The Institutional Ethical Review Board of Aichi Medical University Hospital approved this project (2020-H122). The study was performed in compliance with the Helsinki Declaration.

Formalin-fixed paraffin-embedded (FFPE) samples of primary colorectal tumors were obtained as previously described (Inoue et al., 2019; Inoue et al., 2020; Koshino et al., 2021; Nagano-Matsuo et al., 2021). Immunohistochemistry data for PBK, PHH3 and other cellular proliferation markers in 269 CRCs were obtained from our previous reports (Koshino et al., 2021; Nagano-Matsuo et al., 2021). Briefly, FFPE samples of 269 CRCs resected at the Aichi Medical University Hospital from 2009 to 2012 were collected, along with patients’ clinical information (Supplementary Table S1). After surgery, patients were followed for up to 90 months. All CRCs were diagnosed as invasive and naïve to chemotherapy or radiotherapy. A single 4.5-mm core tumor tissue sample derived from FFPE specimens was assembled into multiple blocks containing up to 30 samples. All the cores were taken from invasive areas, and approximately 20% of the cores contained the invasive front.

Immunohistochemistry was performed using a Ventana BenchMark XT Automated ISC/ISH staining instrument (Roche Diagnostics, Basel, Switzerland). Primary antibodies are summarized in Supplementary Table S2. Signals were visualized using 3,3′-diaminobenzidine (DAB). PBK and cytokeratin AE1/AE3 immunoreactive areas were evaluated using ImageJ software (NIH, Bethesda, MD, United States). PBK positivity was determined as follows: the PBK-positive area was divided by cytokeratin AE1/AE3–positive area (%). The number of PHH3-positive cells was counted in a high-power field (×400). Sequential double staining was performed using Leica Bond-Max (Leica Biosystems, Bannockburn, IL, United States), and additional antibody was visualized by HistoGreen.

FHC, a human colon epithelial normal cells, were purchased from the American Type Culture Collection (ATCC, Manassas, VA) and cultured according to the recommended procedure by the supplier. HEK293T and CRC cells (Caco-2, COLO205, SW480, CW-2, LoVo, SW48 and HCT116) were obtained as reported previously (Inaguma et al., 2017; Inoue et al., 2020). CRC cells were maintained in Dulbecco's Modified Eagle’s Medium (DMEM) supplemented with 10% fetal bovine serum (FBS).

The lentivirus vectors for PBK and the control LacZ were constructed using the CSII-EF-MCS plasmid, which was kindly provided by Dr. H. Miyoshi (RIKEN BioResource Center, Tsukuba, Japan). The pcDNA3.1 vector (Invitrogen) was used to express PBK, CDH1^cyt (CDH1 cytoplasmic domain, R733 to D882) or firefly luciferase (Luc2CP). The CDH1^cyt mutants (CDH1^{cyt/S840,846,847A} and CDH1^{cyt/S840,846,847D}) were generated by PCR-based mutagenesis. The pGEX-5X-1 vector was used for the synthesis and purification of glutathione S-transferase (GST)-tagged proteins.

OTS514, a selective PBK inhibitor, was purchased from Selleck Biotech (Tokyo, Japan) and used to treat cells for 24–72 h. Colchicine was obtained from FUJIFILM Wako Pure Chemical Corporation (Osaka, Japan) and added at a concentration of 1 × 10^–7 M for 8 h. MG132, a proteasome inhibitor, was from Selleck Biotech (Tokyo, Japan) and applied at 1 µM for 8 h.

5 × 10⁴ CRC cells were seeded on 6-well plates. After colchicine treatment at 1 µM for 8 h, cells were harvested, washed and fixed by 70% ethanol for 24 h at −20°C. After saining by the Guava® Cell Cycle Reagent (Guava Technologies, Inc., Hayward, CA), cell cycle analyses were performed using Guava® easyCyte™ systems (Guava Technologies, Inc., Hayward, CA) according to the manufacturer’s protocol.

5 × 10³ LoVo and CW-2 cells with or without exogeneous PBK expression were seeded on 12-well plates. After incubation, cell numbers were measured using CellTiter 96® Aqueous One Solution (Promega, Madison, WI, United States) according to the manufacturer’s protocol. To assess the suppressive effects of OTS514, 5 × 10³ CRC cells on 12-well plates were treated by OTS514 as follows: SW48 for 24 h; HCT116 and LoVo for 48 h.

To assess the apoptosis induction by OTS514, 5 × 10⁴ CRC cells on 6-well plates were treated by 20 nM of OTS514 as follows: SW48 for 8 h; HCT116 and LoVo for 48 h. Annexin V assays were performed using the Guava Nexin® Reagent (Guava Technologies, Inc., Hayward, CA) and Guava® easyCyte™ systems according to the manufacturer’s protocol.

Migration and invasion assays were performed using the Falcon® Permeable Support for 24-well Plate with 8.0 µm Transparent PET Membrane and Corning® BioCoat™ Matrigel® Invasion Chambers with 8.0 µm PET Membrane (Corning, NY, United States) according to the manufacturer’s procedure. Cell numbers and incubation times were as follows: migration assays, 2 × 10⁴ cells per chamber and 24 h; invasion assays, 5 × 10⁴ cells per chamber and 48 h. We used 10% FBS as a chemoattractant. After incubation, the migrated or invaded cells from upper chamber to the opposite side of the PET Membrane were fixed using 100% methanol at room temperature and stained with a Giemsa stain. The number of migrated or invaded cells was counted under a microscope.

TOP/FOP reporter vectors along with the control hRluc vector (Promega) were transfected using Lipofectamine 3000 reagent (Invitrogen). Dual luciferase reporter assay was performed using the Dual-Luciferase® Reporter Assay System (Promega) according to the manufacturers’ protocol.

Transfected HEK293T cells were lysed with buffer containing 50 mM Tris, pH 7.5, 5 mM EDTA, 150 mM NaCl, and 0.1% Triton X-100 with protease and phosphatase inhibitor cocktail (Roche, Nutley, NJ, United States). FLAG-tagged proteins were immunoprecipitated using anti-DYKDDDDK tag antibody magnetic beads (FUJIFILM). Whole cell lysates or immunoprecipitated proteins were subjected to immunoblot analyses as previously described (Inaguma et al., 2011; Inaguma et al., 2013). In brief, whole cell lysates were prepared using 1x Sodium Dodecyl Sulfate (SDS) sample buffer, containing 50 mM Tris-HCl and 2% SDS. The SDS polyacrylamide gel electrophoresis was performed using polyacrylamide gel and separated proteins were transferred to a PVDF membrane. Antibody dilutions are summarized in Supplementary Table S2. For sequential detection, antibody stripping buffer (0.1 M Glycine-HCl pH 2.5) was used. Signal intensity was measured by ImageJ software (National Institutes of Health, Bethesda, MD).

Recombinant PBK and HH3.1 proteins were purchased from Carna Biosciences, Inc. (Kobe, Japan) and New England Biolabs Japan Inc. (Tokyo, Japan), respectively. GST-tagged proteins used in the kinase assays were produced in Escherichia coli and purified by the MagneGST™ Protein Purification System (Promega). In vitro kinase reactions were performed in 1x kinase buffer (Cell Signaling) with adenosine triphosphate (ATP) labelled by γ-³²P (PerkinElmer Japan Co., Ltd., Yokohama, Japan) for 90 min at 30°C. After SDS polyacrylamide gel electrophoresis, the signals from phosphorylated proteins were detected by using Fujifilm BAS-5000 (FUJIFILM, Tokyo, Japan).

All statistical analyses were performed with EZR software version 1.41 (Kanda, 2013). The Mann–Whitney test, the Kruskal–Wallis test with a post-hoc test, or Spearman’s rank correlation coefficient was performed to analyze the significance. p < 0.05 indicated statistical significance.

Immunohistochemical staining analyses revealed that PBK expression showed a significant correlation to PHH3 expression in CRC cells (ρ = 0.238, p < 0.0001, Figure 1A). PBK expression was also significantly associated with the expression of other cellular proliferation markers (Supplementary Figure S1). The sequential immunohistochemical staining revealed co-expression of PBK and PHH3 in cells in M phase (Figure 1B and Supplementary Figure S2). All of the cultured CRC cells expressed PBK with correlation to PHH3. FHC, a human colonic epithelial cells, expressed PBK and PHH3 at lower levels than CRC cells (Figure 1C). To test the cell cycle-specific expression of PBK and PHH3, we performed colchicine treatment in actively proliferating CRC cells. Colchicine treatment significantly increased M phase around 2.5 folds (p < 0.01, Table 1). Cell cycle arrest at M phase by colchicine treatment resulted in up-regulation of both PBK and PHH3 (Figures 1D,E).

To uncover the effects of PBK on cellular proliferation, we established CRC cells over-expressing PBK. Forced expression of PBK in LoVo and CW-2 cells significantly enhanced cell proliferation, and these cells also showed PHH3 up-regulation and phospho-ERK accumulation (Figures 2A,B). In contrast, OTS514, a selective PBK inhibitor, significantly suppressed cellular proliferation of CRC cells (Figure 2C). Annexin V assay revealed apoptotic induction in OTS514-treated CRC cells: 1.6 to 2.3 fold increase in early apoptosis (p < 0.01) in SW48, HCT116 and LoVo; 1.9 and 4.2 fold increase in late apoptosis (p < 0.01) in SW48 and HCT116, respectively (Table 2 and Figure 2D). PHH3 was down-regulated by OTS514 treatment. In contrast, cleaved PARP and cleaved caspase 3 were up-regulated with BCL2 down-regulation or BAX up-regulation by OTS514 (Figure 2E).

Transwell migration and invasion assays revealed that PBK significantly suppressed the migration and invasion activity of CRC cells (Figures 3A,B). To uncover the mechanism by which PBK suppressed the migration and invasion of CRC cells, we analyzed the expression of CDH1 and VIM, those are markers for epithelial-mesenchymal transition (EMT). As shown in Figure 3C, PBK up-regulated CDH1 expression without altering CDH1 levels, while no changes in VIM or VIM was observed. These results indicated that PBK suppressed the migration and invasion of CRC cells through other than EMT. In agreement with CDH1 stabilization, TOP/FOP reporter activities, reflecting internal Wnt/β-catenin signaling activity, was significantly reduced in PBK-transfected cells (Figure 3D). The selective PBK inhibitor OTS514 suppressed CDH1 expression with inducing apoptosis (Figure 3E).

PBK was initially identified as a kinase that binds to the PDZ domain of DLG1, a member of the MAGUK protein family (Gaudet et al., 2000), which was reported to interact with CDH1 through APC and regulates epithelial cell polarity (Zhang et al., 2011; Roberts et al., 2012). Therefore, we assessed the possibility that PBK stabilizes CDH1 via phosphorylation of the CDH1 cytoplasmic domain (CDH1^cyt) (McEwen et al., 2014). To test this possibility, an expression vector for CDH1^cyt was co-transfected with vectors encoding Luc2CP and PBK. PBK stabilized CDH1^cyt with accumulation of its phosphorylated form (Figures 4A,B, and Supplementary Figure S3A). The expression level of co-transfected Luc2CP was not affected by PBK, supporting the notion that PBK stabilized CDH1^cyt protein not but activated CMV promoter of the expression vectors. In addition, CDH1^cyt accumulation was not from the down-regulation of proteasome activity (Supplementary Figure S3A).

To assess the direct CDH1^cyt phosphorylation by PBK, in vitro kinase assays were performed using GST-tagged proteins synthetized in E. coli (Supplementary Figure S3B). As shown in Figure 4C left panel and Supplementary Figure S3C, both GST-tagged CDH1^cyt/WILD and CDH1^{cyt/S840,846,847A} were not phosphorylated by recombinant PBK. Furthermore, no band shifts were detected in both CDH1^cyt/WILD and CDH1^{cyt/S840,846,847A} after the kinase reaction (Figure 4C, right panel). These results indicated that PBK accumulates CDH1 with S840, S846, and S847 phosphorylation; however, PBK itself may not phosphorylate CDH1 directly (McEwen et al., 2014). In contrast to CDH1, in vitro kinase assay revealed that recombinant PBK directly phosphorylated recombinant HH3.1 (Figure 4C, left panel and Supplementary Figure S3C) consistent with the previous report (Park et al., 2006).