BEST4 inhibits CRC cell proliferation, clonogenesis, migration, and invasion in vitro. Overexpression of BEST4 decreased proliferation of HCT 116 (A) and Caco2 (B) cells as determined by the IncuCyte confluence assay. (C) Viability of three individual BEST4-deleted clones as determined by a CCK-8 assay. (D) BEST4 deletion significantly enhanced cell viability compared with cells transduced with EV-gRNA control; the reintroduction of BEST4 resulted in suppressed cell growth as determined by IncuCyte confluence analysis. (E) Expression of BEST4-HA in HCT116 and Caco2 cells resulted in more colonies than cells transduced with EV plasmids (left panel); colonies were counted and analysed as shown (right panel). (F) A colony formation assay was performed to evaluate the responses of HCT-15 cells following BEST4 deletion and reintroduction (left panel); cell colonies were counted and analysed as shown (right panel). (G) Heterologous expression of BEST4 in HCT116 cells suppressed invasion and migration (upper panel, scale bar, 200 μm); cells that passed through the membrane were counted and compared (lower panel). (H) BEST4 deletion in HCT-15 cells promoted invasion, whereas reintroduction of BEST4 expression resulted in the inhibition of HCT-15 cell invasion and migration (upper panel, scale bar, 200 μm). Cells that passed through the membrane were counted and compared (lower panel). Data shown are the result of at least three independent experiments, with the mean ± SEM; *P<0.05, **P<0.01 vs EV or WT.

BEST4 deters EMT in CRC in vitro and in vivo. (A) Expression of mRNAs encoding EMT-related genes as determined by qPCR in HCT116 cells transfected with BEST4-HA or EV. (B) As determined by qPCR, the expression of mRNAs encoding EMT markers in BEST4-deleted and BEST4-restored HCT-15 cells. (C) Protein expression of EMT-related genes was evaluated by western blotting of lysates of HCT116 cells transfected with BEST4-HA or EV. (D) Western blotting determined the levels of immunoreactive EMT markers in BEST4-deleted and BEST4-restored HCT-15 cells. (E) Growth of xenograft tumours in BALB/c nude mice resulting from the injection of HCT116 cells stably-transduced with BEST4-HA was significantly diminished compared with those resulting from HCT116 cells transduced with EV alone (eight mice per group). (F) The histogram represents the mean weights of tumours isolated from the HCT116-BEST4-HA and HCT116-EV groups. (G) Tumours from the HCT116-BEST4-HA and HCT116-EV groups were sectioned and subjected to immunohistochemical staining to detect HA and human Ki67; scale bar=50 μm. (H) Mean percentage of Ki67-positive staining in each group. (I) Representative western blot documenting protein levels of EMT markers in individual HCT116-BEST4 and HCT116-EV xenograft tumours. *P<0.05, **P<0.01. Data from at least three independent experiments are presented as the mean ± SEM. (J-L) Intrasplenical injections of HCT-15 cell lines with EV-gRNA control, BEST4-gRNA or BEST4-Rescued (n = 3 per group). After 28 days, the animals were sacrificed and numbers of metastatic nodules in the liver were counted, and the liver tissues were fixed for sectioning and H&E staining; T, Tumour.Data are presented as the mean ± SEM. **P<0.01 vs EV-gRNA.

Molecular regulation of BEST4 expression by Hes4. (A) Endogenous BEST4 mRNA and protein levels in Hes4-overexpressing HCT116 cells were detected by qPCR and western blotting, respectively. (B) Endogenous BEST4 mRNA and protein levels in LS174T cells after shRNA-mediated Hes4 knockdown as detected by qPCR and western blotting, respectively. (C) Detection of BEST4 promoter activity in Hes4-overexpressing HCT116 cells using a dual-luciferase reporter assay. (D) As determined by ChIP-qPCR, Fold-enrichment of the BEST4 promoter region in Flag-ChIP samples from Hes4-overexpressing HCT116 cells. Data were normalized using a fold-enrichment method (i.e., ChIP signals divided by control IgG signals). The DHFR 5’ UTR was used as a negative control (upper panel). The primer sets that target the BEST4 promoter in the ChIP-qPCR assays are as illustrated; +1 marks the transcriptional start site of BEST4, and P1, P2, P3, and P4 represent primer locations (lower panels). (E) Immunofluorescence staining documents the colocalization of BEST4 (green) and Hes4 (red) in HCT116 cells based on detection of their specific tags; scale bar, 20 μM. (F) Co-immunoprecipitation of nuclear extracts with specifically-tagged BEST4 and Hes4 as determined by immunoblotting. Lamin B1 served as a nuclear protein control. (G) Detection of BEST4 promoter activity in Hes4-overexpressing HCT116 cells after BEST4 knockdown through a dual-luciferase reporter assay. Data shown are the result of at least three independent experiments, with the mean ± SEM; *P<0.05, **P<0.01 vs EV

BEST4 relays the Hes4 signal and downregulates Twist1 expression to counteract EMT induction in CRC (A) Detection of Twist1 promoter activity in Hes4-overexpressing or BEST4-overexpressing HCT116 cells through a dual-luciferase reporter assay. (B) Detection of Twist1 promoter activity in Hes4-overexpressing HCT116 cells after BEST4 knockdown through a dual-luciferase reporter assay. (C) Detection of Twist1 promoter activity in HCT116 cells after shRNA-mediated Hes4 knockdown and combination with BEST4-overexpressing using a dual-luciferase reporter assay. (D) Co-immunoprecipitation of nuclear extracts with specifically-tagged BEST4 and Twist1 as determined by immunoblotting. Lamin B1 served as a nuclear protein control. (E) Co-immunoprecipitation of nuclear extracts with specifically-tagged Hes4 and Twist1 as determined by immunoblotting. Lamin B1 served as a nuclear protein control. (F) Expression of EMT markers in Hes4-overexpressing HCT116 cells. (G) Expression of EMT markers in LS174T cells after shRNA-mediated Hes4 knockdown. (H) Expression of EMT markers in Hes4-overexpressing HCT116 cells after BEST4 knockdown. (I) Expression levels of EMT markers in BEST4-overexpressing LS174T cells after Hes4-knockdown; SCR, scrambled siRNA.

Correlation of the BEST4 mRNA expression with clinicopathology and outcomes of patients diagnosed with CRC. (A) Heatmap based on the top 100 differentially expressed genes (downregulated and upregulated) in CRCs compared with ANTs. (B) Volcano plots of RNA seq results showing all differentially expressed genes in CRC tissues compared with ANTs based on the fold-change (|FC|≥L1) and P-value (P≤0.05). (C) Expression levels of BEST4 mRNA in 50 pairs of adenoma tissues compared with ANTs. (D) Expression levels of BEST4 mRNA in 124 pairs of CRC tissues compared with ANTs. (E) BEST4 protein levels in CRC and their matched counterparts were measured using western blotting. (F) Kaplan-Meier survival analysis according to BEST4 expression in 124 patients with CRC. The difference was statistically significant based on the log-rank test (P<0.01). (G) A working model delineates the mechanism by which BEST4 is transcriptionally activated by Hes4 and suppressed EMT through Twist1 inhibition: a. the expression of BEST4 is regulated by interactions of its upstream promoter region with Hes4. BEST4 positively promotes the transcriptional activity of Hes4 on BEST4 by physically binding to Hes4 in a positive feedback loop; b. The complex formed by BEST4 and Hes4 interacts with Twist1 and inhibits the promoter activity of Twist1. Hes4 alone is insufficient to inhibit Twist1 promoter activity, and BEST4 is required. Inhibition of Twist1 promoter activity by Hes4 is BEST4-dependent c. The complex formed by BEST4 and Hes4 destabilized Twist1; the same effect caused by Hes4 required BEST4 intermediation. The Hes4-BEST4-Twist1 axis counteracts EMT induction. *P<0.05, **P<0.01; CRC, colorectal cancer; ANT, adjacent normal tissue; T, tumour.

Correlation of BEST4 mRNA level with clinicopathological factors.

Univariate and Multivariate analysis of prognostic factors for overall survival of patients with CRC.

Construction of BEST4-overexpressing /BEST4-depetion CRC cell lines, and cell proliferation measured by CCK-8 assay. (A) HA protein expression in BEST4-HA-transfected GFP+ HCT116 individual clonal lines. (B) HA protein expression in BEST4-HA-transfected GFP+ Caco2 individual clonal lines. (C) BEST4 mRNA and HA protein expression was detected by real-time PCR and western blotting. (D) Cell viability in BEST4-HA- and EV-transfected HCT116 cells was measured by the CCK-8 assay. (E) Cell viability in BEST4-HA- and EV-transfected Caco2 cells was measured by the CCK-8 assay. (F) DNA sequencing results revealed an excision of a 44-bp genomic fragment in exon 3 of the BEST4-knockout lines. (G) DNA fragmentation from each sample was analysed by agarose gel electrophoresis. (H) The protein level of BEST4 was determined by western blotting. (H) Cell proliferation measured by CCK-8 assay showed that knockout of BEST4 obviously promoted cell growth; however, rescue of BEST4 expression in the knockout cells significantly abrogated the promotive effect of BEST4 knockout on cell growth. *P<0.05, **P<0.01. All the data come from three independent experiments and are presented as the mean ± standard deviation.

Western blot signals were quantified using ImageJ software. (A-D) Each band’s mean pixel intensities were determined and normalized to mean β-actin pixel intensity. *P<0.05, **P<0.01. Data from at least three independent experiments are presented as the mean ± SEM.

BEST4 interacted with Hes4. (A) Hes4 mRNA level in each CRC cell line. (B) Western blot for BEST4 expression of Figure 3A is quantified by ImageJ, each band’s mean pixel intensities were determined and normalized to mean β-actin pixel intensity. (C) Hes4 mRNA and protein levels in HCT116 cells with Hes4 overexpression by Flag-Hes4 transfection were examined by real-time PCR and western blot. (D) Western blot for BEST4 expression of Figure 3B is quantified by ImageJ, each band’s mean pixel intensities were determined and normalized to mean β-actin pixel intensity. (E) Hes4 mRNA and protein levels in LS174T cells with Hes4 knockdown by shRNA transfection were examined by real-time PCR and western blot. (F) No significant change of Hes4 mRNA levels in BEST4-overexpression HCT116 cells as detected by real-time PCR. *P<0.05, **P<0.01. All the data come from three independent experiments and are presented as the mean ± SEM. SRC, scramble siRNA.

BEST4 interacted with Hes4 to inhibit Twist1. (A) BEST4 mRNA levels in HCT116 and LS174T cells transfected with BEST4-specific siRNAs as detected by real-time PCR. (B) Co-immunoprecipitation of nuclear extracts with specifically-tagged BEST4 and Twist1 as determined by immunoblotting, Lamin B1 served as a nuclear protein control. (C) Co-immunoprecipitation of nuclear extracts with specifically-tagged Hes4 and Twist1 as determined by immunoblotting, Lamin B1 served as a nuclear protein control. (D) Vimentin mRNA levels in Hes4-overexpression HCT 116 cell lines as detected by real-time PCR. (E) Vimentin mRNA levels in Hes4-overexpression HCT 116 cell lines after BEST4 knockdown as detected by real-time PCR. *P<0.05, **P<0.01. All the data were generated from three independent experiments and are presented as the mean ± SEM. SRC, scramble siRNA.

Western blot signals were quantified using ImageJ software. (A-D) Each band’s mean pixel intensities were determined and normalized to mean β-actin pixel intensity. *P<0.05, **P<0.01. Data from at least three independent experiments are presented as the mean ± SEM.

BEST4 suppresses EMT independently of its channel functions. (A-C) Incubation of HCT116 cell lines expressed BEST4 or EV with DIDS (300μM), a gradient increased CaCCinh-A01 and ionomycin (1μM), respectively, overnight. The levels of chloride efflux were determined by calculating relative fold to the EV control. Data are shown as means ± SEM, p < 0.05 vs EV or DIDS, or CaCCinh-A01 or ionomycin, with 3 independent experiments. (D-F) Following the same treatment, cell lysates were prepared for immunoblotting with antibodies to BEST4, ZO-1, E-cadherin, Twist1 and β-actin. Representative 3 independent experiments. (G) Western blot for BEST4 expression of Figure 5E is quantified by ImageJ, each band’s mean pixel intensities were determined and normalized to mean β-actin pixel intensity. *P<0.05, **P<0.01. Data from at least three independent experiments are presented as the mean ± SEM

BEST4 mRNA expression in colorectal tumours and matched non-tumour tissues and the clinical relationship between BEST4, Hes4 and EMT markers. (A) Correlation between BEST4 and CDH1 in normal and colorectal cancer samples analysed from the TCGA and GTEx databases using the GEPIA tool (http://gepia.cancer-pku.cn/). (B) RNA-seq data revealed decreased expression of and Hes4 in CRC tissues compared with ANTs. (C) Heatmap of Hes4 expression in CRC and ANTs. (D) The correlation between BEST4 and Hes4, in CRC analysed from the TCGA and GTEx databases using GEPIA (http://gepia.cancer-pku.cn/). (E) Correlation between BEST4 and CDH1 in normal and colorectal cancer samples analysed from the TCGA and GTEx databases using the GEPIA tool (http://gepia.cancer-pku.cn/). (F) Correlation between BEST4 and vimentin in normal and colorectal cancer samples. (G) Correlation between BEST4 and Twist1 in normal and cancer colorectal samples. ANT, adjacent normal tissues; COAD, colon adenocarcinoma; READ, rectum adenocarcinoma; other tumour abbreviations are available at GEPIA: http://gepia.cancer-pku.cn/.

The target sequence of shRNAs, siRNAs and primers used in this study.