin. RKO cells transfected with PTPRD showed increased motility in a phosphatase activity-dependent manner compared to the control cells (Fig. 3C). In contrast, when β-catenin-S33Y, a constitutively activated form of β-catenin, was co-expressed, PTPRD suppressed cell migration in a phosphatase activity-independent manner, as observed with DLD-1 cells in Fig. 1E. Taken together, these results suggest that the effect of PTPRD on cell migration depends on the status of β-catenin/TCF signaling.

We therefore attempted to identify Wnt target genes that are involved in the regulation of PTPRD-dependent migration. Knockdown experiments were performed using shRNAs targeting known Wnt target genes. Similar to cells transfected with PTPRD, cells transfected with a shRNA targeting CD44 (shCD44) showed decreased motility compared to control cells (Fig. 3D). However, cells transfected with PTPRD along with shCD44 showed increased migratory activity compared to cells transfected with shCD44 alone. These observations suggest that CD44 is involved in PTPRD-dependent migration.

To investigate the clinical importance of PTPRD, gene expression profiles of human colon cancers were obtained from a public microarray database. The expression levels of PTPRD were significantly lower in the highly invasive cancers (T stage 3 or 4) than in the less invasive cancers (Fig. 4A). There were no significant differences in either lymph node and distal metastasis (data not shown). In most colon cancers, APC or β-catenin is mutated and β-catenin/TCF signaling is constitutively activated (25,26). However, the levels of β-catenin/TCF signaling differ among cancers according to the type and/or the combinations of mutations (27). We therefore divided the dataset into two subgroups based on the expression levels of known Wnt target genes and termed them the ‘Wnt High group’ and ‘Wnt Low group’, respectively (Fig. 4B). Samples from normal colons were classified as the Wnt Low group. An association of β-catenin/TCF signaling and PTPRD expression was noted in the Wnt High group, but not in the Wnt Low group (Fig. 4C). We also performed Cox’s regression analysis and found that both overall survival and disease-free survival were correlated with the expression level of PTPRD (p=0.008 and p=0.0025, respectively; Fig. 4D, upper panel). Furthermore, a significant correlation between PTPRD expression and survival was observed only in the Wnt High group, but not in the Wnt Low group (p=0.036 and p=0.22, respectively; Fig. 4D, lower panel). Taken together, these results suggest the possibility that the down-regulated expression of PTPRD contributes to the invasive properties and poor patient outcome in human colon cancers in coordination with β-catenin/TCF signaling.