Fig. 8: FOXM1 induces DKK1 expression independently of Wnt signaling.

A The mRNA levels of DKK1, CTNNB1, and LEF1 in WT S2-CP8 or S2-CP8/ΔFOXM1 BS #1 and #2 cells were measured by quantitative RT-PCR and normalized to GAPDH. The results are shown as fold-changes compared to WT S2-CP8 cells and are expressed as means ± SD from three independent experiments. B HEK293T cells were transfected with the indicated amounts of HA-β-catenin^SA expression vector (middle panel) or FLAG-FOXM1c expression vector (right panel) and the indicated reporter constructs shown in the left panels, and luciferase activities were measured. The results are shown as fold-changes compared to cells without β-catenin^SA or FOXM1c expression. Luciferase activities of cells transfected with the reporter construct containing the full-length DKK1 upstream sequence (FL) are indicated by the solid line and closed circles, those with the mutant DKK1 upstream sequence in which the FOXM1-binding site was deleted (ΔFOXM1 BS) are indicated by the dotted line and open triangle, and those with the mutant DKK1 upstream sequence in which the TCF-binding site was deleted (ΔTCF BS) are indicated by the dotted line and open square. The results are expressed as means ± SD from three independent experiments. C and E Serial sections of DKK1 and FOXM1 double positive PDAC tissues (n = 17) (C) and ESCC tissues (n = 32) (E) were stained with anti-DKK1, anti-FOXM1, or anti-β-catenin antibody and hematoxylin. Black boxes show enlarged images of tumor lesions that are positive for DKK1 and FOXM1 and negative for β-catenin. Numbers of β-catenin positive or negative cases are shown in a pie chart (D and F). β-catenin positivity in DKK1 and FOXM1 double positive PDAC and ESCC tissues is shown in the bar graph (G) and was statistically analyzed using the chi-square test. *P < 0.05; **P < 0.01 (Student’s t test); scale bars, 50 µm (C and E).