Epstein–Barr Virus LMP1 Induces Soluble PD-L1 in Nasopharyngeal Carcinoma

Nasopharyngeal carcinoma (NPC) is an Epstein–Barr virus (EBV)-associated malignancy. The principal oncogene of EBV, latent membrane protein 1 (LMP1), induces the expression of programmed death-ligand 1 (PD-L1), which is an immunosuppressive transmembrane protein and a promising therapeutic target for various malignancies. Recent studies have revealed an association between the level of soluble PD-L1 (sPD-L1) and disease progression. However, the role of sPD-L1 in NPC or its relevance to LMP1 has not been elucidated. This study aimed to examine whether LMP1 induces sPD-L1 in vitro and analyze the clinical relevance of LMP1, PD-L1, and sPD-L1 in NPC patients. Analysis of nasopharyngeal cell lines revealed that LMP1 induces both cellular PD-L1 and sPD-L1. Analysis of biopsy specimens from 32 NPC patients revealed that LMP1 expression was significantly correlated with PD-L1 expression. Finally, the serum sPD-L1 level in NPC patients was higher than that in the controls. Moreover, the sPD-L1 level in the advanced stage was higher than that in the early stage. However, LMP1 expression, PD-L1 expression, and sPD-L1 levels were not associated with prognosis. These results suggest that LMP1 induces both sPD-L1 and PD-L1, which are associated with NPC progression.


Introduction
Epstein-Barr virus (EBV) is strongly associated with the etiology of nasopharyngeal carcinoma (NPC) [1,2]. Latent membrane protein 1 (LMP1), a principal oncogene of EBV [3], contributes to the promotion of invasion and metastasis as well as initial oncogenesis of NPC [4]. Moreover, LMP1 promotes escape from various immune recognition processes, especially cytotoxic T lymphocyte response that kills tumor cells and virus-infected cells [5]. Programmed death ligand 1 (PD-L1), a transmembrane glycoprotein, binds to PD-1 on T lymphocytes and suppresses T lymphocyte activation [6]. Tumor cells express PD-L1 and utilize the PD-L1/PD-1 checkpoint for immune evasion [7]. Fang et al. reported that LMP1 induced cellular PD-L1 expression in vitro, and their data confirmed that blocking both LMP1 and PD-L1 expression is a potential therapeutic approach for NPC [8]. Moreover, Yang et al. showed that a high level of soluble PD-L1 (sPD-L1) is significantly associated with advanced-stage NPC [9]. In addition, sPD-L1 is a prognostic biomarker of malignant melanoma [10]. Asanuma et al. reported that high sPD-L1 levels are associated with a poor prognosis in soft tissue sarcomas, and their findings suggested that sPD-L1 is released (M0897, 1:200, Dako, Glostrup, Denmark), and α-tubulin (T9026, 1:5000, Sigma Aldrich, St. Louis, MO, USA). After washing with TBST (Santa Cruz Biotechnology, Santa Cruz, CA, USA), the membranes were exposed to horseradish peroxidase-conjugated secondary antibodies (Bio-Rad Laboratories, Hercules, CA, USA). The signals were detected using enhanced chemiluminescence reagent (GE Healthcare, Tokyo, Japan).

Serum Samples
Serum samples were obtained from 32 patients with primary NPC and 10 controls. Because 3 of the 35 patients with NPC had double cancer (one patient with stomach cancer, one with lung cancer, and one with prostate cancer), we excluded them from this analysis. All patients did not receive any treatments before the serum collection and all samples were not derived from a tumor recurrence. Control samples were obtained from 10 patients who did not have any pathologies that may affect sPD-L1 levels, including carcinomas, infectious disease, and autoimmune diseases at diagnosis. The serum was frozen at −80 • C before further processing.

sPD-L1 Detection Using ELISA
For detecting sPD-L1 in serum samples, PD-L1 ELISA kit (KE00074, Proteintech, Manchester, UK) was used. Serum samples of 32 patients with NPC and 10 controls were analyzed according to the manufacturer's protocol.

Statistical Analysis
SPSS statistics package ver. 23 (IBM, New York, NY, USA) was used for data analysis. Mann-Whitney U-test was used to analyze the clinical characteristics of the patients. The Spearman rank correlation coefficient was used to analyze the association among LMP1, PD-L1, and sPD-L1. Kaplan-Meier curves were used to examine the overall survival (OS) and progression-free survival (PFS) according to LMP1, PD-L1, and sPD-L1 levels in patients with NPC. We excluded 3 patients who had a double cancer from survival analyses. LMP1 or PD-L1 expression score of 10% and more was considered positive [13], and sPD-L1 levels of 0.1 ng/dL and higher were considered positive [14]. The differences in survival were analyzed using the log-rank test. Cox proportional hazard regression analysis was used to assess univariate and multivariate analysis results of clinicopathological factors such as sex, smoking status, alcohol consumption, TNM classification, clinical-stage, TPS status, CPS status, LMP1 status, and sPD-L1 status. The differences were considered statistically significant when the p-value was less than 0.05.

LMP1 Induces Both Cellular PD-L1 Protein and sPD-L1 in Nasopharyngeal Cells
Fang et al. reported that LMP1 induces cellular PD-L1 expression in NPC cell lines [8]. However, whether LMP1 also induces sPD-L1 remains unknown. Thus, to clarify the association between LMP1 and PD-L1, we examined the cell extracts of the non-malignant EBV-negative nasopharyngeal cell line NP69T. The expression of PD-L1 protein increased in LMP1-expressing NP69T cells (NP69T-LMP1) compared with that in NP69T cells ( Figure 1). This result indicates that LMP1 induced the expression of the PD-L1 protein. Finally, to investigate whether the PD-L1 protein was released from nasopharyngeal cells, we performed western blotting with a culture medium. PD-L1 protein was detected in the concentrated medium of NP69-LMP1 cells ( Figure 1). Therefore, LMP1 induces the sPD-L1 and PD-L1 in nasopharyngeal cells in vitro.

Expression of PD-L1 and LMP1 are Correlated in NPC Tissues Samples
The results of the cell culture model analysis suggested that LMP1 promoted both cellular PD-L1 expression and sPD-L1 levels. Therefore, we examined LMP1 and PD-L1 expression in the 32 NPC tissue samples using immunohistochemical analysis. The characteristics of the patients are shown in Table 1. There were 32 men and 3 women, and 8 patients had stages I-II NPC and 27 had stages III-IV NPC. LMP1 expression was observed mainly in the tumor cell membrane, as indicated by brown staining (Figure 2A). In contrast, PD-L1 expression was positive in both tumor and inflammatory cell membranes ( Figure 2B). Next, we performed dual fluorescence immunostaining and assessed the expression of LMP1 and PD-L1 in the same tissue sample. There were several cells that expressed both LMP1 and PD-L1 in the NPC tissue sample ( Figure 2F). LMP1 expression score was significantly associated with PD-L1 TPS (p = 0.023, r = 0.402, Figure 3A). These results support our hypothesis that LMP1 promotes the expression of PD-L1 in NPC tissue samples, as well as in cell culture in vivo.

Expression of PD-L1 and LMP1 Are Correlated in NPC Tissues Samples
The results of the cell culture model analysis suggested that LMP1 promoted both cellular PD-L1 expression and sPD-L1 levels. Therefore, we examined LMP1 and PD-L1 expression in the 32 NPC tissue samples using immunohistochemical analysis. The characteristics of the patients are shown in Table 1. There were 32 men and 3 women, and 8 patients had stages I-II NPC and 27 had stages III-IV NPC. LMP1 expression was observed mainly in the tumor cell membrane, as indicated by brown staining (Figure 2A). In contrast, PD-L1 expression was positive in both tumor and inflammatory cell membranes ( Figure 2B). Next, we performed dual fluorescence immunostaining and assessed the expression of LMP1 and PD-L1 in the same tissue sample. There were several cells that expressed both LMP1 and PD-L1 in the NPC tissue sample ( Figure 2F). LMP1 expression score was significantly associated with PD-L1 TPS (p = 0.023, r = 0.402, Figure 3A). These results support our hypothesis that LMP1 promotes the expression of PD-L1 in NPC tissue samples, as well as in cell culture in vivo.    34.3 ± 28.5 OS (mean month ± SD) 49.5 ± 31 NPC, nasopharyngeal carcinoma; SD, standard deviation; PFS, progression-free survival; OS, overall survival.

Relationship between LMP1 and PD-L1 Expression in NPC Tissue Samples
We next assessed the characteristics of the patients with NPC and the LMP1 and PD-L1 status in the tissue samples. The characteristics of patients based on LMP1 and PD-L1 expression (PD-L1 TPS and CPS) are shown in Table 2. There was a significant increase in PD-L1 TPS in advanced-stage patients (p = 0.03) compared to that in early-stage patients. In addition, LMP1 expression was higher in females (p = 0.009) than in males. This result suggests that the number of PD-L1-positive tumor cells was associated with the NPC stage. Moreover, LMP1 expression differed between males and females.

Relationship between LMP1 and PD-L1 Expression in NPC Tissue Samples
We next assessed the characteristics of the patients with NPC and the LMP1 and PD-L1 status in the tissue samples. The characteristics of patients based on LMP1 and PD-L1 expression (PD-L1 TPS and CPS) are shown in Table 2. There was a significant increase in PD-L1 TPS in advanced-stage patients (p = 0.03) compared to that in early-stage patients. In addition, LMP1 expression was higher in females (p = 0.009) than in males. This result suggests that the number of PD-L1-positive tumor cells was associated with the NPC stage. Moreover, LMP1 expression differed between males and females.

Elevated Serum sPD-L1 Level in Patients with NPC
Because PD-L1 expression was elevated in NPC tissues, we next investigated the sPD-L1 levels in serum samples. sPD-L1 levels significantly differed between the controls and patients with NPC (p = 0.031, Figure 4). In addition, we found a significant increase in sPD-L1 levels in NPC patients with advanced tumor stage (p = 0.036, Table 3) compared with that in patients with the early tumor stage. However, sPD-L1 level was not associated with LMP1 expression score (p = 0.49, r = 0.13), PD-L1 TPS (p = 0.92, r = 0.02), and PD-L1 CPS (p = 0.88, r = 0.03) in patients with NPC. Taken together, these results showed that sPD-L1 levels were elevated in patients with NPC. Moreover, high sPD-L1 levels were associated with advanced tumor stage in NPC.  There was a significant difference in sPD-L1 levels between the controls and the patients with NPC (p = 0.031). p-value was calculated using the Mann-Whitney U-test. NPC: nasopharyngeal carcinoma, sPD-L1: soluble PD-L1. Table 3. Characteristics of 32 patients with NPC according to sPD-L1 levels.

Expression of LMP1, PD-L1, and sPD-L1 Level Was Not Related to Prognosis of Patients with NPC
Finally, we examined whether LMP1 and PD-L1 expression served as a prognostic factor in patients with NPC because both PD-L1 expression and sPD-L1 levels are prognostic factors in other malignancies [10,15,16]. PFS of LMP1-positive patients with NPC (LMP1 expression score: more than 10%) was not significantly different from that of LMP1-negative patients (p = 0.965, Figure 5A). Similarly, PD-L1 TPS and CPS and sPD-L1 status (positive or negative) were not prognostic factors (p = 0.178, p = 0.089, p = 0.657, respectively, Figure 5B-D). Univariate analysis showed that smoking and metastasis were significantly associated with PFS, and multivariate analysis showed that both smoking and metastasis were independent variables associated with PFS (Table 4). In addition, there was no significant difference in OS ( Figure S1). Univariate analysis showed that sex was significantly associated with OS (Table S1). These results suggest that neither PD-L1 expression nor sPD-L1 level was a prognostic factor for patients with NPC.
Microorganisms 2021, 9, x FOR PEER REVIEW 9 of 13 LMP1-negative patients (p = 0.965, Figure 5A). Similarly, PD-L1 TPS and CPS and sPD-L1 status (positive or negative) were not prognostic factors (p = 0.178, p = 0.089, p = 0.657, respectively, Figure 5B-D). Univariate analysis showed that smoking and metastasis were significantly associated with PFS, and multivariate analysis showed that both smoking and metastasis were independent variables associated with PFS (Table 4). In addition, there was no significant difference in OS ( Figure S1). Univariate analysis showed that sex was significantly associated with OS (Table S1). These results suggest that neither PD-L1 expression nor sPD-L1 level was a prognostic factor for patients with NPC.

Discussion
NPC is a virus-related cancer. In particular, the EBV principal oncogene LMP1 plays an important role in the pathogenesis of NPC [17,18]. Some studies have reported that PD-L1 expression is upregulated in virus-associated malignancies, such as EBV-related Hodgkin's lymphoma and human papilloma-virus related head and neck cancers [19,20]. These studies suggested that induction of PD-L1 expression was associated with immune tolerance in virus-associated tumors and blocking PD-L1 expression was a potential therapeutic approach. There are few treatment options for recurrent NPC [21]. Therefore, blocking PD-1/PD-L1 expression has attracted attention as a new therapeutic approach for NPC [22]. Ma et al. reported that PD-1 blockade has clinical significance for the treatment of recurrent NPC after heavy pretreatment [23]. In addition, a previous study found that LMP1 upregulated PD-L1 expression in NPC cell lines [8]. A similar result was found in a study on natural killer/T cell lymphoma, which showed that LMP1 mediated PD-L1 expression through the MAPK/NF-κB pathway [24]. In this study, we found that there was a slight correlation between LMP1 and PD-L1 expression ( Figure 2). Thus, our results were consistent with previous results and confirmed that LMP1 might contribute to the induction of PD-L1 expression in NPC.
In contrast, sPD-L1 has received much attention because some studies revealed its clinical significance in malignancies [25][26][27]. sPD-L1 suppresses T cell activation and induces T cell apoptosis [10,28,29]. These analyses indicated that tumor cells escape the host immune system via circulating sPD-L1 levels. We found that sPD-L1 level was induced by LMP1 in the nasopharyngeal cell lines (Figure 1). To our knowledge, this is the first study to demonstrate an association between LMP1 and sPD-L1 levels. Our findings suggested that both sPD-L1 and PD-L1, which are induced by LMP1, inhibited T cell activation. Similar to most membrane-bound protein soluble forms, an upregulation mechanism of sPD-L1 involves shedding from the surface membrane of PD-L1-positive cells. Chen et al. suggested that matrix metalloproteinase (MMP) produces sPD-L1 from the surface membrane of PD-L1-positive cells [30]. Previously, we reported that LMP1 enhanced the expression of MMP1, MMP3, and MMP9 in NPC [31,32]. Taken together, LMP1 may be associated with the production of sPD-L1 by inducing MMP in NPC. Alternatively, sPD-L1 is related to alternatively spliced variants of PD-L1 [10,33].
Recent studies have reported that PD-L1 expression is associated with factors affecting malignant potential, including immune evasion [34][35][36]. Moreover, Li et al. showed that serum levels of sPD-L1 increase with tumor stage in breast cancer [37]. Similarly, in our study, NPC patients with advanced tumor stage showed elevated sPD-L1 levels ( Table 3). In addition, Chang et al. reported that high PD-L1 expression in lung cancers was associated with an advanced clinical stage [38]. Our current results also revealed that higher PD-L1 expression was associated with an advanced stage (PD-L1 TPS, Table 2). These results suggest that the aggressive behavior of certain malignancies could be partially related to high PD-L1 expression as well as high sPD-L1 levels. However, unlike previous reports [24,39], serum sPD-L1 levels were not significantly associated with PD-L1 expression in the tissue samples in our analysis. Some studies reported sPD-L1 as a candidate serum biomarker [10,40,41]. Similarly, our findings showed that pretreatment sPD-L1 levels were higher in NPC patients than in control patients (Figure 4). Unlike previous studies, we found no evidence to support the hypothesis that sPD-L1 levels, LMP1 expression, and PD-L1 expression were associated with prognosis, and only four patients died in this study. Some of the reasons were differences in techniques and protocols between previous studies and our analysis and heterogeneity of PD-L1 expression using immunohistochemistry [42]. In addition, there may be some factors that affect the dissociation of sPD-L1 and PD-L1, such as endemic NPC and non-endemic NPC. Important limitations of this analysis were that the sample size was small and some cases had a short follow-up time after treatment. Future studies should focus on the predictive value of sPD-L1 and PD-L1 for determining the efficacy of immune checkpoint inhibitors required.
In summary, we found that serum sPD-L1 was detectable and there was a significant difference between sPD-L1 levels and tumor stage in NPC. In contrast, sPD-L1 levels were not associated with the outcome of primary treatment because there were few events of death as per OS. However, about 50% of recurrent patients with NPC have distant metastasis, and it is often difficult to obtain a biopsy tumor tissue sample and evaluate PD-L1 expression [43]. sPD-L1 levels could replace tumor PD-L1 expression as a surrogate biomarker. Therefore, further assessment is needed to determine whether sPD-L1 could predict the outcome of treatment with immune checkpoint inhibitors such as PD-1 inhibitors.

Conclusions
The findings of this study will be useful for a better understanding of the pathophysiology of NPC. Our findings suggest that LMP1 induces sPD-L1 and PD-L1 expression. Further analysis is necessary to apply our findings regarding potential biomarkers in NPC to clinical practice.