Prognostic Value of PLXND1 and TGF-β1 Coexpression and Its Correlation With Immune Infiltrates in Hepatocellular Carcinoma

Hepatocellular carcinoma (HCC) is the most common primary liver malignancy with no curative treatments. Plexin D1 (PLXND1) is a cellular receptor whose functions have been explored in several human cancers; however, the critical roles of PLXND1 in HCC have rarely been probed. Therefore, the present study attempted to elucidate the expression pattern, prognostic significance, and potential roles of PLXND1 in HCC. We found that PLXND1 expression in HCC tissues was significantly higher compared with normal liver tissue from Gene Expression Profiling Interactive Analysis (GEPIA) and Integrative Molecular Database of Hepatocellular Carcinoma (HCCDB) databases. This result was further validated by immunohistochemistry staining (IHC) using tissue microarrays, which contained 216 HCC cases collected from our hospital. Additionally, PLXND1 expression showed a significant correlation with several clinical characteristics, including tumor grade and tumor hemorrhage (TH). Moreover, TISIDB and GEPIA databases were used to investigate the roles of PLXND1 in tumor-immune system interactions in HCC. As an immunoinhibitor, transforming growth factor-beta (TGF-β1) displayed the greatest correlations with PLXND1 in HCC. Finally, Kaplan-Meier curves and Cox analysis were conducted to further examine the potential clinical value of PLXND1 in HCC. We described a subclassification of HCC based on PLXND1 and TGF-β1 expression, which could be used to predict clinical outcomes and patient prognosis. Taken together, the results of this study indicate that PLXND1 might be a promising prognostic biomarker and potential therapeutic target in HCC.


INTRODUCTION
Hepatocellular carcinoma (HCC) is the most common primary liver malignancy and the fourth leading cause of cancer death worldwide (1)(2)(3). More than 70% of patients are diagnosed at an advanced stage with a dismal prognosis. Surgical resection and chemotherapy are the conventional treatment for HCC, but the mortality rate of this cancer remains high (4). In recent years, immunotherapy is increasingly considered as one of the most promising approaches for identifying drug targets for therapeutic interventions (5,6). The precise estimation of patient prognosis plays an important and irreplaceable role on the treatment decision of HCC patients. Hence, the exploration and identification of novel therapeutic biomarkers and reliable prognostic biomarkers are urgently required.
Plexins are a family of transmembrane proteins that serve as cellular receptors for semaphorins and have a range of critical functions in vascular patterning, axonal guidance, cell proliferation, migration, and immune cell regulation (7)(8)(9). Plexin D1 (PLXND1), a newly identified member of the plexin family of molecules, has been reported to be dysregulated in several cancers (7,10). Emerging data have shown that PLXND1 could function as a tumor suppressor or a classical oncogene. Serving as a tumor suppressor, PLXND1 was reported to promote apoptosis in the absence of semaphorin 3E in breast cancer (10). Conversely, as an oncogene, PLXND1 could promote tumor development and progression by aiding in tumor metastasis and epithelial mesenchymal transition (EMT) (7,11,12). In addition, PLXND1 was found to be highly correlated with tumor hemorrhage (TH) in this study and plays a critical role in the tumor immune microenvironment (13,14). Mumblat Yelena et al. showed that PLXND1 was involved in inducing the collapse of the cytoskeleton of lymphatic endothelial cells (LEC), and impacted tumor lymphangiogenesis and metastasis (13): however, until recently, little was known about the abnormal expression and function of PLXND1 in HCC. Thus, the detailed function of PLXND1 in the tumorigenesis and immune infiltration of HCC warrants further investigation. Due to high mortality rate, limited treatment effects and poor prognosis of HCC, a novel therapeutic strategy with different mechanisms from conventional treatments are urgently needed (15). Immunotherapy has recently emerged as one of the most promising frontiers of cancer treatment, especially the development of immune checkpoint inhibitors, and has yielded promising therapeutic outcomes in many types of solid tumors, including HCC (16,17). In spite of these advances, fundamental differences in patient responsiveness to these treatments are inexplicable (18). Hence, comprehensive research on the immune microenvironment of HCC needs to be further investigated.
Prior research on PLXND1 in cancer is limited to cancer types apart from liver cancer. For liver cancer, neither abnormal PLXND1 expression nor its correlation with the clinicopathological parameters of HCC have been reported. In this study, we first reported a higher expression of PLXND1 in HCC and its correlation with clinicopathological features. Additionally, PLXND1 expression was related with tumor-infiltrating lymphocytes (TILs) and immunoinhibitors, and TGF-b1 displayed the greatest correlations with PLXND1 expression in HCC. Then, Kaplan-Meier curves and Cox analysis were conducted to further examine the potential clinical significance of PLXND1 in HCC. Taken together, these results highlight the critical roles of PLXND1 in tumorigenesis, the tumor immune microenvironment, and clinical opportunities for cancer research.

Patient Samples
Written informed consent was obtained from all recruited patients prior to the study. All procedures in this study were approved by the Ethics Review Committee of Xiangya Hospital of Central South University (N201303033). The 216 paraffinembedded archived liver cancer tissues used in this study were histopathologically and clinically diagnosed at the Xiangya Hospital between 2011 and 2013. No patients in this study received chemotherapy before surgery.

Immunohistochemistry
IHC staining was performed on tissue microarrays as previously described (19). The following antibodies were used: anti-PLXND1 (ab28762; Abcam, Cambridge, MA) and anti-TGF-b1 (ab25121; Abcam, Cambridge, MA). Expression of PLXND1 were semiquantitatively classified according to the immunoreactive Hscore (HS; range 0-300). Subsequently, stratification scoring was conducted according to H-score as follow: HS < 80, scored as 0; 80 < HS <120, scored as 1; 120 < HS < 200, scored as 2; HS > 200, scored as 3. If the score was equal to or greater than 2, the tumor was considered to have high PLXND1 expression; otherwise, low PLXND1 expression was classified. The expression levels of target proteins in tissue were examined by two independent pathologists blinded to the clinical characteristics of the patients.

Bioinformatics Analysis
Gene Expression Profiling Interactive Analysis (GEPIA, http:// gepia.cancer-pku.cn/) is an interactive web server for cancer and normal gene expression profiling and interactive analyses (20). In this database, we compared the PLXND1 mRNA expression between tumors and their paired normal tissues. Additionally, GEPIA enabled us to analyze the link between PLXND1 expression and gene markers of different types of tumor infiltrating immune cells.
Integrative Molecular Database of Hepatocellular Carcinoma (HCCDB, http://lifeome.net/database/hccdb) was developed at Tsinghua University and was used to verify the repeatability of identification of HCC gene expression patterns and the annotation of transcripts (21). We use this tool to analyze the expression of PLXND1. TISIDB (http://cis.hku.hk/TISIDB) is an integrated repository portal which integrates multiple types of data resources in oncoimmunology (22). By using the multidimensional profiling data for 30 cancer types from TCGA, TISIDB pre-calculated the relation between the immune features (e.g., lymphocytes, immunomodulator) and expression, copy number, methylation, and mutation of any gene. Therefore, users can check "lymphocyte," "immunomodulator," or "chemokine" tab to evaluate whether the selected gene might regulate immune features across cancers. The heatmap graphic outputs which show Spearman correlations between expression of the selected genes and immune features across human cancers are implemented with the heatmap module of ECharts (https://echarts.apache.org/). Visualization scatter plot for a Spearman correlation between the selected genes and a specific member of immune features in a specific cancer type can be generated through geom_point function of R package ggplot (R script, https://www.r-project.org/). In this database, we analyzed the correlation between PLXND1 expression and TILs and immunoinhibitors in HCC.

Statistical Analysis
The statistical analyses were performed with SPSS 23.0 for Mac (SPSS Inc., Chicago, IL). Quantitative values are presented as the mean ± SEM. Paired t tests and Student's tests were used for paired and unpaired continuous data, respectively. The c2 test was used for categorical data. The Mann-Whitney test was applied to analyze correlations between PLXND1 expression and clinicopathologic characteristics in HCC. Survival analysis was performed by Kaplan-Meier analysis and log-rank test. A cox proportional hazards regression model was used for univariate and multivariate survival analyses. Significant changes were indicated with asterisks: *= (P < 0.05); **= (P < 0.01); *** = (P < 0.001).

Elevated Expression of PLXND1 Messenger RNA in Hepatocellular Carcinoma Tissues
For liver cancer, neither abnormal PLXND1 expression nor its association with clinicopathological parameters of HCC has been reported. Here, we investigated the differential expression of PLXND1 using the GEPIA and HCCDB databases. In GEPIA, PLXND1 mRNA expression between tumors and their paired normal tissues across 32 TCGA cancer types were compared. Significantly differential expression was found in 12 cancer types ( Figure 1A), and elevated expression of PLXND1 mRNA was found in HCC compared with normal liver tissues ( Figures 1A, B). Next, in HCCDB, we validated PLXND1 mRNA expression in the large cohorts of HCC patients. Compared with the adjacent normal liver tissues, significantly upregulated PLXND1 expression in HCC was found in 10 out of 12 clinical cohorts, suggesting that the upregulation of PLXND1 is mainly attributed to the transformation of normal hepatocytes to cancer cells ( Figure 1C, Figure S1). Moreover, PLXND1 was expressed in various tissues, but its expression was low in normal liver tissues ( Figure S2).

PLXND1 Protein Expression and Its Association With Clinicopathological Features
To verify the overexpression of PLXND1 protein in HCC, we explored the protein expression level of PLXND1 in HCC by immunohistochemistry staining (IHC) using tissue microarrays and analyzed its associations with clinicopathological variables. Representative images of PLXND1 IHC staining are shown in Figure 2A. In the clinical samples, significantly higher PLXND1 expression was found in higher-grade HCC samples (Figure 2), which was consistent with the HCC samples from the TCGA database ( Figure S3). In addition, PLXND1 immunoreactivity was also found to be significantly associated with stage, histology grade, TH, tumor number, and satellite nodules, suggesting the potential oncogenic activity of PLXND1 in HCC ( Table 1).

The Regulation of Immune Molecules by PLXND1
Based on the tumor hemorrhage (TH) status, we classified a cohort of patients with HCC into two subgroups, TH-negative and TH-positive. HCC tumors with TH showed even higher expression of PLXND1 as determined by the H-scores ( Figure  3A). Our previous study reported that TH indicates a worse tumor immunological microenvironment, especially in HCC (23,24). Thus, we further explored whether PLXND1 is a significant factor that relates to immune infiltration using the TISIDB and GEPIA databases. In TISIDB, we found that PLXND1 expression was correlated with several tumor-infiltrating lymphocytes ( Table S1). Moreover, as shown in Figure 3C, PLXND1 expression was associated with several immunoinhibitors in HCC. TGF-b1, a multifunctional cytokine involved in tumor invasion and progression, displayed the greatest correlations with PLXND1 expression in HCC (Spearman: r = 0.62, P < 2.2e−16). Therefore, these findings suggest that PLXND1 might exert a significant effect on immune fingerprinting in HCC.

The Prognostic Value of PLXND1 and TGF-b1 in Patients With Hepatocellular Carcinoma
We have found that in HCC, PLXND1 expression was correlated with TGF-b1 ( Figure 3C, Table S1), which is a well-known modulator involved in tumor formation, invasion, and metastasis. Here, we further explored whether the coexpression of PLXND1 and TGF-b1 was associated with patient prognosis. We quantified the expression of PLXND1 and TGF-b1 in HCC tissues by IHC, and a varied expression pattern and intensity among the HCC tissues was observed. Representative images of PLXND1 and TGF-b1 IHC staining are shown in Figure 4A. We then grouped patients into 3 categories: group I, double negative (PLXND1 neg /TGF-b1 neg ); group II, PLXND1 pos /TGF-b1 neg and PLXND1 neg /TGF-b1 pos ; and group III, double positive (PLXND1 pos /TGF-b1 pos ). Survival analysis showed that the overall survival (OS) rates in group III were significantly lower than those in group II and group I ( Figure 4B). A consistent result was obtained for the recurrence-free survival (RFS) analysis ( Figure 4C). In addition, univariate analysis revealed the following features as prognostic factors related with both OS and RFS: the expression of PLXND1 and TGF-b1, stage, TH, liver cirrhosis, tumor size, tumor number, and satellite nodules ( Figures 5A, B). Multivariate analysis further identified that only the expression of PLXND1 and TGF-b1, stage, TH, tumor number, and satellite nodules behaved as independent predictors for both OS and RFS ( Figures 5C, D).

DISCUSSION
In this study, we aimed to investigate critical and novel biomarkers involved in the development and progression of HCC. PLXND1 belongs to a family of single-pass transmembrane proteins called plexins and serves as a receptor for semaphorins (25)(26)(27). Semaphorin-plexin signaling is known to be essential for many cellular aspects of organogenesis, including cell proliferation, invasion, and survival (28,29). Emerging data have shown that the expression of PLXND1 is often dysregulated in several types of cancer and acts either as a tumor suppressor or an oncogene.  (10): however, the detailed roles of PLXND1 in HCC have rarely been studied. In this study, we first revealed an abnormal expression level, prognostic value, and potential roles of PLXND1 in HCC biology. We found that PLXND1 expression in HCC tissues was significantly higher compared to normal liver tissues using data from the GEPIA and HCCDB databases. This result was further validated by IHC using tissue microarrays, which contained 216 HCC cases collected from our hospital. A high expression of PLXND1 was significantly correlated with several markers of HCC progression, including tumor grade and tumor hemorrhage (TH). These data suggest that PLXNDA may be a potential biomarker and therapeutic target in HCC. Recently, increasing evidence has shown that the tumor microenvironment and immune infiltration plays a critical role in tumor development and progression (32)(33)(34). Our previous study revealed that tumor hemorrhage (TH) indicated a worse tumor immunological microenvironment, especially in HCC (23,24). In this study, PLXND1 was found to be highly correlated with TH in HCC. These above findings inspired us to further investigate the correlation between PLXND1 and immunity in HCC development. We assessed the correlation between PLXND1 and the immune system using the TISIDB and GEPIA databases. The results showed that PLXND1 had the most significant correlation with tumor-infiltrating lymphocytes including Tfh cells, macrophages, Tcm_CD4 cells, and mast cells. Tfh cells are a subset of CD4+ helper T cells that play a crucial role in humoral immunity and protection against pathogens and in the generation of autoimmunity (32,35). Macrophages, as the forefront of host defense, also play pivotal roles in tumor immunity (36,37). Mast cells consistently infiltrate tumors and can be both a driving or an opposite force for tumor progression depending on the tumor microenvironment (38,39). Additionally, TGF-b1 was revealed to be the most likely immune inhibitor associated with PLXND1 in HCC. Transforming growth factor betas (TGF-bs) belong to a family of dimeric peptide growth factors and exists in three isoforms, TGF-b1, TGF-b2, and TGF-b3 (40)(41)(42). As a multifunctional cytokine, TGF-b1 was demonstrated to be both a tumor inhibitor, wherein it plays growth inhibitory and anti-inflammatory roles in the early stages of cancer; and a tumor promoter, wherein it promotes the development of aggressive growth characteristics and metastatic spread in late stages of cancer (43,44). TGF-b1 is also involved in the cancer immune microenvironment and promotion of tumor progression (45). In cancers, especially HCC, immune responses and their modulation play critical roles on the outcome of liver malignancies (46)(47)(48). TGF-b1-induced epithelial-mesenchymal transition (EMT) was reported to maintain CD4+ CD25+ regulatory T cells (Tregs) and thus promote HCC invasion and metastasis (49). Interestingly, the present study found that TGF-b1 was the most likely immune inhibitor associated with PLXND1 in HCC. Thus, we further analyzed the patient samples and explored the value of the coexpression of PLXND1 and TGF-b1 as a marker for prognosis. We subclassified liver tumors based on PLXND1 and TGF-b1 expression, and the results showed that this classification could be used to predict clinical outcomes and patient prognosis. Survival analysis showed that the patients with double negative expression (PLXND1 neg /TGF-b1 neg ) had the best outcome in HCC: H=however, patients with double positive expression (PLXND1 pos /TGF-b1 pos ) had the worst survival prognosis in HCC. Both univariate and multivariate analyses revealed that the expression of PLXND1 and TGF-b1 could behave as prognostic predictors for both OS and RFS. These findings could be a step toward the development of new biomarkers and therapeutic approaches in HCC.
Nevertheless, several limitations in this study need to be noted. First, the PLXND1 mRNA expression data was extracted from the TCGA database, where ethnicities were mainly white and black, and thus more studies including other ethnic groups should be performed. Additionally, several attractive findings were revealed in this study, but more mechanistic experiments remain to be conducted to further validate the functional value of PLXND1 and TGF-b1 expression in HCC, these aspects are currently under investigation in our lab. Moreover, we mainly focused our analysis on the clinic value of PLXND1 and TGF-b1 for HCC. The analysis of PLXND1 and TGF-b1expression should be further investigated across multiple cancer types.
In conclusion, we first report that PLXND1 is overexpressed in HCC and shows a significant correlation with several clinicopathological features, including tumor stage and TH. Moreover, PLXND1 expression is related with tumor-infiltrating lymphocytes and immunoinhibitors, and TGF-b1 displays the greatest correlations with PLXND1 expression in HCC. Moreover, we further reveal that PLXND1 and TGF-b1 expression are associated with patient prognosis. Taken together, these findings provide a critical understanding of PLXND1 deregulation in HCC and reveal that PLXND1 likely displays an important role in the tumor immune microenvironment. Finally, the results show that PLXNDI is a promising prognostic biomarker in HCC.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by the Ethics Review Committee of Xiangya Hospital of Central South University (N201303033). The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
Conception and design: YT and ZW. Writing, review, and/or revision of the manuscript: JL, KH, YT, and ZW. Administrative, technical, or material support: JL, LZ, and DH. All authors contributed to the article and approved the submitted version.

FUNDING
This study is supported by grants from the Natural Science Foundation of Hunan Province (2018JJ3820, 12JJ3118) and the National Natural Science Foundation of China (81372630).