Activation of GATA binding protein 6 (GATA6) sustains oncogenic lineage-survival in esophageal adenocarcinoma

Abstract

Gene amplification is a tumor-specific event during malignant transformation. Recent studies have proposed a lineage-dependency (addiction) model of human cancer whereby amplification of certain lineage transcription factors predisposes a survival mechanism in tumor cells. These tumor cells are derived from tissues where the lineage factors play essential developmental and maintenance roles. Here, we show that recurrent amplification at 18q11.2 occurs in 21% of esophageal adenocarcinomas (EAC). Utilization of an integrative genomic strategy reveals a single gene, the embryonic endoderm transcription factor GATA6, as the selected target of the amplification. Overexpression of GATA6 is found in EACs that contain gene amplification. We find that EAC patients whose tumors carry GATA6 amplification have a poorer survival. We show that ectopic expression of GATA6, together with FGFR2 isoform IIIb, increases anchorage-independent growth in immortalized Barrett's esophageal cells. Conversely, siRNA-mediated silencing of GATA6 significantly reduces both cell proliferation and anchorage-independent growth in EAC cells. We further demonstrate that induction of apoptotic/anoikis pathways is triggered upon silencing of GATA6 in EAC cells but not in esophageal squamous cells. We show that activation of p38α signaling and up-regulation of TNF-related apoptosis-inducing ligand are detected in apoptotic EAC cells upon GATA6 deprivation. We conclude that selective gene amplification of GATA6 during EAC development sustains oncogenic lineage-survival of esophageal adenocarcinoma.

Fig. 1.

Integrative genomic analysis of the recurrent amplification at chromosomal 18q11.2. (A) Array-CGH analyses of two representative EACs. A confined region with DNA copy number increase at 18q11.2 is identified. Yellow line highlights the core amplified-domain. (B) qPCR analyses of five genes spanning 1.5 Mb of the 18q11.2 amplicon in 85 EACs. The y axis shows an algorithm of 2^-ΔΔCt indicating the fold-change of a 2N genome and the x axis lists the tumor ID, of which sample 1 is a mean normal value. Numbers in parentheses represent amplification percentiles of the genes examined in 85 EACs. Yellow line highlights the cutoff value. All qPCR reactions were repeated in triplicate. (C) Boxplot of the qPCR data. GATA6 demonstrates higher interquartile range, a larger upper whisker and more extreme upper-outliers than other genes within the amplicon and shows significant difference from the other four genes (**P < 0.01, ***P < 0.001; two-tailed, paired t test). (D) Recurrent amplifications of chromosome 18q11.2 in 73 EACs from SNP array data visualized in hg18 genome build using the IGV software. The y axis shows a descending log₂ copy number ratio in 73 EACs. Horizontal bars represent individual tumor samples. The boxed area shows a 4-Mb region in the vicinity of the 18q11.2 amplicon with the arrow indicating the location of GATA6. (E) Magnified view of the 4-Mb region of the 18q11.2 amplicon from D. Boxed region with yellow lines shows the smallest amplified unit defined in 73 EACs. (F) Kaplan–Meier survival plots estimate a poorer clinical outcome (P = 0.0292) in EAC patients bearing the 18q11.2 amplicon in their tumors.

Fig. 2.

Overexpression of GATA6 driven by gene amplification in EACs. (A) qRT-PCR analysis of 30 EACs. The y axis shows fold-changes (2^-ΔΔCt) of gene expression relative to the normal intestinal tissue (IntN) as GATA6 expression was found to be extremely low or absent in esophageal squamous epithelia (e.g., 43N, and N27 in C). Overexpression of GATA6 was detected in 13 of 14 EACs containing the GATA6 amplicon. GATA6 up-regulation was also observed in a subset of dysplastic Barrett's samples (e.g., 19B). All qRT-PCR reactions were repeated three times. (B) Boxplot analysis of the qRT-PCR data. The y axis represents fold-changes in gene expression relative to the expression of normal intestinal RNA (***P < 0.001, Student's t test). (C) Western blot analysis. Only a small set of primary tissues were examined because of sample availability. Overexpression of GATA6 is shown in a GATA6-amplified EAC (T27) but not in EACs without the GATA6 amplicon (T34 and T78). Samples T27 (EAC), G27 (normal gastric), and N27 (normal esophageal squamous mucosa) were derived from the same patient. (D) Immunohistochemistry of GATA6 and MIB1 in EAC TMAs. Overexpression of GATA6 protein was detected in EACs with amplified GATA6 (representative T27, T70, and T83) compared with EAC without GATA6 amplification (T9) (Magnification ×10). MIB1 expression was only observed in tumor T83 that contains MIB1 amplification (Magnification ×20).

Fig. 3.

Colony formation assays following ectopic expression of GATA6. (A) Enhanced cell transformation in immortalized Barrett's CP-A cells following transient cotransduction of GATA6 and FGFR2IIIb compared with KRAS^12v positive control. (B) CP-A/FGFR2IIIb stable cells transduced with GATA6 pBMN6 compared with LacZ control. Significant increase of anchorage-independent growth was observed in CP-A/FGFR2IIIb stable cells transduced with GATA6. Colony count was performed using ImageJ software. qRT-PCR of FGFR2 and GATA6 was used to monitor transduction efficiency (Magnification ×1.5; **P < 0.01, ***P < 0.001, Student's t test). All transduction and colony formation assays were conducted in triplicate.

Fig. 4.

Cell proliferation, anchorage-independent growth, and DNA fragmentation assays following siRNA-mediated silencing of GATA6. (A) Significant reduction of cell proliferation upon silencing of GATA6 was observed in both immortalized Barrett's cells (CP-A and CP-B) and EAC OE33 and Flo-1/GATA6 stable cells. WST-1 assays were conducted in quadruplicate (see SI Appendix, Fig. S8 for the nonlineage TE13 and Het-1A cells). (B) qRT-PCR of the matched experiments was performed to monitor the knockdown efficiency (up to 85–90%). (C) Significantly decreased colony formation was observed in siGATA6-06-treated OE33 cells compared with siNonTarget controls in soft-agar assays performed in triplicate (Magnification ×1.25). The x axis reflects number of colonies. (D) Brightfield microscopic images of the siRNA-mediated knockdown of GATA6 in esophageal cells at 72 h (Magnification ×10). (E) A significant increase in DNA fragmentation upon GATA6 knockdown, assayed by BrdU/TUNEL flow cytometry, was observed in OE33 cells in both 0.2% and 10% FBS media compared with esophageal squamous TE13 and Het-1A cells. (F) Representative images of BrdU/TUNEL flow cytometry assays in OE33 cells. An increased upper right quadrant cell population is shown in GATA6 knockdown cells. (G) Quantitative verification of GATA6 knockdown using qRT-PCR (*P < 0.05, **P < 0.01, and ***P < 0.001).

Fig. 5.

Induction of apoptosis in GATA6-silenced esophageal cells. (A) Western blot analysis of PARP cleavage following GATA6-silencing was indicative of anoikis. Flo-1/GATA6 stable cells were cultured on agar-coated plates followed by GATA6 knockdown. (B) PARP cleavage by Western blot analysis was observed in OE33 but not in TE13 cells transfected with various GATA6 siRNA fragments against three different GATA6 coding sequences (SI Appendix, Table S7). (C) Caspase-Glo3/7 assays demonstrated that transfection of all three siRNA fragments targeting GATA6 caused significant increases in caspase activity in OE33 cells compared with squamous TE13 cells (*P < 0.05, **P < 0.01, and ***P < 0.001). (D) qRT-PCR assays to monitor knockdown efficiency of all three siRNA fragments targeting GATA6 in OE33 and TE13 cells.

Fig. 6.

Differentially regulated pro- and antiapoptotic signals following GATA6 modulation in EAC cells. (A) Three proapoptotic genes that were up-regulated upon silencing of GATA6 in OE33 cells using U133A array assays. (B) Validation of expression profiling using real-time RT-PCR in an independent set of experiments. Ratios represent comparisons of siGATA6-treated cells to matched cells treated with siNonTarget control at the same time point. (C) Up-regulation of TRAIL and down-regulation of BCL-2 in cells treated with either siGATA6 or pBMN6. OE33* and ** represent two independent experiments. (D) Analyses of procaspase 3 (casp3) and cleaved caspase 3 (c-casp3) expression in an apoptosis antibody array (Left) and Western blot of procaspase 9 (casp9) and cleaved caspase 9 (c-casp9) expression (Right) in OE33 cells treated with either siNonTarget control or siGATA6 for 60h (M, mock). (E) Representative kinase activation from the analysis of 46 phosphorylated kinases in EAC cells with either ectopic expression (FloA) or silencing (OE33) of GATA6 for 60 h using human phospho-kinase antibody array. Each phospho-kinase antibody is dotted in doublet. Sample layout is numbered in the upper panel and listed underneath. (F) Western blot analysis of p38α activation. Both OE33 and TE13 cells were treated for 60 h with siGATA6 or controls and FloA cells were transduced with either pBMN-Z or pBMN6 for 60 h. Total p38α and phospho-p38α (p-p38α) were examined and protein extracted from UVC-irradiated Flo-1 cells was used as a positive control. (N, not treated; M, mock).