Construction of a ceRNA network combining transcription factors in eutopic endometrial tissue of tubal factor infertility and endometriosis-related infertility

Purpose Although tubal factor infertility (TFI) and endometriosis-related infertility all can result in female infertility, the pathogenesis of TFI and endometriosis-related infertility were different. The pathophysiologic mechanisms of TFI and endometriosis-related infertility have not been investigated thoroughly. Thus, the aim of the study is to identify the potential crucial genes, pathways, transcription factors (TFs) and long non-coding RNAs (lncRNAs) associated with TFI and endometriosis-related infertility, and further analyze the molecular mechanism implicated in genes. Methods 3 patients with TFI and 3 patients with endometriosis-related infertility were recruited, and microarray hybridization of the eutopic endometrial tissue during the window of implantation (WOI) was performed to examine the expression of mRNAs and lncRNAs. First, differentially expressed genes (DEGs) and differentially expressed lncRNAs (DELs) were screened out based on P < 0.05 and fold change (FC) ≧ 2. Second, gene ontology, pathway and TFs enrichment analyses and PPI network construction of DEGs were performed. Then, important modules and Hub genes of PPI network were further analysed and a ceRNA network combining TFs based on DEGs and DELs was constructed. Finally, we selected the lapping genes of the important modules, Hub genes and the ceRNA network as potential key genes associated with TFI and endometriosis-related infertility and constructed a new ceRNA network on the base of potential key genes. Results 508 DEGs and 576 DELs were screened out. The gene ontology and pathway of were mainly enriched in transmembrane transporter activity and the immune system (eg, passive transmembrane transporter activity, Intestinal immune network for IgA production and so on). In addition, a ceRNA network (based on potential key genes) combining TFs including LDLR), and hsa-miR-17-5p), 8 lncRNAs (eg. LMO7-AS1, ITFG3 and MGST1, etc.) and 10 TFs (eg. SRF, FOX, which target to mRNA and eg. POU3F2, HNF1A, which target to miRNA) was successfully built. Conclusions In conclusion, pathophysiologic mechanisms of TFI and endometriosis related infertility may be related to the transmembrane transporter activity and the immune system. These potential key RNAs and TFs might have clinical utility for the diagnosis and prognosis prediction in TFI and endometriosis-related infertility. The results of the current study might lay the foundation for future basic and clinical research.


Introduction
Tubal factor infertility (TFI) and endometriosis-related infertility are common gynecological reproductive system diseases in the world, both of which seriously affect women's quality of life. 1,2 Although the two diseases have different pathogenesis, they can result in female infertility by affecting the uterine endometrium. 3 The fallopian tube is one of the important components of the female reproductive system. It has the important role of transporting sperm, ingesting eggs and transporting fertilized eggs to the uterine cavity. 4 TFI refers to women who are infertile caused by tubal adhesions, blockages, insu cient ovulation of the ovaries and other reasons. Recent studies have shown that patients with TFI are signi cantly thinner in endometrial thickness, which may be one of the pathological mechanisms leading to infertility. 5,6 Endometriosis is a common gynecological disease in women and it refers to the presence of the endometrium outside the uterus. There are various mechanisms for the pathogenesis of this disease, and the endometrial implantation theory is generally accepted. Recent studies have shown that pathological changes in the eutopic endometrium of endometriosis may be associated with cell proliferation, immunity, and implantation. 7 The endometrium plays an important role in the implantation of blastocysts. Under normal circumstances, the endometrium is a tissue that undergoes periodic changes in hormone regulation 8 , and it can only accept embryo implantation during the window of implantation (WOI). During the WOI, the activated blastocyst interacts with the endometrium in the receiving state to promote blastocyst implantation into the uterus. This process begins on days 4-5 after fertilization and ends approximately on days 9-10, which is equivalent to the 19th to 24th day of the menstrual cycle. During this period, a series of morphological and biochemical changes in the endometrium are inseparable from changes in maternal secretion and changes in the expression of numerous molecules in endometrial cells. 9 Recent studies have shown that with the rapid development of gene chips, high-throughput microarray have been applied to various aspects of disease pathogenesis, such as transcription, translation, and epigenetic regulation. [10][11][12] However, in the study of TFI and endometriosis-related infertility, the microarray technology of the two has not yet been popularized. In addition, the eutopic endometrium is also an important tissue in genetic analysis, which plays an important role in TFI and endometriosis-related infertility. However, the pathophysiologic mechanisms of TFI and endometriosis-related infertility have not been investigated thoroughly.
Here we selected patients who had been diagnosed with TFI and endometriosis-related infertility respectively in the Department of Obstetrics and Gynecology, the First A liated Hospital of Xiamen University (n=6, 3 patients per disease). In the endometrial window phase, the eutopic endometrial tissue was selected for microarray hybridization, and the DEGs and DELs of the two diseases were screened out, and the GO, KEGG pathway and TFs of DEGs were further analyzed by comprehensive bioinformatics methods. In addition, a ceRNA network was further constructed to identify the potential key genes. Excitingly, it is anticipated that the novel DEGs and pathways between FTI and endometriosis-related infertility identi ed in this study may help to seek potential molecular mechanisms. The work ow of this study was shown in Fig. 2A.

Ethics Statement
All procedures were performed according to standard protocols or the manufacturers' instructions and the study was approved by the ethics committees of the First A liated Hospital of Xiamen University.

Sample Collection
A total of 6 subjects (3 TFI patients and 3 endometriosis-related infertility patients) were recruited to participate, and all of the patients were of reproductive age with normal menstrual cycles. Herein, 3 patients with endometriosis were all patients whose endometrium occur in the ovary (group EMS).
Meanwhile, 3 patients with TFI refer to women who have been diagnosed with tubal obstruction by fallopian tube angiography before (group TFI). In addition, all patients in our study gave informed consent. The follicular were monitored by ultrasonography from the ninth day of the menstrual cycle. On the 6th-7th day after ovulation (the LH test showed double line and the ovulation sign was detected by B ultrasonic examination), endometrium was obtained after these checks performed. The tissue was divided into two parts: the rst part was xed by 10% formaldehyde and was sent for pathological examination and endometrial lesions were excluded. The second part was stored in liquid nitrogen, and then the total RNA was extracted to explore lncRNA and mRNA expression pro les by human lncRNA Array v4.0.

RNA Extractionand Quality Control
According to the manufacturer's instructions, total RNA extracted from tissue was performed using Trizol reagent (Invitrogen). RNA quantity and quality were assessed by spectrophotometric measurement (Nanodrop ND-1000). Denaturing agarose gel electrophoresis was used to evaluate the RNA integrity.
RNA was stored at -80°C for further experiments.

Microarray Analysis and Data Processing
Human LncRNA Microarray V4.0 (lncRNAs (40173) + coding genes (20730), 8 × 60 K Array Star, Rockville, MD, USA) designed for the global pro ling of human lncRNAs and protein-coding transcripts was used in our study. The microarray can detect about 40,173 lncRNAs and 20,730 coding transcripts, which are derived from authoritative databases (RNA-seq, GENECODE, RefSeq, UCSC, UCR, ect) and other landmark publications. Sample labeling and array hybridization were performed in terms of the Agilent One-Color Microarray-Based Gene Expression Analysis protocol (Agilent Technology) with minor modi cation. In brief, mRNA samples from endometrium were puri ed to remove rRNA and were ampli ed and transcribed into uorescent cRNA along the entire length of the transcripts without 3' bias utilizing a random priming method (Arraystar Flash RNA Labeling Kit, Arraystar). RNeasy Mini Kit (Qiagen) was employed to purify the labeled cRNA. After hybridization and washing, the arrays were scanned with Agilent DNA Microarray Scanner (part number G2505C). Raw data were extracted using Agilent Feature Extraction software (version 11.0.1.1). Furthermore, quantile normalization and subsequent data processing were done through GeneSpring GX v12.1 software package (Agilent Technologies). Log2 transformation was performed on the mRNA and lncRNA expression data. The average RNA expression value was used when duplicated data were found. Microarray hybridization and expression data collection were performed by Kangchen Bio-tech, Shanghai, China. The microarray analysis was conducted in School of Medicine, Xiamen University, Xiamen, Fujian, China. Finally, we obtained lncRNA and mRNA expression pro les.

Identi cation of DEGs
Herein, R software and limma package (http://www.bioconductor.org/packages/release/bioc/html/limma.html) in Bioconductor were used to identify DEGs by comparing mRNA expression differences in eutopic endometrium tissues from TFI and endometriosis-related infertility respectively. 13 Normalized data of mRNAs in two groups of samples (TI and EMS) was shown in Fig. 1C. DEGs were identi ed by fold-change screening at a threshold of 2.0-fold or greater, and a p-value < 0.05.

Enrichment Analyses of DEGs
To understand the biological roles of these DEGs associated with TFI and endometriosis-related infertility, pathway analysis (based on Kyoto Encyclopedia of Genes and Genomes, http://www.genome.jp/kegg/) and Gene Ontology (GO) (http://www.geneontology.org) analysis were assessed by the clusterPro ler package in R software to explore the signi cant pathways related to DEGs14. adjust. P < 0.05 was used as thresholds to de ne markedly enriched GO terms/pathways. GO is a series of semantics used to describe the characteristics of genes. These semantics are mainly divided into three kinds: Biological Process (BP) , an orderly biological process with many steps, such as cell growth, differentiation and maintenance, apoptosis and signal transduction, Molecular Function (MF), which mostly refers to the functions of individual gene products, such as binding activity or catalytic activity, and Cellular Component (CC), which is used to describe the location of gene in cells, such as cytoplasm, nucleus, endoplasmic reticulum or mitochondria. KEGG is a database for systematic analysis of metabolic pathways of gene products in cells, and it is the most commonly used metabolic pathway analysis.

TFs-DEGs Network Construction
Genes are often regulated by TFs at the transcription level. Therefore, DAVID (Database for Annotation, Visualization and Integrated Discovery) 6.8 (https://david.ncifcrf.gov/) was used for TFs enrichment analyses to investigate DEGs at the transcriptional level. Herein, the p-value (Benjamini) ≥ 0.05 was taken as the exclusion standard and the TFs-mRNAs network was constructed by Cytoscape 15 software. DAVID is a gene functional classi cation tool that integrates a set of functional annotation tools for investigators to analyze biological functions behind massive genes. Currently, the DAVID database is mainly used for the analysis of functions, pathways and TFs enrichment of genes.

Analysis of the Important Modules and Hub Genes
To analysis the important modules and Hub genes associated with endometriosis-related infertility, we constructed a PPI network of DEGs through the STRING 11.0 (https://string-db.org/) online database with the basic settings (meaning of network edges: evidence, minimum required interaction score: medium con dence) and visualized the results by Cytoscape software. Then, the important modules in the PPI network was analyzed by Molecular Complex Detection (MCODE) plug-in with the max depth, haircut on, node score cut-off, degree cut-off and k-core set as 100, 0.2, 0.2, 10 and 2, respectively. Gene ontology and pathway enrichment analysis of genes in each module (module 1, module 2 and module 3) were performed by DAVID database and the cut-off criterion of p-value (Benjamini) was set as < 0.05.
Moreover, in this study, we screened out the top 50 Hub genes in the PPI network by cytoHubba 16 plug-in and predicted their TFs by iRegulon 17 plug-in. The top 5 TFs in the NES scores were displayed. STRING, a database that predicts interactions between proteins or genes, includes both direct physical interactions and indirect functional correlations between proteins. MCODE plug-in can discover closely related regions in PPI network, which may represent molecular complexes. The MCC algorithm of CytoHubba plug-in can help us discover key genes and subnetwork in PPI network. IRegulon, a plug-in that predict TFs of gene sets, predicts TFs by calculating motif enrichment analysis and uses multiple position weight matrices (PWM) to score each motif.

Identi cation of DELs
The normalized expression of lncRNAs in two groups of samples (TI and EMS) was shown in Fig. 1D. Moreover, normalized data of lncRNAs was used for differential analysis by limma package in R software and the DELs were identi ed by fold-change screening at a threshold of 2.0-fold or greater, and a p-value < 0.05.

CeRNA Network Construction
It is well known that the DEGs is regulated by miRNAs and lncRNAs at transcriptional level. The combination of miRNAs and mRNAs can promote the degradation of mRNAs. At the same time, lncRNAs can competitively combine with mRNAs to further inhibit the degradation of mRNAs by miRNAs. The interaction network of lncRNA, miRNAs and mRNAs are called competing endogenous RNA (ceRNA) network. Herein, in order to better understand the physiological changes of eutopic endometrium between patients with TFI and endometriosis-related infertility, we used R software and Perl software to construct the ceRNA network (DEGs-miRNAs-DELs) through the miRDB database (miRNA-mRNA, http://mirdb.org/), miRTarBase database (miRNA-mRNA, http://mirtarbase.mbc.nctu.edu.tw/index.html), TargetScan database (miRNA-mRNA, http://targetscan.org/vert_72/) and miRcode database (lncRNA-miRNA, http://www.mircode.org/index.php) database.
Herein, rst, DELs-miRNAs interaction pairs were integrated using the miRcode database. Second, target genes of miRNA signatures were obtained using three databases: miRDB, miTarBase, and TargetScan. Genes present in all three databases were regarded as target genes of miRNAs. Comparing predicted target genes of miRNAs with the DEGs, only the remaining overlapping genes and their interaction pairs were used for further analysis. Then, the ceRNA networks (DEGs-miRNAs-DELs) were constructed by Cytoscape software according to DELs-miRNAs pairs and miRNAs-mRNAs (overlapping genes) pairs.

TFs Prediction of MRNA in CeRNA Network
MRNA and miRNA can be regulated by TFs at the transcription level. Herein, in order to predict the upstream regulatory mechanism of mRNAs related to endometriosis-related infertility, we predicted the TFs of mRNAs in ceRNA network by iRegulon plug-in in Cytoscape software and we only showed the top 10 TFs in the NES scores in this study.

Screening of Potential Key Genes
In order to screen out the potential key genes related to endometriosis-related infertility. In this study, we selected the lapping genes of important modules, Hub genes and ceRNA network as the potential key genes. At the same time, we also predicted the TFs of potential key genes and miRNAs (target to potential key genes) by iRegulon, respectively. The top 5 TFs in the NES scores were displayed.

TFBs Predition of Potential Key Genes
In this study, we predicted transcription factor binding sites (TFBs) for these potential key genes that bind to TFs by JASPAR 2020 online software (http://jaspar.genereg.net/). JASPAR 18 is a collection of transcription factor DNA-binding preferences, modeled as matrices. These can be converted into Position Weight Matrices (PWMs or PSSMs), used for scanning genomic sequences. JASPAR is an open-access database of curated, non-redundant transcription factor (TF) binding pro les stored as position frequency matrices (PFMs) and TF exible models (TFFMs) for TFs across multiple species in six taxonomic groups. The higher the score, the greater the likelihood that the TF will bind to the DNA binding site (motif).

MRNA and LncRNA Expression Pro les
Through comparing the quality of expression values of microarray before and after normalization, we found that the medians of expression value of 6 samples were in a straight line (Fig. 1A and Fig. 1B). Microarray analysis was used to assess the expression levels of mRNAs in eutopic endometrial tissue of TFI and endometriosis-related infertility. We identi ed 17965 mRNAs: 404 were up-regulated, 549 were down-regulated (FC ≥ 1.0, p-Value < 0.05), while 17012 were not differential expressed. The variation of mRNA expression between the TFI and endometriosis-related infertility is shown with a scatter plot (Fig.   1C).
Using microarray analysis, we identi ed 18912 lncRNAs in eutopic endometrial tissue of TFI and endometriosis-related infertility: 498 were up-regulated, 705 were down-regulated (≥ 1.0 fold, p < 0.05), while 17709 were not differential expressed. The variation of lncRNA expression between the TFI and endometriosis-related infertility is shown with a scatter plot (Fig. 1D).

Identi cation of DEGs
In this study, 508 DEGs (FC ≥ 2.0, p-Value < 0.05) between TFI and endometriosis-related infertility were screened out, including 203 up-regulated mRNAs and 305 down-regulated mRNAs. The volcano plots display the expression pro le for all detected mRNAs and the top 10 genes with p-value from small to large are shown in Fig. 2B. In addition, the top 50 most signi cantly differentially expressed mRNAs are shown in Fig. 2C, and the clustering analysis revealed the relationships among mRNA expression patterns in different samples.

Enrichment Analyses of DEGs
GO functional enrichment analysis showed that DEGs were mainly involved in GO terms about transmembrane transporter activity (eg, passive transmembrane transporter activity, organic anion transmembrane transporter activity, G protein-coupled receptor binding, cation transmembrane transporter activity and so on; Fig. 3A). The gene transcription pro le was signi cantly different between TFI and endometriosis-related infertility. KEGG pathway enrichment analysis were mainly enriched in immune system (eg, Cell adhesion molecules (CAMs), Intestinal immune network for IgA production, Viral protein interaction with cytokine and cytokine receptor and so on; Fig. 3B).

TFs-DEGs Network Construction
In this study, 52 TFs were enriched, including RSRFC4, LHX3, SRY, S8, and so on (Table 1). So, the visualization of TFs-mRNAs network was shown in Fig. 3C.

Analysis of the Important Modules and Hub Genes
A total of 495 nodes and 1040 edges were mapped in the PPI network of the identi ed DEGs and these DEGs demonstrated signi cant interactions of DEGs each other. In addition, 3 modules were selected by MCODE plug-in (Fig. 4A). As presented in Fig. 4B, GO functional enrichment analysis and KEGG pathway enrichment analysis of genes in each module showed that these genes were mainly enriched in immunity (eg, chemokine activity, cell chemotaxis, chemotaxis, in ammatory response as well as immune response and so on). The top 50 Hub genes screened out by cytoHubba were selected and sequentially ordered as Screening of DELs 576 DELs between TFI and endometriosis-related infertility were screened out, including 225 up-regulated lncRNAs and 351 down-regulated lncRNAs. Among these DELs, the top 10 DELs with p-value from small to large were shown in Fig. 2D. In addition, the top 50 most signi cantly differentially expressed lncRNAs are shown in Fig. 2E, and the clustering analysis revealed the relationships among mRNA expression patterns in different samples.

Construction of the CeRNA Network commbining TFs and Screen of Potential Key Genes
A ceRNA network (21 mRNAs, 13 miRNAs and 16 lncRNAs) was constructed and is presented in Fig. 5A. At the same time, the top 10 TFs of mRNAs in ceRNA network were visualized in the ceRNA network combining TFs by Cytoscape software according to NES ranking. After counting the interaction numbers of each RNA, we found the lapping genes (potential key genes) of important modules, Hub genes and ceRNA network include 2 mRNAs (PLAU and LDLR). At the same time, the top 10 TFs of potential key genes and miRNAs were visualized according to NES ranking respectively. Herein, a new ceRNA network combining TFs based on potential key genes, including 2 mRNAs (PLAU and LDLR), 10 TFs (eg. SRF, FOS, which target to mRNA And HNF1A, POU3F2, which target to miRNA), 5 miRNAs (hsa-miR-301b-3p, hsa-miR-27a-3p, hsa-miR-20b-5p, hsa-miR-193a-3p and hsa-miR-17-5p) and 8 lncRNAs (eg. LMO7-AS1, ITFG3 and MGST1, etc.) ( Fig. 5B and Fig. 5C) was shown in Fig. 5D, and the Transcription Factor Binding site (TFBs) of potential key genes was shown in table 3. Herein, we only show the binding sites with the highest score for each TF.

Discussion
TFI and endometriosis-related infertility are well known to all. However, the underlying molecular mechanisms between TFI and endometriosis-related infertility remain well unknown. In this study, we made use of eutopic endometrial tissue from TFI and endometriosis-related infertility patients with active disease to screen out the DEGs and DELs of the two diseases, and subsequently analyzed for their biological relevance by using GO, KEGG pathway. In addition, a ceRNA network was further constructed to identify the potential key genes. As a result, we found the DEGs between TFI and endometriosis-related infertility to identify the factors for the pathogenetic or therapeutic relevance in endometriosis-related infertility, including "PLAU" "LDLR". At the same time, the TFs of these 2 potential key genes were "IKZF2", "TFAP4", "HSF1", "FOS", "SRF".
Studies are discussed that the high expression of "LDLR" gene leads to the occurrence of hyperlipidemia, 19,20 which further interferes with endothelium-related vasomotor contraction, leading to the disorder of vascular hemodynamic regulation. 21 By stimulating the release of platelet-related pselectin, it induces leukocyte adhesion, and increases blood viscosity and blood ow resistance. 22 Secondly, the activation of neutrophils can extend pseudopodia and adhesion with other cells (such as white blood cells, endothelial cells, platelets, etc.), blocking the microvessels. 23,24 A large number of studies have shown that in ammation response plays an important role in the pathogenesis of endometriosis-related infertility. [25][26][27][28] For example, endometriosis leads to pelvic in ammatory response and produces a variety of cytokines, which can gobble up sperm or fertilized eggs, inhibit oviduct motility, and thus interfere with fertility. Another key gene "PLAU" gene located on 10q22.2 encodes urokinase type plasminogen activator (uPA), transforming plasminogen into plasminogen. 29 And previous studies have shown that uPA protein is mainly located in the cytoplasm, and both ectopic and eutopic endometrium express uPA protein, 30 which is positively expressed in vascular endothelial cells. 31 It is involved in the occurrence and development of endometriosis.
Making mention of the TFs of these 2 potential key genes, IKZF2 is one of the huge TFs that regulate the development of the lymphatic system as well as plays an important role in the development of T, B lymphocytes and NK cells. 32 FOS can effectively reduce the amount of serum cholesterol, triglycerides, free fatty acids. 33 TFAP4 is an important downstream factor of c-MYC and widely regulates cell proliferation and differentiation, cell cycle, apoptosis, immune response and other processes. There have been growing evidences that elevated TFAP4 expression signi cantly correlates with tumor progression and poor prognosis in a number of malignancies, including colorectal cancer, 34,35 gastric cancer 36 and non-small cell lung cancer. 37 As for the mechanism of TFAP4 in different diseases, studies showed that TFAP4 overexpression promoted Wnt/β-catenin pathway activation, 38 PI3K/AKT signaling pathway activation, 39 p53 pathway activation 40 and WNT signaling activation 41 etc. However, whether TFAP4 plays a role in endometriosis-related infertility is challenging.
HSF1 is the master transcriptional regulator of the heat shock response in eukaryotes, 42 which is a signaling pathway in cells that regulates the expression of molecular chaperones in the presence of thermal and other environmental stresses. 43 Research clearly shows that Hsp90 is a major repressor of HSF1 gene. HSF1 has been shown to contribute to cancer progression, 44 organismal lifespan 45 and protect against obesity. 46 By contrast, a lack of HSF1 activity has been suggested to contribute to neurodegenerative diseases. 47 Meanwhile, some studies suggested that heat shock protein may be attached importance to the pathogenesis of EMS SRF, a member of the MADS transcription factor family, regulates the transcription of a series of genes by binding to serum response element (SRE). It mainly shows rapid gene response to ischemia, hypoxia and other stimuli, and affects cell development, differentiation, proliferation and apoptosis, including mesoderm formation, 48 cardiac development, 49 angiogenesis, oligodendrocyte differentiation, 50 neuronal migration, 51 and circadian regulation. 52 Mitogen-activated protein kinase (MAPK) and RhoA signaling pathway are two important signaling pathways that activate intracellular SRF expression and can promote the transcription of growth-related genes and cell cycle genes. In addition, SRF also plays an important role in the biodynamics of the actin skeleton and participates in the regulation of cell migration ability.
Those crucial genes and TFs could re ect the key biomarkers with mechanistic relevance for endometriosis-related infertility pathogenesis and progression, some of them well established and already exploited for therapeutic purposes. Herein, those DEGs and DELs could contribute to promote the diagnostic and therapeutic interest in endometriosis-related infertility on relevant and innovative molecules or pathways, which could indicate a new direction of the acquaintance of endometriosisrelated infertility. Based the DAVID database, in the GO and KEGG pathway enrichment analyses of DEGs the domain terms including transmembrane transporter activity and immune system, indicating that endometriosis-related infertility is characterized by transmembrane transporter and immune processes in the process of disease. Besides, the TFs of miR in the ceRNA were mainly enriched in "MKX", "EN1", "HNF1A", "POU3F2" and "HDAC2".
MKX is a member of the three-amino-acid loop (TALE) superclass of atypical homeobox genes belonging to the Iroquois family. 53 MKX is expressed in the embryonic progenitor cell populations of the cartilage, skeletal muscle, tendons and bones, in addition to the tips of the ureteric buds in the metanephric kidneys and the testis cords of the male gonad testis in the mouse embryo. [54][55][56][57] Recently, research revealed that MKX was a central transcription factor that regulated AF development, maintenance and regeneration. Meanwhile, MKX is required for the maturation of collagen brils and plays a crucial role in tendon maturation by regulating the expression of type I collagen. 58 Additionally, activation of Wnt/β-catenin signaling reduces gene expressions of Mkx. 59 EN1, a neural-speci c transcription factor, plays a crucial role in the development of many tissues and organs. 60 It can promote cell survival and cell resistance to apoptotic stress, thereby promoting dopaminergic neuronal-cell longevity throughout adulthood. 61 EN1 is highly expressed in many tumors, such as breast tumors, salivary gland adenoid cystic carcinoma, and adenoid cystic carcinoma. 62 HNF1A, hepatocyte nuclear factor 1 homeobox A, was subsequently veri ed to be expressed in several organs including the liver, pancreas, kidney, and intestine. 63 In addition, HNF1A has been reported to inhibit Wnt and NF-κB signalling during hepatocarcinogenesis and hepatocellular carcinoma metastasis by transcriptionally regulating the expression of miR-194. 64 Study also revealed that STAT3-induced upregulation of HNF1A-AS1 promoted OSCC progression by activating NOTCH signaling pathway. 65 POU3F2 belongs to a large family of brain-speci c homeobox transcription factors and is expressed in the central nervous system (CNS) during neuronal development and in adult brain. POU3F2 has been shown to regulate the expression levels of critical genes at different stages of neural differentiation, the migration of cortical neuron, the neurogenesis and positioning of cortical neurons, etc. 66 In addition, it is also involved in different types of cancer, such as melanoma, lung cancer, stomach cancer. 67 It is concluded that POU3f2 affects the NF-κB, MAPK/JNK and Myc/Max pathways in cancer pathway reporter array.
HDAC2, a subtype of histone deacetylase (HDAC) that is a group of enzymes that regulate lysine acetylation and thereby protein function, mainly plays a positive role in cardiac hypertrophy. Gradually, HDAC2 was found to be aberrantly expressed in several types of cancer including gastric, colorectal, prostate and Hodgkin's Lymphoma. Later, study has reported that HDAC2 was recruited to CDKN1A promoter by FOXO3a and regulates p21 expression in cerebellar granule neuron. 68 In hepatocellular carcinoma, it also was shown to be related to P21. Meanwhile, some researchers have found that sumoylation of HDAC2 promotes NF-κB-dependent gene expression. 69 In conclusion, according to integrated bioinformatical analyses of multiple cohort's pro le datasets in the eutopic endometrial tissue between TFI and endometriosis-related infertility patients, we identi ed the 508 DEGs. Other studies could also use some methods to explore the releation of other genes and endometriosis-related infertility, and futher gain new insights into important biological processes. But our study erplored the potential key gene associated with endometriosis-related infertility and the relevant and innovative molecules or pathways, which could indicate a new direction of the acquaintance of endometriosis-related infertility. As for the relationships between these TFs and endometriosis-related infertility, little is known and more studies are needed. In a word, these candidate genes and pathways might provide novel insight into its occurrence of endometriosis-related infertility, and be therapeutic targets and biomarkers for endometriosis-related infertility treatment.

Conclusion
In conclusion, physiological of TFI and endometriosis related infertility may be related to the transmembrane transporter activity and the immune system. These potential key RNAs and TFs might have clinical utility for the diagnosis and prognosis prediction in TFI and endometriosis-related infertility.
The results of the current study might lay the foundation for future basic and clinical research.

Availability of Data and Materials
The expression data associated with this study were sequenced by endometrial transcriptome microarray in 6 volunteers.
Authors' contributions JZL and LLR contributed equally to this work. They mainly achieved all the experimentation. CY, RFW and ZXH participated in data analysis and discussion. QHC organized and supervised the project. All authors read and approved the nal manuscript.

Ethics Approval and Consent to Participate
All procedures were performed according to standard protocols or the manufacturers' instructions and the study was approved by the ethics committees of the First A liated Hospital of Xiamen University.   Figure 1