Gα15 in early onset of pancreatic ductal adenocarcinoma

Abstract

The GNA15 gene is ectopically expressed in human pancreatic ductal adenocarcinoma cancer cells. The encoded Gα15 protein can promiscuously redirect GPCR signaling toward pathways with oncogenic potential. We sought to describe the distribution of GNA15 in adenocarcinoma from human pancreatic specimens and to analyze the mechanism driving abnormal expression and the consequences on signaling and clinical follow-up. We detected GNA15 expression in pre-neoplastic pancreatic lesions and throughout progression. The analysis of biological data sets, primary and xenografted human tumor samples, and clinical follow-up shows that elevated expression is associated with poor prognosis for GNA15, but not any other GNA gene. Demethylation of the 5' GNA15 promoter region was associated with ectopic expression of Gα15 in pancreatic neoplastic cells, but not in adjacent dysplastic or non-transformed tissue. Down-modulation of Gα15 by shRNA or CRISPR/Cas9 affected oncogenic signaling, and reduced adenocarcimoma cell motility and invasiveness. We conclude that de novo expression of wild-type GNA15 characterizes transformed pancreatic cells. The methylation pattern of GNA15 changes in preneoplastic lesions coincident with the release a transcriptional blockade that allows ectopic expression to persist throughout PDAC progression. Elevated GNA15 mRNA correlates with poor prognosis. In addition, ectopic Gα15 signaling provides an unprecedented mechanism in the early steps of pancreas carcinogenesis distinct from classical G protein oncogenic mutations described previously in GNAS and GNAQ/GNA11.

Figure 1

GNA15 differential expression in normal tissue and cancer. Median gene expression across several tumor samples and paired normal tissues utilizing GEPIA2.

Figure 2

GNA15 expression and methylation in cancer cell lines. GNA15 mRNA expression (white bar showing RNAseq on the left axis) and DNA methylation (black bars, showing reduced representation bisulfite sequencing (RRBS) on the right axis) according to data from the Cancer Cell Encyclopedia. Among one thousand cancer cell lines, pancreas and biliary tract are the only normal tissues that do not express GNA15 but the derived cancer cell lines had high GNA15 expression levels, comparable to leukemia and upper aero-digestive tract. Box plots are shown as median and 25th and 75th percentiles. ALL acute lymphocytic leukemia, CML chronic myelocytic leukemia, AML acute myelocytic leukemia, DLBCL diffuse large B cell lymphoma, BNHL B cells non-Hodgkin lymphoma. The number below to the lineage name indicates how many cell lines are in the lineage.

Figure 3

Demethylation, ectopic GNA15 expression and survival. (A) GNA15 gene methylation was mapped at single CpG in apparently normal tissue, and xenografted tumors, matched with primary tumors when available (gray bars represent standard deviation. n = 4 for normal tissues and primary tumors and n = 22 for xenografts). (B) Spearman’s correlation between Gα subunits expression levels and survival according to the TCGA-PAAD cohort (n = 59 samples). The isotypes associated to a significant p value (< 0.05) are in bold. No apparent correlation with tumor stages was found for any Gα gene. (C) The methylation level reported in the TCGA-PAAD court plotted against survival data distinguishing the NET and the PDAC patients further subdivided based on low (bottom 30%) vs high (top 30%) GNA15 expression levels. (D) GNA15 high expression is associated with poor survival. Kaplan survival curve in patients in the cytology group. All 20 patients were divided in two groups based on GNA15 expression (log RNA Seq V2 RSEM, log-rank test; z = 2.67, p = 0.00754).

Figure 4

de novo GNA15 gene expression in PDAC. (A–C) Human pancreas biopsies were analyzed by eosin hematoxylin (upper panels) or ISH (middle and lower panels) developed in Fast-RED (two examples are indicated by the red arrows) utilizing a GNA15 specific probe in addition to positive and negative controls (Supp. Fig. S1). Preneoplastic and neoplastic lesions were evidently positive. A PanIN1 lesion is shown in panel A (left micrograph), and a PanIN2 (right micrograph). A superficial fragment of an IPMN is shown in panel B, the yellow arrows indicate papillary structures in the non invasive region. Panel C shows two examples of PDAC. (D) The expression level of GNA15 was quantified by ISH. In the abscissa each number corresponds to a patient, and each circle represents the average value (dot density by ISH) of all lesions in a FFPE section, one circle per section, 2–4 sections per patient, 37 total patients. Red or green dots were used alternatively to differentiate each patient. (E) GNA15 relative expression in the primary tumor was quantified by ISH and compared in patients that relapsed versus patients that remained disease free one year after pancreas resection. Box plots are shown as median and 25th and 75th percentiles: points are displayed as outliers if they are above or below the 1.5 times the interquartile range. A T-test was applied to compare the two groups, p < 0.05. (F) Tumor tissue fragments of resected pancreata were split in two parts. Half was analyzed by ISH to select for variable GNA15 expression levels going from absent (apparently healthy tissue from NET patients) to high (PDAC with abundant positive lesions). The other half of the fragment was analyzed by TaqMan RT-PCR measuring GNA15 and GNA14 expression levels.

Figure 5

Functional consequences of Gα15 signaling. (A) In PT45 cells Gα15 was down-modulated by IPTG inducible shRNA expression or knocked out by CRISPR/Cas9 in all alleles of 3 independent clones, as confirmed by DNA sequencing. Protein down-modulation was confirmed by immunoblot as shown and quantified as 70% ± 9% of control (n = 4). (B) The impact of preventing Gα15 expression was evaluated by immunoblot analysis of phospho-amino acids diagnostic of the activation state of PKD1 (n = 10). (C) The impact of down-modulating Gα15 expression was evaluated as in B (n = 10). (D) The impact of preventing Gα15 expression in response to IGF was evaluated by immunoblot analysis of phospho-amino acids diagnostic of the activation state of AKT (n = 6). (E) The impact of down-modulating Gα15 expression in response to 100 ng/ml IGF and 100 nM insulin was evaluated as in D (n = 8). (F) PT45 cell motility assessed in scratch assays. Representative images are shown in Supp. Fig. S12. (G) PT45 cell invasion of collagen-coated membranes. Data are presented as mean ± SEM. Direct comparison between cells with GNA15 downmodulated and controls was performed applying t-test and *, p < 0.05; **, p < 0.01; *** P < 0.0005. Extended blots are presented in Supp. Fig. S13.