In this study, we found that high expression of AEG-1 was correlated with
advanced tumor stages and regional lymph node metastasis in a large cohort of OSCC
samples. The association between AEG-1 and distant metastasis was not statistically
significant; evidence for an association may be confounded by the relatively low
incidence (10%) of distant metastasis at initial presentation, a feature intrinsic to
HNSCC [13]. In addition, our research has demonstrated that silencing of AEG-1
mitigates the malignant phenotypes of HNSCC cell lines in vitro and attenuates tumor
growth and pulmonary metastasis in vivo. Our results provide the first strong evidence
that AEG-1 is overexpressed in at least a subset of HNSCC and contributes to adverse
clinical outcomes. Moreover, we found that AEG-1 upregulates the expression of
MMP1, thereby uncovering a novel mechanism underlying the invasiveness of
HNSCC.
Metastasis, defined as the detachment of daughter cells from the primary site of
lesions and subsequent colonization of preferential target organs, is one of the
hallmarks of malignancies [76, 77]. The metastasis cascade can be divided into steps
of local invasion, intravasation, survival, extravasation and colonization. The tumor
cells disseminate as individual cells, referred to as “individual cell migration”, or
expand in solid cell strands, sheets, files and clusters, called “collective cell
migration” [78]. Base on cell type, integrin engagement, cytoskeletal structures and
protease production, single-cell migration can occur in mesenchymal type and
amoeboid type. Mesenchymal type single cell migration in carcinomas is driven by
epithelial-mesenchymal transition. Cells that undergo mesenchymal migration assume
a spindle fibroblast-like morphology that is dependent on integrin-mediated adhesion
dynamics and the presence of high traction force on both cell pole [79, 80]. Focal
contacts form and turnover in the range of 10-120 min, resulting in relatively slow
migration speeds of 0.1-2 μm/min in 3D models [80, 81]. However, oral squamous
cell carcinomas exhibit predominantly collective cell invasion [82]. Our data also
revealed though knockdown of AEG-1 impeded migration and invasion of HNSCC
cells, the EMT-related genes were unaffected in majority. There are three hallmarks
that characterize collective cell migration. First, the cells remain physically and
functionally connected such that the integrity of cell-cell junctions is preserved during
movement [83, 84]. Second, multicellular polarity and organization of actin generate
traction and protrusion force for migration and maintain cell-cell junctions. Finally,
moving cell groups structurally modify the tissue along the migration path, either
clearing the track or by causing secondary modification of extracellular matrix (ECM).
Degradation and remodeling of ECM are essential for neoplastic permeation into
adjacent stomal tissue, as well as for breaching the perivascular basement membrane
to initiate metastasis. MMPs are zinc-dependent enzymes, consisting of a propeptide,
catalytic domain and a hemplexin-like C terminal domain. MMPs acquire enzymatic
activity after peptidyl cleavage of the propeptide that interacts with the zinc ion in the
catalytic domain [85]. The role of MMPs was traditionally believed to be primarily
restricted to degradation of the ECM; however, mounting evidence suggests that
MMPs are also involved in development, angiogenesis, inflammation and cancer
progression, with the latter of which occurs through promoting migration and survival
of cancer cells, orchestrating release of growth factors from extracellular reservoirs,
and modulating recruitment of inflammatory cells to the tumor [86-88].
MMP1 is the stereotypical secreted collagenase of the MMP family, with its
principle interstitial substrates consisting collagen I, II, III, VII, VIII, X, and gelatin.
During collective cellular migration, which is the predominant pattern adopted by
squamous cell carcinoma [89], MMP1 interacts with integrin α2β1 at the leading edge
[90], and degrades native matrix macromolecules into fragments that are subsequently
processed by the gelatinases MMP2 and MMP9 [91]. Analyses of clinical specimens
revealed that expression of MMP1 correlated with lymphatic invasion and lymph
node metastasis [92, 93]. Furthermore, it is thought that MMP1 not only plays a
pivotal role in vascular extravasation of metastatic neoplastic lesion, but it also
contributes to the vascular remodeling at distant target sites, such as lung and bones
[94-96]. Promotion of osteotropic metastasis can be accomplished in part by
activation of the RANKL pathway through cleavage of EGF-like ligand by MMP1
[97]. In addition to classical function of matrix remodeling, MMP1 was shown to
manipulate biological behavior through direct interaction with receptors on neoplastic
cells. Protease-activated receptors (PARs) are tethered-ligand, G protein-coupled
receptors that are activated by proteolytic cleavage of their extracellular domains and
participate in tumor invasion by inducing cancer cell migration [98]. PAR1, the
prototypic member of the PAR family, has been known to respond to serine proteases
including thrombin [99], plasmin [100], and activated protein C [101]. PAR1
expression is increased in a number of cancers including breast, colon, and lung.
Nontheless, MMP1 was identified as a nonserine protease agonist of PAR1 through
cleaving the receptor between amino acid residues arginine 41 and serine 42 to
generate PAR1-dependent Ca2+ signals and subsequent cell migration of breast cancer
cells [86]. This cleavage can not be performed by MMP2, 3, 7, or 9. In summary,
MMP1 contributes to tumor invasion and metastasis by remodeling the matrix, and
triggering the signaling cascades and crosstalk between neoplastic cells and adjacent
interstitium. In our study, we have shown that AEG-1-knockdown SAS and FaDu
cells reduce both the invasive ability of cancer cells (Figure 4D) and the expressions
of MMP1 (Figure 7). Furthermore, MMP inhibitor is able to inhibit the invasive
abilities of cancer cells (Figure 8) to the level comparable to those observed in
AEG-1-knockdown SAS and FaDu cells (Figure 4D). Taken together, these results
indicate that AEG-1 is able to increase the invasive ability of cancer cells by
increasing MMP1 expression. It is also intriguing that MMP1 was demonstrated to
converse pro-TNF-α into the soluble cytokinetic form through proteolytic cleavage
[102]. TNF-α is a well-established positive upstream regulator of AEG-1 and it
promotes cancer cell survival in an NF-κB-dependent manner [103]. Combined with
our data, it is suggested that AEG-1 function as a link in the vicious loop of TNF-α-
NF-κB axis.
To our knowledge, the present article is the first to report that AEG-1 regulates
p65 phosphorylation at serine 536 and the subsequent MMP1 expression in HNSCC.
MMP activity is modulated at various levels, including transcription, subcellular
compartmentalization, proteolytic activation and inhibition. Data from our luciferase
reporter assay indicate that AEG-1 regulates MMP1 transcription, primarily by acting
on a region in the promoter upstream of nucleotide -2269. Previous studies reported
the presence of NF-κB and AP-1 binding sites at nucleotides -2886 and -3471 of the
MMP1 promoter, respectively [75, 104]. Although AEG-1 has previously been
reported to regulate MMP9 expression [64] through c-jun in human glioma cells [71],
we found that AEG-1 does not affect phosphorylation of serines 63 and 73 of c-jun.
On the other hand, AEG-1 knockdown in HNSCC cell lines attenuated
phosphorylation of the p65 subunit (RelA) of NF-κB at serine 536. Phosphorylation at
this amino acid residue is required for perinuclear localization of p65 to facilitate
nuclear import [42, 105]. This observation was supported by the finding that AEG-1
expression correlates with phosphorylation of p65 at serine 536, both intertumorally
and intratumorally, in OSCC clinical specimens (p < 0.001, Fisher’s exact test). It was
thought that the activation of NF-κB requires degradation of IκB. However,
post-translational modifications of the subunits of NF-κB have also been found to
determine the functional activity to a great extent. Serine 536 of p65, which is located
within the C-terminal transactivation domain, is a target of multiple protein kinases,
including IκB kinase α/β (IKKα/β), IKKε, TANK binding kinase 1 (TBK1), and
ribosomal S6 kinase 1 (RSK1). Phosphorylation at this amino acid residue has also
been reported to suppress nuclear export of NF-κB and to increase transactivation of a
variety of downstream genes through positive interaction of p65 with co-activators
(CBP and p300) [106-110]. Phosphorylation of serine 536 also activates the
survival-promoting pathway when cells are challenged with chemotherapeutic
cytotoxic agents such as doxorubicin and etoposide. [105, 111]
As mentioned previously, AEG-1 contains no known functional domains,
making it unlikely that AEG-1 phosphorylates p65 directly. Based on our findings, it
seems plausible that AEG-1 enhances p65 phosphorylation by recruiting associated
protein kinases to the AEG-1-p65 complex. More detailed experiments are required to
confirm this hypothesis. Our ChIP data also revealed that AEG-1 enhances the
binding of p65 to the MMP1 promoter, thereby activating downstream genes by
functioning as a linker between p65 and CBP to form the basal transcriptional
machinery. Although p65 has been suggested to bind to AEG-1 through amino acid
residues 101 to 205 of the latter [46], the interacting regions of CBP and AEG-1
remain unknown. Elucidation of the exact binding epitopes will require examination
of the crystal structure of AEG-1 and its associated proteins.
Microarray analysis of the gene expression profile in SAS cell line revealed that
MMP1 was not the sole gene under the influence of AEG-1. SNORD3B-1 (small
nucleolar RNA, C/D box 3B-1), the gene that is up-regulated with high fold change
after AEG-1 knockdown, is believed to function in RNA transport and ribosome
biogenesis in eukaryocytes. However, no correlation between SNORD3B-1 and
cancer biology has been documented in literatures. Initially isolated in 2001, Elastin
microfibril interface located protein 2 (EMILIN2) is a relatively new member of the
EMILIN family [112]. It is composed of the cysteine-rich EMI domain at the
N-terminus, which is the hallmark of the family, α-helical domains with high
probability for coiled-coil structure formation, and a proline-rich motif adjacent to a
collagenous stalk preceding the globular gC1q domain at the C-terminus. It was
shown that expression of EMILIN2 triggered the apoptosis of different cancer cell
lines including HT1080 cells (a cell line of fibrosarcoma) and HeLa cells but not
normal dermal fibroblasts [113]. Cell death depends on the activation of the extrinsic
apoptotic pathway following EMILIN2 binding to the TRAIL receptors DR4 and, to a
lesser extent, DR5. Binding is followed by receptor clustering and colocalization with
lipid rafts occurred subsequently and lead to assembly of death-inducing signaling
complex as well as caspase activation. The knockdown of EMILIN2 increased
transformed cell survival, and overexpression impaired clonogenicity in soft agar and
three-dimensional growth in natural matrices due to massive apoptosis. The domain
responsible for triggering apoptosis of neoplastic cells was mapped to be between
amino acid residues 286 to 436, a coiled-coil fragment toward the N-terminus of the
molecule [114]. In addition to apoptosis-inducing cabability on cancer cells, an
increased tumor vessels density in the tumors treated with EMILIN2 was observed
while tumor growth was significantly suppressed. Further studies showd that
treatment with EMILIN2 failed to alter endothelial tubules formation in Matrigel, the
protein significantly increased proliferation rate and migration of endothelial cells
both in transwell systemand in scratch tests. Combined the current findings with
previous studies [74], it is evident that AEG-1 plays a role in remodeling
tumor-associated extracellular matrix and endothelial function in scenarios of
angiogenesis and vascular damage.
Anactamin-1 (ANO1), also known as discovered on gastrointestinal stromal
tumors protein 1 (DOG1), oral cancer overexpressed 2 (ORAOV2), tumor-amplified
and overexpressed sequence 2 (TAOS2), and transmembrane proteins with unknown
function 16 (TMEM16A), is upregulated in different cancer types, including
gastrointestinal stromal tumor, squamous cell carcinoma of the upper aerodigestive
tract and the esophagus (ESCC), and pancreatic cancer. ANO1 is a calcium-activated
homodimerized chloride channel and the gene is located in chromosome 11q13
amplicon, where CCND1 (encoding cyclin D1) resides [115]. Amplification of a
discrete, approximately 5-Mb region of 11q13 occurs in approximately one third of
HNSCC, and it is even more prevalent in HPV-negative tumors [116, 117]. Although
cyclin D1 has been considered to be the main driver of the 11q13 amplicon, it is not
sufficient for malignant transformation of normal breast cells and lacks predictive
value for the survival of HNSCC or breast cancer patients [118, 119]. A recent study
demonstrated that ANO1 is amplified and highly expressed in cell lines and clinical
specimen of breast cancer and HNSCC [120]. Amplification of ANO1 correlated with
disease grade and poor prognosis in both disease entities. ANO1 is sufficient to
promote cell viability in the absence of 11q13 amplification. Knockdown of ANO1 in
ANO1-amplified breast cancer cell lines and HNSCC cell lines (Te11 and FaDu,
specifically) bearing 11q13 amplification inhibited proliferation, induced apoptosis,
and reduced tumor growth in established cancer xenografts. Moreover, ANO1
chloride channel activity was indispensable for its prosurvival properties. Furthermore,
ANO1 knockdown or pharmacological inhibition of its chloride-channel activity
reduced EGF receptor (EGFR) and calmodulin-dependent protein kinase II (CAMKII)
signaling, which subsequently attenuated AKT, v-src sarcoma viral oncogene
homolog (SRC), and extracellular signal-regulated kinase (ERK) activation in breast
cancer and HNSCC in vitro and in vivo. Validation of AEG-1 regulation upon ANO1
and elucidation of the underlying mechanism seems to be a promising topic for
research and requires more detailed experiments.