Targeting FAT1 Inhibits Carcinogenesis, Induces Oxidative Stress and Enhances Cisplatin Sensitivity through Deregulation of LRP5/WNT2/GSS Signaling Axis in Oral Squamous Cell Carcinoma

FAT atypical cadherin 1 (FAT1) regulates cell-cell adhesion and extracellular matrix architecture, while acting as tumor suppressor or oncogene, context-dependently. Despite implication of FAT1 in several malignancies, its role in oral squamous cell carcinoma (OSCC) remains unclear. Herein, we document the driver-oncogene role of FAT1, and its mediation of cell-death evasion, proliferation, oncogenicity, and chemoresistance in OSCC. In-silica analyses indicate FAT1 mutations are frequent and drive head-neck SCC, with enhanced expression defining high-risk population and poor prognosis. We demonstrated aberrant FAT1 mRNA and protein expression in OSCC compared with non-cancer tissues, whereas loss-of-FAT1-function attenuates human primary SAS and metastatic HSC-3 OSCC cell viability, without affecting normal primary human gingival fibroblast cells. shFAT1 suppressed PCNA and upregulated BAX/BCL2 ratio in SAS and HSC-3 cells. Moreover, compared with wild-type cells, shFAT1 concomitantly impaired HSC-3 cell migration, invasion, and clonogenicity. Interestingly, while over-expressed FAT1 characterized cisplatin-resistance (CispR), shFAT1 synchronously re-sensitized CispR cells to cisplatin, enhanced glutathione (GSH)/GSH synthetase (GSS)-mediated oxidative stress and deregulated LRP5/WNT2 signaling. Concisely, FAT1 is an actionable driver-oncogene in OSCC and targeting FAT1 in patients with erstwhile cisplatin-resistant OSCC is therapeutically promising.


Introduction
Oral cancer, comprising cancers of the oral cavity (International Statistical Classification of Diseases and Related Health Problems, ICD-10: C00-06) and oropharyngeal regions (ICD-10: C09-10), is one of that the therapeutic targeting of FAT1 re-sensitized cisplatin-resistant OSCC cells to cisplatin through deregulation of LRP/Wnt signaling, thus projecting FAT1 as a novel therapeutic target for anticancer treatment of therapy-resistant OSCC.

High FAT1 Expression Drives HNSC, Defines High Risk Population and is Associated with Poor Prognosis
In the light of the divergent roles of FAT1 in different malignancies, seeking to understand the pathocytological relevance of FAT1 and determine its molecular dynamics in highly metastatic and recurrent OSCC cells, we examined FAT1 expression and mutational profile in the TCGA HNSC cohort (n = 502). Results of our bioinformatics analysis showed that of the 131 mutated/mutant cancer drivers detected in the TCGA HNSC cohort, the most mutated drivers included TP53, FAT1, NOTCH1, CDKN1, CDKN2A and PIK3CA, in decreasing order of mutational frequency ( Figure 1A). In addition, we observed that of the principal 97 mutations in FAT1, the truncating (otherwise known as nonsense) gene mutation accounting for 75% of all FAT1-associated mutations, was most frequent ( Figure 1B). Because of the implication of driver mutation burden in poor clinical outcome [22], we evaluated the effect of altered FAT1 expression in the TCGA HNSC cohort, and demonstrated that compared to the FAT1 low group, high FAT1 (FAT1 high ) expression confers significant survival disadvantage on the patients with HNSC ((hazard ratio (HR) = 1.34, 95% CI: 1.02-1.77; p = 0.038) ( Figure 1C). Consistent with the above, the FAT1 high group exhibited strong association with high risk of disease-specific mortality compared with the FAT1 low group ( Figure 1D), suggesting a prognostic role for FAT1 in patients with HNSC.

High FAT1 Expression is Positively Correlated with Disease Progression and Poor Clinical Outcome in Patients With HNSC
In the light of our data suggesting a driver and prognostic role for FAT1 mutations and expression, respectively, in patients with HNSC, probing further for validation of these function, we

High FAT1 Expression Is Positively Correlated with Disease Progression and Poor Clinical Outcome in Patients with HNSC
In the light of our data suggesting a driver and prognostic role for FAT1 mutations and expression, respectively, in patients with HNSC, probing further for validation of these function, we observed a strong association between increased expression of FAT1 (probe ID: 201579_at) and disease progression, as demonstrated by mean expression transformed counts of 2.31, 1.07, 1.2, and 1.46 in stage I (T1), stage II (T2), stage III (T3) and stage IV (T4), compared with 0.97 in normal samples (Figure 2A), based on analysis of the GSE3524/GDS1584 dataset of gene expression profiles in OSCC cells from 16 patients with 20 samples, isolated by laser capture microdissection. Interestingly, FAT1 expression spikes in this dataset were markedly associated with nodal involvement as seen in T1N2bM0, T3N2bM0, T4N1M0, and T4N2bM0 (Figure 2A, also see Supplementary Figure S4). Our analysis of the OSCC cohort (n = 291) of the TCGA HNSC dataset showed that aberrant expression of FAT1 in patients with OSCC was strongly associated with worse overall survival in the FAT1 high , compared to the FAT1 low patients [HR: 1.12 (0.97-1.3), p = 0.121] with the median survival for FAT1 high and FAT1 low patients being 1081 and 1591 days, respectively ( Figure 2B). Moreover, analysis of the E-MTAB-1328 dataset from methylome, transcriptome and miRNome array and high throughput sequencing profiling of 89 patients with 104 samples also showed that FAT1 high patients exhibited shorter metastasis-free survival than their FAT1 low counterparts [HR:1.1 (0.74-1.63), p = 0.653] ( Figure 2C) and shorter relapse-free survival [HR:2.35 (0.48-11.52), p = 0.290] in the FAT1 high , compared to the FAT1 low patients, based on analysis of 44 primary head and neck tumor samples using the GSE10300 dataset ( Figure 2D). These data are indicative of a critical role of high FAT1 expression in disease progression and poor clinical outcome in patients with HNSC.  (Figure 2A), based on analysis of the GSE3524/GDS1584 dataset of gene expression profiles in OSCC cells from 16 patients with 20 samples, isolated by laser capture microdissection. Interestingly, FAT1 expression spikes in this dataset were markedly associated with nodal involvement as seen in T1N2bM0, T3N2bM0, T4N1M0, and T4N2bM0 ( Figure 2A, also see Supplementary Figure S4).  Figure 2D). These data are indicative of a critical role of high FAT1 expression in disease progression and poor clinical outcome in patients with HNSC.

FAT1 is Aberrantly-Expressed in OSCC Clinical Samples and Cell Lines
Since the focus of the present study is more specifically OSCC, tapering down from our evolving generic understanding of the role of FAT1 in HNSC, using the Peng Head-Neck OSCC cohort (n = Figure 2. High FAT1 expression is positively correlated with disease progression and poor clinical outcome in patients with HNSC. (A) Histograms of the stage-associated differential expression of FAT1 in the GPL96/GDS1584 gene expression profiling by array data-set, transformed count, 20 samples. Kaplan-Meier plots of the correlation between differential FAT1 expression and (B) overall survival of OSCC cohort of the TCGA-HNSC, n = 291, (C) metastasis-free survival of patients with HNSC, n = 89 using the E-MTAB-1328 dataset, and (D) relapse-free survival of patients undergoing surgery or biopsy for HNSC at University of North Carolina, Chapel Hill and Vanderbilt University using the GSE10300 dataset, n = 44. High/Low expression bifurcation was based on median FAT1 expression in all cases.

FAT1 Is Aberrantly-Expressed in OSCC Clinical Samples and Cell Lines
Since the focus of the present study is more specifically OSCC, tapering down from our evolving generic understanding of the role of FAT1 in HNSC, using the Peng Head-Neck OSCC cohort (n = 79), we further demonstrated a 2.82-fold increase in FAT1 expression in the OSCC compared to the 'normal' non-tumor oral cavity samples (t-test = 12.0, p = 4.57 × 10 -18 ) ( Figure 3A). Similarly, analysis of the Estilo Head-Neck tongue squamous cell carcinoma (TSCC) cohort data (n = 58) showed a 2.99-fold increase in the expression of FAT1 in TSCC samples compared to normal tongue samples (t-test = 5.384, p = 8.77 × 10 −7 ) ( Figure 3B). Immunohistochemical (IHC) staining of samples from our OSCC tissue archive demonstrates significant overexpression of FAT1 protein in the plasma membrane and nucleus of cancerous cells, compared to the neighboring non-cancerous cells ( Figure 3C). Consistent with our IHC data, results of our western blot analyses of our OSCC samples indicate the FAT1 protein is preferentially highly expressed in the cancerous cells compared to their non-cancerous counterparts ( Figure 3D). On the transcript level, we also demonstrate a 2.04-fold enhancement in the mRNA expression of FAT1 in our OSCC samples compared with the non-cancerous samples (p < 0.01) ( Figure 3E). In addition, using in vitro OSCC models, we demonstrated that compared to its expression in adult human primary normal gingival fibroblast cells (HGF, ATCC®PCS-201-018™), the expression of FAT1 protein was significantly up-regulated in the highly metastatic human OSCC HSC-3 (2.14-fold, p < 0.05), and SAS (2.76-fold, p < 0.01) cells ( Figure 3F). These data do indicate that FAT1 is aberrantly expressed in OSCC clinical samples and cell lines. of the Estilo Head-Neck tongue squamous cell carcinoma (TSCC) cohort data (n = 58) showed a 2.99fold increase in the expression of FAT1 in TSCC samples compared to normal tongue samples (t-test = 5.384, p = 8.77 × 10 −7 ) ( Figure 3B). Immunohistochemical (IHC) staining of samples from our OSCC tissue archive demonstrates significant overexpression of FAT1 protein in the plasma membrane and nucleus of cancerous cells, compared to the neighboring non-cancerous cells ( Figure 3C). Consistent with our IHC data, results of our western blot analyses of our OSCC samples indicate the FAT1 protein is preferentially highly expressed in the cancerous cells compared to their non-cancerous counterparts ( Figure 3D). On the transcript level, we also demonstrate a 2.04-fold enhancement in the mRNA expression of FAT1 in our OSCC samples compared with the non-cancerous samples (p < 0.01) ( Figure 3E). In addition, using in vitro OSCC models, we demonstrated that compared to its expression in adult human primary normal gingival fibroblast cells (HGF, ATCC® PCS-201-018™), the expression of FAT1 protein was significantly up-regulated in the highly metastatic human OSCC HSC-3 (2.14-fold, p < 0.05), and SAS (2.76-fold, p < 0.01) cells ( Figure 3F). These data do indicate that FAT1 is aberrantly expressed in OSCC clinical samples and cell lines.

Loss-of-FAT1 Function Impairs OSCC Cell Proliferation and Enhances Cell Death
To proffer a solution, we investigated if and to what degree the loss-of-FAT1 function negatively affects the survival and proliferation of OSCC cells. Our results indicate that shFAT1 transfection markedly suppressed the proliferation cum viability of SAS (62.5%, p < 0.05) or HSC-3 (58.1%, p < 0.05) cells, compared to their wild-type counterparts; interestingly the repressive effect of shFAT1 on the

Loss-of-FAT1 Function Impairs OSCC Cell Proliferation and Enhances Cell Death
To proffer a solution, we investigated if and to what degree the loss-of-FAT1 function negatively affects the survival and proliferation of OSCC cells. Our results indicate that shFAT1 transfection markedly suppressed the proliferation cum viability of SAS (62.5%, p < 0.05) or HSC-3 (58.1%, p < 0.05) cells, compared to their wild-type counterparts; interestingly the repressive effect of shFAT1 on the proliferation/viability of normal gingival fibroblast cell line HGF, was barely apparent and statistically insignificant ( Figure 4A), indicating the non-lethality of shFAT1 to normal oral cavity cells, and at least in part, the oncoselectivity of the loss-of-FAT1 function in patients with OSCC. Using Ki-67 protein immunofluorescence staining, we demonstrated further that shFAT1 induced 80.6% (p < 0.01) and 78.3% (p < 0.01) reduction in the number of viable SAS and HSC-3 cells, respective, while conversely having no effect on the HGF cells which increased by 6% ( Figure 4B). In parallel assays, against the background of cell-cycle regulated and apoptosis stimuli-mediated existent interconnection between cell proliferation and apoptosis [23], we demonstrated that shFAT1 significantly down-regulated the expression levels of marker of proliferation Ki-67, the proliferating cell nuclear antigen (PCNA) and B cell lymphoma 2 (Bcl2) proteins, while up-regulating Bax protein expression in the transfected SAS and HSC-3 cells, compared to their wild-type counterparts ( Figure 4C), which is suggestive of a role for FAT1 in the promoting OSCC cell proliferation and evasion of cell death.
Using Ki-67 protein immunofluorescence staining, we demonstrated further that shFAT1 induced 80.6% (p < 0.01) and 78.3% (p < 0.01) reduction in the number of viable SAS and HSC-3 cells, respective, while conversely having no effect on the HGF cells which increased by 6% ( Figure 4B). In parallel assays, against the background of cell-cycle regulated and apoptosis stimuli-mediated existent interconnection between cell proliferation and apoptosis [23], we demonstrated that shFAT1 significantly down-regulated the expression levels of marker of proliferation Ki-67, the proliferating cell nuclear antigen (PCNA) and B cell lymphoma 2 (Bcl2) proteins, while up-regulating Bax protein expression in the transfected SAS and HSC-3 cells, compared to their wild-type counterparts ( Figure  4C), which is suggestive of a role for FAT1 in the promoting OSCC cell proliferation and evasion of cell death.
Furthermore, gene expression-based heatmap generated from re-analysis of the A-AFFY-44, AFFY_HG_U133_PLUS_2, E-GEOD-30784 dataset of the gene expression profile of OSCC cohort (n = 229 samples, 54675 genes), originally to identify potential biomarkers for early detection of invasive OSCC using OSCC (n = 167), oral dysplasia (n = 17) and normal oral (n = 45) samples, demonstrated that FAT1 expression was marginal in the non-cancerous control samples, but high FAT1 expression was associated with high PCNA, BIRC5/Survivin, BCL2, and BCL2L1, while low FAT1 expression correlated with high FADD, BAX, PARP1, CASP3, CASP8, and CASP9 in the cancer tissues ( Figure  4D) suggesting an association between reduced FAT1 expression and induction of cell death. For a balanced perspective, we also observed that BAX was high in some of the FAT1 high samples, however the median BAX/BCL2 ratio in the FAT1 high samples was ~2.83-fold higher than in the FAT1 low samples ( Figure 4D). Singular value decomposition (SVD)-calculated principal component analysis (PCA) of same E-GEOD-30784 OSCC cohort further shows that while most (95.8%) OSCC sample were FAT1 high , 4 were FAT1 low (2.4%), and 3 were ambiguous ( Figure 4E). These results indicate that the loss-of-FAT1 function significantly impairs the proliferation of OSCC cells and enhances apoptosis, but spares the normal or non-malignant cells of the oral cavity.  used to calculate principal components. X and Y axis show principal component 1 and principal component 2 that explain 25.4% and 14.3% of the total variance, respectively. Prediction ellipses are such that with probability 0.95, a new observation from the same group will fall inside the ellipse. n = 229 data points. β-actin served as loading control. * p < 0.05. Furthermore, gene expression-based heatmap generated from re-analysis of the A-AFFY-44, AFFY_HG_U133_PLUS_2, E-GEOD-30784 dataset of the gene expression profile of OSCC cohort (n = 229 samples, 54675 genes), originally to identify potential biomarkers for early detection of invasive OSCC using OSCC (n = 167), oral dysplasia (n = 17) and normal oral (n = 45) samples, demonstrated that FAT1 expression was marginal in the non-cancerous control samples, but high FAT1 expression was associated with high PCNA, BIRC5/Survivin, BCL2, and BCL2L1, while low FAT1 expression correlated with high FADD, BAX, PARP1, CASP3, CASP8, and CASP9 in the cancer tissues ( Figure 4D) suggesting an association between reduced FAT1 expression and induction of cell death. For a balanced perspective, we also observed that BAX was high in some of the FAT1 high samples, however the median BAX/BCL2 ratio in the FAT1 high samples was~2.83-fold higher than in the FAT1 low samples ( Figure 4D). Singular value decomposition (SVD)-calculated principal component analysis (PCA) of same E-GEOD-30784 OSCC cohort further shows that while most (95.8%) OSCC sample were FAT1 high , 4 were FAT1 low (2.4%), and 3 were ambiguous ( Figure 4E). These results indicate that the loss-of-FAT1 function significantly impairs the proliferation of OSCC cells and enhances apoptosis, but spares the normal or non-malignant cells of the oral cavity.

Molecular Targeting of FAT1 Re-Sensitizes Cisplatin-Resistant OSCC Cells to Cisplatin through Enhanced Oxidative Stress and Deregulation of LRP/Wnt Signaling
Having demonstrated the implication of enhanced FAT1 expression in OSCC cell proliferation, oncogenicity, metastatic phenotype, and evasion of cell death, we sought to understand if, how and to what extent FAT1 expression is implicated in resistance to cisplatin treatment using adaptive cisplatin resistant (CispR) OSCC cells.
Our results indicated that compared to the wild-type (WT) cells, the HSC-3 CispR and SAS CispR were significantly less responsive to cisplatin treatment, with concentration as high as 40 μM eliciting only a ~38.9% (vs. 68.1% in WT, p < 0.05) or 41.5% (vs. 87.4% in WT, p < 0.01) reduction in the viability of the HSC-3 CispR and SAS CispR, respectively ( Figure 6A). Thereafter, we demonstrated that FAT1 is implicated in this cisplatin resistance phenotype with a 7.4-fold higher mRNA expression in the HSC-3 CispR cells compared to the HSC-3 WT cells (p < 0.01) (Figure 6B), and a 3.7fold (p < 0.01) or 4.8-fold (p < 0.01) upregulated FAT1 protein expression in the SAS CispR or HSC-3 CispR cells compared to their parental/wild-type counterparts ( Figure 6C). Consistent with the data above, and in line with contemporary knowledge implicating increased glutathione (GSH) levels in tumor initiation, disease progression, increased metastasis, and the chemoresistant stem cell-like phenotype of cancerous cells [24][25][26], we showed that the median intracellular GSH level in the FAT1rich HSC-3 CispR cells was 2.38-fold (p < 0.01) fold higher than in the HSC-3 wild-type cells ( Figure  6D). Conversely, shFAT1 induced ~ 60% reduction in the antioxidant GSH protein expression of HSC-3 CispR cells, compared to their wild-type control counterparts ( Figure 6E), and was associated with significant enhancement of cisplatin-induced inhibition of OSCC cell viability, as demonstrated by a 50%, ~80%, and ~96% loss of viability of the HSC-3 CispR shFAT1 cells at concentrations of 10 μM, 25 μM, and 40 μM, respectively ( Figure 6F). Moreover, in a bid to gain some mechanistic insight into the anticancer activity of loss of FAT1 function, we probed molecular components of our predicted FAT1 interactome (Supplementary Figure S1), our results showed that while 5 μM cisplatin downregulated the expression of low-density lipoprotein (LRP)5, p-GSK3β, GSK3β and active β-catenin proteins by 26% (p < 0.05), 31% (p < 0.05), 5% (p = 0.869), and 47% (p < 0.01), respectively, combining shFAT1 with

Molecular Targeting of FAT1 Re-Sensitizes Cisplatin-Resistant OSCC Cells to Cisplatin through Enhanced Oxidative Stress and Deregulation of LRP/Wnt Signaling
Having demonstrated the implication of enhanced FAT1 expression in OSCC cell proliferation, oncogenicity, metastatic phenotype, and evasion of cell death, we sought to understand if, how and to what extent FAT1 expression is implicated in resistance to cisplatin treatment using adaptive cisplatin resistant (CispR) OSCC cells.

Discussion
OSCC remains a principal cause of morbidity and mortality in patients with HNSC, especially considering that in spite of documented decrease in OSCC incidence over the last decade, improvement in overall survival as a primary clinical outcome has remained at a dismal 5% over the last 2 decades, and while the treatment of advanced OSCC requires an integrative multimodal approach, including surgical resection, radiation therapy, and cisplatin-based chemotherapy, surgery remains the treatment modality of choice, howbeit beleaguered with several foci of controversy pertaining to preoperative work-up, management of the primary tumors, and initiation of (neo)adjuvant therapy [1-7]. More so, since the response of tumors to neoadjuvant chemotherapy is predictive of probable complete response to radiation therapy or absence of disease recurrence, strong evidence of chemosensitivity or inherent ability to impair innate or adaptive therapy-resistance is pivotal in deciding type of conservative treatment, thus, necessitating the identification of novel molecular targets or predictors of tumor response to treatment, and/or development of new therapeutic strategies to overcome primary resistance.
Cell-cell adhesion and communication molecules, such as the cadherins are constitutively essential for facilitation of cellular signaling pathways and maintenance of physiological functions. Impaired regulation and resultant aberration in the expression and/or activity of cadherins have been associated, or better put, implicated in several pathological condition, including nephropathies, autoimmune diseases, and malignancies [27][28][29]. "As a member of the Ca 2+ -dependent adhesion super-family, FAT proteins were first described in the 1920s as an inheritable lethal mutant phenotype in Drosophila, consisting of four member proteins, FAT1, FAT2, FAT3, and FAT4, all of which are highly conserved in structure [29]." Against the background of conflicting roles ascribed to FAT1 in different malignancies, this present study demonstrates that: (i) high FAT1 expression drives HNSC, defines high risk population and is associated with poor survival, as indicated by strong positive correlation with disease progression and poor clinical outcome in patients with HNSC. Moreover, the first time to the best of our knowledge we showed that (ii) FAT1 is aberrantly-expressed in OSCC clinical samples and cell lines, exhibiting strong association with an activated LRP5 signalosome, dampened oxidative stress, and poor prognosis in patients with OSCC, while the (iii) loss-of-FAT1 function impairs OSCC cell proliferation, enhances cell death, and attenuates the oncogenicity and metastatic phenotypes of OSCC cells. In addition, we provided pre-clinical evidence that (iv) the molecular targeting of FAT1 re-sensitizes cisplatin-resistant OSCC cells to cisplatin through deregulation of LRP5/WNT2/GSS signaling axis and enhanced oxidative stress.
Findings documented herein complement and are consistent with accruing evidence of the tumor initiating and tumor driving roles of FAT1 in various cancer types, such as the demonstrated role of FAT1 as an upstream master regulator of HIF1α expression and activity, and its ability to enhance the invasiveness of GBM cells under hypoxic conditions [19], while conversely, the small interfering RNA-mediated suppression of FAT1 induced upregulated expression of the tumor-suppressor programmed cell death 4 (PDCD4) gene, with concomitant inhibition of c-Jun phosphorylation and activator protein-1 (AP-1) transcription, resulting in reduced migration and invasiveness of GBM cells [30]. Similarly, there is evidence that relative to its marginal expression in normal human colorectal cancer (CRC) samples, FAT1 is broadly expressed in primary and metastatic CRC cells, mainly accumulating in the plasma membrane, regardless of KRAS and BRAF mutations, and positively correlated with enhanced cell invasiveness, conversely, when targeted by the FAT1-specific monoclonal antibody, mAb198.3, reduction in cancer growth ensues in colon cancer xenograft model, in vivo [31]. Moreover, aside its ability to induce metastasis and progression of hepatocellular carcinoma upon interaction with POU2F1/OCT1 [32], FAT1 expression has been shown to be up-regulated in metastatic gastric cancer, and patients with FAT1 high gastric cancer had worse prognosis [33].
Regardless of the aforementioned consistencies between our present findings and several other published works, we do acknowledge our findings contradict those of Lin et al. [17], suggesting FAT1 acts as a tumor suppressor, with lower FAT1 protein expression bearing significant correlation with lymph node metastasis, lymphovascular permeation, tumor recurrence, and shorter disease-free survival (DFS) in patients with HNSC. We cannot fully explain this contradiction, however, we cautiously attribute this to tumor heterogeneity and/or cohort constitution, based on demonstrated association of high expression of FAT1 with high grade cancerous cells and low FAT1 with low grade cancerous cells in previous works [19,30], as observed in the present work with the use of the HSC-3 and SAS cells which are poorly differentiated human squamous carcinoma of the tongue cells with high lymph node metastasis potential [34][35][36], as well as with our use of OSCC cohorts with more high grade, advanced stage or metastatic cases. As rightly noted by Soussi and Wiman for p53 (TP53) [37], it is not impossible that while the standard criteria for definition of various cancer genes may confine the tumor protein FAT1 to the role of a tumor suppressor, accruing evidence across multiple cancer types with diverse histology, indicate that FAT1 does indeed act as an oncogene.
The present study thus provides rationale to look outside the box of classical and traditional classification of FAT1 as a tumor suppressor, by highlighting various oncogenic properties that make FAT1 a putative therapeutic target that should not be underestimated in OSCC. Consistent with rationalizations on genes with both oncogenic and tumor-suppressor functions by Shen et al. [38], it is also probable that the function-altering mutations in FAT1 are the main driving force in the FAT1-facilitated oncogenesis and cisplatin resistance in OSCC documented herein. Interestingly, like TP53, the mode of FAT1 inactivation is quite unique, compared with most tumor suppressors;~22% and 75% of FAT1 genomic alterations are missense and truncating/nonsense mutations, respectively, both of which facilitate the synthesis of a stable mutant FAT1 protein which accumulates in the plasma membrane and nucleus of the aggressive and/or cisplatin-resistant OSCC cells. Comparatively, this high frequency of amino acid substitution (missense) or premature termination of translation (truncating/nonsense) is highly analogous with various cancer types, regardless of the difference in mutation spectrum [37]. Finally, consistent with Muller's exposition on the nature and causes of gene mutations, based on the classification of mutations hinged on genophenotypic analyses [39], genomic alterations in FAT1 are akin to the 'amorph' or 'hypomorphic' mutation which are more characteristic of so-called 'tumor suppressors', wherein the tumor suppressor function is totally impaired or a partial reduced, resulting in its acquisition of oncogenicity and ability to drive cancer. While it is conceivable that missense and nonsense/truncating mutations resulting in true amorphic variants elicit complete loss of tumor suppressor function, in many instances it is difficult to exclude residual activities that result in heterogeneous hypomorphic variants with context-dependent functional duality as documented in the present study for the protocadherin FAT1. This rationalization based on tumor heterogeneity and mutational status are therapeutically relevant as they go beyond the initial biological function ascribed to FAT1, and can inform discovery or development of novel anti-OSCC therapeutic strategies with high efficacy.
Contextually, our FAT1 findings are, at least in part, corroborated by data indicating that the aberrant expression of the male-specific protocadherin-PC (PCDH-PC) in prostate cancer cells facilitate their acquisition of an apoptosis-evading and hormone therapy-resistant phenotype through enhanced nuclear accumulation of β-catenin and increased WNT-signaling [40]. Also, the observed strong co-expression of FAT1, molecular components of the LRP5 signalosome, namely LRP8, GSK3β, WNT2, β-catenin, casein kinase 1 gamma 1 (CSNK1G1), CSNK1G2, CSNK1G3, AXIN1, caveolin (CAV)1, CAV2, and glutathione (GSS), with weak expression of glutathione-disulfide reductase (GSR) in patients with OSCC, compared with the dysplasia or normal control group, is partially in concordant with reports demonstrating a positive correlation between expression of CAV1 and LRP5-analogous LRP6 in human primary and metastatic prostate cancer tissues, and that the interaction between CAV1 and LRP6 plays an important role in the regulation of Wnt/β-catenin signaling [41].
Moreover, our finding demonstrating that the ablative targeting of FAT1 re-sensitizes cisplatin-resistant OSCC cells to cisplatin through suppressed glutathione (GSH) expression, enhanced oxidative stress and deregulation of LRP/Wnt signaling, become therapeutically relevant when put in the context of contemporary knowledge that while resistance to chemotherapeutics constitute a major impediment to treatment success, cisplatin-resistance can be overcome by inhibition of glutathione S-transferase (GSTs), in vitro and in vivo, where the activity of GST is dependent on the steady production or availability of glutathione (GSH) [25,42,43], and deregulated GSH homeostasis is implicated in the pathogenesis and progression of several human diseases including malignancies, especially as impaired GSH production, or decreased GSH/glutathione disulphide (GSSG) ratio, results in enhanced susceptibility to oxidative stress, which in turn is culpable in cancer progression, where elevated GSH levels augment the antioxidant capacity and resistance to oxidative stress characteristic of many cancerous cells, including OSCC [25]. Thus, consistent with Matés et al's exposition on the implication of oxidative stress for cell proliferation, apoptosis and carcinogenesis [44], the present study expounds the role of FAT1 in the modulation of oral carcinogenesis, oxidative stress and cisplatin resistance via deregulation of the LRP5/WNT2/GSS signaling axis, wherein all molecular factors alluded to in the study are molecular effectors of the crosstalk between WNT/β-catenin and GSH oxidative stress signaling pathways.
Put together, as summarized in the Graphical abstract, this present study uncovers a new role for FAT1 in OSCC oncogenesis and cisplatin resistance, with some mechanistic insight into this oncogenic role, thereby highlighting the functional duality of FAT1 in OSCC. Our findings provide pre-clinical evidence for the therapeutic exploitation of FAT1 ambivalence especially in the treatment of metastatic and/or cisplatin-resistant disease.

OSCC Tissue Specimens
We obtained 21 matched OSCC and adjacent non-tumor tissue samples as kind gift from Dr. Chun-Shu Lin, from the National Defense Medical Centre, Tri-Service General Hospital OSCC tissue bank, following ethical approval for their use from the Institutional Review Board of the Tri-Service General Hospital (TSGHIRB 2-102-05-125). Requirement for patients' signed informed consent was waived because tissue samples were obtained retrospectively from the Tri-Service General Hospital OSCC archive.

Immunohistochemistry
After deparaffinizing the paraffin-embedded OSCC tissue sections in xylene, antigen retrieval was performed by rehydration in decreasing concentration of ethanol and heating in citrate buffer (pH 6.0). After washing, the tissue slides were blocked with 3.0% hydrogen peroxide (H 2 O 2 ) and 10% goat serum and then incubated with primary antibody against FAT1 (1:400, FAT-1 3D7/1, #sc-53283; Santa Cruz Biotechnology Inc.) at 4 • C overnight. Thereafter, the slides were sequentially incubated with appropriate biotinylated secondary antibody, streptavidin-horseradish peroxidase (HRP) complex and diaminobenzidine (DAB). The stained slides were then counterstained with hematoxylin, dehydrated, and mounted, followed by visualization and imaging under light microscope.

shFAT1 Transfection and Establishment of Stable Knockdown Cell Lines
The small hairpin RNA (shRNA) specifically targeting human FAT1 (FAT1 shRNA Plasmid (h); #sc-88872-SH) obtained from Santa Cruz Biotechnology Inc. was transfected into HSC3 or SAS cells grown in 6-well plates to 60% confluence using Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer's instructions. For puromycin selection of stably transfected clones, 48 h after shFAT1 transfection, medium was aspirated and replaced with fresh medium containing 5 µg/mL puromycin and incubated for another 48 h in humidified 5% CO 2 atmosphere incubator at 37 • C. Thereafter, shFAT1-transfected cells were harvested for qRT-PCR or western blot analysis.

Cell Migration and Invasion Assays
Cell migration capability was evaluated using the wound healing assay. Briefly, wild-type or shFAT1-transfected HSC-3 or SAS cells were seeded into 6-well plates (Corning Inc., Corning, NY, USA) containing complete growth media supplemented with 10% FBS, and cultured to ≥98% monolayer confluency. The cell monolayers were scratched with sterile yellow pipette tips to denude the culture wells. Images of cell migration were captured at the 0 and 24 h time-points after denudation, under a microscope with a 10× objective lens, and later analyzed with the NIH ImageJ software (https://imagej.nih.gov/ij/download.html).
For invasion assay, using 24-well plate matrigel Transwell®systems, we seeded 3 × 10 4 cells into the upper chamber of the insert (BD Bioscience, pore size = 8 µm) containing FBS-free media, while the lower chamber contained 10% FBS-supplemented media. After incubation for 24 h, all media were carefully discarded, non-invaded cells in the upper surface of the insert were removed carefully with sterile cotton swipes, while invaded cells on the underside of the membrane were stained with crystal violet dye after fixture with 3.7% formaldehyde, and then the average number of invaded cells were determined under microscope, from at least five non-overlapping visual fields selected randomly.

Statistical Analysis
All experiments were performed at least three times in triplicates, and data presented represent mean ± standard deviation (SD). Comparison between two groups was done using 2-sided Student's t-test, and the 1-way analysis of variance (ANOVA) used for comparison between ≥3 groups. All statistical analyses were performed utilizing the GraphPad Prism version 7.00 for Windows (GraphPad Software Inc., La Jolla, CA, USA). p-value < 0.05 was considered statistically significant.

Conclusions
In conclusion, as summarized in the Graphical abstract, this present study uncovers a new role for FAT1 in OSCC oncogenesis and cisplatin resistance, with some mechanistic insight into this oncogenic role, thereby highlighting the functional duality of FAT1 in OSCC. Our findings provide pre-clinical evidence for the therapeutic exploitation of FAT1 ambivalence especially in the treatment of metastatic and/or cisplatin-resistant disease.
Supplementary Materials: The following are available online at http://www.mdpi.com/2072-6694/11/12/1883/s1, Figure S1: FAT1 by association is implicated in the oncogenicity and metastatic phenotype of OSCC cells. Figure S2: Enhanced FAT1 expression is associated with an activated LRP5 signalosome, dampened oxidative stress, and poor prognosis. Figure S3: FAT1 interacts with LRP5 and GSS, to form FAT1/LRP5/GSS signalosome complex in OSCC cells. Figure S4: High FAT1 expression is positively correlated with nodal involvement and poor clinical outcome in patients with HNSC.