Identification of AGR2 Gene-Specific Expression Patterns Associated with Epithelial-Mesenchymal Transition

Abstract

The TGF-β signaling pathway is involved in numerous cellular processes, and its deregulation may result in cancer development. One of the key processes in tumor progression and metastasis is epithelial to mesenchymal transition (EMT), in which TGF-β signaling plays important roles. Recently, AGR2 was identified as a crucial component of the cellular machinery responsible for maintaining the epithelial phenotype, thereby interfering with the induction of mesenchymal phenotype cells by TGF-β effects in cancer. Here, we performed transcriptomic profiling of A549 lung cancer cells with CRISPR-Cas9 mediated AGR2 knockout with and without TGF-β treatment. We identified significant changes in transcripts associated with focal adhesion and eicosanoid production, in particular arachidonic acid metabolism. Changes in transcripts associated with the focal adhesion pathway were validated by RT-qPCR of COL4A1, COL4A2, FLNA, VAV3, VEGFA, and VINC mRNAs. In addition, immunofluorescence showed the formation of stress fibers and vinculin foci in cells without AGR2 and in response to TGF-β treatment, with synergistic effects observed. These findings imply that both AGR2 downregulation and TGF-β have a role in focal adhesion formation and cancer cell migration and invasion. Transcripts associated with arachidonic acid metabolism were downregulated after both AGR2 knockout and TGF-β treatment and were validated by RT-qPCR of GPX2, PTGS2, and PLA2G4A. Since PGE₂ is a product of arachidonic acid metabolism, its lowered concentration in media from AGR2-knockout cells was confirmed by ELISA. Together, our results demonstrate that AGR2 downregulation and TGF-β have an essential role in focal adhesion formation; moreover, we have identified AGR2 as an important component of the arachidonic acid metabolic pathway.

Keywords: AGR2; EMT; RNAseq; TGF-β; arachidonic acid; focal adhesion.

Figure 1

Evaluation of gene expression changes with respect to particular biological processes using the Protein Analysis Through Evolutionary Relationships database (PANTHER visualization tools). (A) Venn diagrams for overlapping genes with significantly decreased (left) or increased (right) expression in comparison with A549 scr cells. (B,C) GO enrichment analysis of biological processes (red graphs) and cellular processes (green graphs) associated with downregulated (left) and upregulated (right) genes common to all three categories.

Figure 2

Schematic representation of samples subjected to RNAseq and the data analysis workflow using DAVID Functional Annotation Clustering tool at the DAVID 2021 Knowledgebase. (A) DAVID clustering of transcripts with significant FC in scr TGF-β samples vs scr control samples. (B) DAVID clustering of significant transcripts from KOAGR2 samples vs scr samples. (C) DAVID clustering of significant transcripts common between KOAGR2 and scr TGF-β. (D) DAVID clustering of significant transcripts from KOARG2 TGF-β vs scr. (E) DAVID clustering of significant transcripts common for KOAGR2, scr TGF-β, and KOAGR2 TGF-β. FC, fold change.

Figure 3

(A) Representative immunofluorescence of F-actin stress fibres (red) and vinculin foci (green). Nuclei are stained in blue by Hoechst. Slides were captured at 100× magnification and the scale bar represents 10 μm. Graphs show quantification of Vinculin foci either as average count per cell or average area taken up by foci per cell in pixels squared. Quantification was performed according to Horzum et al. [56]. (B) Analysis of invasion potential of A549 using transwell assay. (C) Analysis of migratory properties by wound healing assay using IncuCyte. The migratory rate is expressed as the relative wound density in % and is the highest in cells with AGR2 knockout. * p < 0.05; **** p < 0.0001; ns (non-significant).

Figure 4

Validation of mRNA levels of representative genes by RT-qPCR for (A) focal adhesion pathway and (B) arachidonic acid metabolism pathway. Graphs show gene expression normalized to GAPDH as an endogenous control. In parallel, another two endogenous controls, 18S rRNA and HPRT1, were used with similar outputs. * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001; ns (non-significant). (C) Representative immunochemical analysis of COL4A1, VAV3, PTGS2, and AGR2. Beta-actin served as loading control.

Figure 5

(A) ELISA determination of PGE₂ released from A549 cells into culture media. (B) Representative immunochemical analysis of COX-2 and AGR2. Beta-actin served as a loading control and for normalization to determine fold changes calculated by densitometric analysis in cells exposed to prostaglandin E₂ (PGE₂), diclofenac (DIC), celecoxib (CXB), and acetylsalicylic acid (ASA) compared to untreated cells (CTR). Significant changes (p < 0.05) determined from three independent biological experiments are indicated by asterisk.