The significance of Hippo pathway protein expression in oral squamous cell carcinoma

Introduction The Hippo pathway consists of mammalian sterile 20-like kinase 1/2 (MST1/2), large tumor suppressor 1/2 (LATS1/2), and yes-associated protein (YAP)1. Herein, we present the first report on the significance of major Hippo pathway protein expression in oral squamous cell carcinoma (OSCC). Methods The analyses included oral epithelial dysplasia (OED, n = 7), carcinoma in situ (CIS, n = 14), and oral squamous cell carcinoma (OSCC, n = 109). Results Cytoplasmic expression of MST1, LATS1, and LATS2 was low in OED, CIS, and OSCC. The cytoplasmic expression of MST2 was high in OED (5/7 cases), CIS (9/14 cases), and poorly differentiated OSCC (8/8 cases) but was low/lost in a proportion of differentiated OSCC (60/101 cases). The expression of YAP1 was associated with differentiation; low YAP expression was significantly more frequent in well-differentiated OSCC (35/71 cases), compared to moderately and poorly differentiated OSCC (11/38 cases). An infiltrative invasion pattern was associated with a high expression of MST2 and high expression of YAP1. The high expression of YAP1 was associated with features of epithelial-to-mesenchymal transition (EMT), such as the loss of E-cadherin and high expression of vimentin, laminin 5, and Slug. High expression of protein arginine methyltransferase (PRMT) 1 or 5, which positively regulates YAP activity, was associated with the high expression of YAP1 (p < 0.0001). Conclusion Among the major Hippo pathway proteins, MST2 displayed a distinctive expression pattern in a significant proportion of differentiated OSCC, suggesting a possible differential role for MST2 depending on the course of OSCC progression. A high YAP1 expression may indicate aggressive OSCC with EMT via PRMTs at the invasive front.


Introduction
Cancers of the lip and oral cavity are the sixteenth most common neoplasm worldwide, with approximately 378,000 new cases and178,000 deaths in 2020 (1).More than 90% of oral cavity cancer cases are histologically classified as squamous cell carcinoma (SCC) (2).Although advances have been made in the treatment of oral SCC (OSCC), including surgical resection, chemotherapy, and radiotherapy, the 5 years overall survival (OS) rate of OSCC ranges between 24.5% and 50% (3).Comprehensive molecular analyses of head and neck cancers, including OSCC, have identified the amplification of representative target genes, including epidermal growth factor receptor (15%), fibroblast growth factor receptor 1 (10%), and human epidermal growth factor receptor 2 (5%) (4).Therefore, new cancer therapies based on the molecular mechanisms underlying OSCC are required.
The Hippo pathway consists of a large network of proteins that control the end-organ sizes of different tissues by regulating proliferation, cell growth, and apoptosis (5).This pathway comprises a core kinase cascade, starting with the activation of a pair of serine/ threonine kinases, mammalian STE20-like protein kinases (MST1/2), which, in turn, activate another pair of kinases, large tumor suppressor kinases (LATS1/2) (5).LATS1/2 phosphorylates the transcriptional activator yes-associated protein 1 (YAP), inducing its transport from the nucleus to cytoplasm (5).When the Hippo pathway is active, YAP1 is phosphorylated by LATS1/2, resulting in interaction with 14-3-3 and cytoplasmic retention, which also leads to polyubiquitination and degradation of YAP/TAZ (5).
When the Hippo pathway is inactive, YAP1 is present in the nucleus and interacts with transcription factors such as TEAD 1-4.The localization and phosphorylation of YAP1 are often used as measures of Hippo pathway activity (5).Previous studies have reported the overexpression of YAP1 in human cancer and revealed a relationship between higher expression or activity of YAP1 and worse patient prognoses in various tumor entities (6).Furthermore, YAP1 has been shown to drive cancer metastasis (6).In OSCC, the Hippo pathway plays an essential role in disease progression and YAP1 activation, both of which correlate with unfavorable prognosis (7,8).
Protein arginine methyltransferases (PRMT) transfer methyl groups to guanidine nitrogen molecules of arginine residues on their target proteins, which affects gene expression, signal transduction, RNA splicing, and other cellular functions (9).
Recent studies have demonstrated that PRMT 1, a type I PRMT, functions as a positive regulator of YAP1 activity in chondrosarcoma (10), while PRMT5, a type II PRMT, promotes laryngeal cancer cell growth and its invasive capacity through the LATS2 and YAP1 signaling pathways (11).
Although the expression of the major Hippo pathway proteins has been reported in pancreatic cancer (12), their involvement in OSCC remains unclear.In the present study, we examined the expression of major proteins (MST1, MST2, LATS1, LATS2, and YAP1) in the Hippo pathway in OSCC, carcinoma in situ (CIS), and its precursor lesion, oral epithelial dysplasia (OED), and investigated the clinicopathological significance of Hippo pathway protein expression in OSCC.

Patients and tumors
Tumor specimens were obtained from 130 patients who underwent surgery for OSCC without neoadjuvant chemoradiotherapy at the Jichi Medical University Hospital between 2010 and 2015.These cases were derived from our previous study cohort (141 cases) (13).Eleven cases were excluded because the tissue blocks were not eligible for further sectioning.The patients had OED (corresponding to moderate dysplasia) (n = 7), CIS (n = 14), or OSCC (n = 109).All histological diagnoses were made according to the 2005 World Health Organization 2005 criteria (14).Information was collected on the following factors for each case: age, sex, primary site, histological grade, tumor stage, lymph node metastasis, Yamamoto-Kohama (YK) classification (15), lymphovascular invasion, perineural invasion, and patient survival.The patient cohort consisted of 70 males and 60 females, ranging in age from 26 to 89 years (mean age: 63.9 years).The primary sites were the tongue (n = 95), gingiva (n = 18), buccal mucosa (n = 6), and the oral floor (n = 11).Pathological stage was defined according to the TNM classification established by the American Joint Committee on Cancer and the International Union Against Cancer (16), and cut-off values for the depth of invasion (DOI) and tumor thickness were also based on this classification (16).The follow-up period ranged from 81 to 3,441 days (median, 1,520 days).
The morphological pattern of invasion was selected according to the YK classification as follows: grade 1, well-defined borderline; grade 2, less prominent borderline; grade 3, a group of cells with no distinct borderline; grade 4C, diffuse invasion of the cord-like type; grade 4D, diffuse invasion of the widespread type (15).
The study protocol was approved by the Ethics Committee of the Jichi Medical University (approval ID: A15-269).

Immunohistochemistry
Specimens for immunohistochemical evaluation were collected from representative areas of the tumors.Immunohistochemistry was performed using primary antibodies against the following antigens: MST1, MST2, LST1, LST2, YAP1, E-cadherin, vimentin, Slug, laminin 5, PRMT1, and PRMT5.The details of the antibodies used, antigen retrieval methods, and antibody dilutions are listed in Supplementary Table S1.
The immunoperoxidase polymer method (Histofine Simple Stain MAX PO, Nichirei, Tokyo, Japan) was used to detect all primary antibodies.3,3′-Diaminobenzidine tetrahydrochloride was used as a chromogen, and hematoxylin was used as a light counterstain.

Evaluation of immunohistochemical findings
Immunohistochemical staining was independently analyzed by two pathologists (YA and DM) who were blinded to the clinical data obtained through microscopic examinations.LATS1, LATS2, and YAP1 immunoreactivity was evaluated as previously described (17)(18)(19).

Statistical analyses
The chi-squared test was used to examine the relationships between Hippo pathway protein expression, clinicopathological factors, epithelial-to-mesenchymal transition (EMT) marker expression, and PRMT expression.
The relapse-free survival (RFS) and OS rates were calculated according to the Kaplan-Meier method, and differences between groups were assessed using the log-rank test.Univariate and multivariate analyses were performed using the Cox proportional hazards model [described as hazard ratios with 95% confidence intervals (95% CIs), together with p-values].Statistical significance was set at p < 0.05.All statistical analyses were performed using the statistical software Bell Curve for Excel (Social Survey Research Information Co., Ltd.).

Results
Expression of MST1 and MST2 in the normal squamous epithelium, OED, CIS, and SCC We examined the MST1 staining patterns in the normal epithelium (the unaffected area of the tongue in OED cases), OED, and CIS.The cytoplasm of the basal cells of the normal squamous epithelium was weakly positive for MST1 (Figure 1A).In all OED cases, uniformly positive MST1 staining was detected in the cytoplasm of the dysplastic cells (Figure 1B).In all CIS and SCC cases, uniformly positive MST1 staining was detected in the cytoplasm of the neoplastic cells (Figures 1C-E).The intensity of MST1 staining did not significantly differ among normal epithelium, OED, CIS, SCC, and clinicopathological factors.
The cytoplasm of the basal cells of the normal squamous epithelium was positive for MST2 (Figure 1F).In five OED cases, the cytoplasm of the dysplastic cells was strongly positive for MST2 (Figure 1G).In the remaining two cases of OED, uniformly positive MST2 staining was weakly detected in the cytoplasm of dysplastic cells (Table 1).In nine CIS cases, the cytoplasm of the dysplastic cells was strongly positive for MST2 (Figure 1H).In the remaining five cases of CIS, uniformly positive MST2 staining was weakly detected in the cytoplasm of the dysplastic cells (Table 1).MST2 was differentially expressed in SCC, depending on tumor differentiation and mode of invasion.In a significant proportion of differentiated SCC (60/101 of grade 1 and 2 cases), MST2 expression was largely low/absent, while it was highly expressed in all poorly differentiated OSCC (Table 1).Similarly, the proportion of MST2 high cases were significantly low in SCC with well-defined invasive pattern (28/40 of YK-1-3 cases) as compared to SCC with a diffuse invasion pattern (32/69 of YK-4C and 4D cases) (Figures 1I,J and Table 1).
The relationship between MST2 expression and other clinicopathological factors of OSCC is shown in Table 1.The intensity of MST2 staining was not associated with age, sex, tumor location, pathological T (pT) stage, DOI, or lymphovascular invasion (Table 1).

Expression of LATS1 and LATS2 in the normal squamous epithelium, OED, CIS, and SCC
The cytoplasm of basal cells of normal squamous epithelium was weakly positive for LATS1 and LATS2 (Figures 2A,F).Staining was Expression of MST1 and MST2 in the normal epithelium, oral epithelial dysplasia (OED), carcinoma in situ (CIS), and squamous cell carcinoma (SCC).(G) Cytoplasmic expression of MST2 in dysplastic cells that occupy the basal 2/3 of the epithelium.In 5 of 7 OED cases, the MST2 expression was especially high in dysplastic cells in the basal layer.(H) Diffuse high cytoplasmic expression of MST2 was observed in neoplastic cells in the majority of CIS (9 of 14 cases).(I) The MST2 expression was largely lost in neoplastic cells of YK-2 OSCC, with a few MST-positive cells remaining (arrows).This pattern was observed in the large majority of YK-1-3 cases (28 of 40 cases).(J) Diffusely high cytoplasmic expression of MST2 in neoplastic cells of YK-4C OSCC.Such high expression of MST2 was observed in 37 of 69 YK-4C or 4D cases, with statistical difference as compared to YK-1-3 cases (p < 0.0169).Note that inflammatory cells in the stroma were also stained positive for MST1 in (C-E) (encircled).Bar: 100 μm.2B-E).
In the remaining cases of OED, CIS, and SCC, positive LATS1 staining was strongly detected, mainly in the cytoplasm of dysplastic or neoplastic cells (Supplementary Table S2).
In the remaining cases of OED, CIS, and SCC, positive LATS2 staining was strongly detected, mainly in the cytoplasm of dysplastic or neoplastic cells (Supplementary Table S3).

Expression of YAP1 in the normal squamous epithelium, OED, CIS, and SCC
The cytoplasm of the basal cells of the normal squamous epithelium was very weakly positive for YAP1 (Figure 2K).In all OED cases, uniformly positive YAP1 staining was detected in the cytoplasm of the dysplastic cells (Figure 2L).In most CIS and YK-2 SCC cases, uniformly positive YAP1 staining was detected in the cytoplasm of the neoplastic cells (Figures 2M,N).In YK-4C SCC cases, cancer cells often displayed diffuse strong positivity for YAP1 in the nucleus and cytoplasm (Figure 2O).High YAP1 expression was correlated with nuclear and cytoplasmic localization (hereinafter referred to as YAP1 high), while low YAP1 expression was correlated with cytoplasmic localization (hereafter referred to as YAP1 low).These findings are consistent with those of the previous report (20,21).
We then compared the significance of Hippo pathway protein expression in OED, CIS, and OSCC to clarify the differences according to the histological grade (Figure 3).High MST2 expression was detected in 71.4% of OED cases (5/7) and 64.3% of CIS cases (9/14).The concomitant high expression of MST2 and YAP1 in OSCC cases was associated with a higher histological grade, that is, in 17/71 grade 1 cases (23.9%), 11/30 grade 2 cases (33.3%), and 8/8 grade 3 cases (100%).The present results suggest that the co-expression pattern of MST2 high and YAP1 is highly correlated with aggressive histologic features of OSCC.

Relationships among the expression of MST2 and YAP1 and that of EMT markers and PRMT1 and 5
In OSCC, the mode of invasion at the invasive front is a histological predictor of recurrence and prognosis (15).An infiltrative growth pattern was shown to be associated with the loss of the epithelial marker E-cadherin and the upregulated expression of vimentin, Slug, and laminin 5, and was considered to be indicative of EMT (22).We previously reported that the expression of PRMT5, a type II arginine methyltransferase, was associated with EMT in OSCC (23).PRMT1, another arginine methyltransferase, is also associated with EMT (24).PRMT1 and PRMT5 have recently been associated with activation of YAP1 activity (10,11).Therefore, we stained adjacent sections for E-cadherin, vimentin, Slug, laminin 5, PRMT1, and PRMT5 and investigated whether the expression levels of these molecules and/or those of MST2 and YAP1 were associated with the cancer cell invasion pattern.The results are shown in Figure 4 and Supplementary Table S4.

Relationships among the expression of YAP1, MST2, PRMT1, and PRMT5 and the prognosis of OSCC
We performed univariate and multivariate Cox proportional hazards analyses to identify parameters affecting the RFS and OS in patients with OSCC.
Univariate analysis showed that YK-4C, 4D (p = 0.003), and high PRMT5 expression (p = 0.0389) were correlated with OS in patients with OSCC.In multivariate analysis, only YK-4C and 4D correlated with RFS (p = 0.001) (Table 6).The univariate analysis further showed that the MST2 expression was not significantly associated with the OS or RFS in patients with OSCC (Tables 5, 6).
Kaplan-Meier analyses revealed that YAP1 and MST2 expression was not associated with patient prognosis (Figure 5A; Supplementary Figure S1) and that PRMT1 and PRMT5 expression correlated with RFS (both p = 0.035) (Figures 5B,C).
Next, we divided the patients into groups according to the expression of YAP1, PRMT1, and PRMT5: group 1, PRMT1 high and YAP1 high; group 2, PRMT1 low and YAP1 high; group 3, PRMT1 high and YAP1 low; group 4, PRMT1 low and YAP1 low; group 5, PRMT5 high and YAP1 high; group 6, PRMT5 low and YAP1 high; group 7, PRMT5 high and YAP1 low; and group 8, PRMT5 low and YAP1 low.
The RFS was significantly shorter in groups 1 and 2 than in group 4 (p = 0.020 and p = 0.049, respectively) (Figure 5D).RFS was also significantly shorter in group 5 than in groups 6 and 7 (p = 0.038 and 0.043, respectively) (Figure 5E).

Discussion
In this study, we immunohistochemically analyzed the expression of major proteins in the Hippo pathway in OED, CIS, and OSCC.Cytoplasmic expression of MST1, LATS1, and LATS2 was low in OED, CIS, and OSCC.MST2 was highly expressed in the cytoplasm of OED, CIS, and poorly differentiated OSCC but was lost in differentiated OSCC.YAP1 expression was low in well-differentiated OSCC, and high in moderately and poorly differentiated OSCC.An infiltrative invasion pattern was associated with a high expression of MST2 in the cytoplasm and YAP1.High YAP1 expression was associated with loss of E-cadherin and high expression of Slug, vimentin, PRMT1, and PRMT5.Although high YAP1 expression alone did not correlate with OS or RFS, YAP1-high/PRMT1-high and YAP-high/PRMT5-high expression correlated with shorter RFS (both p < 0.05).These results suggest that the high co-expression patterns of YAP1/PRMT1 and YAP1/PRMT5 represent aggressive OSCC with EMT.
The kinase cascade forming the core of the Hippo pathway includes MST1/2 and LATS1/2 protein kinases as well as Salvador family WW domain-containing protein 1 as necessary activators of MST1/2 and MOB1 as downstream effectors (5).These core components lead to phosphorylation of the primary downstream effectors YAP1 and TAZ, thereby preventing their translocation to the nucleus (5).We found that MST1, LATS1, and LATS2 were consistently expressed at low levels in the cytoplasm of basal cells of the normal epithelium, dysplastic cells of the OED, and cancer cells.
These results suggest that MST1, LATS1, and LATS2 have physiological functions in normal cells and that these functions are retained through the progression from OED to CIS and then to OSCC.MST1/2 are kinases that regulate many biological processes including apoptosis, cell cycle progression, and migration/invasion (5).The loss of MST1/MST2 has been shown to results in hyperproliferation and tumorigenesis, whereas the overexpression of  MST1 prevents tumor growth and proliferation and induces apoptosis (5).Collectively, these findings are consistent with the tumorsuppressive roles of these two kinases.In contrast, several recent studies have reported a tumorpromoting role of MST2. Park et al. (25) recently reported that overall survival was longer in estrogen receptor (ER)-positive breast cancer patients with low MST2 expression than in those with high MST2 expression using Kaplan-Meier survival analyses, which implies an important interaction between ER and MST2 in breast cancer.Schirmer et al. (26) demonstrated that MST2 is frequently amplified in prostate cancer and that high MST2 expression is associated with a worse prognosis.Pharmacological inhibition of MST2 reduced the growth of prostate cancer cells, indicating its potential of MST2 as a therapeutic target for the treatment of prostate cancer (26).Chen et al. (27) showed that MST2 promotes gastric carcinogenesis by activating the RAS-MAPK signaling pathway and functions as an independent prognostic marker.
In our analyses, the MST2 expression showed a distinctive pattern in OED, CIS, and the different types of OSCC.MST2 was highly expressed in the cytoplasm of OED and CIS cells.In contrast, MST2 expression was lost in a significant proportion of differentiated OSCC, whereas MST2 was highly expressed in all poorly differentiated OSCC.Similarly, the proportion of MST2 high cases were significantly low in SCC with well-defined invasive pattern, as compared to SCC  The chi-squared test was used to evaluate the associations between PRMT1 expression and PRMT5 expression.Bold, p < 0.05.with a diffuse invasion pattern.Although the MST2 expression was not significantly associated with the prognosis in our cohort, a high expression was associated with a poor prognosis in the TCGA data cohort of head and neck cancer (Supplementary Figures S1, S2).These results suggest that MST2 plays a tumor-suppressive role in a subset of low grade differentiated OSCC and well-defined invasive fronts (YK-1-3), but it may be a tumor-promoting factor in OSCC showing poor differentiation and in a subset of OSCC with an aggressive mode of invasion (YK-4C and 4D).The possibility that MST2 is a tumor-promoting factor in OSCC may be highly speculative, but given the above experimental data (25-27), this possibility may warrant further experimental studies in OSCC.
YAP1 functions as an oncogene; its overexpression inhibits cell contact and promotes cancer cell proliferation and invasion (6).High expression of YAP1 has frequently been reported in various cancers, such as cervical cancer (7), head and neck cancer (7), pancreatic cancer (28), breast cancer (29), intrahepatic cholangiocarcinoma (30), non-small cell lung cancer (31), and liposarcoma (32), and is correlated with poor prognosis, while the loss of YAP1 has potential as a clinical marker in small cell lung cancer (33).Cytoplasmic expression of YAP1 was low in the basal cells of the normal epithelium, dysplastic cells of OED, and cancer cells.These results are largely consistent with previous findings (20) and suggest that YAP1 not only plays a physiological role in basal cells but is also involved in the oncogenesis of OSCC from an early stage.
High/nuclear YAP1 expression was observed in some cases, particularly in those with an advanced pT stage (pT4), an infiltrative invasion pattern (YK-4C or YK-4D), loss of epithelial markers (E-cadherin), and upregulated expression of mesenchymal markers (vimentin, laminin 5, Slug, PRMT1, and PRMT5).Furthermore, high YAP1 expression was associated with worse prognosis.Collectively, these results are somewhat consistent with those of a previous report (8,34,35) and suggest that YAP1 is involved in oncogenesis and  aggressiveness involving EMT in OSCC.EMT is a complex and reversible biological process, in which an epithelial tumor changes its polar, adhesive phenotype to a mesenchymal phenotype characterized by an increase in cell migration and invasion potential, cytoskeleton remodeling, and resistance to apoptosis (22).In OSCC cell lines, microRNA-27a-3p was found to regulate EMT via the YAP1-OCT4/ Sox2 signal axis (36).In non-small lung cancer, silencing of YAP1 inhibit tumorigenesis and EMT and downregulates Slug expression in vivo (37).In colon cancer, upregulated expression of YAP1 was shown to promote EMT and tumor aggressiveness (38).These findings are consistent with those of the present study.
We previously reported that nuclear localization of PRMT1 and PRMT5 is involved in EMT in OSCC (13,23).PRMTs, including type I (e.g., PRMT1) and type II (e.g., PRMT5), transfer methyl groups to the guanidine nitrogen molecules of arginine residues on their target proteins, and this process affects gene expression, signal transduction, RNA splicing, and other cellular functions (9).PRMT1 has been identified as a positive regulator of YAP1 activity in chondrosarcoma (10).In laryngeal cancer, PRMT5 promotes EMT through YAP1 signaling (11).In the present study, the downregulated expression of E-cadherin and high expression of vimentin, laminin 5, Slug, PRMT1, and PRMT5 correlated with high YAP1 expression.These results are consistent with previous findings on chondrosarcoma and laryngeal cancer and suggest the potential of the YAP1/PRMTs axis as a therapeutic target for the treatment of OSCC.
The present study has several limitations.We were unable to examine the expression of all the proteins involved in the Hippo pathway, including their phosphorylated forms.In addition, changes in the expression do not necessarily indicate changes in the function.YAP target genes may negatively regulate YAP and TAZ; thus, it is possible that the increased expression of MST2 in aggressive OSCC may merely indicate a negative feedback loop caused by an elevated YAP function.Furthermore, the molecular pathways by which MST2, YAP1, and PRMTs contribute to the EMT remain unknown.
During EMT, some cancer cells simultaneously exhibit both mesenchymal and epithelial characteristics, a phenomenon termed partial EMT (pEMT).pEMT is a plastic state in which cells co-express epithelial and mesenchymal markers.In squamous cell carcinoma (SCC), pEMT is regulated via multiple pathways, including the Hippo (YAP being the effector molecule), Notch, and TGF-β pathways and the microenvironment (39).
Cheung et al. (40) showed that the loss of MST1/2 decreased the phosphorylation of MOB1 and YAP1, but did not affect that of LATS1/2, whereas Lu et al. (41) reported that the levels of phosphorylated YAP1 and LATS1/2 were reduced in MST1/2 double knockout livers.However, the precise molecular mechanisms of each Hippo pathway protein interaction in OSCC are yet to be elucidated.Hasegawa et al. (42) showed that YAP1 signaling induces Piezo-type mechanosensitive ion channel component 1 to promote the proliferation of OSCC.
Saladi et al. (43) showed that p63 interacts with co-amplified ACTL6A, encoding an SWI/SNF subunit linked to stem cell and progenitor cell functions.Chromatin remodeling by the ACTL6A/p63 complex suppresses WWC1 to activate YAP1 and promotes tumorigenesis (43).Ahmad et al. (44) showed that DSG3 and YAP activity exhibit mutually exclusive dependence, and desmoglein-3 acts as an important component of the Hippo network in the control of contact inhibition of cell locomotion.
A recent paper proposed that cancer is a multidimensional spatiotemporal "unity of ecology and evolution." Therefore, the initiation and progression of human cancer can be considered an ecological disease (45).This viewpoint may be helpful for understanding tumor progression from the normal epithelium to OSCC.
In conclusion, a high co-expression pattern of YAP1 and PRMTs indicates aggressive OSCC with EMT at the invasive front.Further studies are needed to elucidate the precise molecular mechanisms by which YAP1 promotes EMT via PRMTs in OSCC.

( A )
Cytoplasmic expression of MST1 was weakly observed in the basal and parabasal cell layers of the normal epithelium (arrows).(B) Cytoplasmic expression of MST1 was weakly observed in dysplastic cells that occupy the basal 2/3 of the epithelium.(C) Cytoplasmic expression of MST1 in neoplastic cells.(D) Cytoplasmic expression was observed MST1 in YK-2, OSCC.(E) Cytoplasmic expression of MST1 in YK-4C.(F) Cytoplasmic expression of MST2 in the basal and parabasal cell layers of the normal epithelium.The MST2 expression was especially high in the basal layer (arrows).

FIGURE 2
FIGURE 2Expression of LATS1, LATS2, and YAP1 in the normal epithelium, oral epithelial dysplasia (OED), carcinoma in situ (CIS), and squamous cell carcinoma (SCC).(A) Weak expression of LATS1 mainly in the cytoplasm of the basal and parabasal cells of the normal epithelium (arrows).(B) Weak expression of LATS1 mainly in the cytoplasm of dysplastic cells of most OED (5 of 7 cases).(C) Weak expression of LATS1 mainly in the cytoplasm of neoplastic cells of most CIS (9 of 14 cases).(D) Weak expression of LATS 1 mainly in the cytoplasm of neoplastic cells of most YK-2 OSCC (34 of 40 cases).(E) Weak expression of LATS1 mainly in the cytoplasm of neoplastic cells of most YK-4C OSCC (60 of 69 cases).(F) Weak expression of LATS2 mainly in the cytoplasm of the basal and parabasal cells of the normal epithelium (arrows).(G) Weak expression of LATS2 mainly in the cytoplasm of dysplastic cells of OED (4 of 7 cases).(H) Weak expression of LATS2 mainly in the cytoplasm of neoplastic cells of CIS (9 of 14 cases).(I) Weak expression of LATS2 mainly in the cytoplasm of neoplastic cells of most YK-2 OSCC (38 of 40 cases).(J) Weak expression of LATS2 mainly in the cytoplasm of most YK-4C OSCC (63 of 69 cases).(K) Vary faint cytoplasmic expression of YAP1 in the basal cell layer of the normal epithelium.(L) Weak expression of YAP1 mainly in the cytoplasm of dysplastic cells of OED (6 of 7 cases).(M) Weak expression of YAP1 mainly in the cytoplasm of neoplastic cells of most CIS (12 of 14 cases).(N) Weak expression of YAP1 mainly in the cytoplasm of neoplastic cells of most YK-2 OSCC (32 of 40 cases).(O) High expression of YAP1 in both nuclei and cytoplasms of neoplastic cells of most YK-4C OSCC (55 of 69), with statistical difference as compared to YK-1-3 OSCC (8 of 40) (<0.0001).Note that stromal cells (mainly inflammatory cells and fibroblasts) were also stained positive for LATS1 (B-D), LATS2 (G-J), and YAP1 (N) (encircled).Bar: 100 μm.

FIGURE 3
FIGURE 3The comparison of the semiquantitative immunohistochemical expression of each Hippo pathway protein in OSCC, carcinoma in situ (CIS), and oral epithelial dysplasia (OED).A red-filled box indicates a high expression; an open box indicates a low expression.The concomitant high expression of MST2 and YAP1 was associated with a higher histological grade.

TABLE 1
Relationships between the localization of MST2 and clinicopathological factors.

TABLE 2
Relationships between YAP1 localization and clinicopathological factors.

TABLE 4
Relationship between the expression of PRMT1 and PRMT5.

TABLE 5
Univariate and multivariate analyses of relapse-free survival in patients.

TABLE 6
Univariate and multivariate analyses of overall survival in patients.