Gene Expression in Barrett's Esophagus Cell Lines Resemble Esophageal Squamous Cell Carcinoma Instead of Esophageal Adenocarcinoma.

Abstract

Simple SummaryStable cell lines derived from primary tissues and tumors are widely used in medical research. This study presents interesting findings from evaluation of publicly available gene expression profiles (GEPs) of primary tissues derived from the normal esophagus, Barrett’s esophagus (BE), esophageal adenocarcinoma (EAC), esophageal squamous cell carcinoma (ESCC), as well as available esophageal cell lines. We observed that the GEPs of currently available BE cell lines deviate from the GEPs of primary BE tissues (columnar) and EAC tumors (glandular), and are unexpectedly similar to the GEPs of normal esophageal mucosa and ESCC tumors. In vitro exposure to an acid and bile environment was not sufficient to reverse this “squamous like” GEP adopted by a BE cell line, BAR-T. This incomprehensible change in the GEP may result in ambiguous changes in the phenotype of the BE cell lines, and needs careful consideration during experimental design.Esophageal adenocarcinoma (EAC) is strongly associated with Barrett’s esophagus (BE), a pre-malignant condition resulting from gastric reflux. Esophageal squamous cell carcinoma (ESCC), the other major subtype of esophageal cancer, shows strong association with smoking and alcohol intake and no association with gastric reflux. In this study, we constructed and validated gene expression signatures of EAC vs. ESCC tumors using publicly available datasets, and subsequently assessed the enrichment levels of these signatures in commonly used EAC and ESCC cell lines, normal esophageal tissues and normal esophageal cell lines, and primary BE tissues and BE cell lines. We found that unlike ESCC cell lines which were quite similar to primary ESCC tumors, EAC cell lines were considerably different from primary EAC tumors but still more similar to EAC tumors than ESCC tumors, as the genes up in EAC vs. ESCC (EAChi) had considerably lower expression in EAC cell lines than EAC tumors. However, more surprisingly, unlike various normal cell lines (EPC2, Het-1A) which were very similar to various tissues from normal esophagus, BE cell lines (BAR-T, CP-A) were extremely different from primary BE tissues, as BE cell lines had substantially lower levels of EAChi and substantially higher levels of ESCChi gene expression. This ESCC-like profile of the BAR-T remained unaltered even after prolonged exposure to an acidic bile mixture in vitro resulting in malignant transformation of this cell line. However, primary BE tissues had EAC-like gene expression profiles as expected. Only one EAC case from the Cancer Genome Atlas resembled BE cell lines, and while it had the clinical profile and some mutational features of EAC, it had some mutational features, the copy number alteration profile, and the gene expression profile of ESCC instead. These incomprehensible changes in gene expression patterns may result in ambiguous changes in the phenotype and warrants careful evaluation to inform selection of appropriate in vitro tools for future studies on esophageal adenocarcinoma.