Hypoxia-induced MIR31HG expression promotes partial EMT and basal-like phenotype in pancreatic ductal adenocarcinoma based on data mining and experimental analyses

Abstract

Background: Pancreatic ductal adenocarcinoma (PDAC) is the most common and aggressive type of pancreatic cancer, with a five-year survival rate below 8%. Its high mortality is largely due to late diagnosis, metastatic potential, and resistance to therapy. Epithelial-mesenchymal transition (EMT) plays a key role in metastasis, enabling cancer cells to become mobile. Partial EMT, where cells maintain both epithelial and mesenchymal traits, is more frequent in tumors than complete EMT and contributes to cancer progression. The long non-coding RNA MIR31 host gene (MIR31HG) has recently emerged as a critical factor in PDAC oncogenesis. This study aimed to investigate MIR31HG's role in partial EMT and its association with the basal-like PDAC subtype.

Methods: We analyzed the relationship between MIR31HG expression, partial EMT, and the basal-like subtype of PDAC by integrating data from public databases. We reanalyzed public data from PDAC patient-derived organoids to assess MIR31HG expression and gene signatures under hypoxic and normoxic conditions. RNA sequencing and bioinformatics analyses, including gene set enrichment analysis (GSEA), were used to investigate differentially expressed genes and pathway enrichments. EMT, partial EMT, and hypoxia scores were calculated based on the expression levels of specific gene sets.

Results: We observed that MIR31HG overexpression strongly correlates with higher partial EMT scores and the stabilization of the epithelial phenotype in PDAC. MIR31HG is highly expressed in the basal-like subtype of PDAC, which exhibits partial EMT traits. Hypoxia, a hallmark of basal-like PDAC, was shown to significantly induce MIR31HG expression, thereby promoting the basal-like phenotype and partial EMT. In patient-derived organoids, hypoxic conditions increased MIR31HG expression and enhanced basal-like and partial EMT gene signatures, while normoxia reduced these expressions. These findings suggest that hypoxia-induced MIR31HG expression plays a crucial role in driving the aggressive basal-like subtype of PDAC.

Conclusions: Our results indicate that MIR31HG is crucial in regulating PDAC progression, particularly in the aggressive basal-like subtype associated with hypoxia and partial EMT. Targeting the MIR31HG-mediated network may offer a novel therapeutic approach to combat hypoxia-driven PDAC.

Keywords: Epithelial-mesenchymal transition; Hypoxia; Long non-coding RNA; MIR31HG; Pancreatic cancer.

Fig. 1

The correlation between MIR31HG and partial EMT in PDAC patients. (A) Data from PDAC patients (TCGA-PAAD dataset; n = 177) were obtained from the cBioPortal website (https://www.cbioportal.org/). EMT and partial EMT (pEMT) scores were calculated as described in the Materials and Methods section, and their relationship to MIR31HG gene expression was analyzed using Pearson’s correlation. (B) GSEA was performed to compare the gene sets (from left to right: Hallmark_EMT, KS_Epithelial, KS_Mesenchymal, and Partial_EMT) with MIR31HG expression in PDAC patients. NES, normalized enrichment score

Fig. 2

The correlation between MIR31HG and partial EMT in PANC-1 stable clones. (A) PANC-1 stable clones with minimal and robust MIR31HG overexpression (MIR31HG-2X and − 5X, respectively) were established. The relative MIR31HG expression was quantified by real-time qPCR. (B) Total RNA from PANC-1-Vector, PANC-1-MIR31HG-2X, and PANC-1-MIR31HG-5X cells was subjected to RNA-Seq analysis. The heat map shows the relative gene expressions for partial EMT (pEMT), epithelial, and mesenchymal markers. The minimum (blue) and maximum (red) expression values for each gene were used to map the values to colors. (C) GSEA was performed to compare the gene sets (from left to right: Hallmark_EMT, KS_Epithelial, KS_Mesenchymal, and Partial_EMT) with MIR31HG expression in PANC-1 stable clones. NES, normalized enrichment score

Fig. 3

The correlation between partial EMT and PDAC subtypes. (A) Data from PDAC patients (TCGA-PAAD dataset; n = 177) were obtained from the cBioPortal website (https://www.cbioportal.org/). Patients were classified into classical (n = 142) and basal-like (n = 35) subtypes to compare their overall survivals. (B) GSEA was performed to compare the gene sets (from left to right: Hallmark_EMT, KS_Epithelial, KS_Mesenchymal, and Partial_EMT) in PDAC subtypes. NES, normalized enrichment score

Fig. 4

The correlation between MIR31HG and PDAC subtypes. (A) Data from PDAC patients (TCGA-PAAD dataset; n = 177) were obtained from the cBioPortal website (https://www.cbioportal.org/). The basal-like probability was calculated using the PurIST classifier and plotted against MIR31HG expression. Their relationship was analyzed using Pearson’s correlation. (B) PDAC patients were classified into classical (n = 142) and basal-like (n = 35) subtypes to compare their MIR31HG expressions. (C, D) GSEA was performed to compare the PDAC subtype gene signatures (left: classical; right: basal-like) with MIR31HG expression in PDAC patients (C) and in PANC-1 stable clones (D). NES, normalized enrichment score

Fig. 5

The association of hypoxia with MIR31HG and PDAC subtypes. (A) The overlapping cancer hallmarks enriched in MIR31HG-high vs. MIR31HG-low group and basal-like vs. classical group in PDAC patients. (B) The enrichment plot for relationship between the hypoxia cancer hallmark and MIR31HG expression in PANC-1 stable clones. NES, normalized enrichment score. (C,D) Data from PDAC patients (TCGA-PAAD dataset; n = 177) were obtained from the cBioPortal website (https://www.cbioportal.org/). The correlation between hypoxia scores and MIR31HG expression was plotted in C, and their relationship was calculated using Pearson’s correlation. The hypoxia scores in classical and basal-like subtypes were plotted in D

Fig. 6

The role of hypoxia in MIR31HG expression, partial EMT and PDAC subtypes. (A) PDAC patient-derived organoid data were obtained from the NCBI GEO database (GSE240649). Organoids were established under normoxia (Normo-PDO) and hypoxia (Hypo-PDO) conditions, then switched to hypoxia and normoxia, respectively. The expression levels of MIR31HG and selected HIF-1α-target genes in each condition were plotted. (B-D) GSEA was performed to compare the enrichment of classical (left), basal-like (middle), and partial EMT (right) gene signatures in the Hypo-PDO vs. Normo-PDO group (B), hypoxia vs. normoxia in Normo-PDO (C), and normoxia vs. hypoxia in Hypo-PDO (D). NES, normalized enrichment score