Expression of Oct3/4 and Nanog in the head and neck squamous carcinoma cells and its clinical implications for delayed neck metastasis in stage I/II oral tongue squamous cell carcinoma

The clinical role of CSC-specific molecular markers is not only the detection and
isolation of the CSC-like population but also the prediction of aggressive tumor behavior,
which is achieved by evaluating the density and/or spatial distribution of the CSCs.
Additionally, they serve as ideal targets for the development of new therapeutic agents.
Therefore, the identification of a more?selective CSC-specific marker that defines
the stem cell phenotype alone is desirable so that the clinical relevance of the marker’s
expression in each type of cancer can be assessed. In the present study, we successfully
identified and isolated SP cells in all three HNSCC cell lines examined at proportions
of 0.9 % to 10.2 %. This result is comparable with previous studies that examined
other HNSCC cells, in which the proportion of the SP ranged from 0.02 % to 17.1 %
37]-40]. Moreover, we found that the SP cells expressed notably higher levels of Oct3/4, Nanog, and ABCG2 than the MP cells, which is partially consistent with the findings of the aforementioned
studies, especially with regard to Oct3/4 37] and ABCG2 37], 38], 40]. In a previous study that examined the HSC-4 cells, we found comparable results concerning
Oct3/4 and Nanog, but not ABCG2, with much lower proportion of the SP (0.37 %) than that shown in the present study
(0.9 %) 41]. In addition, although increased expression of Bmi-1 in the SP of HNSCC cells was also previously reported 37], 40], this was not observed in our cells. Such inconsistencies as mentioned above are
thought to be attributable not only to differences in the baseline expression levels
of genes among the examined cell lines but also to differences in the conditions of
SP analysis, including factors such as Hoechst concentration, incubation time, a type
of cell sorting machine, and the stringency of SP selection 62].

Both Oct3/4 (also termed as Oct3, Oct4, or POU5F1) and Nanog are known to play an
essential role in the maintenance of self-renewal and pluripotency at an undifferentiated
state, i.e., the fundamental features that define embryonic stem (ES) cells 63], 64]. In simple terms, Oct3/4 collaborates with Sox2 and Nanog to form a regulatory circuit
that maintains ES cell pluripotency 63]. Interestingly, Oct3/4 and Nanog are reportedly two of the four defined factors that
give rise to the reprogramming of human somatic cells into germ-line-competent induced
pluripotent stem (iPS) cells 65], 66]. In comparative analyses using a naive Bayes network methodology focused on self-renewal
of both human and murine ES cells, Oct3/4 was ranked top among 17,342 genes evaluated
in both networks and Nanog was ranked within the top 1 % in each network 67].

Oct3/4, which is a member of the Pit-Oct-Unc domain transcription factors family,
is normally expressed in both adult and embryonic stem cells 68], 69]. Genomic profiling studies revealed that Oct3/4 regulates the transcription of a
large number of its downstream target genes, either positively or negatively, to sustain
the stemness of ES cells 70], 71]. Intriguingly, Oct3/4-mediated gene regulation in ES cells is unique because Oct3/4
expression levels must remain within an appropriate narrow range to maintain the undifferentiated
state, and either up- or downregulation of Oct3/4 to levels outside this range triggers
the differentiation of ES cells 68]. Of the specific pathways downstream of Oct3/4, Tcl1 (T cell leukemia/lymphoma 1,
a product of proto-oncogene) was found to be regulated transcriptionally by Oct3/4,
and it is involved in the control of ES cell proliferation (but not differentiation)
via enhancement of the kinase activity of Akt1 71]. Furthermore, Oct3/4 is reportedly essential for the survival and anti-apoptosis
activity of murine ES cells in response to stress, with effects being mediated through
the STAT3/survivin pathway 72]. In addition, the potential functions of Oct3/4 expressed in malignant tumors have
also been investigated. For example, the oncogenic potential of ES cells was shown
to be directed by Oct3/4 in a dose-dependent manner; higher Oct3/4 expression increases
the malignant potential of ES cell-derived tumors, whereas Oct3/4 inactivation induces
regression of the malignant component 73]. Moreover, Oct3/4 was revealed to maintain the survival of CSC-like cells of Lewis
lung carcinoma 3LL cells and breast cancer MCF7 cells, partly by inhibiting apoptosis
through the Oct4/Tcl1/Akt1 pathway 74]. In hepatocellular carcinoma cells, Oct3/4 was shown to upregulate BIRC5 (survivin)
and CCND1 expression by increasing their promoter activity, thereby promoting cell
proliferation and resisting cell apoptosis 75]. These findings suggest that expression of Oct3/4 contributes to the transformation
of non-tumorigenic cells as well as the maintenance of the CSC-like and malignant
properties of cancer cells.

Nanog, the variant homeobox transcriptional factor, was identified as one of the primary
downstream targets of Oct3/4 by a functional cDNA screen in ES cells 76], 77]. Although Oct3/4 cannot maintain the undifferentiated state of ES cells without leukemia
inhibitory factor (LIF) 68], overexpression of Nanog allows ES cells to remain undifferentiated and to self-renew
independently of LIF/STAT3 signaling 77], 78]. Interestingly, while Oct3/4 is obviously required for Nanog to function, expression
of Nanog itself can be sustained in the absence of Oct3/4 76], 78]. Taken together, the aforementioned studies imply that the functions of Oct3/4 and/or
Nanog aberrantly expressed in CSCs of human cancer may be implicated in the malignant
behavior of cancer cells, including tumorigenicity and metastasis. However, the functional
and clinical significance of these molecules in human cancers, including HNSCC, have
remained largely unknown.

In our in vitro experiments, we found that proliferation rates were similar between SP and MP cells.
This is in agreement with previous studies including those that examined HNSCC cells,
where SP cells expressed higher levels of Oct3/4 and/or Bmi-1 than MP cells, while
no difference was shown in proliferation rate between them 37], 38]. These results indicate that Oct3/4 and Nanog, which were expressed at markedly higher
levels in SP cells than MP cells, have a negligible contribution to the proliferation
of the individual HNSCC cells examined here. In contrast, we confirmed that SP cells
were more capable of migration and invasion than MP cells. This result is similar
to that of a previous study in which SP cells were shown to be more invasive than
both non-SP cells and parental HNSCC cells; however, these SP cells expressed higher
levels of Bmi-1, and Oct3/4 and Nanog expression were not examined 38]. Other than in SP cells, increased migration and/or invasion capacities have also
been observed in oral cancer stem-like cell populations enriched through sphere formation
in serum-free cultivation, which showed high expression of Oct3/4, Nanog, and ABCG2
79], as well as in CD44+ cell populations of HNSCC cells isolated by cell sorting 19]. Therefore, in combination with previous studies, our findings suggest that certain
CSC marker molecules including Oct3/4 and Nanog may contribute to the enhanced cell
motility and invasiveness of HNSCC cells.

In the clinical specimens examined here, a significant positive correlation was found
between Oct3/4 and Nanog in terms of their immunopositivity. This finding substantiates
the notion that Oct3/4 transcriptionally regulates the expression of Nanog, and that
the two collaborate to maintain stemness properties in ES cells 63], 71], 76], 77]. Although the percentage of Oct3/4- or Nanog-positive cells was no more than 10 %
at the most, such a small population of potential CSC-specific marker-positive cells
appears to match the basic concept of CSC.

More importantly, multivariate analysis of clinicopathological and immunohistochemical
data demonstrated, for the first time, that expression of Oct3/4, as well as vascular
invasion, are independently correlated with DNM in stage I/II TSCC following partial
glossectomy. Hence, both Oct3/4 and vascular invasion can be considered dependable
markers for prediction of DNM development, which was further supported by the favorable
diagnostic performance as shown in Table 4, although their reliability requires further validation based on larger independent
cohorts. Although, in our cohort, Nanog was excluded from the independent predictors
of DNM at the final step of the logistic regression analysis, this seems rather reasonable
because the significant correlation found in expression between Oct3/4 and Nanog inevitably
means these two factors are confounding factors of each other. Therefore, we suggest
that Nanog could be an alternative independent predictor of DNM where Oct3/4 cannot
be evaluated. Taken together, these results imply that certain stemness properties
of CSCs that exist in primary cancer cells are closely involved with the development
of DNM. Regarding the association between DNM and vascular invasion, our result lends
support to a significant correlation in a previous study, which was identical except
that a smaller sample size was analyzed 55].

Apart from the association with DNM, our results regarding the clinical significance
of Oct3/4 and Nanog expression are seemingly inconsistent with a previous study of
OSCC, in which increased expressions of Oct3/4 and Nanog were correlated with advanced
stage and worse overall survival, but not with lymph node metastasis at the time of
diagnosis 79]. However, such inconsistency could be attributable to the unequal distribution of
the stage of the patients and/or to differences in the protocols for immunohistochemistry
and criteria for positive staining. As a comprehensive molecular approach, gene expression
profiling based on cDNA microarray is reportedly advantageous for discriminating between
N0 and N+ patients, with clustering analysis of at least 100 to 1,500 metastasis-associated
genes being employed 80], 81]. However, because these profiling strategies are technically complex and not currently
cost-effective, they have not necessarily been incorporated into routine diagnosis.

Based on the results of our in vitro experiments, and those of previous studies, the existence of Oct3/4- and Nanog-expressing
CSCs in cancer cells in primary lesions is presumed to contribute to development of
DNM, at least in part, by enhancing the cell motility and invasiveness of these CSCs.
Specifically, given the assumption that cancer cells have already migrated and invaded
from the primary lesion into nearby lymphatic vessels, even in the early stage before
complete resection of the primary tumor, we can speculate that the vast majority of
these cells, which lack the stemness properties associated with Oct3/4 and Nanog expression
(i.e., non-CSCs), cannot survive and establish lymph node metastasis, in part due
to insufficient cell motility and invasiveness. In contrast, we can postulate that
certain cells expressing Oct3/4 and Nanog (i.e., CSCs) can form metastatic lesions
because of their enhanced migration and invasion abilities as well as their other
stemness properties. However, further studies will be necessary to clarify the actual
difference in behavior between the CSCs and non-CSCs during the processes required
for metastasis in human cancer.

One of the crucial mechanisms that regulate cell motility is epithelial-to-mesenchymal
transition (EMT), a morphogenetic change in epithelial cells associated with various
phenotypic modulations, which include loss of epithelial features such as cell polarity
and cell-to-cell contact, and acquisition of mesenchymal traits such as cell motility
82]–84]. EMT and its reverse process, mesenchymal-to-epithelial transition (MET), are considered
to be involved in embryonic development and a variety of pathological events such
as wound healing, tissue fibrosis, and cancer progression including metastasis 82]–84]. Recent studies have implicated EMT in the emergence of CSCs; for example, induction
of EMT resulted in acquisition of stemness properties in immortalized mammary epithelial
cells and, inversely, stem-like cells isolated from both mammary glands and carcinomas
expressed a number of EMT markers 85]. These findings support the concept of “migrating CSCs” that are assumed to be derived
from “stationary CSCs” by undergoing transient EMT and are responsible for metastasis
86]. Such an EMT-induced stem cell-like phenotype was presumed to be acquired through
developmentally regulated signaling pathways that induce EMT, such as Wnt, Notch1,
and Hedgehog, which also drive the stemness properties of both normal cells and CSCs
87]. Intriguingly, a recent study showed that co-overexpression of Oct4 and Nanog in
lung adenocarcinoma cells increased CSC-like properties and induced EMT by upregulating
Slug and Snail expression, and thereby promoted tumorigenesis, drug resistance, and
metastasis 88]. Collectively, our results and those of previous studies suggest that enhancement
of the CSC’s stemness properties by Oct3/4 and Nanog expression may also contribute
to cancer progression, including the development of DNM, together with the elevated
cell motility associated with EMT. Additional studies are required to fully understand
the mechanism by which Oct3/4 and Nanog regulate the malignant behaviors of CSCs,
as well as to establish the optimal criteria for estimating the latent malignant potential
of cancer at the early stage using CSC-related molecular markers.