Our recent publications showed that multidrug resistance protein 2 (MRP2, encoded by the ABCC2 gene) conferred oxaliplatin resistance in human liver cancer HepG2 cells. However, the contribution of MRP2 to oxaliplatin resistance remains unclear in colorectal and pancreatic cancer lines. We investigated the effects of silencing MRP2 by siRNA on oxaliplatin accumulation and sensitivity in human colorectal cancer Caco-2 cells and pancreatic cancer PANC-1 cells. We characterized the effects of oxaliplatin on MRP2 ATPase activities using membrane vesicles. Over-expression of MRP2 (endogenously in Caco-2 and PANC-1 cells) was associated with decreased oxaliplatin accumulation and cytotoxicity, but those deficits were reversed by inhibition of MRP2 with myricetin or siRNA knockdown. Silencing MRP2 by siRNA increased oxaliplatin-induced apoptotic rate in Caco-2 and PANC-1 cells. Oxaliplatin stimulated MRP2 ATPase activity with a concentration needed to reach 50% of the maximal stimulation (EC₅₀) value of 8.3 ± 0.7 µM and Hill slope 2.7. In conclusion, oxaliplatin is a substrate of MRP2 with possibly two binding sites, and silencing MRP2 increased oxaliplatin accumulation and cytotoxicity in two widely available gastrointestinal tumour lines (PANC-1 and Caco-2).

Gastric cancer metastasis is a major cause of mortality worldwide. Inhibition of RUNX3 in gastric cancer cell lines reduced migration, invasion, and anchorage-independent growth in vitro. Following splenic inoculation, CRISPR-mediated RUNX3-knockout HGC-27 cells show suppression of xenograft growth and liver metastasis. We interrogated the potential of RUNX3 as a metastasis driver in gastric cancer by profiling its target genes. Transcriptomic analysis revealed strong involvement of RUNX3 in the regulation of multiple developmental pathways, consistent with the notion that Runt domain transcription factor (RUNX) family genes are master regulators of development. RUNX3 promoted “cell migration” and “extracellular matrix” programs, which are necessary for metastasis. Of note, we found pro-metastatic genes WNT5A, CD44, and VIM among the top differentially expressed genes in RUNX3 knockout versus control cells. Chromatin immunoprecipitation sequencing and HiChIP analyses revealed that RUNX3 bound to the enhancers and promoters of these genes, suggesting that they are under direct transcriptional control by RUNX3. We show that RUNX3 promoted metastasis in part through its upregulation of WNT5A to promote migration, invasion, and anchorage-independent growth in various malignancies. Our study therefore reveals the RUNX3-WNT5A axis as a key targetable mechanism for gastric cancer metastasis. Significance: Subversion of RUNX3 developmental gene targets to metastasis program indicates the oncogenic nature of inappropriate RUNX3 regulation in gastric cancer.

The excel file shows active RUNX3 binding sites enriched with CDX2, LHX2 and TCF3 motifs and their corresponding GO biological processes at the Pages 1, 2 and 3, respectively.

Higher RUNX3 expression in gastric cancer patients is associated with advanced stage and disease progression A, B, TCGA gastric cancer patient dataset visualized by XENA software (UCSC, n=591). Comparison of the mRNA expression of RUNX3 between normal tissues (n=38) and primary tumors (n=411) (A) as well as between stage Ia (n=17) and stage IV (n=49) (B) are shown; **, P < 0.01 and *, P < 0.05 by a two-tailed Student t test. C, TCGA gastric cancer patient dataset from microarray (cBioPortal) was obtained to determine RUNX3 mRNA expression in patients with disease-free (n=22) or with recurrence and disease progression (n=5); *, P < 0.05 by a two-tailed Student t test. D, comparison of survival data in RUNX3 high (red, n=349) or low (black, n=526) patients samples using microarrays obtained from KM-plotter database; P < 0.001.

RUNX3 drives a metastatic transcriptional program. A, A volcano plot obtained from RNA-seq. A total of 1,005 genes significantly downregulated (P < 0.05) in RUNX3 KO cells were identified, including CD44, Vimentin, IGFBP3, and WNT5A (marked in red). A total of 554 genes significantly upregulated (P < 0.05) in RUNX3 KO cells were identified, including GPNMB (marked in blue). All dots are visualized on the basis of expression alteration by ≥2-fold change by RUNX3 KO. B, Analysis of enriched pathway obtained from MSIgDB Hallmark 2020 and Gene Ontology Biological Process of significantly downregulated genes in RUNX3 KO cells (n = 1,005). C, Heat maps representing read densities of RUNX3, H3K27ac, H3K4me3 (promoter mark), and H3K4me1 (enhancer mark) in ±5-kb regions centered on RUNX3 binding sites. RUNX3 binding sites were classified into 642 sites around TSS and 2,492 distal to TSS (left). Enriched de novo motifs at the 642 RUNX3 binding site around TSS and the 2,492 distal regulatory regions are presented (right). D, Diagrams of H3K27ac HiChIP loops for prediction of RUNX3 target genes. Significant interactions were calculated (P < 0.05) in 10-kb (left) and 25-kb resolution (right). In 10-kb resolution, 2,301 loops between TSS's, 4,768 loops between TSS and distal region and 5,569 loops between distal regions were found using H3K27ac HiChIP data. Of them, 827 loops containing RUNX3 binding sites in distal region were identified. In 25-kb resolution, 6,734 loops between TSS's, 13,092 loops between TSS and distal region, 12,466 loops between distal regions were found using H3K27ac HiChIP data. Of them, 2,357 loops containing RUNX3 binding sites in distal region were identified. E, Rate of differentially expressed genes in among the RUNX3 target genes predicted by H3K27ac HiChIP. A total of 178 and 58 genes were significantly downregulated and upregulated, respectively, by RUNX3 KO. F, Enriched pathways obtained from MSIgDB Hallmark 2020 2 and Gene Ontology Biological Process, analyzing significantly repressed RUNX3 target genes in RUNX3 KO cells.

Higher WNT5A expression in gastric cancer patients is associated with advanced stage TCGA gastric cancer patient dataset was visualized by XENA software (UCSC, n=591). The RNA expression level of WNT5A between stage I (n=25) and stage IV (n=41) are shown; **, P < 0.01 by a two-tailed Student t test.