Identification of core genes in the progression of endometrial cancer and cancer cell-derived exosomes by an integrative analysis

Endometrial cancer is one of the most prevalent tumors of the female reproductive system causing serious health effects to women worldwide. Although numerous studies, including analysis of gene expression profile and cellular microenvironment have been reported in this field, pathogenesis of this disease remains unclear. In this study, we performed a system bioinformatics analysis of endometrial cancer using the Gene Expression Omnibus (GEO) datasets (GSE17025, GSE63678, and GSE115810) to identify the core genes. In addition, exosomes derived from endometrial cancer cells were also isolated and identified. First, we analyzed the differentially expressed genes (DEGs) between endometrial cancer tissues and normal tissues in clinic samples. We found that HAND2-AS1, PEG3, OGN, SFRP4, and OSR2 were co-expressed across all 3 datasets. Pathways analysis showed that several pathways associated with endometrial cancer, including “p53 signaling pathway”, “Glutathione metabolism”, “Cell cycle”, and etc. Next, we selected DEGs with highly significant fold change and co-expressed across the 3 datasets and validated them in the TCGA database using Gene Expression Profiling Interactive Analysis (GEPIA). Finally, we performed a survival analysis and identified four genes (TOP2A, ASPM, EFEMP1, and FOXL2) that play key roles in endometrial cancer. We found up-regulation of TOP2A and ASPM in endometrial cancer tissues or cells, while EFEMP1 and FOXL2 were down-regulated. Furthermore, we isolated exosomes from the culturing supernatants of endometrial cancer cells (Ishikawa and HEC-1-A) and found that miR-133a, which regulates expression of FOXL2, were present in exosomes and that they could be delivered to normal endometrial cells. The common DEGs, pathways, and exosomal miRNAs identified in this study might play an important role in progression as well as diagnosis of endometrial cancer. In conclusion, our results provide insights into the pathogenesis and risk assessment of endometrial cancer. Even so, further studies are required to elucidate on the precise mechanism of action of these genes in endometrial cancer.


Results
Identification of DEGs in endometrial cancer. We used basic packages implemented in R to identify potential DEGs across the three data sets. Based on a cut-off criteria of adjust P-value <0.05 and |logFC | >1.0, we identified a total of 2346, 340, and 233 genes in GSE17025, GSE63678 and GSE115810, respectively. The top 10 DEGs with the largest fold changes in the three GEO data sets are shown in Tables 1-3, while the top 100 DEGs in each dataset, represented by heat maps are shown in Fig. 1b-d. After expurgating duplicate genes and expression values which lacked specific gene symbols, we combined the top 100 DEGs in GSE17025, GSE63678 and GSE115810 to create a Venn diagram. Five genes were found to be co-expressed across all three data sets, including HAND2-AS1, PEG3, OGN, SFRP4, and OSR2. All these genes were, however, down-regulated in endometrial cancer tissues (Fig. 2). Three genes, including FOSB, FOS and LTBP4, were co-expressed in GSE17025 and GSE115810. In addition, a total of 10 genes, including CFB, RBM47, WT1, ADAMTS5, TSPYL5, HTR2B, ANK2, FOXL2, CXCL12 and EFEMP1, were co-expressed in GSE63678 and GSE115810. On the other hand, 13 genes, including BCHE, LMOD1, TPX2, CXCL9, LIPG, TFAP2A, DLGAP5, MMP12, CEP55, RRM2, MELK, TOP2A and ASPM were co-expressed in GSE17025 and GSE63678,.

Rank
Probe set ID Gene symbol Description Regulation  www.nature.com/scientificreports www.nature.com/scientificreports/ Biological process analysis of DEGs. To gain more insights into the function of the DEGs, we performed a GO analysis. Sorting the transcripts by false discovery rate (FDR) revealed that there are 15, 20 and 10 GO terms of biological processes in GSE17025, GSE63678 and GSE115810, respectively ( Fig. 3a-c). There was no common biological processes were found across the 3 datasets. Nevertheless, we found 6 common biological processes including " cell division ", "mitotic nuclear division", " cell proliferation ", " sister chromatid cohesion ", " chromosome segregation ", and "mitotic sister chromatid segregation" in GSE63678 and GSE17025. Other biological processes, such as "epithelial cell differentiation", "apoptotic process ", "extracellular matrix organization", " cell development ", "blood vessel development" and " extracellular structure organization ", were also identified very vital in each dataset.
Signaling pathway enrichment analysis. The potential pathways regulated by the DEGs in endometrial cancer were explored using KEGG analysis, We found that specific pathways were enriched in each dataset ( Fig. 4a-c). Bubble plots from this analysis presented the most significantly enriched pathways (Fig. 4a-c). Among them, "Pathways in cancer" was the common pathway in the three datasets. The top 10 pathways are as outlined in Table 4. Of these, 7 pathways, including those involved in p53 signaling pathway, Glutathione metabolism, Cell cycle, ECM-receptor interaction, PI3K-Akt signaling pathway, Focal adhesion and MAPK signaling were notable because of their association with cancer progression. Particularly, these pathways were implicated in processes such as cell proliferation, cell migration and invasion. Pathways involved in HIF-1 signaling pathway, protein digestion and absorption as well as FoxO signaling pathway were also other interesting pathways following pairwise comparisons.
Validation of DEGs in TCGA database. To further validate and determine the core genes highly associate with endometrial cancer from these DEGs, eight down-regulated and five up-regulated genes that showed the most significant fold changes (top10) between endometrial cancer and normal tissue in GSE17025, GSE63678 and GSE115810, were selected. Meanwhile, these 13 genes are derived from DEGs that overlap across the three data sets (Fig. 2). These included HAND2-AS1, PEG3, OGN, SFRP4, TFAP2A, MMP12, TOP2A, ASPM, FOSB, CFB, WT1, FOXL2 and EFEMP1. Hypothesis test analysis proves that these 13 genes have statistical significance (data not shown). Next, we explored the Gene Expression Profiling Interactive Analysis (GEPIA) database for identification and retrieval of the expression profiles of these DEGs in both endometrial cancer and normal tissues. We did not find any significant differences in FOSB expression in the GEPIA database between the treatments. However, expression profiles of the remaining twelve DEGs were consistent with our analysis results in GEO database. Particularly, expression of TFAP2A, MMP12, TOP2A, ASPM and CFB were higher in endometrial cancer relative to normal tissues. On the other hand, HAND2-AS1, PEG3, OGN, SFRP4, WT1, FOXL2 and EFEMP1 were significantly lower in endometrial cancer tissues than in normal tissues (Fig. 5).
Survival analysis of core genes. We then used prognostic information of the 12 core genes, available in the free online TIMER database (Tumor IMmune Estimation Resource) (https://cistrome.shinyapps.io/timer/), to perform a survival analysis. Aberrant expression levels of four genes out of 12 core DEGs showed a significant (P < 0.05)association with poor prognosis of endometrial cancer, and this included two up-regulated (TOP2A and ASPM) and two down-regulated (FOXL2 and EFEMP1) genes ( Fig. 6a-d). This indicated that the four genes might play important roles in the progression of endometrial cancer.

Validation of core genes expression in endometrial cancer cells. Given the lack of clinical samples,
we only used three cell lines in this study to validate the results of endometrial cancer from bioinformatics analyses. Ishikawa and HEC-1-A cell lines were derived from endometrioid adenocarcinoma, while 4003 was the   www.nature.com/scientificreports www.nature.com/scientificreports/ normal endometrial cells and was used as a control. Real-time qPCR assay was employed to quantify expression levels of 4(TOP2A, ASPM, EFEMP1, and FOXL2) core DEGs across the three types of cells. Results showed a significant down-regulation of EFEMP1 and FOXL2 in endometrial cancer cells (Fig. 7c,d). A comparison with normal cells showed that TOP2A was highly expressed in endometrial cancer cells (Fig. 7a), which was consistent with the resultsfrom bioinformatics analysis above. However, we noted that ASPM was only significantly expressed in HEC-1-A cells (Fig. 7b).

Exosomes secretion by endometrial cancer cells.
Cancer progression is always related to the microenvironment. Recently, exosomes have been found to modulate cancer cell communication as a novel way and almost all types of cancer cells can release exosomes into their microenvironment 10 . In this study, we hypothesized that endometrial cancer cells could secrete exosomes and pass them on to recipient cells, thereby promoting cancer development. Consequently, we isolated exosomes from the culture supernatants of Ishikawa and HEC-1-A cells (Fig. 8a). The concentration and particle distribution of exosomes were first characterized using NTA analysis, and this revealed average sizes of 127.2 nm and 146.9 nm for Ishikawa and HEC-1-A, respectively (Fig. 8b). Results from Western blotting analysis showed presence of CD63, an exosome marker, in tumor cell-derived exosomes. Conversely, no endoplasmic reticulum protein CANX was detected in the exosomes (Fig. 8c). TEM results showed that the vesicles from endometrial cancer cells were either round or oval, with a complete lipid www.nature.com/scientificreports www.nature.com/scientificreports/ bilayer membrane structure (Fig. 8d,e). Previous studies have demonstrated presence of miRNAs in exosomes, which are the main source of circulating miRNAs 24 , and play a vital role in cancer development 25 . In this study, we observed that FOXL2 was down-regulated in endometrial cancer tissues and was significantly associated with endometrial cancer. Previous studies have shown that FOXL2 is regulated by miR-133a 26 (Fig. 8f). Based on this, we detected the expression of miR-133a in all three cell lines. Compared with normal endometrial cancer cells, we found that miR-133a was highly expressed in endometrial cancer cells (Fig. 8g). At the same time, we also detected miR-133a expression in Ishikawa and HEC-1-A-derived exosomes (Fig. 8h). In addition, exosomes derived from endometrial cancer cells could be taken up by normal endometrial cells (Figure S), indicating that endometrial cancer cells release exosomes, which might be then transported to normal endometrial cells.

Discussion
Endometrial cancer occurs due to abnormal expression of many cancer-related genes 27 , many of which are associated with the susceptibility and development of endometrial cancer 22,23 . Numerous molecular approaches have been performed to identify biomarkers for endometrial cancer 28 . Nonetheless, most of them only focused on a single genetic aspect, limiting the reliability of the biomarkers 29 . In the present study, we used three GEO datasets to perform an integrative analysis with a view of identifying DEGs between normal and cancerous endometrial tissues. The results were validated using the TCGA database. Key biological processes and pathways based on the DEGs between endometrial cancer and normal tissues were also identified. Particularly, we focused on the common DEGs across the three GEO datasets as well as those that were significantly expressed in each dataset. Moreover, we identified exosomes derived from endometrial cancer cells.
Activation of cell proliferation process and p53 signaling pathways have been previously associated with cancers development 30,31 . In this study, GO and KEGG analysis showed that DEGs are involved in progression of endometrial cancer through these biological processes and pathways. A total of 12 DEGs, includingHAND2-AS1, PEG3, OGN, SFRP4, TFAP2A, MMP12, TOP2A, ASPM, CFB, WT1, FOXL2, and EFEMP1, were identified as candidate genes for endometrial cancer. Furthermore, a cross comparison of common DEGs across the three datasets revealed that TOP2A, ASPM, FOXL2 and EFEMP1 were potential biomarkers for distinguishing www.nature.com/scientificreports www.nature.com/scientificreports/ endometrial cancer, and this was significantly associated with the survival ratios among patients. Further validation in endometrial cancer cells showed that indeed TOP2A, ASPM, FOXL2 and EFEMP1 were significantly regulated. Particularly, TOP2A was significantly up-regulated in endometrial cancer tissues and cells, consistent with Lapinska-Szumczyk et al. 32 . These results further indicated that high levels of TOP2A can lead to EC progression. Previous studies have reported that aberrant expression of TOP2A is related to the stage of endometrial cancer and its progression 33 . Functionally, TOP2A encodes a DNA topoisomerase that controls and alters the topologic states of DNA during transcription 34 , and is involved in processes such as chromosome condensation, DNA transcription and replication 35 . High levels of TOP2A have also been detected in other types of cancers, including ovarian cancer 36 and prostate cancer 37 . Based on these results, therefore, we hypothesized that abnormal expression of TOP2A could be a positive tumor metastasis marker and a poor biomarker for prognosis. In addition, inhibition of TOP2A has been found to mediate a decrease in migration, proliferation or invasion of cells 38 . These results might lead to development of a new treatment strategy for endometrial cancer. Abnormal spindle-like microcephaly-associated (ASPM), which encodes the ASPM protein, plays an essential role in mitotic spindle function during cell replication 39 . ASPM is associated with cell cycle and mediates the tightly coordinated Ubiquitin-Cyclin E-Retinoblastoma-E2F bistable-signaling pathway to control restriction point progression and stem cell maintenance 40 . Previous studies have demonstrated that overexpression of the ASPM gene is associated with progression and poor outcomes in various types of cancer, such as bladder cancer 41 , prostate cancer 42 , and breast cancer 43 . However, none of these studies have elucidated the relationship between ASPM and endometrial cancer. In this study, our results first showed that high levels of ASPM were related to endometrial cancer. When combined with the function of the gene, ASPM is expected to become a new molecule for detection and treatment of this disease although its mechanism in endometrial cancer is yet to be studied.
In addition to the above findings on DEGs, EFEMP1 and FOXL2 were found to be down-regulated in endometrial cancer tissues. The epidermal growth factor-containing fibulin-like extracellular matrix protein 1(EFEMP1), functions as an oncogene or a suppressor, depending on the types of tumor. With previous studies showing that EFEMP1 is a new candidate tumor suppressor gene in endometrial carcinoma 44 . Our findings showed that EFEMP1 was down-regulated in endometrial cancer tissues, as well as endometrial cancer cell lines. Functionally, EFEMP1 inhibits migration of tumor cells by regulating MMP2 and MMP9 via ERK1/2 activity 45,46 . The findings www.nature.com/scientificreports www.nature.com/scientificreports/ on EFEMP1 in endometrial cancer revealed that the gene could inhibit tumor growth and invasion both in vitro and in vivo consistent with what has been reported in other types of tumors 44 . A separate studies found that EFEMP1 was elevated in highly invasive ovarian cancer, compared with the low invasive types 47 . Combined, these studies suggest that EFEMP1 plays different roles in the development of different types of carcinoma. FOXL2 on its part encodes a fork head transcription factor. Although many RNA-sequencing studies have been performed to understand the genetic regulation of endometrial cancer, the relationship between FOXL2 and the disease is still lacking 48 . In the current study, analysis on GSE115810 showed that FOXL2 was one of the top10 DEGs found to be down-regulated during the development of endometrial cancer. On the other hand, we observed a significantly low FOXL2 expression in endometrial cancer relative to normal cells. A related study showed that that FOXL2 could suppress cells proliferation and enhance cells apoptosis in cervical squamous cancer, mainly by decreasing Ki67 expression 49 . Various conclusions have been drawn regarding the mechanism of action of FOXL2 in other types of tumors 50 . However, its function remains unclear in endometrial cancer thereby necessitating the need for further clarification using molecular experiments.
A vast array of researches have demonstrated that almost all types of cells can release exosomes that mediate the cell-cell communication 51 , especially tumor cells 52 . Exosomal cargos, particularly those associated with development of cancer, have been considered the best biomarkers for non-invasive diagnostic 53 . Furthermore, studies have revealed that miRNAs are mainly packaged in exosomes. In our study, results from the GO analysis showed that the DEGs were associated with extracellular exosomes. Here, we assumed that the progression of endometrial cancer was mediated by exosomal miRNAs derived from cancer cells. Meanwhile, we observed a down-regulation of FOXL2 in endometrial cancer cells while the expression of miR-133a, which targets FOXL2 54 , was high in the cancerous cells. We found that endometrial cancer cells could secrete exosomes, which contained miR-133a. Exosomal miR-133a may regulate the down-regulation of FOXL2 in endometrial cancer tissues. Furthermore, the progression of endometrial cancer was not only achieved by malignant epithelial cells, but also through involvement of stromal cells. Previous research evidence indicated that extracellular vesicles mediate he communication www.nature.com/scientificreports www.nature.com/scientificreports/ between tumors and stroma 55 . In the current study, it was found that exosomes derived from endometrial cancer cells could be taken up by stromal cells, indicating that exosomal miRNAs may contribute to the progression of endometrial cancer.
In summary, four core genes (TOP2A, ASPM, EFEMP1 and FOXL2) and several interesting pathways involved in endometrial cancer were identified. Analysis of qRT-PCR assay revealed that expression of mRNA for TOP2A and ASPM are up-regulated in endometrial cancer cells, whereas those of EFEMP1 and FOXL2 are down-regulated. Currently, there are no studies on the role of ASPM and FOXL2 in endometrial cancer. Based on the findings of this study, we propose that multiple molecules can be effectively used to diagnose endometrial cancer. Accordingly, further studies into the functions of these genes can guide development of treatment strategies for EC. In addition, exosomes secreted by endometrial cancer cells, contain miRNAs that regulate FOXL2 and can be transported to normal stromal cells. These results open up new frontiers for the development of approaches for detection and treatment of endometrial cancer. However, further studies are required to elucidate the function and underlying mechanisms of action of these potential biomarkers in the progression of endometrial cancer.

Materials and Methods
Study design. The aim of this study was to identify core genes involved in the development of endometrial cancer, using bioinformatics analysis on GEO datasets. A flow diagram of the study design is shown in Fig. 1a. Initially, three GEO datasets (GSE17025, GSE63678 and GSE115810) were reasonably searched in published databases, and then bioinformatics tools used to screen DEGs in the datasets. Analyses on the functions of GO and KEGG pathways were also performed on the DEGs. Next, Genes co-expressed in the three datasets were identified and verified in TCGA database. Survival analyses were performed to further identify core genes related Isolation and identification of exosomes from endometrial cancer cells culture medium. To purify exosomes from tumor cell culture medium, cells were cultured in medium supplemented with 10% exo-free FBS and cell culture supernatants were collected 24-48 h after culture. A standard differential centrifugation protocol was used to isolate exosomes. Briefly, cell cultures were centrifuged at 500 g for 10 min to remove dead cells, followed by a second centrifugation at 2,000 g for 20 min to remove cell debris. Micro and macro vesicles were depleted by centrifugation at 10,000 g for 45 min. Collected supernatants were then centrifuged at 120,000 g for 1.5 h at 4 °C. The pelleted exosomes were resuspended in PBS and washed again, and re-ultracentrifuged at 120,000 g for 1.5 h to collect exosomes.
To study the characteristics of exosomes, transmission electron microscopy was performed. Summarily, exosomes were suspended in PBS and placed onto formvar carbon-coated coper grids at room temperature for 1 min. The exosomes were then stained by 2% uranyl acetate at room temperature for 1 min. and the grids were dried. The stained exosomes were visualized under a TecnaiG2 Spirit120KV transmission electron microscope operating at 120 kV (Thermo FEI).
Particel sizes distribution and concentration of exosomes were examined by Nanoparticle Tracking Analysis (NTA) analysis using a ZetaView PMX 100 (Particle Metrix, Meerbusch, Germany). For each condition analysis, at least three independent experiments were performed.
A western blot analysis was performed to detect exosomal surface protein markers. The exosomal proteins were separated using 10% SDS PAGE, and then transferred onto the polyvinylidene difluoride membranes (Millipore). The membranes were blocked with QuickBlockTM western buffer (Beyotime, China) for 20 min, and incubated with primary antibody(diluted with manufacturer's instructions) at 4 °C for overnight, followed by a 2 h incubation with HRP-conjugated secondary antibodies at room temperature. The membranes were detected with ECL reagents (Beyotime, China). CD63 was exosome protein marker. CANX, an endoplasmic reticulum protein, was used as the negative control.

Exosomes tracking analysis.
Here, 50 μg of exosomes from supernatants of cultured endometrial cancer cells were incubated for 15 min at room temperature with 20 μM DiI dye containing 5% BSA, added to avert over staining. Exosomes were diluted with PBS and then centrifuged at 120,000 g for 1.5 h at 4 °C. The labeled exosomes were suspended in PBS and collected by ultracentrifugation at 120,000 g for 1.5 h. Next, the labeled exosomes were added to endometrial stroma cell culture medium and incubated for 12 h. The cells were fixed with 4% paraformaldehyde for 20 min, permeabilized with 0.1% Triton X-100 for 15 min, then incubated for 50 min