miRNA-221 and miRNA-483-3p Dysregulation in Esophageal Adenocarcinoma

Abstract

Alterations in microRNA (miRNA) expression have been reported in different cancers. We assessed the expression of 754 oncology-related miRNAs in esophageal adenocarcinoma (EAC) samples and evaluated their correlations with clinical parameters. We found that miR-221 and 483-3p were consistently upregulated in EAC patients vs. controls (Wilcoxon signed-rank test: miR-221 p < 0.0001; miR-483-3p p < 0.0001). Kaplan-Meier analysis showed worse cancer-related survival among all EAC patients expressing high miR-221 or miR-483-3p levels (log-rank p = 0.0025 and p = 0.0235, respectively). Higher miR-221 or miR-483-3p levels also correlated with advanced tumor stages (Mann-Whitney p = 0.0195 and p = 0.0085, respectively), and overexpression of miR-221 was associated with worse survival in low-risk EAC patients. Moreover, a significantly worse outcome was associated with the combined overexpression of miR-221 and miR-483-3p (log-rank p = 0.0410). To identify target genes affected by miRNA overexpression, we transfected the corresponding mimic RNA (miRVANA) for either miR-221 or miR-483-3p in a well-characterized esophageal adenocarcinoma cell line (OE19) and performed RNA-seq analysis. In the miRNA-overexpressing cells, we discovered a convergent dysregulation of genes linked to apoptosis, ATP synthesis, angiogenesis, and cancer progression, including a long non-coding RNA associated with oncogenesis, i.e., MALAT1. In conclusion, dysregulated miRNA expression, especially overexpression of miR-221 and 483-3p, was found in EAC samples. These alterations were connected with a lower cancer-specific patient survival, suggesting that these miRNAs could be useful for patient stratification and prognosis.

Keywords: esophageal adenocarcinoma; miR–221; miR–483–3p; microRNA.

Figure 1

Discovery of deregulated miRNAs in EAC. (A) Heatmap target centric, showing the differential expression of hsa (Homo sapiens) miR−221 and miR−483−3p in EAC samples and control samples (pool CTR). For each target, the middle expression level is set as the mean of all of the ΔCT values from all samples for that assay. Data can only be compared across a particular row. The color and intensity of the boxes are used to represent changes (not absolute values). Red indicates upregulated with a ΔCT value below the middle level (thus a negative value compared to the middle expression level), and green indicates downregulated with a ΔCT value above the middle level (thus a positive value compared to the middle expression level). Scale bar represents ΔCT values. (B) Expression of miR–221 in different human tissues, image derived from the Genotype−Tissue Expression (GTEx) portal.

Figure 2

miR–221 and miR–483–3p were significantly upregulated in the EAC replication group. (A) miR–221 and miR–483–3p expression levels in a cohort of 124 EAC patients (All EACs). The values are expressed as the fold increase compared to control FFPE–derived healthy gastric tissue samples (green baseline) (Wilcoxon signed–rank test, p < 0.0001). (B,C) miR–221 expression levels and correlation with clinical outcomes. Kaplan–Meier curves show the cancer–specific survival for EAC groups stratified by miR–221 expression levels. (B) All EAC cases (log–rank p = 0.0025). (C) EACSGE lower–risk subgroup (log–rank p = 0.0065). Blue line: samples with low expression of miR–221; red line: samples with high expression of miR–221. (D) Correlation between miR–221 expression and recurrence in All EACs (Mann–Whitney p = 0.0002) in the lower–risk EACSGE subgroup (Mann–Whitney p = 0.0005). (E) Correlation between miR–221 expression and TNM stages (Mann–Whitney p = 0.0195 stage 1 versus stage 2–3–4). (F) miR–483–3p expression levels and correlation with clinical outcomes. Kaplan–Meier curves show cancer–specific survival for EAC groups stratified based on miR–483–3p expression levels in All EAC patients (log–rank p = 0.0235). (G) Recurrence in All EACs (Mann–Whitney p = 0.0173). (H) Correlation between miR–483–3p and TNM stages (Mann–Whitney p = 0.0085 stage 1 versus stage 2–3–4). (I,J) Combined overexpression of miR–221 and miR–483–3p and correlation with survival. Kaplan–Meier curves for patients overexpressing both miRNAs versus patients not overexpressing both miRNAs showing cancer–specific survival in (I) All EAC cases (log–rank p = 0.0410) and (J) lower–risk EACSGE (log–rank p = 0.0340). (K) Predictive algorithm of cancer–specific survival. By using the decision tree method, a predictive algorithm of cancer–specific survival was developed. * = p ≤ 0.05; ** = p ≤ 0.01; *** = p ≤ 0.001.

Figure 3

Differentially expressed genes (DEGs) in transfected OE19 cells. (A) OE19 cells were transfected with either miR–221 or miR–483–3p mimic and a scramble negative control (CTR). The transfection efficiency was evaluated by RT–qPCR for miR–221 and miR–483–3p (ANOVA test p < 0.0001 and p < 0.0001, respectively). After transfection, RNA–seq analysis of DEGs was performed vs. cells transfected with negative control. Biological processes of genes (GO–Slim Biological processes) differentially expressed in EAC, as identified by PANTHER Functional Classification analysis, are reported for (B) mir–221 and for (C) mir–483–3p. (D) Real–time qRT–PCR data for MALAT1 expression. Data from transfected cells (overexpressing either miR–221 (pink circle) or miR–483–3p (blue triangle) were compared vs. cells transfected with a negative control CTR (black circle) and with untransfected cells (NT, green square) (normalization was performed on a commercial pool of esophageal control tissues); human beta–actin was used as an endogenous control gene (ANOVA test; p = 0.007 and p = 0.0469).* p < 0.05; ** p < 0.01; *** p < 0.0001.