In our two armed, matched collection of CCs we demonstrated that high levels of both SOX2 and nuclear β-catenin expression, respectively, were associated significantly with metastatic disease. This observation might explain the association of SOX2 protein expression with poorer prognosis and reduced overall survival in CC, reported in the literature [19
]. Lymph node and distant metastases are both independent prognostic factors correlating with poor survival in CC [25
]. Patients with node-negative CCs (UICC stage I/II) have a 5 year survival exceeding 75%, whereas patients with node-positive disease only have a 5 year survival rate of less than 50% [26
]. Similar, patients with synchronous- or metachronous liver metastases reveal 5 year survival rates following liver resection ranging between 25% and 55% whereas for non-operated patients this rate lies between 0% and 5% [27
The highest rate of lymph node- and distant metastases could be obtained in the group of CC with combined scores of high SOX2 and high β-catenin expression, whereas tumours with low or absent expression of both markers were associated with a reduced risk for loco-regional- and distant spread. SOX2 negative cases with high nuclear β-catenin expression showed higher rates of distant metastases but no association with the nodal status. Since nuclear β-catenin is involved in epithelial-mesenchymal transition (EMT) and stem cell formation during embryonic development, the combination of EMT and stem cell competence during malignant progression of CC, both regulated by β-catenin, may result in "migrating tumour stem cells", driving tumour invasion and metastases [6
]. β-catenin is considered to be one of the driving forces for the initiation and maintenance of colorectal cancer stem cells (CSCs). In addition, SOX2 is an inducer of embryonic stemness which is part of the signature of colorectal CSCs [28
]. Mechanistically, it might therefore be assumed that both β-catenin and SOX2 drive the malignant progression of CCs by inducing cancer stemness of colorectal tumour cells. CSCs are known to be responsible for the process of metastasis [30
This is also supported by our finding that SOX2 expression can be found within the stem cell niche in a small proportion of normal colonic crypts (Figure ). The fact, that SOX2 plays a major role for the development of the upper but not of the lower digestive tract may explain why SOX2 protein-expression can only be found in a minority of cases within the normal colonic tissue. This finding is concordant to previously published data [19
]. To further investigate this issue we correlated the SOX2 expression with previously published data on CD133, an established marker for CSC which is predictive for poor prognosis and distant spread in CC [31
]. Unexpectedly, no significant correlation between SOX2 and CD133 expression could be demonstrated in our collection on CC (χ2
-test, data not shown). However, similar to β-catenin combined scores of high SOX2 and high CD133 showed strong predictive value according to the risk of distant spread (8 of 10 cases; 80%) and the presence of lymph-node metastases (9 of 10 cases; 90%). In summary, our data underline the importance of stemness-associated markers for the prediction of distant spread and the prognostification of CC. These observations also indicate that markers and marker-combinations for a stemness phenotype could be relevant for identification of CCs with high risk for distant spread and subsequently could have major impact on clinical decisions according to the radicalness of loco-regional surgery and adjuvant chemotherapy.
Through in vitro studies it was shown that SOX2 represses the transcriptional activity of the β-catenin/TCF complex, thus affecting the expression of oncogenic β-catenin driven transcription [18
]. The assumed underlying mechanism, as far as already understood, is complex as SOX2 directly binds to the β-catenin/TCF complex and vice versa SOX2 transcription is regulated by β-catenin [34
]. Moreover, knockdown of SOX2 decreased the growth rate of cultivated colorectal tumour cell lines in vitro and in vivo [37
]. Surprisingly, no correlation of SOX2 and β-catenin protein expression levels could be demonstrated in our collection of CRCs. Nevertheless, it could not be concluded from our results that the reported negative regulatory feedback between β-catenin and SOX2 does not occur in human CRC [34
]. Therefore, to clarify this issue, further immunohistochemical and functional studies involving SOX2 and Wnt-regulated target genes are needed to elaborate the actual impact of the regulatory feedback mechanisms between SOX2 and the β-catenin/TCF-complex in CRC patients.
Taken together, our results suggest that SOX2 protein expression may be a possible prognostic marker in the management of CC patients. Nevertheless, this observation requires validation in larger, randomized patient cohorts including also the left side of the colon and the rectum before SOX2 immunohistochemistry should be used in the clinical setting.