Overexpression Of RNPEPL1 Predicts Poor Overall Survival In Ovarian Carcinoma


 Background. RNPEPL1 (arginyl aminopeptidase-like 1) belongs to the M1 family of zinc metallopeptidases. We aimed to explore the role of RNPEPL1 in the occurrence and progression of ovarian carcinoma (OV).Methods: We analyzed the expression level and prognostic value of RNPEPL1 in OV across Oncomine, GEO, TCGA and Human Protein Atlas (HPA) database and identified its prognostic value by the Cox regression analysis and Kaplan–Meier method. We utilized GESA software to perform GO and KEGG analyses. We also utilized TIMER to evaluate the relevance between RNPEPL1 and tumor-infiltrating immune cells. Finally, relevance between RNPEPL1 and immune-related genes was examined by Pearson correlation analysis.Results: RNPEPL1 was overexpressed in OV tissues compared with normal ovary tissue. Moreover, high RNPEPL1 expression was related with poor OV survival in TCGA. Cox analyses suggested that overexpression of RNPEPL1 was an independent risk factor for overall survival(OS) of OV. Results of KEGG analysis suggested that genes were mainly enriched in T cell receptor signaling pathway, B cell receptor signaling pathway and natural killer cell mediated cytotoxicity when the expression level of RNPEPL1 is low. Besides, RNPEPL1 expression is associated with infiltrating levels of CD4+ T cells, B cells, macrophage, neutrophils and dendritic cells. And, the level of dendritic cells is correlated with OS of OV. RNPEPL1 expression was significantly correlated with the immune signaling pathway of antigen processing and presentation.Conclusions: RNPEPL1 is a novel independently prognostic biomarker of OV and was related to immune infiltration. Especially, the tight correlation between RNPEPL1 and antigen processing and presentation may be the potential hub of anti-tumor immunity and one of the essential factors influencing survival.


Introduction
Ovarian carcinoma (OV) ranks fth among the causes of death in female and ranks rst among the mortality rates of all gynecologic malignant tumors [1]. There were 295,400 women diagnosed with OV and 184,800 women died of ovarian cancer around the world in 2018, [2]. In the United States, there are 21,410 new cases and 13,770 deaths of ovarian cancer predicted in 2021 [3]. Surgery and chemotherapy are the main treatments for OV. Despite development in these treatments, the survival rate of OV has only improved slightly. The 5-year survival rate is only 38.9% in China and 48% in the United States for OV [4]. Epithelial OV accounts for 90% among all pathological types and serous carcinoma is the most common subtype [5]. OV is usually diagnosed in advanced stage when the mortality rate is high due to the anatomical features of the ovary [6]. Exploration for valid adjuvant therapeutics is based on a better comprehending of its progression biology. Reliable biomarkers, as a potential improvement for OV patients including screening, diagnosis, response to therapy, prognosis and survival, are urgently required. RNPEPL1(arginyl aminopeptidase-like 1) belongs to the M1 family of zinc metallopeptidases [7]. Most M1 family of zinc metallopeptidases are aminopeptidases and exopeptidases. They hydrolyze amino acid residues sequentially from the N-terminus region of protein substrate or peptide. Inhibitors of these enzymes are promising in many treatment circumstances, such as in ammation[8], hypertension [9], and even some cancers [10,11]. The intron 5 is an alternative splicing event of RNPEPL1 and was speculated to generate a mRNA isoform including an immature stop codon, and it would nally generate a curtal protein product de cient in catalytic activity [7]. There were only ovary and thymus that did not generate mRNA containing intron 5 [7]. Hence, ovary and thymus would not neither produce the truncated protein of RNPEPL1 nor lose catalytic activity. Whether RNPEPL1 plays a role in the occurrence and progression of OV is still unknown either. With the progression of sequencing technique, bioinformatics can be utilized to understand and discover new biomarkers of cancer [12][13][14]. Therefore, we used GEO, TCGA and other bioinformatic database to explore the relation betweent RNPEPL1 and OV and the possible function of RNPEPL1 in the occurrence and progression of OV.

Data resource and Description
The sequencing data of RNPEPL1 were downloaded from three gene datasets (GSE18521, GSE40595, and GSE38666) containing OV tumor and normal ovary samples from the Gene Expression Omnibus (GEO). RNA-sequencing gene expression and clinical data of 375 OV patients from The Cancer Genome Atlas (TCGA) o cial website were retrieved on October 15, 2020.

Identi cation of RNPEPL1 expression
The expression values of RNPEPL1 were normalized using the quantile method of Robust Multichip Analysis (RMA). Expression values were rendered as log2transformed values using the R affy package. Student's t-test was used to analyze the expression difference of RNPEPL1 by the limma package. In the Oncomine database, studies comparing RNPEPL1 between OV tumor and normal samples of ovary were set with a threshold of fold change ≥ 2, P ≤ 1E-4 and top 10% gene rank. Protein expression of RNPEPL1 of OV tumor, normal ovary samples and normal lymph node detected by immunohistochemical method was acquired from the Human Protein Atlas (HPA) website.

Survival analysis
The clinicopathological characteristics of 375 OV patients relevant to the overall survival (OS) of OV patients were identi ed by the Cox regression analysis and Kaplan-Meier method. The cut-off point of RNPEPL1 expression value was the median value. Based on the median value, patients were divided into RNPEPL1-high and RNPEPL1-low level groups. Log-rank test was used to determine the difference between high and low level groups of RNPEPL1 by R package "survival". The association between RNPEPL1 expression and OS in different clinicopathological groups was also analyzed.
The correlations between clinicopathological characteristics and RNPEPL1 expression Associations between RNPEPL1 expression and clinicopathological characteristics were analyzed by a non-parametric test (the Wilcoxon test was performed when the data were classi ed into two groups; the Kruskal-Wallis test was conducted when the data were classi ed into three or four groups).

GO and KEGG analysis by GSEA
We used GESA software (version 4.1.0) to conduct KEGG and GO analyses. According to the expression level of RNPEPL1, OV samples from TCGA were divided into 2 different groups. Permutation analysis (1,000 permutations) was used to identify the thresholds for signi cance. Metrics for ranking the key mRNAs were computed ground on Pearson's correlation coe cient. The minimum and maximum sizes of gene sets were set as 10 and 500. The results with a nominal p < 0.05 and an FDR value < 0.25 were identi ed as statistically signi cant.
The relevance between RNPEPL1 and in ltrating immune cells Tumor Immune Estimation Resource (TIMER) was used to evaluate the relevance between RNPEPL1 and immune cells [15,16]. "Gene" module of TIMER was performed to analyze the association between enrichment of six main types of immune cells in ltrates and RNPEPL1 expression, including CD8+ T cells, CD4+ T cells, B cells, neutrophils, dendritic cells and macrophages in OV using tumor purity-corrected partial Spearman's correlation. The prognostic value of each immune cells was determined using Kaplan-Meier analysis.
Immune-related genes enrichment analysis All immune-related genes (IRGs) from the ImmPort database were downloaded. Pearson correlation analysis was used to select the RNPEPL1-related IRGs.
The IRGs that are signi cantly related to the expression of RNPEPL1 in TCGA dataset were remained and used as the candidates for the enrichment analyses. The most signi cantly enriched IRGs gene ontology terms and the corresponding genes were exhibited using a cross-link analysis.

Statistical analysis
Paired t test was used to analyze the expression difference by R software. In R software survival package (version 3.1-11), "coxph" method was used to conduct Multivariate Cox regression. RNPEPL1 expression in different clinicopathological groups were analyzed by a non-parametric test using GraphPad Prism software. Pearson correlation analysis and cross-link analysis were performed using R software. The scatter plots, box plots and survival plots were graphed using GraphPad Prism software (version 5.0). A p value less than 0.05 was considered as signi cantly different.

Survival analysis
There were 375 OV patients included in the survival analysis. The main characteristics of them were summarized in Table 1 Table 2). Overexpression of RNPEPL1 was an independent risk factor for OS of OV.  Overexpression of RNPEPL1 in OV tumor tissues was signi cantly correlated with poor OS (P = 0.0016, Figure 5A) in OV patients. Besides, a subgroup analysis suggested that the overexpression of RNPEPL1 in OV tumor tissue was correlated with reduced 5 year (P = 0.0179, Figure 5B) and 10 years survival In different clinicopathological groups, overexpression of RNPEPL1 was correlated with poor survival in patients with age ≤ 58 years (P = 0.0450, Figure 6A), while no signi cant difference was found in patients with age 58 years (P = 0.191, Figure 6B). Moreover, overexpression of RNPEPL1 signi cantly correlated with poorer OS in patients with late stage (P = 0.0009, Figure 6D) , high grade (P = 0.0032, Figure 6E) and ≥1mm residue disease (P = 0.0038, Figure 6H), but not determined as risk factors for OV patients with early stage (P = 0.3819, Figure 6C), low grade (P = 0.4486, Figure 6F) and no macroscopic disease(P = 0.1839, Figure 6G).

The correlations between RNPEPL1 expression and clinicopathological characteristics
The analysis of clinical data and gene expression data downloaded from TCGA-OV showed that the expression of RNPEPL1 in younger patients (≤ 58 years old) was signi cantly higher than older patients (> 58 years old) (P < 0.0001, Figure 7A). RNPEPL1 was highly expressed in early stage (P < 0.0001, Figure 7B) and low grade (P < 0.0001, Figure9C) patients. RNPEPL1 increased with venous invasion (P < 0.0001, Figure 7D) and lymphatic invasion(P < 0.0001, Figure  7E). RNPEPL1 was signi cantly distributed differently in different groups of residue disease and highly expressed in patients with 1mm-10mm and 10mm-20mm residue disease classi cation (P < 0.0001, Figure 7F).

GO and KEGG analyses by GSEA
The results of KEGG analysis showed that genes were mainly enriched RNA degradation, RNA polymerase, spliceosome, oxidative phosphorylation and ribosome when the expression level of RNPEPL1 is high, genes were mainly enriched in T cell receptor signaling pathway, B cell receptor signaling pathway, apoptosis, natural killer cell mediated cytotoxicity and cell adhesion molecules cams when the expression level of RNPEPL1 is low (Figure 8). The results of GO analysis suggested that genes were mainly enriched in spliceosomal snrnp assembly, U2 type spliceosomal complex, nuclear transcribed mRNA catabolic process, viral gene expression, and polysome when the expression level of RNPEPL1 is high, genes were mainly enriched in speci c granule, positive regulation of NF kappab transcription factor activity, macrophage activation, actin cytoskeleton reorganization, negative regulation of toll like receptor signaling pathway when the expression level of RNPEPL1 is low (Figure 9).

Immune-related genes enrichment analysis
There are 2483 immune-related genes (IRGs) extracted from the ImmPort website. A total of 249 IRGs are signi cantly related with the expression of RNPEPL1 in TCGA dataset ( Table 3). The most signi cantly correlated IRGs gene ontology terms and the corresponding genes were exhibited using a cross-link analysis. The expression of RNPEPL1 was signi cantly correlated with the immune signaling pathway of antigen processing and presentation, antimicrobials, TCR signaling pathway, BCR signaling pathway and natural killer cell cytotoxicity ( Figure 11).  Immunohistochemistry staining results showed that RNPEPL1 was positive in OV samples but was not detected in normal ovary tissues and normal lymph node (Fig. 4). This is in accordance with the expression difference of RNPEPL1 at mRNA level. Furthermore, RNPEPL1 may be used as a new pathologic diagnostic index to distinguish normal ovary and OV. The negative expression of RNPEPL1 in normal lymph node may help doctors judge the lymph node metastasis of OV which is important in the de nition of tumor stage.
RNPEPL1 was an independent prognostic biomarker in OV. Overexpression of RNPEPL1 can predict poor OS for OV. This also implies that RNPEPL1 may take part in the progression of OV. RNPEPL1 could be used as a prognostic biomarker in OV patients with age ≤ 58 years, late stage, high grade and ≥ 1mm residue disease. Usually, there is no obvious symptom of OV in the early stages. The occurrence of OV is hidden [17]. Most serous OV patients are diagnosed at stage III (51%) or IV (29%) [5], the prognostic value of RNPEPL1 in late stage patients is more meaningful due to the high proportion and low survival rate of late stage. Patients with less residue disease experience longer survival time [18,19]. How to predict survival time for patients with residue disease is of signi cant importance due to the low survival rate of them. The high expression of RNPEPL1 in patients with venous invasion and lymphatic invasion re ects that RNPEPL1 may be correlated with the invasion of OV.
The results of GO and KEGG analysis implies that RNPEPL1 may take part in the process of gene transcription and tumor immunity in OV. Genes were mainly enriched in T cell receptor signaling pathway, B cell receptor signaling pathway and natural killer cell mediated cytotoxicity when the expression level of RNPEPL1 is low (Fig. 6B). It implies that overexpression of RNPEPL1 may inhibit both speci c immunity and nonspeci c immunity. And, that may explain the poor prognosis of OV patients with high expression level of RNPEPL1. RNPEPL1 belongs to the M1 family of zinc metallopeptidases. The family has functions of breakdown of peptides and recycle of amino acids from usual cellular proteolysis [20], adjustment of signaling cascades through hydrolyzing peptide hormones [21] and management of peptides exhibited on class I MHC molecules [22,23]. RNPEPL1 may have a similar function of the processing of peptides exhibited on class I MHC molecules and plays a role in tumor immune escape of OV.
The CD4 + T cells, dendritic cells, neutrophils and macrophage all play a positive role in anti-tumor immunity process. RNPEPL1 was associated with all the six main in ltrating immune cells which indicate that RNPEPL1 is strongly immune-related, suggesting that RNPEPL1 may inhibit the process of anti-tumor immunity. Moreover, we found that when the amount of dendritic cells is high, patients will have longer survival time. Whereas, overexpression of RNPEPL1 was correlated with low amount of dendritic cells. RNPEPL1 may affect the progression of OV through inhibiting the process of anti-tumor immunity. Expression of RNPEPL1 was signi cantly correlated with the immune signaling pathway of antigen processing and presentation, antimicrobials, TCR signaling pathway, BCR signaling pathway and natural killer cell cytotoxicity (Fig. 11). The results further proved the immune-related function of RNPEPL1.
The result further indicates that RNPEPL1 may affect the progression of OV through inhibiting the process of anti-tumor immunity.
Under normal circumstances, tumor cells could be recognized by the immune system based on tumor-associated antigens and eliminated [24]. One of the features of cancer is that tumor cells get the ability of avoiding attacks from the immune system [25]. Tumor immune escape is indispensable for the development of tumor. There are also some successful cases of cancer immunotherapy by activating immune system [26]. Tumor with high tumor mutation burden is more sensitive to immunotherapy [27,28] The accumulation of mutation induced the expression of tumor-speci c antigen. Antigen processing and presentation is one of the most important steps of anti-tumor immunity. Our results suggested that the expression of RNPEPL1 is signi cantly correlated with antigen processing and presentation. RNPEPL1 may affect the progression of OV through affecting antigen processing and presentation. The speci c antigen of tumor will bind with MHC1 molecules on the surface of tumor cell and be recognized by CD8 + lymphocyte to trigger anti-tumor immune response [29].
Although there is no report about the impact of RNPEPL1 on MHC1 molecules, Members of this family have been implicated in the processing of peptides exhibited on class I MHC molecules [22,23]. RNPEPL1 may have similar function of processing of peptides exhibited on class I MHC molecules, play a role in tumor immune escape of OV, promote the progression and affect survival of OV patients.
There are also some limitations in our research. The data of this study was retrospective data. The clinical value of RNPEPL1 needs to be veri ed by more prospective data. Experimental data is lacking in this study to explain the correlation between RNPEPL1 and the possible mechanism in OV. More research is required to illuminate the potential value of RNPEPL1 in the diagnosis and treatment of OV.

Conclusions
RNPEPL1 is a novel gene correlated with the prognosis of OV. It has rarely been reported before and even never been reported in cancer. We rst explore the expression and possible function of RNPEPL1 in OV. RNPEPL1 is a new diagnostic and prognostic biomarker of OV. Further experiment on the mechanism of promoting OV progression and whether it could inhibit anti-tumor immunity through regulating immunity is needed. If the antigen processing and presentation of OV could be promoted through inhibiting RNPEPL1, RNPEPL1 may also be a potential immunotherapy target for OV. Figure 1 Flow chart of study design   Gene expression of RNPEPL1 was assessed using immunohistochemistry in normal ovary, OV tissues and lymph node. Protein levels of Protein levels of RNPEPL1 in tumor tissue (staining: Medium; intensity: Moderate; quantity: > 75%; Location: Nuclear). RNPEPL1 in normal tissue (staining: Not detected;

Figure 5
Overall survival of OV patients grouped by RNPEPL1 median cutoff in TCGA database; overall (A), 5-year (B) and 10-year(C) survival comparison between high and low RNPEPL1 groups. OV, Ovarian carcinoma.