Gene Expression Analysis of Label-Retaining Pancreatic Adenocarcinoma Cells: Defining a New Class of Cancer Stem Cell

Introduction: A new paradigm of treatment for patients with pancreatic adenocarcinoma is needed given the dismal prognosis. Recent identification of cancer stem cells using cell surface markers has offered a new approach to therapy. We hypothesize that identification of cancer stem cells should capitalize on their unique functional properties such as self-renewal by asymmetric division. Label-retaining cells have been described to represent adult tissue stem cells and to self-renew by asymmetric division. Thus, we undertook this study to investigate whether pancreatic cancers contain label-retaining cancer cells (LRCC), and to investigate whether LRCC have a unique gene signature amenable to novel therapeutic approaches. Methods: Three pancreatic adenocarcinoma cell lines (PaTu8988T, BxPC-3, and Panc-1) were labeled with fluorescent nucleotide analogues (Cy5-dUTP) to isolate live LRCC. Labeled cells were sorted using FACS after one cell cycle to identify a Cy5+ population of cells which had incorporated the fluorescent nucleotide into their genome. the Cy5+ cells were placed back in culture and grown for a total of 8 cell cycles. FACS was then used to identify the Cy5+ cells within the expanded population, referred to as LRCC. Total RNA was isolated from the LRCC, non-LRCC (NLRCC) and normal cells. Following amplification and in vitro transcription, cRNA was loaded onto Illumina HumanHT-12 v4.0 Expression BeadChips. Microarray analysis was performed using Agilent GeneSpringGX software package, v11.5 (Agilent Technologies, Santa Clara, CA). Results: The proportion of cells retaining the fluorescent nucleotides was higher than mathematically expected in each group. Using FACS, the percent of LRCC isolated from the total cell mass ranged from 2.97%-5.02% compared to the expected 0.3% (p=0.02, Student's t-test). Cluster analysis of microarray data generated from LRCC, NLRCC and control cell RNA demonstrated that the NLRCC population was similar to the control cells and the LRCC population segregated separately. Gene expression analysis revealed a distinct subset of genes up-regulated > 2 fold change in the LRCC compared to the NLRCC. Canonical pathway analysis of the upregulated subsets of genes noted increased activity of the Wnt-β-catenin and Hedgehog pathway. Gene ontology analysis revealed increased expression of negative regulators of apoptosis in the LRCC population. Conclusions: We show that 3 pancreatic cancer cell lines contain LRCC. Based on the fact that LRCC are hypothesized to be putative stem cells, and we have demonstrated that pancreatic LRCC have a gene signature consistent with stem cells, we propose that pancreatic LRCC may represent a novel class of pancreatic cancer stem cells. Additionally, the unique gene signature of LRCC may provide novel targets for therapy specifically directed towards pancreatic cancer stem cells.