A potential role for Periostin in Barrett’s carcinogenesis

B28

Introduction: Periostin is a secreted extracellular matrix protein important in embryonic development and wound repair. It is a potential inducer of epithelial-mesenchymal transition and cell motility in normal development and cancer; although, whether it has a tumor suppressive or tumor promoting role is controversial. Overexpression of periostin mRNA was recently reported in Barrett’s (BE) and oesophageal adenocarcinoma (OAC) tissues (Hao Y, Gastroenterology 2006).
 >Aims: 1. Use gene expression data to determine whether periostin is overexpressed in the Barrett’s-dysplasia- carcinoma sequence. 2. Validate array data in independent samples at the protein level. 3. Investigate the capacity for periostin to promote invasion in vitro.
 >Methods: RNA expression was determined using Agilent microarrays from samples representing different stages of OAC development (normal squamous-oesophagus (NE); n=6, intestinal metaplasia (IM); n=21, Low-grade-dysplasia (LGD); n=17, High-grade-dysplasia (HGD); n=13, OAC; n=8). Periostin protein expression was examined immunohistochemically on an independent data-set (abcam14041). Oesophageal stromal fibroblasts were isolated from normal (NAFs), Barrett’s (BAFs), and tumor-associated patient tissues (CAFs) and characterized using vimentin and CK8/18. Periostin expression was determined by Western blotting and QPCR in primary fibroblasts, oesophageal adenocarcinoma (SEG, FLO, OE33, BIC) and dysplastic epithelial cell lines (GohTert, ChTert, GihTert). Effects of recombinant periostin (100-400ng/ml) and addition of primary fibroblasts on FLO invasiveness and motility were investigated in vitro using matrigel and wound healing assays.
 >Results: Periostin was significantly overexpressed in OAC compared to NE (4.36 fold, p=0.0035), IM (2.76 fold, p=0.0275), LGD (3.61 fold, p=0.0027) and HGD (14 fold, p=8.83E-08), (One-Way ANOVA). IHC showed that periostin protein expression was restricted to the stroma of BE and cancer tissues, with an apparent concentration at the epithelial-stromal interaction front in OAC. Purified cell types (primary stromal fibroblasts and established epithelial cell lines) were used to further characterize the cell specific expression of periostin. Apart from SEG cells, dysplastic and OAC epithelial cell lines expressed lower levels of periostin mRNA compared to oesophageal fibroblast lines derived from patients with dysplasia and cancer (2 fold, p<10^-4, t-test). Periostin was detectable at the protein level in all cell types. Recombinant Periostin (100ng/ml) promoted invasion of FLO cells in vitro (1.7 fold. P=0.0092), but had no significant effect on cell motility. 2/3 CAFs increased invasiveness of co-cultured FLO cells compared to corresponding NAFs (p= n.s), but this was not directly attributable to periostin.
 >Conclusion: Overexpression of periostin in primary fibroblasts and in stroma of cancer related tissues, combined with periostin-induced promotion of invasion in vitro, suggests a role for fibroblast derived periostin in cancer. The isolation of stage-specific primary stromal fibroblasts offers a valuable tool for further functional studies of stromal-epithelial interactions in Barrett’s carcinogenesis.