Biphasic regulation of tumorigenesis by PTK7 expression level in esophageal squamous cell carcinoma

Protein tyrosine kinase 7 (PTK7), also known as colon carcinoma kinase 4 (CCK-4), is a member of the catalytically defective receptor protein tyrosine kinase family and is upregulated in various cancers, where it is known to act as either an oncoprotein or a tumor suppressor. To understand the contrasting roles of PTK7 in tumorigenesis, we analyzed the tumorigenic characteristics of esophageal squamous cell carcinoma (ESCC) cells with low levels of endogenous PTK7 expression (TE-5 and TE-14 cells) and high levels of expression (TE-6 and TE-10 cells) after transfections with a PTK7 expression vector. PTK7 overexpression increased the proliferation of TE-5 and TE-14 cells but decreased the proliferation of TE-6 and TE-10 cells. In the ESCC cells, proliferation, migration, and invasion were initially increased and then decreased according to PTK7 expression levels, which were mirrored by initial increases and then decreases in the tyrosine phosphorylation of cellular proteins and phosphorylation of Src, Akt, and ERK. In ESCC patients included in The Cancer Genome Atlas database, those with higher PTK7 mRNA levels had a longer overall survival and lower relative risk than those with lower PTK7 mRNA levels. These results demonstrate that PTK7 biphasically regulates tumorigenesis in ESCC.

carcinoma 28 . In lung squamous cell carcinoma cells, the overexpression of PTK7 decreases cell proliferation, migration, and invasion by suppressing ERK and Akt phosphorylation 29 .
To clarify the role of PTK7 in ESCC tumorigenesis, we compared PTK7 expression levels with physiologic changes in these cells. We also analyzed the relationship between PTK7 mRNA levels and the survival of ESCC patients using transcriptome sequencing (RNA-seq) data from The Cancer Genome Atlas (TCGA) database.

Results
PTK7 is expressed at high levels in numerous cancers. We analyzed RNA-seq data from TCGA to assess PTK7 mRNA levels in various cancer tissues. In samples of endometrial, head and neck, lung, ovary, cervical, prostate, breast, pancreatic, bladder, thyroid, and esophageal cancers, the mean PTK7 mRNA levels were higher than the overall mean level from all cancer tissues (Fig. 1a). Thirteen pairs of esophageal cancer and matched normal tissue samples were found in the database. A paired analysis showed that the PTK7 mRNA level was higher in esophageal cancer tissues than in adjacent normal tissues (P < 0.001) (Fig. 1b). Among the 198 esophageal cancer samples in the database from TCGA, we extracted 96 EAC and 96 ESCC samples; six samples were undefined. The PTK7 mRNA level was higher in ESCC samples than in EAC samples (Fig. 1c). These data demonstrate that in most cancer types, including esophageal cancer, PTK7 expression is upregulated.

PTK7 overexpression increases the proliferation of PTK7-low ESCC cells but decreases the proliferation of PTK7-high ESCC cells.
To examine the oncogenic role of PTK7, we ectopically expressed it in two ESCC cell lines with low endogenous levels of expression (TE-5 and TE-14) and two cell lines that highly express PTK7 (TE-6 and TE-10) (Fig. 2a). Ectopic expression of PTK7 increased the proliferation of TE-5 and TE-14 cells (i.e., PTK7-low ESCC cells) (Fig. 2b), whereas the proliferation of the PTK7-high TE-6 and TE-10 cells decreased with ectopic expression (Fig. 2c). This result was unexpected, as previous evidence showing PTK7 knockdown decreased the proliferation of TE-10 and TE-11 cells indicated that PTK7 is an oncogene in ESCC 4 .

Biphasic oncogenic regulation of ESCC cells by PTK7 expression.
To further examine the unexpected findings regarding PTK7 expression levels and ESCC cell proliferation, we next analyzed the oncogenic properties of ESCC cells expressing low to high levels of PTK7. Increasing amounts of the PTK7 expression vector were transfected into PTK7-low TE-5 and TE-14 cells ( Fig. 3a and Supplementary Fig. S1a), and increasing amounts of the expression or a knockdown vector were transfected into PTK7-high TE-10 and TE-6 cells ( Fig. 3b and Supplementary Fig. S1b). Inverted U-shaped dose-effect curves were observed with regard to the effects and PTK7-high TE-6 and TE-10 cells (c) were transfected with a PTK7 expression vector (pcDNA3-PTK7-FLAG; PTK7) or an empty vector (pcDNA3; Vector) as the control. Twenty-four h later, Western blotting and proliferation assays were performed. Each bar represents the mean ± standard deviation from three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001 vs. vector control. Samples derived from the same experiment and gels/blots were processed in parallel. The blots were cropped to focus upon the specific proteins indicated. Uncropped images of blots are shown in Supplementary Fig. S4 Biphasic regulation of signaling protein phosphorylation by PTK7 expression. We next investigated whether PTK7 expression biphasically regulates the phosphorylation of signaling proteins in ESCC cells. We found that, in accordance with the tumorigenic effects, increasing levels of PTK7 biphasically regulated the tyrosine phosphorylation of cellular proteins as well as the phosphorylation of Src, Akt, and ERK, with maximal phosphorylation achieved with 2 μg of the PTK7 expression vector in PTK7-low TE-5 and TE-14 cells ( Fig. 4a and Supplementary Fig. S2a) or 1 μg of that in PTK7-high TE-10 and TE-6 cells ( Fig. 4b and Supplementary  Fig. S2b).

High PTK7 expression in ESCC is associated with increased patient survival. The associations
between PTK7 mRNA levels and relative risk and survival were assessed from 93 ESCC samples in the database from TCGA; three outliers were removed (see Supplementary Fig. S3) using the LPEseq tool 30 . Our analysis using a Cox proportional hazards model showed that the relative risk in ESCC patients decreased with increasing PTK7 mRNA levels (Fig. 5a, left). As PTK7 mRNA levels are already high in tumor tissues of ESCC patients, we were not able to detect a change in risk with lower levels of expression. However, relative risk was not associated with PTK7 expression in EAC patients (Fig. 5a, right).
To further explore how PTK7 mRNA levels affect ESCC patients, we categorized ESCC samples as either "PTK7 high" (n = 62) or 31 "PTK7 low" (n = 31) using a log-rank test of overall survival. There were no significant associations between relative PTK7 expression and clinical parameters, such as age at diagnosis, sex, pathologic stage, neoplasm histologic grade, and lymph node metastasis (Table 1). However, a Kaplan-Meier survival plot showed that ESCC patients in the PTK7-high group had a significantly longer overall survival than those in the PTK7-low group (P = 0.011) (Fig. 5b). We also compared the relative risk between PTK7-high and PTK7-low groups with regard to the age at diagnosis in a spline survival analysis. As in other cancers, the relative risk increased with an increase in age at diagnosis in both groups of patients with ESCC. Interestingly, the PTK7-high group showed a lower relative risk than the PTK7-low group over the range of ages analyzed (P = 0.024) (Fig. 5c).

Discussion
We analyzed samples comprising a broad range of cancer types in the database from TCGA and found that PTK7 mRNA levels are high in many cancers, including endometrial and esophageal cancers. A further examination in esophageal cancers revealed that PTK7 mRNA level is higher in cancer tissues than that in adjacent normal tissues, and the level is higher in ESCC than that in EAC. Thus, we hypothesized that PTK7 plays a role in tumorigenesis of ESCC.
The oncogenic role of PTK7 has been reported in various cancer cells including ESCC cells. PTK7 knockdown decreased proliferation, survival, wound healing and invasion by inhibiting ERK, JNK, p38, Akt and FAK activation in ESCC TE-10 and TE-11 cells 4 . Overexpression of PTK7 induced cell proliferation and invasion and suppressed apoptosis in ESCC TE-5 and TE-9 cells 24 . Consistently, we found here that overexpression of PTK7 increased proliferation in PTK7-low ESCC TE-5 and TE-14 cells and that PTK7 knockdown decreased it in PTK7-high ESCC TE-6 and TE-10 cells. However, overexpression of PTK7 decreased proliferation in PTK7-high ESCC TE-6 and TE-10 cells, suggesting that the PTK7 expression level does not correlate linearly with the oncogenic potential. To further understand the relationship between the level and the oncogenic potential of PTK7, PTK7 expression level was modulated by transfection of a PTK7 knockdown vector or a PTK7 expression vector. As PTK7 expression levels increased, the proliferation, migration, and invasion of ESCC cells were enhanced. However, when PTK7 expression levels exceeded an optimal level, these characteristics were suppressed. Thus, the oncogenic properties of ESCC cells are biphasically regulated with increasing PTK7 expression levels. In accordance with the oncogenic effects of PTK7 expression, we also demonstrated that tyrosine phosphorylation of cellular proteins and phosphorylation of Src, Akt, and ERK was also biphasically modulated with regard to PTK7 expression in ESCC cells.
Our findings for a biphasic regulation of oncogenicity by PTK7 in ESCC cells correspond with those of our previous study showing a biphasic regulation of angiogenesis by PTK7. In that study, the VEGF-induced phosphorylation of KDR and tube-formation in human umbilical vein endothelial cells (HUVECs) and in vivo angiogenesis initially increased and then decreased with increasing PTK7 expression levels 25 . We showed that PTK7 interacts with KDR but not with FLT-1 in HUVECs, even though KDR and FLT-1 are both VEGF receptors. It appears that when levels of PTK7 are lower than that of KDR, PTK7 helps KDR molecules to oligomerize and activate, whereas high amounts of PTK7 inhibit this. As shown in the analysis of the functional mechanism of PTK7 in angiogenesis, we hypothesized that PTK7 would bind to some RPTKs in ESCC cells. However, KDR, which binds to PTK7 in endothelial cells, is not expressed in ESCC cells (data not shown). Further studies are needed to determine which binding partners such as catalytically active RPTKs are responsible for these effects in ESCC cells.
An analysis of the patient data provided in the database from TCGA revealed that the overall survival of ESCC patients was higher in those with higher PTK7 levels. Of note, the relative expression of PTK7 in ESCC is comparable to that in lung cancer samples (Fig. 1a and c), and PTK7 overexpression was also shown to decrease oncogenic effects in lung squamous cell carcinoma cells 29 . Thus, in some cancer types with comparable, or even higher, PTK7 expression, the suppressive effect of PTK7 on oncogenic characteristics may be apparent.
The inhibition of PTK7 has been suggested as a novel therapeutic target for various cancers such as colorectal cancer and atypical teratoid rhabdoid tumors 31,32 , and some anticancer agents under development involve PTK7-targeted antibody-drug conjugates 33 or PTK7 aptamers 34 . However, caution should be used with this strategy when considering the evidence showing PTK7 can biphasically regulate tumorigenesis, as the inhibition of PTK7 activity or removal of PTK7-positive cells may induce tumor progression. Therefore, PTK7-targeted therapy should be applied to only for cancers with either low PTK7 expression or where the inhibition of PTK7 has been shown to reduce tumorigenicity.
Taken together, our data show that tumorigenesis in ESCC cells initially increases and then decreases with increasing expression levels of PTK7 and that lower PTK7 expression in ESCC tissues is associated with a poorer prognosis. Therefore, a better understanding of the biphasic regulation of tumorigenesis by PTK7 will provide important information for the development of therapeutic means to control ESCC and other cancers that express PTK7.

Materials and Methods
Analysis of public dataset. A preprocessed RNA-seq count matrix and the associated clinical features of 10,048 samples, including 198 esophageal cancer samples were downloaded using recount2 35 from TCGA (https:// cancergenome.nih.gov/). To compare the PTK7 mRNA levels among cancer tissue types, a scaling (library size) normalization was first conducted using LPEseq. 30 . Then log-transformed normalized mRNA levels for the 20 different cancer tissues were used for further analyses. were maintained in Dulbecco's modified Eagle's medium (DMEM; Gibco of Thermo Fisher Scientific) supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin, and 100 μg/ml streptomycin at 37 °C in 5% CO 2 .
Western blotting and antibodies. Western blotting was performed as described previously 25 . The anti-PTK7 antibody was described previously 36 . Antibodies against phospho-ERK, ERK2, and β-actin were purchased from Santa Cruz Biotechnology (Santa Cruz, Dallas, TX, USA). Antibodies against phospho-Src (Tyr416), Src, phospho-Akt (Ser473), and Akt were purchased from Cell Signaling Technology (Danvers, MA, USA). The anti-phospho-tyrosine antibody (clone 4G10) was purchased from Millipore (Billerica, MA, USA). Horseradish peroxidase-conjugated goat anti-mouse IgG and goat anti-rabbit IgG were purchased from KOMA Biotech (Seoul, Korea). Chemotactic migration and invasion assays. The chemotactic migration assay was performed as described previously 36 . Briefly, detached cells (1 × 10 5 cells/0.2 ml) were placed on a Transwell filter of which the bottom surface had been coated with 10 μl of 0.1% gelatin. The lower compartment of the well below the filter was filled with 0.6 ml DMEM with 5% FBS as the chemoattractant. After a 24-h incubation at 37 °C, cells that had migrated to the bottom surface of the filter were fixed with 3.7% formaldehyde for 15 min and stained with 5% crystal violet for 5 min. Invasion assays were identical to the chemotactic migration assay, except that the top surface of the filter was coated with 25 μg of growth factor-reduced Matrigel and dried overnight. After the removal of non-migrated or invaded cells with a cotton swab, the stained cells were solubilized with 1% SDS and the absorbance was measured at 600 nm.

P-value
Statistical analysis. Statistical analyses were performed using R (http://www.r-project.org) and Excel (Microsoft, Redmond, WA, USA) software. For group comparisons, Student's t-tests were performed to analyze the differences in PTK7 mRNA levels between esophageal cancer samples and adjacent normal tissue samples and to assess the proliferation, migration, and invasion of ESCC cells transfected with various amounts of PTK7 expression vectors. A Fisher's exact test was conducted to provide the significance of the association between clinicopathologic features and PTK7 mRNA levels. A Kaplan-Meier plot and its corresponding P value from a log-rank test were used to provide the significance of survival between PTK7-high and -low groups. Cox proportional hazards regressions including nonlinearity of PTK7 expression and age at diagnosis were performed using "survival" and "splines" R packages to estimate the relative risk associated with PTK7 mRNA level and age at diagnosis. For all tests, P values less than 0.05 were considered statistically significant.