Differential gene expression profile reveals deregulation of pregnancy specific beta1 glycoprotein 9 early during colorectal carcinogenesis

Abstract

Background: APC (Adenomatous polyposis coli) plays an important role in the pathogenesis of both familial and sporadic colorectal cancer. Patients carrying germline APC mutations develop multiple colonic adenomas at younger age and higher frequency than non-carrier cases which indicates that silencing of one APC allele may be sufficient to initiate the transformation process.

Methods: To elucidate the biological dysregulation underlying adenoma formation we examined global gene expression profiles of adenomas and corresponding normal mucosa from an FAP patient. Differential expression of the most significant gene identified in this study was further validated by mRNA in situ hybridization, reverse transcriptase PCR and Northern blotting in different sets of adenomas, tumours and cancer cell lines.

Results: Eighty four genes were differentially expressed between all adenomas and corresponding normal mucosa, while only seven genes showed differential expression within the adenomas. The first group included pregnancy specific beta-1 glycoprotein 9 (PSG9) (p < 0.006). PSG9 is a member of the carcinoembryonic antigen (CEA)/PSG family and is produced at high levels during pregnancy, mainly by syncytiotrophoblasts. Further analysis of sporadic and familial colorectal cancer confirmed that PSG9 is ectopically upregulated in vivo by cancer cells. In total, deregulation of PSG9 mRNA was detected in 78% (14/18) of FAP adenomas and 75% (45/60) of sporadic colorectal cancer cases tested.

Conclusion: Detection of PSG9 expression in adenomas, and at higher levels in FAP cases, indicates that germline APC mutations and defects in Wnt signalling modulate PSG9 expression. Since PSG9 is not found in the non-pregnant adult except in association with cancer, and it appears to be an early molecular event associated with colorectal cancer monitoring of its expression may be useful as a biomarker for the early detection of this disease.

Figure 1

Genomic structure of PSG9. The human PSG9 gene is located on chromosome 19q13.2 and encodes at least three variants. The exon structures of each isoform are indicated in the figure. The largest PSG9 variant, variant 1, consists of 1282 nucleotides and encodes for a protein of 427 amino acids. The location and size of each primer and riboprobes used in this study are indicated (a). Statistical analysis of microarray data using SAM program revealed up to two-fold over-expression of PSG9 in adenoma 1, 2 and 3 (A1 to A3) compared to normal (N). The log² ratio of fluorescence intensity is shown on the y-axis while PSG9 expression in different samples is shown on the x-axis. At least four microarray replicates including dye switches (Cy3- or Cy5-labelled) were performed to account for possible labeling and hybridization bias (b). The data have further been verified in adenoma 1, 2 and normal tissue by Real Time PCR (c).

Figure 2

PSG9 is exclusively expressed by placental cells. Quantitative RT-PCR analysis of PSG9 transcript (a) and northern blot analysis of different normal tissues (1 μg mRNA/lane) probed with the labelled PSG9 cDNA (b) shows specific expression of PSG9 in placenta. (a-b). Colorectal cancer cell lines were also examined for PSG9 expression level by quantitative- (c) and semi-quantitative RT-PCR (d). Highest expression level of PSG9 was detected in SW480 cells and was lowest in the RKO cell line (c-d).

Figure 3

PSG9 is ectopically expressed by cancer cells. A multiple-tumour Northern blot (10 μg RNA/lane) revealed over-expression of PSG9 in colon, rectal and uterus cancer. At least three different transcripts were observed (a). Expression of PSG9 "isoform a" was examined in a panel of colorectal cancer cases. Forty-nine percent (15/27) of cases showed up-regulation of PSG9 (isoform a) in the tumours compared to corresponding normal tissue (b). However, the expression levels were different between different cases (T; tumour, A; adenoma).

Figure 4

RNA in situ hybridization and IHC analysis of colorectal cancer cases. Sections from sporadic/familial colorectal cancer and placenta (as positive control) were hybridised with Dig-labelled PSG9 RNA probes. Both PSG2 (b) and PSG9 (c) were expressed at a high level in placental tissue. Sense-probes were used as a negative control on placental tissue (a). In microscopic normal epithelial cells from FAP cases, PSG9 expression was detected at the top of crypt (d) (see discussion). PSG9 transcripts (shown as dark blue) were detected at very low levels in normal mucosa (e), adenomas (f), while high expression was detected in tumour cells from the same FAP case (g). In contrast to sporadic cases, PSG9 was detected in normal appearing mucosa in some FAP cases with APC germline mutations, suggesting that dose and level of APC have an impact on PSG9 levels in cells (e, i). A high level of PSG9 was detected in a sporadic case (k), while corresponding normal tissue was negative (i). Tumours and corresponding normal tissue were also examined for β-catenin stabilization by immunostaining (h, j, l). As expected, high levels of β-catenin were detected in all sporadic colorectal tumours (l), while the protein level was less intense in FAP cases (h) where PSG9 up-regulation could be measured (g).

Figure 5

PSG9 expression analysis in Wnt stimulated RKO cells. To determine whether induction of Wnt signaling in cells with wild type APC could induce PSG9 expression, RKO cells with wild type APC and β-catenin were stimulated with either Wnt3a or Kenpaullone (Kenp). After 23 hrs treatment RNA and protein were extracted and processed for PSG9 transcripts expression level by semi-quantitative RT-PCR (a) and β-catenin accumulation by western blot analysis (b). Neither of the treatments nor the RKO-β-catenin^S37A (βcat-S37A) stable cell line which expressed constitutively active β-catenin in RKO cells caused expression of PSG9 (a). The SW480 colorectal cancer cell line was used as a positive control (a-b). The RKO cells responded to Wnt stimulation, since cells treated with Wnt3a showed 4-fold induction in luciferase activity compared to untreated cells (c). Axin2, a known downstream target of Wnt signaling, showed 2.4 fold up-regulation in expression as determined by quantitative PCR (d). No PSG9 transcript up-regulation was detected in these cells under these conditions (e). Each sample was analyzed in triplicate.