Integrative analysis of gene expression profiles reveals specific signaling pathways associated with pancreatic duct adenocarcinoma

Abstract

Background: Pancreatic duct adenocarcinoma (PDAC) remains a major health problem because conventional cancer treatments are relatively ineffective against it. Microarray studies have linked many genes to pancreatic cancer, but the available data have not been extensively mined for potential insights into PDAC. This study attempted to identify PDAC-associated genes and signaling pathways based on six microarray-based profiles of gene expression in pancreatic cancer deposited in the gene expression omnibus database.

Methods: Pathway network methods were used to analyze core pathways in six publicly available pancreatic cancer gene (GSE71989, GSE15471, GSE16515, GSE32676, GSE41368 and GSE28735) expression profiles. Genes potentially linked to PDAC were assessed for potential impact on survival time based on data in The Cancer Genome Atlas and International Cancer Genome Consortium databases, and the expression of one candidate gene (CKS2) and its association with survival was examined in 102 patients with PDAC from our hospital. Effects of CKS2 knockdown were explored in the PDAC cell lines BxPC-3 and CFPAC-1.

Results: The KEGG signaling pathway called "pathway in cancer" may play an important role in pancreatic cancer development and progression. Five genes (BIRC5, CKS2, ITGA3, ITGA6 and RALA) in this pathway were significantly associated with survival time in patients with PDAC. CKS2 was overexpressed in PDAC samples from our hospital, and higher CKS2 expression in these patients was associated with shorter survival time. CKS2 knockdown substantially inhibited PDAC cell proliferation in vitro.

Conclusions: Analysis integrating existing microarray datasets allowed identification of the "pathway in cancer" as an important signaling pathway in PDAC. This integrative approach may be powerful for identifying genes and pathways involved in cancer.

Keywords: CKS2; GEO; Pancreatic cancer; Pathway network.

Fig. 1

Genes differentially expressed between pancreatic cancer tissues and normal tissues. We analyzed differentially expressed genes in six GEO microarray datasets. Up-regulated: the number of genes whose expression levels were higher in pancreatic cancer tissues than in normal tissues. Down-regulated: the number of genes whose expression levels were lower in pancreatic cancer tissues than in normal tissues. GEO Gene Expression Omnibus, DEG differentially expressed gene

Fig. 2

Core pathway analysis of six GEO datasets. a Overlap of significant pathways in each GEO dataset used in this study revealed 32 common significant pathways. Left, Venn diagram of the result. Right, names of common significant pathways. b Overlap of the core pathways of each GEO dataset used in this study revealed 13 common core pathways. Left, Venn diagram of the result. Right, names of common core pathways. c Relationship between the common significant pathways and common core pathways

Fig. 3

Identification of an association of BIRC5, CKS2, ITGA3, ITGA6 and RALA with survival time of patients with PDAC. a Kaplan–Meier survival curves of BIRC5, CKS2, ITGA3, ITGA6 and RALA in The Cancer Genome Atlas (Internal ID: 811) or International Cancer Genome Consortium database (Internal ID: 812). Survival data were obtained from the SurvExpress database. b Association of CKS2 expression with survival time in a cohort of 102 patients with PDAC at our hospital

Fig. 4

Overexpression of CKS2 in PDAC. CKS2 mRNA levels were significantly higher in PDAC tissues than in normal tissues according to the Oncomine database. Red color indicates CKS2 expression levels in normal pancreatic tissues, while blue color indicates CKS2 expression levels in pancreatic cancer

Fig. 5

Genomic alterations of CKS2 in human cancers. a Summary of copy number variations of CKS2 across 49 cancer genomic studies (http://www.cbioportal.org). The results are displayed as a histogram of the frequencies of CKS2 copy numbers across cancer studies. b Correlation between CKS2 copy number and CKS2 expression level in 87 cancer cell lines derived from 15 types of human cancer (http://portals.broadinstitute.org/ccle/home). Pearson’s correlation coefficient (r) is shown. c Correlation of CKS2 expression and DNA methylation in 149 pancreatic cancer tissues in The Cancer Genome Atlas database. Pearson’s correlation coefficient (r) is shown

Fig. 6

CKS2 promotes pancreatic cancer cell proliferation. a Significant pathways enriched in genes whose expression profile matches that of CKS2, based on DAVID software. b CKS2 mRNA levels in pancreatic cancer tissues and paired normal tissues from 102 patients with PDAC at our hospital, based on real-time PCR analysis. Results are mean ± SEM. *P < 0.05. c Western blotting of CKS2 in PDAC cell lines BxPC-3 and CFPAC-1, as well as in non-cancer pancreatic duct epithelial cell line HPDE6-C7. d Western blotting of CKS2 in PDAC (T) and paired normal tissues (N). Representative results of 20 pairs of clinical specimens are shown. e CKS2 knockdown in BxPC-3 and CFPAC-1 cells. Levels of mRNAs were determined by qRT-PCR. Results are mean ± SEM from three independent experiments. f CKS2 knockdown substantially reduced proliferation of BxPC-3 and CFPAC-1 cells, as determined using the CCK-8 assay. Results are mean ± SEM from three experiments, each conducted with six replicates of each sample. ***P < 0.001 vs. control