Introduction
CEACAM6 is a member of the cacinoembryonic antigen (CEA) family of immunoglobulin glycoprotein cell adhesion molecules (CAM) comprising at least 12 CEACAM members[
1]. CEACAMs are a diverse group of proteins which play major roles in cell-cell and cell-ECM adhesion and have been implicated in the control of cell proliferation, angiogenesis and tissue remodelling[
1]. More recently, CEACAMs have also been implicated in mediating tissue responses to pathogens[
1]. CEACAM6 is expressed at low levels in normal epithelial, endothelial and hematopoetic cells including granulocytes, T-cells and NK cells[
2‐
4]. In contrast, CEACAMs are up-regulated in many epithelial malignancies including pancreatic, colorectal and breast cancers[
5,
6]. The expression of CEACAM6 also correlates with the metastatic potential of some epithelial malignancies, suggesting that the altered expression of CEACAM6 may contribute to tumour progression[
7]. However, a definitive role for CEACAMs in tumourigenesis has not been formally proved. For example, CEACAM6 appears to affect the release of cytochrome-c from the mitochondria in response to cell detachment leading to the inhibition of caspase activation and hence, suppression of caspase induced apoptosis or anoikis in pancreatic cancer cells[
8,
9]. These apoptotic-suppressive effects have been shown to be AKT-dependent in pancreatic cancer cells[
9]. Moreover, transgenic mice which overexpress members of the CEA family display colonic dysplasia[
10]. In contrast, CEACAM6 up-regulation is associated with an increase in apoptosis in acute lymphoblastic leukaemia (ALL), indicating that the apoptosis-modulating effects of CEACAM6 may be tumour-type-specific[
4].
A recent transcriptomic profiling study comparing highly tumourigenic clonal variants of an established head and neck cancer squamous cell carcinoma (HNSCC) cell line with poorly tumourigenic clonal variants, identified a strong association between CEACAM6 expression and tumourigenic potential[
11]. Since an association between HNSCC and CEACAM6 expression has not been previously reported we now examine whether the over-expression of CEACAM6 is also present in human HNSCC samples.
Discussion
In this study we report, for the first time, on the role of CEACAM6 in HNSCC. Previous work with keratinocytes and keratinocyte-derived SCC cells has shown that CEACAM6 is selectively expressed in differentiated keratinocytes[
2] and is highly expressed in a tumourigenic clonal variant of the Detroit 562 HNSCC cell line[
10]. In addition, other workers have reported that i) CEACAM6 overexpression occurs in variety of epithelial malignancies[
5‐
7], ii) that CEACAM6 overexpression is associated with increased metastases, proliferation and the suppression of annoikis[
7‐
9], iii) that CEACAM6 overexpression induces a
src-dependent increase in AKT activity that suppresses gemcitabine sensitivity in pancreatic cancer cells[
9] and finally, iv) a transgenic model of CEA-overexpression suggests CEACAM6 overexpression can contribute to the development of colonic dysplasia[
10]. We now extend these findings and report that CEACAM6 is focally overexpressed in a large fraction of human HNSCCs
in situ. The heterogeneous pattern of CEACAM6 overexpression is also evident in established HNSCC cell lines
in vitro and
in vivo. Moreover, we show that over-expression of CEACAM6 increases tumour growth and tumour initiating activity by suppressing PI3K/AKT-dependent apoptosis of HNSCC in a xenotransplant model of HNSCC. Finally, we show that foci of CEACAM6 expressing cells are selectively ablated by treatment of xenotransplant tumours with pharmacological inhibitors of PI3K/AKT
in vivo.
A novel finding in the present study is the observation that CEACAM6 is focally overexpressed in the majority of HNSCCs examined. Whilst the sample size examined was small it highlights an important issue that has important biological and clinical implications. Specifically, intratumoural heterogeneity is a major contributor to the emergence of drug resistance and tumour recurrence[
17]. Consistent with this, our data suggest that focal overexpression of CEACAM6 is indicative of sensitivity of human HNSCC to selective cytotoxic drugs. In this regard two observations relating to CEACAM6 are relevant. Firstly, knockdown or overexpression of CEACAM6 resulted in a decrease and increase in tumourigenic activity in SCC cells
in vivo respectively. Secondly, CEACAM6 has been shown to modulate the cytotoxic effects of conventional chemotherapeutics such as gemcitabine in pancreatic cancer cell lines[
9] and in the present study we showed that CEACAM6 could mediate sensitivity to new targeted agents such as the PI3K inhibitor, BGT226. It is noteworthy that the modulation of gemcitabine sensitivity is also mediated
via a
src and PI3K/AKT-dependent pathway[
9]. These data indicate that whilst CEACAM6 may invoke pro-survival responses in cancer cells by activating the PI3K/AKT pathway this same pathway could be selectively targeted by specific cytotoxic drugs. Thus, the presence of CEACAM6
+ve foci would be predicted to bestow selective sensitivity against certain chemotherapeutic treatments (eg: gemcitabine or PI3K inhibitors). Proof of principle for this hypothesis is shown by the reduction in phospho-S437 AKT induced by knockdown of CEACAM6 and the loss of CEACAM6
+ve foci in tumours treated with cytotoxic doses of PI3K inhibitors. Thus, CEACAM6 could be used to predict PI3K inhibitor sensitivity. Moreover, the observation that CEACAM6 expression correlates with metastatic potential[
8,
20‐
22] would suggest that, in chemotherapy-naive tumours, the presence of CEACAM6
+ve foci could serve as a prognostic marker of poor outcome and in this instance targeting CEACAM6/PI3K/AKT pathways could be exploited therapeutically. Supporting this, is a recent study, by Blumenthal et al.[
20], demonstrating that the addition of antibodies that inhibited the binding of CEACAM6
+ve breast cancer cells to endothelial cells reduced tumour cell invasion[
20]. Finally, intratumoural heterogeneity can arise through a number of mechanisms such as the evolution of variant cells from a common clonal precursor, micro-environmental influences, stochastic processes or tissue/cell plasticity[
17]. The present study suggests that the focal pattern of CEACAM6 expression, in tumours, is derived from a specific clonal progenitor within the tumour rather than being transiently induced by the local environment. This is based on the observation that CEACAM6
+ve and
–ve cells persist in long term tissue culture models, consistent with an heritable mechanism (eg: genetic or epigenetic).
Whilst CEACAM6 clearly has the capacity to contribute to drug resistance and tumour recurrence it is clear that other factors also contribute to drug resistance and tumour recurrence. This is supported by our observation that targeted inhibition of the CEACAM6/PI3K/AKT pathway in SCC cells induced killing of 50% of the total HNSCC cells. Similarly, we have identified clonal variants of HNSCC cells that express very low levels of CEACAM6 yet still retain tumourigenic potential[
11]. Moreover, we show that the knockdown of CEACAM6 results in a decrease, but not an ablation, of tumour initiating activity or tumour growth. Thus, CEACAM6 likely represents one factor, of many, that can modulate tumour growth and tumour initiating activity. This is entirely consistent with the emerging importance of intratumoural heterogeneity[
17]. We previously reported that HNSCC display intratumoural heterogeneity that was reflected in histomorphologically and transcriptomically distinct clonal variants[
11,
14]. We showed that clonal variants of HNSCC cells could persist
in vitro in established cell lines and displayed significant differences in tumour initiating activity and drug resistance[
11,
13,
14]. Several groups have now definitively shown, by single cell sequencing, that tumours comprise multiple genetically distinct clonal populations[
23‐
27]. Emerging, clinical and molecular data unequivocally show that the presence of intratumoural heterogeneity, exemplified by focal CEACAM6 overexpression in HNSCC cells, is a major contributor to tumour drug responses and patient outcomes[
17].
Earlier work by Duxbury[
28], suggests that the major contribution of CEACAM6 to tumour growth and tumour initiating activity is mediated
via suppression of anoikis. Anoikis is a form of apoptosis induced by loss of cell-cell/EMC contact. Thus, anoikis may be more relevant to a 3 dimensional tumour environment rather than an
in vitro cell monolayer system.[
29]. Supporting this, we found that the
in vivo effects of CEACAM6 over-expression/knockdown were not reflected by the
in vitro effects of CEACAM6. For instance, CEACAM6 over-expression/knockdown had modest and inconsistent effects on apoptotic rates
in vitro. However, over-expression of CEACAM6 significantly reduced caspase-3 dependent apoptosis of HNSCC cells in a xenotransplant model. Anti-apoptotic activity is commonly viewed as tumour promoting and hence the anti-apoptotic activity of CEACAM6 would suggest it has tumour promoting (oncogenic) activity[
30]. CEACAM6-mediated inhibition of apoptosis
in vivo therefore contributes in part, or wholly, to the ability of HNSCC cells to initiate a tumour in a xenotransplant model of HNSCC. In addition, CEACAM6 over-expression also contributes in part, or wholly, to the increased tumour growth in a xenotransplant model of HNSCC. Based on these findings, it is reasonable to speculate that focal patches of CEACAM6 expressing cells within HNSCC may reflect the presence of a subpopulation of cells with a greater potential for recurrence/metastasis than CEACAM6
-ve subpopulations of HNSCC cells.
Conclusions
In conclusion, our study shows that CEACAM6 is focally overexpressed in a large fraction of human HNSCCs in situ and contributes to tumour growth and tumour initiating activity. The effect of CEACAM6 on tumour growth and initiation is mediated via suppression of PI3K/AKT-dependent apoptosis of HNSCC in a xenotransplant model of HNSCC. Finally, our studies show that CEACAM6+ve tumours, or tumour foci, are selectively sensitive to treatment with pharmacological inhibitors of PI3K/AKT in vivo.
This work was supported by a PhD scholarship awarded to SC by the Garnett Passe & Rodney Williams Memorial Foundation. NS is supported by a senior research fellowship awarded by the Cancer Council Queensland. This work was also supported by a research grant awarded to NS (#455929, #569689) from the Australian NHMRC: Cancer Council QLD, #631479 and a practitioner fellowship to AG from the Cancer Collaborative Group.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
SC drafted the manuscript and performed in vitro growth and death assays. LMDL performed the animal experiments. MHR, ET and LEM performed immunohistochemistry and cloning experiments. AC and OG performed the drug toxicity studies. AG and NS supervised the project and contributed to the draft of manuscript. All authors read and approved the final manuscript.