Expression and role of regulator of G‐protein signaling 5 in squamous cell carcinoma of the tongue

Abstract Regulator of G‐protein signaling (RGS) 5 acts as a GTPase‐activating protein to negatively regulate G‐protein signaling. RGS5 is reportedly related to the invasion and metastasis of cancers, such as nonsmall lung cancer and hepatocellular carcinoma. We examined RGS5 expression and its relationship with invasion in squamous cell carcinoma (SCC) of the tongue. For immunohistochemical analysis of RGS5, we used SCC tissues of the tongue obtained from 43 patients. We examined the relationship between RGS5 expression in the deepest point of invasion and clinicopathological features. Because the invasion and metastasis of cancers are related to epithelial‐mesenchymal transition (EMT), we carried out staining for N‐cadherin, vimentin, and E‐cadherin to examine the relationship between EMT and RGS5. RGS5 expression in the deepest point of invasion in SCC of the tongue was observed in 32 cases (75%). Immunohistochemical analysis revealed a significant correlation between RGS5 expression in the aggressive invasion pattern, invasion depth, and lymphovascular invasion. Kaplan–Meier analysis revealed that high RGS5 expression was associated with postoperative early lymph node metastasis. Further, a significant positive correlation was observed between RGS5 and N‐cadherin (P = 0.0003) and vimentin (P < 0.0001). In contrast, E‐cadherin and RGS5 or vimentin were significantly negatively correlated (P < 0.0001–0.005). The findings indicate that RGS5 expression is related to tumor invasion and EMT in SCC of the tongue and that RGS5 may predict postoperative early lymph node metastasis. Therefore, RGS5 may be a useful prognostic biomarker of the surgically resected SCC and a potential target of molecular therapy for treating SCC of the tongue.

growth pattern. Invasion patterns of OSCC were classified as described previously by Li et al. (2013), in which the pattern of invasion was evaluated based on worst pattern of invasion (WPOI) and classified as nonaggressive (WPOI Types 1-3) and aggressive (WPOI Types 4 and 5) patterns. Types 1 and 2 show clear expansive growth pattern. On the other hand, Types 4 and 5 often show infiltrative growth pattern. Type 3 is intermediate between Types 1 and 2 and Types 4 and 5. In cases with infiltrative growth, the tumor cells show poor cell adhesiveness with frequent lymph node metastasis. To examine the effects of infiltrative growth on prognosis, previous studies have focused on the histological characteristics of this type (Li et al., 2013).  (Kalluri & Weinberg, 2009). Therefore, many associations have been detected between factors involved in invasion and factors connecting and activating EMT. Hu et al. (2013) reported that regulator of G-protein signaling (RGS) 5 is related with tumor invasion by inducing EMT of hepatocellular carcinoma (HCC) cells. We also demonstrated that RGS5 expression is closely related with portal vein invasion and intrahepatic metastasis in HCC (Umeno et al., 2018). The RGS family is a group of multifunctional proteins that regulate cellular signaling events downstream of G-protein coupled receptors (Hurst & Hooks, 2009). RGS5 is a member of the RGS family and acts as a GTPase-activating protein (GAP) composed of heterotrimeric G-protein α-subunits that negatively regulate G-protein signaling.
RGS5 expression has been detected in the heart, lung, skeletal muscle, and small intestine and is involved in tumor angiogenesis and gestational hypertension. RGS5 is reportedly related to the invasion and metastasis of cancers, such as HCC and nonsmall lung cancer (Hu et al., 2013;Huang, Song, Wang, Han, & Chen, 2012).
Involvement of Gαq and/or Gαi subunits of G protein in EMT was reported in these cancers, but their results were inconsistent (Hu et al., 2013;Huang et al., 2012). There have been no studies on the role of RGS5 in OSCC. In this study, we investigated EMT-related factors and RGS5.

| Tissue samples
We selected patients with primary tongue cancer who had not undergone preoperative treatments, such as chemotherapy and/or radiotherapy, at Kurume University Hospital between 2011 and 2015 ( Invasion patterns of OSCC were classified previously by Li et al. (2013), in which the pattern of invasion was evaluated based on the WPOI and classified into the following categories: nonaggressive (WPOI Types 1-3) and aggressive (WPOI Types 4 and 5) patterns.
Types 1 and 2 present pushing border and finger-like growth, respectively, and show a clear expansive growth pattern. Types 4 and 5 present small tumor islands (<15 cells per island) and tumor satellites (≥1 mm from the main tumor or next closest satellite), respectively, and often show an infiltrative growth pattern. Type 3 presents large (>15 cells) separated islands and shows a growth pattern that is a mixture of those shown by Types 1 and 2 and Types 4 and 5. On the basis of these findings, we defined Types 1 and 2 as "expansive type" and Types 4 and 5 as "infiltrative type." We defined Type 3 as "intermedi-

| Immunohistochemical analysis
We performed immunohistochemical analysis of paraffin-embedded sections using RGS5, N-cadherin, vimentin, and E-cadherin antibodies. For RGS5, staining intensity of the invasive portion was scored on a 0-3 scale compared with the noncancerous squamous epithelium as follows: 0, intensity in the invasive portion was equal to that in the noncancerous squamous epithelium; 1, intensity in the invasive portion was slightly higher than that in the noncancerous squamous epithelium; 2, moderately higher; and 3, strongly higher. Furthermore, an RGS5 score of 0 or 1 was considered as low expression, whereas an RGS5 score of 2 or 3 was considered as high expression. The staining intensity of N-cadherin was evaluated as follows: Nonstaining was scored as 0, weak staining intensity was 1, medium staining intensity was 2, and equivalent staining intensity was 3, as compared with the staining intensity of nerve cells. The staining intensity of vimentin was evaluated as follows: Staining intensity less than that of mesenchymal cells, such as vascular endothelial cells and fibroblasts, was scored as 0, weak staining intensity was scored as 1, moderate staining intensity was scored as 2, and equivalent staining intensity was scored as 3. The staining intensity of E-cadherin was evaluated as follows: Staining intensity less than that of noncancerous squamous epithelium was scored as 0, weak staining intensity was scored as 1, moderate staining intensity was scored as 2, and equivalent staining intensity was scored as 3. Histological and immunohistochemical analyses were conducted independently by two pathologists (Y. A. and R. K.). If the results were inconsistent, the decisions were made based on discussion and consensus. At least 1,000 tumor cells were counted to evaluate staining intensity. All tests were two-sided, and a P-value <0.05 was considered significant.  There were no deaths among the 43 cases. Two cases had localized recurrence (2 and 48 months), and 16 cases (37.2%) had regional recurrence. The median time to recurrence was 6.8 ± 1.7 months (range 2-24 months).  (Table 2; P = 0.0037). In terms of invasion distance, a significantly larger number of high expression cases showed a depth of invasion ≥4 mm compared with low expression cases (Table 2; P = 0.0463). Additionally, seven of the eight cases with lymphatic invasion showed significantly high expression compared with the low expression cases (Table 2; P = 0.0238). We performed immunostaining of p16 protein, which is a marker of human papillomavirus infection, for 43 cases. Seven and 16 cases with high RGS5 expression were positive and negative, respectively, for p16 expression, whereas three and 17 cases with low RGS5 expression were positive and negative, respectively, for p16 expression. No significant difference was observed in p16 expression between high and low RGS5 expression cases (Table 2;

| Expression pattern of RGS5 in SCC of the tongue
Although there was no significant difference in lymph node metastasis curve between high and low RGS5 expression groups (Figure 4; P = 0.1534), recurrence occurred in all cases by 25 months after surgery.
Lymph node metastasis after surgery means removed when neck failed subsequent to surgery. In addition, lymph node metastasis was observed in 13 of 14 cases (92%) in the high expression group by 10 months after the operation. Patients with N0 were followed up for 3 to 7 years. Low expression of E-cadherin was observed in the invasive portion of infiltrative type (Figure 6o,t), whereas RGS5 and vimentin showed strong expression (Figure 6l,n,q,s), and N-cadherin showed moderate expression (Figure 6m,r).

| Correlation between RGS5 expression and expressions of N-cadherin, vimentin, and E-cadherin
A negative correlation was detected between RGS5 and E-cadherin, between N-cadherin and E-cadherin, and between vimentin and E-cadherin (P = 0.0005, 0.0002, and <0.0001, respectively). A positive

| DISCUSSION
In this study, we found that RGS5 appears to be involved in the aggressive biological features of OSCC, which can be mediated by properties (Huang et al., 2012;Wang et al., 2010). In contrast, high RGS5 expression in HCC was reported to be associated with various aggressive properties, such as vascular invasion, intrahepatic metastasis, and EMT (Hu et al., 2013;Umeno et al., 2018). These reports indicate that the role of RGS5 in carcinoma is organ specific. In this study, we found that RGS5 expression in the invasive portion of tongue cancer was significantly higher in infiltrative type than in expansive type. Given that RGS5 expression is high in the infiltrative type, RGS5 expression may reflect tumor invasiveness and infiltration patterns. Additionally, high expression of RGS5 is associated   Several studies revealed that the tumor invasion pattern is very important for predicting patient outcomes. Locoregional recurrence occurs at an earlier stage in aggressive invasion patterns than in nonaggressive invasion, decreasing disease-specific survival (Li et al., 2013). Furthermore, lymph node metastasis is more common in the aggressive invasion pattern (Chang et al., 2010;Khwaja, Tayaar, Acharya, Bhushan, & Muddapur, 2018;Tanaka, Odajima, Ogi, Ikeda, & Satoh, 2003). In this study, we assessed the invasion patterns as described by Li et al. (2013).   In this study, we found that RGS5 is closely associated with tumor invasion patterns, tumor invasion depth, and lymphatic invasion. In terms of RGS5 expression and postoperative lymph node metastasis, early postoperative lymph node metastasis tended to occur in the RGS5 high expression group, but the lack of a significant difference may be related to the small number of cases (n = 43) and short postoperative observation period. Therefore, RGS5 may be a useful prognostic biomarker of the surgically resected SCC and a potential target of molecular therapy for treating SCC of the tongue.