High Expression of PPFIA1 Is Associated With Tumor Progression and a Poor Prognosis in Esophageal Squamous Cell Carcinoma

Background: PTPRF interacting protein alpha 1 (PPFIA1) is reportedly related to the occurrence and progression of several types of malignancies. However, its role in esophageal squamous cell carcinoma (ESCC) remains unknown. We aimed to investigate the expression and clinical value of PPFIA1 in ESCC. Methods: The Oncomine, Gene Expression Proling Enrichment Analysis (GEPIA), and Gene Expression Omnibus (GEO) databases were utilized to explore PPFIA1 mRNA expression in esophageal cancer. The associations of PPFIA1 expression with clinicopathological variables and prognosis were evaluated in the GSE53625 dataset and veried in quantitative real-time polymerase chain reaction (qRT-PCR)-based cDNA array and immunohistochemistry (IHC)-based tissue microarray (TMA) datasets. The interactions between PPFIA1 and other genes based on the protein-protein interaction (PPI) network was analyzed via the STRING website. Results: PPFIA1 expression was obviously upregulated in ESCC tissues versus adjacent normal tissues according to online database analyses (all P<0.05). High PPFIA1 expression was signicantly associated with several clinicopathological features, including tumor size, histological grade, tumor invasion depth, lymph node metastasis, and tumor-node-metastasis (TNM) stage. High PPFIA1 expression was related to worse outcomes and was identied as an independent prognostic indicator of overall survival (OS) in ESCC patients GSE53625 dataset, P=0.004; cDNA array dataset, P<0.001; TMA dataset, P=0.039). PPI analysis demonstrated that PPFIA1 was highly correlated with multiple genes, including UNC13B, RAB3A, PTPRD, and SYT1. Conclusion: PPFIA1 and used as a biomarker for prognostic evaluation in PAAD: pancreatic adenocarcinoma.

In the present research, we rst investigated the expression of PPFIA1 in ESCC tissues and paracancerous tissues using available online datasets from the Oncomine, Gene Expression Pro ling Interactive Analysis (GEPIA), and Gene Expression Omnibus (GEO) databases. The associations of PPFIA1 expression with the clinicopathological features and outcomes of ESCC patients were analyzed in the GSE53625 dataset and further con rmed in the cDNA array and tissue microarray (TMA) datasets.

Bioinformatics analyses
The following websites and datasets were used:

GEPIA
The difference in the expression of PPFIA1 mRNA and its relationship with OS in human tumors was analyzed through the GEPIA website (http://gepia.cancer-pku.cn/detail.php), which contains RNA sequencing and expression information from data from The Cancer Genome Atlas (TCGA) and the Genotype-Tissue Expression (GTEx) databases [9]. The mRNA expression levels of PPFIA1 in both tumor tissues and adjacent normal tissues are displayed in a bar chart.
Oncomine Oncomine (http://www.oncomine.org) is a public online microarray database and comprehensive data mining platform that can be used to mine cancer genetic information [10]. The expression of PPFIA1 in esophageal cancer tissues and paracancerous tissues was analyzed through Oncomine. The data analysis was carried out based on standardized normalization techniques with statistical calculations supplied by the Oncomine website.

The GEO database
The GEO database is a publicly available database of gene expression data that stores a large amount of microarray data (https://www.ncbi.nlm.nih.gov/geo/) [11]. Six datasets from the GEO database (GSE23400, GSE20347, GSE29001, GSE53625, GSE45670 and GSE26886) were used to determine PPFIA1 mRNA expression in ESCC and normal tissues. The GEO datasets were analyzed online through GEO2R.
Furthermore, the RNA sequencing data, all clinicopathological variable data and the survival data of 179 ESCC patients were also downloaded from the GSE53625 dataset (https://portal.gdc.cancer.gov/). There were 146 (81.6%) males and 33 (18.4%) females, with a median age of 60 (range, 36~82) years. According to the criteria of the American Joint Committee on Cancer (AJCC) tumor-node-metastasis (TNM) staging system, 10 (5.6%), 77 (43.0%), and 92 (51.4%) cases were considered stage I, II, and III, respectively. The OS rates at 1, 3, and 5 years were 77.1%, 48.6%, and 40.8%, respectively, with a median survival time of 42 months. We further used the STRING website (http://string-db.org/) to explore the interactions between PPFIA1 and other genes based on the protein-protein interaction (PPI) network.
Tissue cDNA array data based on quantitative real-time polymerase chain reaction (qRT-PCR) An ESCC cDNA microarray with patient diagnosis information was purchased from Shanghai Outdo Biotech Co., Ltd. (Cat. No.: cDNA-HEsoS095Su01; Shanghai, China); the array contained samples from 67 cancer tissues and 28 adjacent esophageal tissues. These tissues were all from patients who were histologically con rmed to have ESCC and underwent esophagectomy. There were 48 males and 19 females, with a median age of 59 (range, 37~78) years. The study was approved by the Medical Ethics Committee of the Shanghai Outdo Biotech Company.
TMA data based on immunohistochemical analysis TMAs were used for immunohistochemical staining to examine the expression of PPFIA1 protein. TMAs containing samples from 147 ESCC tissues and 40 adjacent normal esophageal tissues were made by Shanghai Outdo Biotech Co. Ltd. All para n block specimens were obtained from patients who underwent radical esophagectomy between 2009 and 2010 at Tianjin Medical University Cancer Institute and Hospital, with reliable information on survival. No patients had received any chemotherapy or radiotherapy before surgery. Among the patients, there were 119 males and 28 females with a median age of 68 years old. All patients were followed up until September 2016 with a median follow-up period of 36 months. The research protocol was approved by the Research Ethics Committee of Tianjin Medical University Cancer Institute and Hospital, and written informed consent was obtained from all patients in accordance with the principles of the Declaration of Helsinki.
The protein expression of PPFIA1 in ESCC samples was detected through IHC. In brief, the tissue sections were depara nized, rehydrated and incubated with an antigen retrieval solution. The activity of endogenous peroxidase was blocked with 0.3% H2O2 and 5% goat serum. Then, the sections were incubated with an anti-PPFIA1 polyclonal antibody (1:100; Cat No. DF12102) at 4 °C overnight and then incubated with a biotinylated secondary antibody for 20 min at room temperature based on the reagent instructions. Diaminobenzidine was used as a chromogen. The stained slide was scanned using an automatic slice scanning system. Two or three experienced pathologists who were blind to the clinicopathological information independently performed the staining evaluation based on the staining intensity (no staining, 0; weak, 1; moderate, 2; strong, 3) and the percentage (0%, 0; 1 ~ 25% positive, 1; 26% ~ 50% positive, 2; 51% ~ 75% positive, 3; 76% ~ 100% positive, 4) of positively stained tumor cells. The immunoreactivity score was calculated as the product of the staining intensity and staining percentage scores, with the nal score ranging from 0 to 12. Patients with a total score of < 4 were considered to have low expression (n = 47), and those with a total score of ≥ 4 were considered to have high expression (n = 100).

Statistical analyses
All statistical analyses were performed using SPSS 17.0 software (SPSS, Inc., Chicago, IL) and GraphPad Prism 7.0 software (GraphPad, Inc., La Jolla, CA, USA). Statistical differences of the Oncomine and GEPIA data were indicated automatically by the program. Differences in PPFIA1 mRNA expression between tumor tissues and adjacent normal tissues were compared using a t test. The patients were divided into two groups according to the best cutoff value of PPFIA1 expression for predicting OS calculated with the X-tile 3.6.1 software (Yale University, New Haven, CT, USA) for analysis of the GSE53625 dataset [12] or according to the median value of PPFIA1 expression for analysis of the cDNA array dataset. The chisquare test or Fisher's exact test was used for the analysis of PPFIA1 expression and clinicopathological variables. Survival analysis was carried out through the Kaplan-Meier method, and differences were assessed via the log-rank test. Univariate and multivariate Cox proportional hazards regression analyses were performed to further evaluate the independent prognostic variables. The correlations between PPFIA1 expression and the expression of other genes were assessed by Pearson's correlation analysis. A P-value <0.05 was deemed statistically signi cant.

Results
The expression of PPFIA1 mRNA in various tumor tissues We rst examined the expression level of PPFIA1 mRNA in various tumors with the GEPIA website, and the results showed signi cantly higher PPFIA1 expression in several types of tumor tissues, including esophageal cancer (ESCA), pancreatic adenocarcinoma (PAAD), and thymoma tissues, compared with adjacent normal tissues at the RNA level (Fig. 1A). The overview of PPFIA1 expression in various tumor specimens and its detailed expression in ESCA specimens, as well as in normal controls, were further analyzed with the Oncomine database. PPFIA1 mRNA expression was obviously increased in the tumor tissues versus normal tissues from most of the datasets for all types of tumors (Fig. 1B). The fold changes in PPFIA1 mRNA expression in different ESCA tissues of 7 datasets are displayed in Supplementary Table S1.

The expression of PPFIA1 was upregulated in ESCC tissues
The expression of PPFIA1 mRNA was markedly higher in ESCC tumor samples than in adjacent normal samples according to the GEPIA website ( Fig. 2A). This nding was further veri ed using the data from six GEO datasets; the results indicated that the relative PPFIA1 mRNA expression was obviously upregulated in ESCC tissues compared with paired or unpaired adjacent normal tissues in the GSE23400 (9.6 ± 1.1 vs. 8.4 ± 0.6, P < 0.001, Fig In addition, the increased expression of PPFIA1 mRNA in ESCC samples was further con rmed through the analysis of cDNA array data based on qRT-PCR (0.12 ± 0.15 vs. 0.06 ± 0.08, P = 0.003, Fig. 2H).

Correlations between PPFIA1 mRNA expression and clinicopathological variables in ESCC patients
The associations of PPFIA1 mRNA expression with the clinicopathological features of patients with ESCC were investigated in the GSE53625 dataset and cDNA array dataset. In the GSE53625 dataset, the 179 ESCC patients were divided into a low-expression group (n = 155) and a high-expression group (n = 24) according to the X-tile cutoff. Our results showed that high PPFIA1 expression was positively related to tumor invasion depth (P = 0.019), lymph node metastasis (P = 0.013), and TNM stage (P = 0.012), but no correlations were found between PPFIA1 expression and sex, age, smoking use, alcohol use, tumor location, or histological grade (P > 0.005) ( Table 1). We then analyzed PPFIA1 levels in the cDNA array of 68 ESCC patients. The patients were split into a lowexpression group (n = 34) and a high-expression group (n = 33) based on the median value of relative PPFIA1 expression. We found that PPFIA1 mRNA expression was signi cantly correlated with histological grade (P = 0.031, Supplementary Table S2). Overall, the above data indicate that the expression of PPFIA1 is associated with various important clinicopathological features of ESCC.
Relationships between PPFIA1 protein expression and clinicopathological variables in ESCC patients in the TMA dataset We further assessed PPFIA1 protein expression using an IHC staining-based TMA containing samples from 147 surgically removed cancer tissues and 40 normal esophageal tissues. PPFIA1 was primarily localized to the cytoplasm and nucleus of cancer cells, and the expression rate was signi cantly higher in cancer tissues than in normal tissues (68.0% vs. 25.0%, P < 0.05). Representative images of low or high PPFIA1 expression are shown in Fig. 3. All patients were divided into the PPFIA1 low-expressing group (n = 47) or the high-expressing group (n = 100). We further analyzed the association between PPFIA1 expression and clinicopathological variables and found that PPFIA1 expression was inversely associated with tumor location (P = 0.011), tumor invasion depth (P = 0.041), lymph node metastasis (P = 0.020), and TNM stage (P = 0.007) in ESCC patients. No signi cant relationships with other variables were found (all P > 0.05) ( Table 2). In view of the high expression of PPFIA1 mRNA in a variety of malignancies, including breast, ovarian, lung, and gastric cancers, we further investigated the prognostic value of PPFIA1 through the Kaplan-Meier plotter website. Patients were grouped based on the automatically selected best cutoff value for PPFIA1 expression. As shown in Fig. 4, high expression of PPFIA1 apparently correlated with the RFS, OS, PPS, DMFS of patients with breast cancer (Fig. 4A, P < 0.05), the PFS, OS, and PPS of patients with ovarian cancer (Fig. 4B, P < 0.05), the FP, OS, and PPS of patients with lung cancer (Fig. 4C, P < 0.05), and the FP, OS, and PPS of patients with gastric cancer (Fig. 4D, P < 0.05). These results suggest that high expression of PPFIA1 is obviously associated with worse outcomes in patients with breast cancer, ovarian cancer, lung cancer, and gastric cancer.

Correlations between PPFIA1 expression and the prognosis of ESCC patients
Survival analyses revealed that ESCC patients with high PPFIA1 mRNA levels had a signi cantly worse 5year OS rate than those with low PPFIA1 mRNA levels in the GSE53625 dataset (45.2% vs. 12.5%, P < 0.001, Fig. 5A) and cDNA array dataset (38.0% vs. 9.2%, P = 0.001, Fig. 5B). High PPFIA1 expression at the protein level was also obviously related to poor outcomes in patients in the TMA dataset (5-year OS rate: 46.0% vs. 23.4%, P < 0.001, Fig. 5C).
The prognosis-related indicators in the 3 datasets were analyzed by univariate and multivariate survival analyses, and the results are presented in  We further analyzed the associations of PPFIA1 mRNA expression with the abovementioned genes in the GSE53625 dataset. Our results showed that PPFIA1 expression was clearly correlated with UNC13B, RAB3A, PTPRD, SYT1, RIMS1, PTPRS, APBA1, PTPRF, and LIN7A expression ( Figure 6B-J) but not with CASK expression (Figure 6K).

Discussion
Early invasion and metastasis are two of the main reasons for the poor prognosis of patients with ESCC [13]. In our current study, we observed an increase in PPFIA1 expression in ESCC, which was associated with malignant biological behavior and a poorer prognosis. PPFIA1 may be a potential molecular indicator for diagnosis and prognosis evaluation and could be regarded as a new therapeutic target for ESCC.
PPFIA1 is a gene located at the 11q13 ampli cation region and mainly encodes the liprin-α1 protein in humans. It was initially discovered to control the formation and function of synapses in neurons [14].
Studies have indicated that PPFIA1 is a key regulator of cell motility, focal adhesion, cell signal transduction, and cytoskeletal organization [8,15]. PPFIA1 is frequently ampli ed and plays a crucial role in cell migration and invasion by affecting cell motility, mediating extracellular matrix degradation and facilitating the formation of lamellipodial protrusions of cancer cells [3,[16][17][18][19]. Although the role of PPFIA1 in tumor cell progression has been well veri ed, it is not clear whether its dysregulation is related to metastasis risk or prognosis in ESCC patients.
To determine the clinical value of PPFIA1 in ESCC, we analyzed the difference in the expression of this novel marker in tumor tissues compared with normal tissues and assessed its correlation with the survival of ESCC patients using multiple datasets. Analysis of datasets from Oncomine, GEPIA and GEO con rmed that PPFIA1 mRNA expression was markedly higher in ESCC tissues than in adjacent normal control tissues, which was further con rmed with cDNA array data based on qRT-PCR and TMA data based on IHC. These results indicated that PPFIA1 may play a vital role during the tumorigenesis of ESCC, which was consistent with the results discovered in other malignant tumors [17,20]. Further correlation analysis revealed that PPFIA1 expression is highly correlated with aggressive biological behaviors of tumors, indicating the important function of PPFIA1 in the progression of ESCC. It is worth mentioning that the expression of PPFIA1 was only correlated with the degree of tumor differentiation in the cDNA chip dataset, which might be due to the limited sample size in the research.
Kaplan-Meier Plotter is an online database containing survival information for patients with several types of tumors. Importantly, we discovered that PPFIA1 can be applied for the prognostic assessment of breast cancer, ovarian cancer, lung cancer and gastric cancer by online analysis. Cho et al [21]. explored the prognostic value of PPFIA1 alone or in combination with TMEM16A and FADD in patients with invasive ductal breast carcinoma, and they found that the combined expression was signi cantly associated with perineural invasion and a low disease-free survival rate. A recent study showed that the expression of PPFIA1 is closely related to poor response to endocrine treatment in luminal breast cancer [22]. Since PPFIA1 was found to be involved in the development of ESCC, we believe that PPFIA1 is likely to have a great impact on the survival of patients with ESCC. To obtain a reliable conclusion, we used the GSE53625, cDNA array and TMA datasets to investigate the prognostic value of PPFIA1 in ESCC. Our results indicated that high PPFIA1 expression was evidently correlated with a poorer prognosis. Notably, PPFIA1 was identi ed as an independent indicator of poor prognosis in all three independent databases in the multivariate analyses. Thus, we propose that PPFIA1 may serve as a potential diagnostic and novel prognostic biomarker, as well as a new therapeutic target for ESCC. that liprin-α1 can determine the polarization and morphological dynamics related to cell migration by forming a complex with the liprin-β1, ERC1/ELKS, and LL5 proteins [23]. To examine the potential mechanisms of PPFIA1 in ESCC progression, we further used the PPI network and Pearson's correlation analysis to identify proteins that may bind to PPFIA1; we identi ed UNC13B, RAB3A, PTPRD, SYT1 and RIMS1 as potential candidates. Some of these interacting genes have already been reported as tumor oncogenes or suppressor genes [24][25][26]. For instance, RAB3A interacting protein (Rab3IP) is a Rabspeci c GEF, and the activation of Rab proteins, including Rab3A and Rab8, has been considered a tumorspeci c marker in colorectal cancer, gastric cancer, and pancreatic cancer [27][28][29]. A recent study reported by Ren et al. showed that Rab3IP interacts with SSX2 and enhances an invasive aggressive phenotype of gastric cancer through epithelial-mesenchymal transition [25]. PTPRT is a phosphatase that can participate in JAK/STAT signal transduction. Deleterious mutations or copy number loss of PTPRT and its related gene PTPRD are potential markers for evaluating the resistance to bevacizumab regimens and are closely associated with shorter PFS in metastatic colorectal cancer patients [24]. However, the biological function and underlying mechanism of PPFIA1 in ESCC need to be further studied [25].
There are several certain limitations to the current study that should be pointed out. First, although the results were analyzed through bioinformatics analysis and three independent databases (GSE53625, cDNA array and TMA), the sample size involved in our study was relatively small. Second, the clinicopathological information from the three datasets was not comprehensive, and part of the information in the cDNA array dataset was incomplete. Third, the exact biological function and detailed molecular mechanism of PPFIA1 in ESCC were not veri ed in in vitro and in vivo experiments. Therefore, our results need to be further veri ed in a large patient cohort with complete clinical pathological data concerning tumor progression and prognosis.

Conclusion
In conclusion, this preliminary study indicates that the expression of PPFIA1 is signi cantly increased and is related to some malignant clinical features and poor outcomes in ESCC patients. PPFIA1 might be a valuable biomarker for early detection, treatment formulation and prognosis evaluation for ESCC.      Abbreviations: PPFIA1, PTPRF interacting protein alpha 1; PPI, protein-protein interaction.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.