Analysis of the circular RNA transcriptome in the grade 3 endometrial cancer

Introduction

Endometrial cancer (EC) is the most common gynecological malignancy of the female reproductive system. Traditionally, EC was divided into type 1 and type 2,¹ grades 1 and 2 are considered “type 1” EC, while grade 3 is considered “type 2”. Type 1 may arise from atypical hyperplasia and is associated with excess progesterone and estrogen receptor. In addition, the prognosis is relatively good. Type 2 may arise from atrophic endometrium and is associated with a lack of estrogen and progesterone receptor.However, it shows the worse prognosis than type 1 because of the aggressive cancerous tissue.

Non-coding RNAs (ncRNAs) play an important role in epigenetics increasingly. NcRNAs can be divided into circular non-coding RNAs (circRNAs) and linear ncRNAs. However, in recent years, we have observed the explosive growth of published research on various aspects of circRNA biology. It indicates that this molecular plays an important role in normal cellular differentiation, tissue homeostasis as well as in the development of the disease. As it is known that circRNAs arise from precursor mRNAs, their biogenesis remains unspecific. circRNAs differ from other RNAs in their extraordinary continuous closed loop structure, which is covalently linked by free 3′ and 5′ ends.² It is also called a “back-splicing” structure,³ and is generated from the joining of an upstream 3′ splice site to a downstream 5′ splice site, which prevents them from degradation by RNA exonucleases.⁴ There are many biological functions of circRNA, and the most important function is to regulate gene expression. An increasing number of studies have found that many circRNAs as miRNA sponges or competitive endogenous RNAs (ceRNAs) can regulate the expression of miRNA targets, hsa_circ_0072309 inhibits proliferation and invasion of breast cancer cells via sponging miR-492,⁵ hsa_circZNF609 promotes breast cancer cell growth, migration, and invasion via sponging miR-145-5p.⁶ In addition to acting as miRNA sponges, circRNAs can also regulate the transcription and alternative splicing. The experiment of co-localization indicated that circRNAs could interact with RBPs. It has been reported the interaction among circ-Foxo3 with E2F1, ID-1, HIFα, and FAKs played a part in anti-aging and anti-stress.⁷ Besides the non-coding function of circRNAs, scientists have also studied the capacity of translation of circRNAs.⁸ Based upon the functions of circRNAs, researchers have studied the role of circRNAs in pathology and physiology. The evidence show that circRNAs exert a regulatory function in different diseases via different mechanisms. Moreover, circRNAs have the potential to serve as clinical diagnosis markers and therapeutic targets.⁹

To date, only one study has shown that circRNAs may play roles in EC, with grade (1–2) type I EC.¹⁰ However, it was known little about the underlying regulatory mechanisms of circRNAs with the grade 3 EC. To our knowledge, this was the first study of circRNAs transcriptome analysis of the grade 3 EC in the worldwide. We made a description about the factor of circRNAs transcriptome analysis of the grade 3 EC, detected the differentially expressed circRNAs, and made a further analysis of the related microRNAs of the differentially expressed circRNAs. In short, the differentially expressed circRNAs might become the diagnosis, treatment, and the potential biomarker of the prognosis of the grade 3 EC.

Materials and methods

Ethics statement

The project was approved and supervised by the Ethics Committee of the First Affiliated Hospital of Jinan University, Guangzhou, People's Republic of China. All samples were collected from the patient who had undergone surgery in the First Affiliated Hospital of Jinan University. All the enrolled patients have signed the informed consent. This study was conducted in accordance with the Declaration of Helsinki.

EC and normal tissue samples

Tissues were collected from the patients with EC undergone surgery at the Department of Surgery at the First Affiliated Hospital of Jinan University, and the nontumorous tissue was collected at the site 10 mm away from the affected region at the same time. The tumor specimen was staged and classified by pathological result. Two pairs of snap-frozen endometrial tissue with the grade 3 EC and adjacent non-cancerous endometrial tissue were collected for circRNA sequencing analysis (Tables 1 and 2) and another 60 samples (15 pairs of the grade 3 EC and adjacent non-cancerous endometrial tissue, 15 pairs of the grade 1–2 EC and adjacent non-cancerous endometrial tissue) were collected for qRT-PCR validation.

Table 1 Patient and tissue sample chart

Table 2 Detailed clinical pathological parameters of patients being sequenced

RNA isolation

We used TRIzol (Invitrogen, Carlsbad, CA, USA) to extract total RNA from each frozen specimen and treating them with DNase (Takara, Dalian, People's Republic of China) according to the manufacturer’s instructions. The RNA integrity was assessed using an Agilent Bioanalyzer2100 (Agilent Technologies, Santa Clara, CA, USA). The quantity (ng/mL) and purity (260/280 and 260/230 ratios) of total RNAs were assessed using NanoDrop 2,000 spectrophotometer (Thermo, Waltham, MA, USA) and denaturing agarose gel electrophoresis. Each RNA sample had an A260:A280 ratio above 1.9 and an A260:A230 ratio above 1.8. The values of RIN of A1, A2, B1, and B2 were 8.8, 8.7, 8.9, and 8.6.

Circrnas sequencing analysis

RNA library preparation and circRNA sequencing were performed by the Beijing Genomics Institute (Shengzheng, People's Republic of China). The sample preparation and microarray hybridization were performed according to the Arraystar’s standard instructions. In short, total RNA samples were digested with RNase R (Epicenter Inc, San Francisco, CA, USA) to remove linear RNAs and enrich circRNAs. The rRNA-depleted RNA was used to build up the RNA libraries with TruSeq Stranded Total RNA Library Prep Kit (Illumina, San Diego, CA, USA). Quality of libraries were checked with BioAnalyzer 2100 system (Agilent Technologies, Richardson, TX, USA). The RNA libraries were denatured as single-stranded DNA molecules. The cDNAs were captured on Illumina Flow Cells (Illumina, San Diego, CA, USA) amplified in situ as clusters and finally sequenced with 150-bp paired reads on HiSeq 4000 sequencing system (Illumina, San Diego, CA, USA).

We used Cluster and Tree View software to perform hierarchical clustering analysis, in order to analyze the differentially expressed circRNAs between the grade 3 EC and adjacent non-cancerous endometrial tissue. The predicted functions of the differentially expressed circRNAs between the grade 3 EC and adjacent non-cancerous endometrial tissue were obtained by GO (http:\\www.geneontology.org\) and KEGG pathway analysis (http:\\www.genome.jp\kegg\).

Delineation of circrna/mirna interactions

The circRNA/microRNA interaction was predicted using Arraystar’s home-made miRNA target prediction software based on TargetScan and miRanda.^11,12 We searched MREs on circRNAs using the software and then selected the miRNAs according to seed-matching sequence to establish circRNA-miRNA network. Then, we used Cytoscape (version 3.6.0, USA) to draw the graph of the circRNA/miRNA network.

Quantitative real‐time polymerase chain reaction (qRT‐PCR) validation

MLV reverse transcriptase (Invitrogen, Carlsbad, CA, USA) was used for synthesizing cDNA according to manufacturer’s instructions after RNA extraction. The expression level of the circRNAs was evaluated by SYBR Green qPCR. Specific divergent primers were designed to amplify the circular transcripts, only primers achieving a single peak in the melting curve were considered proper for qRT-PCR validation. PCR was performed in 10 μL reaction volume, including 2 μL of cDNA, 5 μL 2×Master Mix, 0.5 μL of Forward Primer (10 μM), 0.5 μL of Reverse Primer (10 μM), and 2 μL of double-distilled water. The reaction was set at 95°C for 10 mins for pre-denaturation, then at 95°C for 10 s and at 60°C for 60 s repeating 40 cycles. Housekeeping gene U6 was used as an internal control. Both target and reference were amplified in triplicate wells. The relative expression of candidate circRNAs was normalized to U6 and then calculated using the 2^−ΔΔCt method.Primers are designed using PRIMER5 software (Table 3).

Table 3 Primers allowed for amplification of genes in this work

Statistical analysis

All statistical analyses in this study were accomplished by the GraphPad Prism 7.0 (GraphPad Software, La Jolla, CA, USA) and the SPSS 23.0 software package (IBM, Chicago, IL, USA). Paired t test, independent t test, and chi-square test were used in this study accurately. P-value of 0.05 or less was considered statistically significant.

Results

Identification of differentially expressed circrna profiles

High throughput sequencing analysis facilitated our research on circRNA expression signatures in the grade 3 EC. A total of 75,928 circRNA targets were detected by microarray probes in two pairs of samples. In the initial stage of experiments, circRNA expression profile of two pairs of the grade 3 EC and adjacent non-cancerous tissues were performed. We confirmed the differentially expressed circRNA with statistical significance between the two groups by using Scatter plot filtering and Volcano Plot filtering (Figure 1A and B). Box plots show the normalized intensities from the two groups (Figure 1C). What is more, we analyzed the category of differentially expressed circRNAs. Most of them are transcribed from the introgenic, some are from intergenic (Figure 1D). Furthermore, we analyzed the distribution of the differential expression of circRNAs on all the human chromosomes (Figure 2).

Figure 1 Bioinformatics analysis of differentially expressed circRNAs in patients with the grade 3 EC and adjacent non-cancerous endometrial tissue. Notes: (A) Scatter plot shows the variation of circRNA between the adjacent non-cancerous endometrial (Y-axis) versus the grade 3 EC tissue (X-axis). The upper red squares to the left indicate 2.0 fold changes of circRNA, while the bottom blue squares indicate negative fold changes of 2.0. (B) Volcano plot shows the differential circRNA expression in the grade 3 EC and adjacent non-cancerous endometrial tissue. (C) Box plot shows the distributions of circRNAs in more direct way. (D) The bar diagram shows the circRNA category. Most of them are transcribed from the introgenic, some are from intergenic.

Figure 2 The distribution of differentially expressed circRNAs in human chromosomes (A1, A2, B1, and B2).

Identification of differentially expressed circRNAs in patients with the grade 3 EC and adjacent non-cancerous endometrial tissue

Hierarchical cluster was performed to analyze all the circRNAs and shown the top 10 up and downregulated circRNAs between the grade 3 EC and adjacent non-cancerous endometrial tissues (Figure 3A and B). The data suggested that the expression of circRNAs in the grade 3 EC tissues was different from the matched non-tumor tissue. A total of 62,167 unique circRNAs were significantly altered in women with the grade 3 EC and adjacent non-cancerous endometrial tissue (P<0.05). Among them, 25,735 were significantly upregulated and 36,432 were down-regulated. The top 10 up and downregulated circRNAs between the two groups are listed in Tables 4 and 5, respectively. Among these, the expression levels of hsa_circ_0039569, hsa_circ_0001523, hsa_circ_0001610, hsa_circ_0001400, and hsa_circ_0007905 were up-regulated, and hsa_circ_0000437, hsa_circ_0001776, hsa_circ_0009043, hsa_circ_0000471, and hsa_circ_0014606 were downregulated as top 5, respectively.

Figure 3 Heat map and hierarchical clustering showing the expression values of all target circRNAs and the most up and downregulated circRNAs. Each column represents a sample and each row represents a circRNA. Notes: (A) Hierarchical cluster analysis of all target circRNAs. (B) Hierarchical cluster analysis of the most 10 up and downregulated circRNAs. A2/A1 means A2 divided A1. B2/B1 means B2 divided B1.

Table 4 Comparison of the grade 3 EC with the adjacent non-cancerous endometrial tissue for the top 10 upregulated circRNAs (fold change>2 and P<0.05) ranked by fold changes in microarray date

Table 5 Comparison of the grade 3 EC versus the adjacent non-cancerous endometrial tissue for the top 10 downregulated circRNAs (fold change>2 and P<0.05) ranked by fold changes in microarray data

Construction of the circRNA/miRNA interaction network and prediction of circRNA-miRNA associations

Considered that the circRNA had the potential to serve as miRNA sponges to regulate gene expression, so we made a further exploration about the function of ceRNA of circRNA. We used TargetScan and miRanda database software to predict the target miRNAs of circRNAs based on the base pairing rule in Arraystar. All the differentially expressed circRNAs were predicted according to the complementary miRNA matching sequence. There were a total of 451 miRNAs, the top 10 upregulated and downregulated circRNAs, could be combined with circRNAs. Then, used Cytoscape delineated an entire network of circRNA/miRNA interaction (Figure 4).

Figure 4 circRNAs-miRNAs network diagram. Notes: The red and black nodes represent circRNA, while the blue nodes represent target miRNAs. Red and black nodes represent upregulation and downregulation, respectively.

Validation for the differential expression level of circRNAs

To confirm the data analysis expression of circRNA microarray, six circRNAs (hsa_circ_0039569, hsa_circ_0001523, hsa_circ_0001610, hsa_circ_0000437, hsa_circ_0001776, and hsa_circ_0009043) were validated out using qRT-PCR in 30 pairs of samples including 15 pairs of the grade 3 EC and adjacent non-cancerous endometrial tissue, 15 pairs of the grade 1–2 EC and adjacent non-cancerous endometrial tissue. The results showed that circRNAs were significantly differentially expressed between the grade 3 EC and the grade 1–2 EC.

The qRT‐PCR results are presented in Figure 5. It showed that the expression of the circRNAs hsa_circ_0039569 and hsa_circ_0001610 were upregulated at the grade 3 EC and the grade 1–2 EC tissue versus the adjacent non-cancerous endometrial tissue, compared with the grade 1–2 EC tissue, hsa_circ_0039569 and hsa_circ_0001610 have a higher level of expression in the grade 3 EC. The hsa_circ_0001523 qPCR results were inconsistent with the microarray analysis. The hsa_circ_0009043, hsa_circ_0000437, and hsa_circ_0001776 were significantly downregulated at the grade 3 EC and the grade 1–2 EC tissue versus the adjacent non-cancerous endometrial tissue. And the hsa_circ_0009043 and hsa_circ_0001776 had a higher level of expression in the grade 3 non-cancerous endometrial tissue versus the grade 1–2 non-cancerous endometrial tissue, whereas hsa_circ_0000437 had a lower level of expression.

Figure 5 Quantitative real-time PCR validation for the expression of six circRNAs. Notes: The expression levels of six circRNAs were validated by qPCR in 30 patients. G3 NC, grade 3 non-cancerous endometrial tissue. G1/G2 NC, grades 1–2 non-cancerous endometrial tissue. G3 EC, grade 3 EC. G1/2 EC, grades 1–2 EC.

The relationship between the expression of hsa_circ_0039569 and clinicopathological parameters of the 30 patients

The qPCR results showed that hsa_circ_0039569 was higher expression of the grade 3 EC versus the grade 1–2 EC. To analyze the relationship between the expression of hsa_circ_0039569 and clinicopathological parameters of the 30 patients, the patients were divided into low expression group (n=15) and high expression group (n=15) with the hsa_circ_0039569 relative expression level 2^−ΔΔCт, median is 2.985. The correlation between the hsa_circ_0039569 and clinicopathological data was analyzed. The results showed that the expression level of hsa_circ_0039569 was not correlated with age, lymph node metastasis, tumor size, FIGO stage, and myometrial invasion (P>0.05), and was significantly correlated with tumor differentiation (P=0.001). The difference was statistically significant (Table 6).

Table 6 The relationship between the expression of hsa_circ_0039569 and clinicopathological parameters of the thirty patients

Prediction of circRNA-miRNA associations

In order to explore the mechanism of circRNAs, the circRNA-miRNA axis in cancer-related pathways was predicted. DIANA-miRPath determined all the candidate miRNAs that were involved in possible pathways by p-value cutoff at 0.05. The data showed that hsa_circ_0039569 was associated with cancer-related pathways for the grade 3 EC with the greatest potential. Base on TargetScan and miRanda, the hsa_circ_0039569/microRNAs interaction was predicted using Arraystar’s home-made miRNA target prediction software. KEGG analysis of all microRNAs showed that miR-542-3p and -let-7c-5 was associated with cancer-related pathways for the grade 3 EC with the greatest potential. Firstly, we further analyzed the binding sequence between hsa_circ_0039569 and hsa-miR-542-3p or hsa-let-7c-5p interaction by TargetScan and miRanda analysis. We found circ_0039569 had a perfect match sequence (7-mer-m8 seed type) to bind miR-542-3p (Figure 6A) or let-7c-5p (Figure 6B). To made a confirmation of the relationship between circRNA between hsa_circ_0039569 and hsa-miR-542-3p or hsa-let-7c-5p, hsa-miR-542-3p and hsa-let-7c-5p were validated by using qRT-PCR in 30 pairs of samples.

Figure 6 The interaction between hsa_circ_0039569 and hsa-miR-542-3p or hsa-let-7c-5p was predicted by TargetScan and miRanda. Notes: (A) The interaction between hsa_circ_0039569 and hsa-miR-542-3p. (B) The interaction between hsa_circ_0039569 and hsa-let-7c-5p.

The qRT‐PCR results are presented in Figure 7. It showed that the expression of the microRNAs hsa-miR-542-3p and hsa-let-7c-5p was upregulated at the grade 3 non-cancerous endometrial tissue and the grade 1–2 non-cancerous endometrial tissue versus the the grade1-3 EC tissue, compared with the grade 1–2 non-cancerous endometrial tissue, hsa-miR-542-3p, and hsa-let-7c-5p have a higher level of expression in the non-cancerous endometrial tissue. So, the hsa_circ_0039569-hsa-miR-542-3p/hsa-let-7c-5p axis was predicted as one of the circRNA-miRNA axis. Spearman correlation test was used to analyze the correlation between hsa_circ_003956 and hsa-miR-542-3p and hsa-let-7c-5p in the grade 3 EC. The correlation coefficient R results are -8923 and -0.9027 (Figure 8). So, hsa_circ_003956 has a negative correlation with hsa-miR-542-3p and hsa-let-7c-5p.

Figure 7 Quantitative real-time PCR validation for the expression of two miRNAs. Notes: The expression levels of two miRNAs were validated by qPCR in 30 patients. G3 NC, grade 3 non-cancerous endometrial tissue. G1/G2 NC, grades 1–2 non-cancerous endometrial tissue. G3 EC, grade 3 EC. G1/2 EC, grades 1–2 EC.

Figure 8 The result of spearman correlation test between hsa_circ_003956 and hsa-miR-542-3p/hsa-let-7c-5p in the grade 3 EC.

Discussion

Presently, the mechanism and role of circRNAs in cancer progression haven’t been well identified.¹³ Emerging evidence have found that circRNAs played an important part role in multiple pathological and physiological functions. The available evidence showed that circRNAs are associated with proliferation,¹⁴ cell cycle,¹⁵ aoptosis,¹⁶ and autophagy,¹⁷ involving nervous system disease, cardiovascular system disease, and series of tumors.^18,19 Especially, more and more studies have found that circular RNA plays an increasingly important role in malignant tumors which were related to sex hormones. Accumulating evidence indicates that hsa_circ_006054, hsa_circ_100219 hsa_circ_0006528, and circ-ABCD10 might also play roles in breast cancer.^20–22 Furthermore, hsa_circ_0001982 promotes breast cancer cell carcinogenesis by targeting miR-143.²³ Additionally, the expression of circ-Foxo3 was downregulated in mouse breast tumor tissues compared with normal breast, lipid tissues, and skin, and as an inhibitor of breast cancer tumor cells inhibits tumor cell proliferation and survival.²⁴ It has been proved that circ-SMARCA5 acts as an oncogene in prostate cancer by promoting the cell cycle and inhibiting apoptosis.²⁵

In the field of EC, up to the present time, just one study has shown that circRNAs may play roles in EC, with grade (1–2) type I EC, it reported the circular transcriptome specific for EC as determined by RNA sequencing and found that the abundance of circRNAs was lower in the grade 1–2 EC than in adjacent non-cancerous endometrial tissue. What is more, they found that numerous “hotspot” genes from which circRNAs were transferred that may account for alterations in circRNA expression between the non-cancerous endometrial tissue and the grade 1–2 EC. Most importantly, they have also identified circRNAs which were differentially expressed between the grade 1–2 EC and normal endometrial tissue. More and more researchers acknowledge that the patients with the grade 3 EC have a generally poor prognosis,²⁶ and the specific role of circRNAs in the grade 3 EC remains unclear.

In this study, the differentially expressed profiles of circRNAs in the grade 3 EC tissue and adjacent non-cancerous endometrial tissue were analyzed and validated. Up to now, this is the first study to evaluate the differential expression of circRNAs in the grade 3 EC patients. It indicated that the grade 3 EC-associated circRNAs could be exploited as new candidates for diagnosis and treatment of the grade 3 EC.

Base on the above study, there were 25,735 circRNAs significantly upregulated and 36,432 circRNAs significantly downregulated in the grade 3 EC tissue compared with adjacent non-cancerous endometrial tissue. Among these, the expression levels of hsa_circ_0039569, hsa_circ_0001523, hsa_circ_0001610, hsa_circ_0001400, and hsa_circ_0007905 were upregulated, and hsa_circ_0000437, hsa_circ_0001776, hsa_circ_0009043, hsa_circ_0000471, and hsa_circ_0014606 were downregulated as top 5, respectively. The upregulated hsa_circ_0039569 and hsa_circ_0001610 and downregulated hsa_circ_0000437, hsa_circ_0001776, and hsa_circ_0009043 were validated by qRT-PCR analysis. Hsa_circ_0039569 were selected for further analysis, in the clinicopathological parameters of the 30 patients, the expression level of hsa_circ_0039569 was significantly correlated with tumor differentiation (P=0.001), and was not correlated with age, lymph node metastasis, tumor size, FIGO stage and myometrial invasion (P>0.05). Furthermore, using Cytoscape to delineate an entire network of circRNAs/miRNAs interaction. This was the first time of our study which predicted hundreds of circRNAs/miRNAs interaction. An interaction network of circRNAs/miRNAs was constructed, which provided valuable information for ceRNA research of circRNAs in the grade 3 EC. In addition, the results showed that the hsa_circ_0039569 was highly expressed in the grade 3 EC compared to control. With bioinformatical methods, we found that hsa_circ_0039569 had MREs of hsa-miR-542-3p and hsa-let-7c-5p. Hsa-miR-542-3p and hsa-let-7c-5p were downregulated at the grade 3 EC were validated by qRT-PCR analysis. Predicting that hsa_circ_0039569-hsa-miR-542-3p/hsa-let-7c-5p was one of the circRNA-miRNA axis. Our results indicated that it was worthwhile to investigate these novel dysregulated circRNAs as miRNA sponges and their potential biological functions in the grade 3 EC.

All in all, for the first time, our study revealed the circRNA expression signatures of the grade 3 EC. Furthermore, the regulation of hsa_circ_0039569-hsa-miR-542-3p/hsa-let-7c-5p axis was initially predicted in the grade 3 EC. Next, our laboratory is doing further research on the mechanism of hsa_circ_0039569. The result of our study emphasized that hsa_circ_0039569 could be an important prediction of the grade 3 EC. This study has laid the foundation for the functional studies, further diagnosis, treatment of circRNAs of the grade 3 EC.

Conclusion

This study demonstrated that lots of differentially expressed cricRNAs in endometrial tissue of the grade 3 EC patients compared with adjacent non-cancerous endometrial tissue. Therefore, it may offer novel molecular candidates for diagnosis and clinical treatment of the grade 3 EC.