Multidrug transporter MRP4/ABCC4 as a key determinant of pancreatic cancer aggressiveness

Abstract

Recent findings show that MRP4 is critical for pancreatic ductal adenocarcinoma (PDAC) cell proliferation. Nevertheless, the significance of MRP4 protein levels and function in PDAC progression is still unclear. The aim of this study was to determine the role of MRP4 in PDAC tumor aggressiveness. Bioinformatic studies revealed that PDAC samples show higher MRP4 transcript levels compared to normal adjacent pancreatic tissue and circulating tumor cells express higher levels of MRP4 than primary tumors. Also, high levels of MRP4 are typical of high-grade PDAC cell lines and associate with an epithelial-mesenchymal phenotype. Moreover, PDAC patients with high levels of MRP4 depict dysregulation of pathways associated with migration, chemotaxis and cell adhesion. Silencing MRP4 in PANC1 cells reduced tumorigenicity and tumor growth and impaired cell migration. Transcriptomic analysis revealed that MRP4 silencing alters PANC1 gene expression, mainly dysregulating pathways related to cell-to-cell interactions and focal adhesion. Contrarily, MRP4 overexpression significantly increased BxPC-3 growth rate, produced a switch in the expression of EMT markers, and enhanced experimental metastatic incidence. Altogether, our results indicate that MRP4 is associated with a more aggressive phenotype in PDAC, boosting pancreatic tumorigenesis and metastatic capacity, which could finally determine a fast tumor progression in PDAC patients.

Figure 1

Bioinformatic analysis of MRP4 differential expression in PDAC samples, normal/adjacent pancreatic tissue, PDAC cells lines and circulating tumor cells from PDAC patients. (a) Differential expression of MRP4 according to the Gene Expression Profiling Interactive Analysis (GEPIA) database using TCGA PAAD tumor data and matched data of normal tissue from TCGA. (b) Comparative analysis of MRP4 mRNA expression from three different gene expression datasets comprising pancreatic normal tissue and their paired PDAC samples (GSE15471 and GSE62452), as well as PDAC cell lines (GSE71729). (c) MRP4, vimentin, E-cadherin, and EGFR transcript levels in a panel of 9 pancreatic cancer cell lines systematically arranged from the less aggressive (Lo-G) to the more aggressive (Hi-G) phenotypes (GSE64557). (d) Inverse correlation between MRP4 and GATA6 expression levels in the 9 pancreatic cancer cell lines from the beforementioned dataset. (e) Heatmap showing differentially expressed genes in MRP4 LE and HE PDAC samples performed with MultiExperiment Viewer (MeV). Differentially expressed genes were identified based on a log2(fold change) ≥ 1 and an FDR ≤ 0.05. Hierarchical clustering was based on Pearson correlation coefficients. (f) Functional enrichment analysis of differentially expressed genes showing pathways associated with MRP4 expression (p-adjusted < 0.05). See Supplementary Table S1. (g) Negative depletion fluorescence activated cell sorting (FACS) was used to enrich for CTC from the blood of 6 patients who underwent surgery for PDAC (GSE18670). Positive 7-amino-actinomycin D viability and negative CD45 and CD34 cells were identified as CTC. We considered mRNA levels of ABCC4 1552918 isoform an indicator of MRP4 expression levels. Statistics: t-test ***p < 0.001 and *p < 0.05.

Figure 2

MRP4 silencing reduces tumor growth and tumorigenicity in PANC1 xenografts. Swiss nude mice were subcutaneously injected either with PANC1-MRP4sh2, PANC1-MRP4sh1 clones or PANC1-scramble as a control. Each group contained 8–11 animals and the data shown represents measurements of one of two experiments. (a; left) Tumors were measured with a caliper three times a week for 32 days post-inoculation and tumor volume was calculated as described in “Materials and Methods”. (a; middle) Representative pictures of the dissected xenografts of each group at the end of the experiment. (a; right) Tumor weight was measured with a scale after dissection. (b) PANC1-scramble, PANC1-MRP4sh2 and PANC1-MRP4sh1 xenografts were processed for histological (H&E staining), MRP4, Ki67, and EGFR immunostaining evaluation. Nuclei were counterstained with hematoxylin. Representative images are shown. Cell number per field, proliferative index, MRP4 and EGFR scores were determined by H&E, Ki67, MRP4, and EGFR staining respectively, and are shown as bar plots (right). Data is shown as mean ± SEM. Statistics: Linear regression and one-way ANOVA followed by Tukey’s t-test. ***p < 0.001 and **p < 0.01.

Figure 3

MRP4 silencing impairs cell migration in PANC1 clones. (a) PANC1-scramble, PANC1-MRP4sh1 and PANC1-MRP4sh2 cell migration was assessed by closure of the wound. The area of the scratch was measured at T:0 h and T:48 h in all cell lines, and % reduction of initial scratch area was compared as described in “Materials and methods”. Data is shown as mean ± SEM of six measurements and the experiment was performed three times. (b) Representative images of a wound at starting point (T: 0 h), and PANC1-scramble, PANC1-MRP4sh1 and PANC1-MRP4sh2 cell migration into the wound at 48 h post scratch are shown. Wound areas are delimited in black. Statistics: One-way ANOVA followed by Tukey’s t-test. **p < 0.01.

Figure 4

MRP4 silencing reprograms gene expression, mainly dysregulating pathways related to cell surface interactions, focal adhesion, and extracellular matrix remodeling. (a) Unsupervised clustering of PANC1 cell lines based on transcriptomic profiles. (b) Box plot of MRP4 (ABCC4) mRNA expression between PANC1 cell lines, showing MRP4 is underexpressed in MRP4sh2 compared to control (PANC1 and PANC1-scramble) cells. (c) Volcano plot representing the significance (− Log10 p-value) and magnitude of transcript change (Log2FC) in scramble and MRP4sh2 cells. The red dots represent significantly upregulated transcripts and the green dots represent downregulated transcripts upon MRP4 silencing (FDR < 0.05; Fold Changes > 2). (d) Functional enrichment analysis were performed using the Enrichr resource (https://amp.pharm.mssm.edu/Enrichr) and InnateDB (https://www.innatedb.com) based on the list of dysregulated transcripts between scramble and MRP4sh2 cells (FDR < 0.05; Fold changes > 2). See Supplementary Table S2. (e) Heatmap showing differentially expressed genes related to the enriched signaling pathways. Hierarchical clustering was based on Pearson correlation coefficients.

Figure 5

MRP4 overexpression increases tumor growth in BxPC-3 xenografts. NSG mice were subcutaneously injected with BxPC-3-MRP4 or BxPC-3-mock as a control. Each group contained 5 animals and the data shows measurements of one of two experiments. (a; left) Tumors were measured with a caliper three times a week for 32 days post-inoculation and tumor volume was calculated as described in “Materials and methods”. (a; middle) Representative pictures of the dissected xenografts of each group at the end of the experiment. (a; right) Tumor weight was measured with a scale after tumor dissection. (b) BxPC-3-mock and BxPC-3-MRP4 xenografts were processed for histological (H&E staining), MRP4, Ki67, and EGFR immunostaining evaluation. Nuclei were counterstained with hematoxylin. Representative images are shown. Cell number per field, proliferative index, MRP4 and EGFR scores were determined by H&E, Ki67, MRP4, and EGFR staining respectively, and are shown as bar plots (right). Data is shown as mean ± SEM. Statistics: Linear regression and Student´s t-test. ***p < 0.001, **p < 0.01, and *p < 0.05.

Figure 6

Effect of MRP4 overexpression upon BxPC-3 experimental metastasis. (a) Vimentin and E-cadherin (CDH1) transcript levels in BxPC-3-mock and BxPC-3-MRP4 cells, determined by qPCR as described in “Materials and methods”. Data is shown as mean ± SD of three measurements and the experiment was performed three times. (b) H&E staining, GFP, Ki67, and EGFR immunostaining of hepatic, renal, and pulmonary metastatic foci from BxPC-3-MRP4 inoculated mice. Nuclei were counterstained with hematoxylin. Representative images are shown. (c) Representative liver macro-metastases at the end of the experiment. (d) Representative comparison of GFP expression in liver homogenates, determined by FAC. (e) Number of hepatic metastatic foci per animal in each experimental group. Data is shown as mean ± SEM. Statistics: Student’s t-test. ***p < 0.001, **p < 0.01, and *p < 0.05.