The role of microRNA-200 in progression of human colorectal and breast cancer

Abstract

The role of the epithelial-mesenchymal transition (EMT) in cancer has been studied extensively in vitro, but involvement of the EMT in tumorigenesis in vivo is largely unknown. We investigated the potential of microRNAs as clinical markers and analyzed participation of the EMT-associated microRNA-200-ZEB-E-cadherin pathway in cancer progression. Expression of the microRNA-200 family was quantified by real-time RT-PCR analysis of fresh-frozen and microdissected formalin-fixed paraffin-embedded primary colorectal tumors, normal colon mucosa, and matched liver metastases. MicroRNA expression was validated by in situ hybridization and after in vitro culture of the malignant cells. To assess EMT as a predictive marker, factors considered relevant in colorectal cancer were investigated in 98 primary breast tumors from a treatment-randomized study. Associations between the studied EMT-markers were found in primary breast tumors and in colorectal liver metastases. MicroRNA-200 expression in epithelial cells was lower in malignant mucosa than in normal mucosa, and was also decreased in metastatic compared to non-metastatic colorectal cancer. Low microRNA-200 expression in colorectal liver metastases was associated with bad prognosis. In breast cancer, low levels of microRNA-200 were related to reduced survival and high expression of microRNA-200 was predictive of benefit from radiotheraphy. MicroRNA-200 was associated with ER positive status, and inversely correlated to HER2 and overactivation of the PI3K/AKT pathway, that was associated with high ZEB1 mRNA expression. Our findings suggest that the stability of microRNAs makes them suitable as clinical markers and that the EMT-related microRNA-200-ZEB-E-cadherin signaling pathway is connected to established clinical characteristics and can give useful prognostic and treatment-predictive information in progressive breast and colorectal cancers.

Figure 1. EMT factors in different compartments of clinical samples.

In matched human colorectal cancer samples, the levels of mir-141 were significantly lower in metastatic tissue than in normal tissue (p=0.043) (a). Expression of mir-200s in primary colorectal tumors with known metastatic status revealed a significant association with metastatic disease for low levels of mir-200c, relative to matched normal mucosa (p=0.041) (b) and in the metastatic lesions low levels of CDH1 indicated a trend of shorter time to recurrence (p=0.11) (c). E-cadherin was expressed specifically in epithelial colorectal cancer cells (d). Mir-141 showed strongest expression in epithelial cells of the liver metastases as compared to normal tumor-adjacent mucosa and primary colorectal tumor (e, f). Staining of the probe was negative in stromal cells and in hepatocytes. The integrity of frozen tissue samples from cancer patients was assayed using a bioanalyzer. The Ct values of microRNAs were not correlated with the corresponding RNA integrity numbers (RIN) values (g), whereas the Ct values of mRNA were associated with the RIN values (h).

Figure 2. Metastatic tumors show altered mir-200 expression.

Matched FFPE tissue samples from patients with metastatic colorectal cancer were microdissected (a). Compared to normal epithelial cells, primary colorectal cancer cells, but not metastatic cells, showed a significant decrease in mir-141 expression. Vimentin selectively stain the stromal compartment of primary and metastatic tumors (b). Measurement of mir-141 and mir-200c levels in primary cell cultures from matched primary colorectal cancer and liver metastases from the same patient confirmed an association between the two microRNAs equivalent to that detected in the microdissected tissue samples (c-f).

Figure 3. EMT-related factors act as treatment predictors in primary breast tumors.

The treatment predictive value of mir-200 was analyzed in tumors from women who had been randomized to treatment with radiotherapy (RT) or systemic chemotherapy (CMF: cyclophosphamide, methotrexate, 5-FU) (a, d). After stratifying for mir-200 expression, RT reduced the risk of local recurrences in the group of patients with tumors expressing high levels of mir-200s (b) but not in the group of tumors showing low mir-200 levels (c). The difference in the risk of distant recurrence in relation to treatment (e, f) was significant (test for interaction p = 0.0035).

Figure 4. Prognostic value of EMT-related factors in primary breast tumors.

In the group of breast cancer patients that did not receive chemotherapy, low expression of mir-141/200c was associated with significantly worse outcome in terms of distant recurrence-free survival (HR = 2.74, [CI 95% 1.08–6.99, p = 0.034]) (a). The opposite trend was seen for patients who received systemic chemotherapy (HR = 2.09, [CI 95% 0.95–4.58], p = 0.067) (b).

Figure 5. Associations between EMT and factors in cancer signaling pathways.

Expression of EMT-related factors was associated with components of cancer signaling pathways (significant associations found in this study indicated in red).