Identification of Circular RNAs as a Novel Biomarker for Ovarian Endometriosis.

BACKGROUND: Endometriosis is a challenging disease with symptoms such as dysmenorrhea and infertility. However, its etiology is still vague and there is still no effective markers or treatment. This study aimed to profile the circular RNAs (circRNAs) expressed in eutopic endometrium from patients with ovarian endometriosis and explore potential clues to the pathogenesis of endometriosis, providing an evidence for clinical diagnosis and treatment.
METHODS: A total of 63 clinical samples, including control endometrium (n = 22) and eutopic endometrium (n = 41), were collected from Peking Union Medical College Hospital between May 1, 2016, and December 31, 2016. Of them, four samples in each group were used for circRNA microarray. Then, four upregulated circRNAs were screened out for quantitative real-time polymerase chain reaction (qRT-PCR) validation. After that, bioinformatics analysis was performed to predict miRNAs targeted by validated circRNAs and investigate the circRNA-miRNA-mRNA interactions.
RESULTS: Among 88 differentially expressed circRNAs, 11 were upregulated and 77 were downregulated in eutopic endometrium of patients with endometriosis. qRT-PCR validation results for two upregulated circRNAs (circ_0004712 and circ_0002198) matched the microarray results. The area under the receiver operating characteristic curve of circ_0002198 for distinguishing ovarian endometriosis was 0.846 (95% confidence interval [CI]: 0.752-0.939; P < 0.001) while that of circ_0004712 was 0.704 (95% CI: 0.571-0.837; P = 0.008). On the basis of target prediction, we depicted the molecular interactions between the identified circRNAs and their dominant target miRNAs, as well as constructed a circRNA-miRNA-mRNA network.
CONCLUSIONS: This study provides evidence that circRNAs are differentially expressed between eutopic and normal endometrium, which suggests that circRNAs are candidate factors in the activation of endometriosis. circ_0002198 and circ_0004712 may be potential novel biomarkers for the diagnosis of ovarian endometriosis.

INTRODUCTION

Endometriosis is considered as a challenging disorder. Over 176 million women worldwide have been tormented by endometriosis associated pain and infertility, which severely affects their quality of life.[12] Unfortunately, there are still no specific markers and therapies. The essential reason for such adversity is due to the elusive pathogenesis. It is to date widely assumed that ectopic lesions arise through retrograde endometrial fragments during menstruation. Not all the women suffered endometriosis despite most women undergoing retrograde menstruation. The affected women may have certain susceptible factors, which partially comes down to the eutopic endometrium.[3]

Evidences have indicated that gene expression level in eutopic endometrium of endometriosis is aberrantly altered by comparison with control endometrium.[45] These alterations of eutopic endometrium might be the source of the pathogenesis of endometriosis, which keep endometrial debris easier alive in the ectopic site. Recently, emerging evidences reveal that circular RNAs (circRNAs) are involved in regulating gene expression as competitive endogenous RNAs (ceRNAs).[678] Moreover, circRNAs are characterized by covalently linked terminals and high stability as well as abundant expression level.[910] All abovementioned characteristics make circRNAs become evaluate indicators for diagnosis, prognosis, and therapeutic-response prediction. However, the relationship between circRNAs and endometriosis is unknown.

Given their pivotal biological roles, we investigated and identified dysregulated circRNAs in eutopic endometrium to offering new idea for diagnosis and treatment.

METHODS

Ethical approval

This study was approved by the Institutional Review Board and Hospital Local Ethics Committee (No. JS-875). All participants provided informed consent.

Study population

A total of 63 clinical samples were collected at the Department of Obstetrics and Gynecology of Peking Union Medical College Hospital between May 1, 2016, and December 31, 2016. Eutopic endometrium samples were collected from 41 women who underwent hysteroscopic and laparoscopic surgery for ovarian endometriosis (22–46 years old; proliferative phase: n = 28; secretory phase: n = 13; American Fertility Society [AFS] Stage III–IV). Tissue samples of control endometrium were acquired from another 22 women without endometriosis who underwent hysteroscopy and laparoscopy for other benign ovarian cysts, infertility or uterine septum (no endometriosis, 25–43 years old; proliferative phase: n = 16; secretory phase: n = 6), similarly with a previous study.[11] All patients met the criteria as follows: regular menstruation (25–32 days), no hormonotherapy for at least 6 months, and histopathology confirmation. All obtained samples were excluded from abnormalities of endometrium by the pathological diagnosis, and proliferative and secretory endometrium phases were distinguished by hematoxylin and eosin staining. Tissue collection and storage were conducted as described previously.[12] Four samples for each group were randomly selected for a microarray analysis. Besides, all the samples were validated by quantitative real-time polymerase chain reaction (qRT-PCR).

Total RNA isolation and quality control

Total RNA was isolated with TRIzol reagent (Invitrogen Life Technologies, Carlsbad, CA, USA) according to the manufacturer's protocol. RNA quantity and quality were measured using a NanoDrop ND-1000 (Thermo Fisher Scientific, Wilmington, DE, USA). RNA integrity and genomic DNA contamination were assessed by denaturing agarose gel electrophoresis.

Circular RNA microarray analysis

The Arraystar Human circRNA Array 8 × 15 K V2 Microarray (Arraystar, Rockville, MD, USA) contains 15,000 probes for 13,617 human circRNAs, all of which have been confirmed by Jeck et al.,[10] Salzman et al.,[13] Memczak et al.,[14] Zhang et al.,[15] Zhang et al.,[16] Guo et al.,[17] and You et al.[18] Total RNA was first digested with Rnase R (Epicentre, Madison, WI, USA) to remove linear RNAs and enrich for circRNAs. Then, the enriched circRNA was amplified and transcribed into fluorescent complementary RNA, utilizing random primers according to the Arraystar Super RNA Labeling protocol (Arraystar). After hybridization and washing, processed slides were scanned using an Agilent Scanner G2505C (Agilent Technology, Santa Clara, CA, USA). Thereafter, Agilent Feature Extraction software (version 11.0.1.1, Agilent Technology, Santa Clara, CA, USA) was utilized to analyze acquired array images. Quantile normalization and subsequent data were processed using the R software package (version 3.3.1; R Foundation Inc. Vienna, Austria).

Quantitative real-time polymerase chain reaction validation

During the qRT-PCR validation stage, we recruited 41 endometria from patients with endometriosis and 22 normal endometria for control. Total isolated RNA was reversely transcribed using SuperScript III Reverse Transcriptase (Invitrogen Life Technologies). Subsequently, qRT-PCR was performed by ViiA 7 RT-PCR System (Applied Biosystems, Foster City, CA, USA) in a 10-μl reaction volume, including 5-μl 2 × PCR Master Mix (Arraystar), 0.5 μl/10 μmol/L forward/reverse primers, 2 μl cDNA, and 2 μl RNAase-free H2O. The cycling program was initiated from 95°C for 10 min, followed by 40 cycles of 95°C for 10 s and 60°C for 60 s. Divergent primers were designed and optimized for circRNAs. Supplementary Table 1 listed all the PCR primers, the specificity of which was verified by a single-peak on the melting curve. The threshold cycle method (2−ΔΔCT) was used to calculate relative expression levels, which were normalized to β-actin levels.

Supplementary Table 1

Primers used for qRT-PCR

Gene	Forward, 5’-3’	Reverse, 5’-3’
β-actin	GTGGCCGAGGACTTTGATTG	CCTGTAACAACGCATCTCATATT
circRNA_0004712	AGGGGTGAACCAGCCATTT	GCCAATCTCCCCTGAGTATGTT
circRNA_0002198	GCAAACCTATATCAGGAAACAGC	TTGAAGAGGTGGCACAACAGT
circRNA_0002503	CGTATTCACCTGCTCATCTCC	CTGGTGTTGGGATGATTTGA
circRNA_0000141	TCAGGCCCCATGCAGGTG	ACCATGCCACTCGGATCCTC

qRT-PCR: Quantitative real-time polymerase chain reaction.

Bioinformatics analysis

To further elucidate the role of circRNAs in endometriosis, we first used the Arraystar target prediction software based on TargetScan[19] and miRanda[20] to predict the targeted miRNAs of each circRNA, which described interactions between circRNAs and miRNAs. According to the miRNA support vector regression (mirSVR) algorithm, we ranked the top five predicted miRNA targets for each circRNA. Then, we used TargetScan and miRDB to predict the related target genes according to the miRNAs targeted by circRNAs. Finally, the interaction network was depicted using the validated circRNAs and predicted miRNAs/mRNAs, preliminarily accounting for interactions of circRNAs-miRNAs-mRNAs in pathogenesis of endometriosis. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses further annotated the function and signaling pathways of these genes.

Statistical analysis

Data were first explored by the tests of normality. If the normal distribution was satisfied, Student's t-tests was used; otherwise, nonparametric Mann-Whitney U-test was used. The thresholds for differentially expressed circRNAs were expressed as an absolute value of fold change (FC) >2 (EuE = 1) and P < 0.05. Above statistical analyses were processed with R software version 3.3.1 and GraphPad Prism 5 (GraphPad Software, La Jolla, CA, USA).

RESULTS

Clinical characteristics of the participants

In the study, a total of 63 endometrial biopsies were obtained from women with laparoscopically and histopathologically proven presence (n = 41) or absence (n = 22) of endometriosis. The clinical characteristics of the participants were presented in Supplementary Tables 2 and 3 where the mean (standard deviation) ages of patients in endometriosis and control group were 32 ± 6 years and 32 ± 5 years, respectively. There were no differences in age, cycle phase, parity, and infertility between two groups. Patients with endometriosis were more likely to gain moderate-to-severe dysmenorrhea (visual analogue scale [VAS] ≥4) and high level of CA125 by comparison with controls (Z = −2.627, P = 0.009 and t = 7.529, P < 0.001, respectively). All the patients with endometriosis were judged to be Stage III and IV according to the AFS staging system.

Supplementary Table 2

Clinical characteristics of 41 patients with ovarian endometriosis

Patient	Cycle phase	Age (years)	Parity	Infertility	VAS	CA125	Diameter of cyst (cm)	rAFS stage	Indication for surgery
1	Proliferative	31	0	Yes	6.5	44.1	3.8	IV	Pain/infertility
2	Proliferative	33	0	Yes	8.0	81.8	2.5	IV	Pain/infertility
3	Proliferative	28	0	N/A	7.5	34.6	5.8	III	Pain
4	Proliferative	33	0	N/A	6.5	34.1	4.8	III	Pain
5	Secretory	45	2	No	0	54.1	7.6	IV	Cyst
6	Secretory	28	0	N/A	0	56.6	6.7	IV	Cyst
7	Secretory	28	1	No	4.5	28.1	4.6	III	Infertility
8	Proliferative	33	2	No	5.5	46.9	4.7	IV	Pain
9	Secretory	30	0	N/A	4.0	30.1	4.9	III	Pain
10	Proliferative	30	0	Yes	5.0	24.2	5.8	III	Pain/infertility
11	Proliferative	45	1	No	4.5	108.8	5.9	IV	Pain
12	Secretory	29	0	Yes	4.5	79.0	3.8	IV	Pain/infertility
13	Secretory	29	2	No	4.5	63.2	5.9	IV	Pain
14	Proliferative	32	2	No	0	1.9	6.4	III	Cyst
15	Secretory	27	0	N/A	7.0	60.5	5.3	IV	Pain
16	Proliferative	30	0	Yes	3.5	52.1	3.4	III	Pain/infertility
17	Proliferative	32	2	No	2.0	70.7	5.1	IV	Cyst
18	Proliferative	37	5	No	3.0	51.6	4.2	IV	Pain
19	Proliferative	32	1	No	4.0	23.1	7.0	III	Pain
20	Proliferative	28	0	N/A	0	51.0	5.0	IV	Cyst
21	Proliferative	39	2	No	5.5	37.1	6.5	III	Pain
22	Secretory	38	1	No	6.0	76.4	2.9	IV	Pain
23	Proliferative	25	0	N/A	0	60.5	7.3	IV	Cyst
24	Proliferative	38	2	No	8.5	59.7	4.9	IV	Pain
25	Proliferative	24	0	N/A	4.0	40.3	6.7	III	Pain
26	Proliferative	25	0	N/A	7.5	30.2	5.9	IV	Pain
27	Proliferative	33	1	No	7.5	78.3	5.5	IV	Pain
28	Proliferative	30	0	N/A	0	37.4	5.3	III	Cyst
29	Secretory	25	0	N/A	2.0	42.0	8.3	IV	Cyst
30	Proliferative	34	3	No	0	185.2	6.6	III	Cyst
31	Secretory	36	0	Yes	0	29.3	4.5	IV	Infertility
32	Secretory	25	0	Yes	2.0	31.5	3.5	III	Infertility
33	Proliferative	31	0	N/A	5.5	56.6	4.8	IV	Pain
34	Proliferative	39	1	No	0	96.0	7.1	IV	Cyst
35	Secretory	46	1	No	0	22.4	3.6	III	Cyst
36	Proliferative	32	1	No	10.0	21.9	13.2	IV	Pain
37	Secretory	26	0	N/A	0	96.2	4.5	IV	Cyst
38	Proliferative	22	0	N/A	3.0	66.8	4.6	III	Pain
39	Proliferative	43	1	No	5.0	40.3	5.3	IV	Pain
40	Proliferative	29	0	N/A	5.0	24.4	7.0	IV	Pain
41	Proliferative	25	0	N/A	6.5	79.0	1.8	III	Pain

N/A: Never attempted pregnancy; VAS: Visual analogue scale; rAFS: Revised American Fertility Society.

Supplementary Table 3

Clinical characteristics of 22 patients without ovarian endometriosis

Patient	Cycle phase	Age (years)	Parity	Infertility	Pathologic diagnosis	VAS	CA125	Diameter of cyst (cm)	Indication for surgery
1	Proliferative	27	0	N/A	Teratoma	0	14	4.9	Cyst
2	Proliferative	26	0	N/A	Serous cystadenoma	0	12.3	10.4	Cyst
3	Proliferative	30	1	No	Teratoma	0	12.7	8.0	Cyst
4	Proliferative	27	0	Yes	Simple cyst	3.0	13.7	3.1	Cyst infertility
5	Secretory	34	1	No	Secretory endometrium	2.5	15.2	–	Uterine diverticula
6	Proliferative	25	1	No	Proliferative endometrium	0	9.3	–	Uterine septum
7	Secretory	29	0	N/A	Teratoma	0	14.3	4.6	Cyst
8	Secretory	32	0	N/A	Paraovarian cyst	3.5	18.8	6.8	Cyst
9	Proliferative	30	0	N/A	Teratoma	6.5	20.4	5.0	Cyst
10	Proliferative	31	1	No	Proliferative endometrium	0	22.3	–	Uterine septum
11	Proliferative	34	1	No	Mesosalpinx cyst	0	18.6	6.7	Cyst
12	Proliferative	43	2	No	Simple cyst	0	19.3	–	Cyst
13	Proliferative	32	0	N/A	Mesosalpinx cyst	4.5	20.2	4.0	Cyst
14	Secretory	39	5	No	Serous cystadenoma	3.5	17.7	6.4	Cyst
15	Proliferative	39	3	No	Teratoma	6.5	8.5	4.0	Cyst
16	Secretory	26	0	N/A	Teratoma	5.0	9.2	5.5	Cyst
17	Proliferative	37	6	No	Teratoma	0	19.7	4.1	Cyst
18	Proliferative	36	5	No	Teratoma	0	13.4	3.7	Cyst
19	Proliferative	33	2	No	Teratoma	0	29.8	3.6	Cyst
20	Proliferative	31	0	N/A	Serous cystadenoma	0	17.2	5.1	Cyst
21	Secretory	25	0	N/A	Teratoma	3.5	11.5	5.9	Cyst
22	Proliferative	34	2	No	Proliferative endometrium	3.5	8.7	–	Uterine diverticula

N/A: Never attempted pregnancy; VAS: Visual analogue scale.

Circular RNA expression profiles in eutopic endometrium relative to those in normal endometrium

According to the circRNA microarray, a total of 88 circRNAs, 11 significantly upregulated and 77 significantly downregulated, were differentially expressed between eutopic endometrium and normal endometrium. The two types of endometrial tissue could be clearly distinguished using hierarchical clustering, scatterplot, and volcano plots [Figure 1]. Hierarchical clustering indicated the relative expression level of circRNAs in eutopic and normal endometrium. Scatter plot evaluated the variation in circRNA expression between the two groups. Volcano plot displayed the statistical significance of differentially expressed circRNAs between cases and controls.

Figure 1

circRNA expression patterns in eutopic endometrium (Group T) relative to those in normal endometrium (Group C). (a) Hierarchical clustering of circRNAs. Each group included four individuals. circRNAs are represented by single rows and samples by single columns. The color scale indicates relative expression, upregulation (red), and downregulation (green). Fold change >2 and P < 0.05 were regarded as the differentially expressed circRNAs. (b) Scatter plot of circRNAs. The values corresponding to the X- and Y-axes are the normalized signal values. (c) Volcano plot of circRNAs. The values on the X- and Y-axes represent normalized fold changes and P values, respectively. The red points represent significantly differentially expressed circRNAs. circRNAs: Circular RNA.

Quantitative real-time polymerase chain reaction validation of selected circular RNAs

To confirm the differentially expressed circRNAs in the microarray, four circRNAs were selected for validation based on the following criteria: (1) upregulated in eutopic endometrium, (2) FCs >2, (3) P < 0.05, (4) raw intensity >200, (5) exonic-related circRNAs, and (6) length between 200 and 3000 bp. Of them, two circRNAs (circ_0004712 and circ_0002198) matched the microarray results and met the statistical cutoff by comparison with control endometrium [Z = −2.653, P = 0.008; Z = −4.498, P < 0.001; Figure 2a]. circRNA_0002503 was expressed at a higher level in the eutopic endometrium of patients with endometriosis; however, the difference was not statistically significant. And, circ_0000141 could not be amplified by qPCR % [Figure 2a]. Intriguingly, the two validated circRNAs were not affected by the menstrual cycle [Figure 2b]. We also performed a subgroup analysis between VAS and CA125, respectively, where patients with endometriosis were classified as two groups: one with moderate-to-severe dysmenorrhea (VAS ≥4) or CA125 >35 U/ml and the other with VAS <4 or CA125 ≤35 U/ml. As a result, no significant association between the two validated circRNAs and VAS or CA125 was found [Supplementary Figure 1]. Receiver operating characteristic curve analyses revealed that circ_0004712 and circ_0002198 were valuable biomarkers for distinguishing women with or without endometriosis, with area under curve (AUC) value of 0.704 (95% confidence interval [CI]: 0.571–0.837; P = 0.008) and 0.846 (95% CI: 0.752–0.939; P = 0.000), respectively [Figure 3]. The diagnostic power of circ_0004712 achieved notable improvement when the circ_0004712 and circ_0002198 werecombined, while that of circ_0002198 was not improved (AUC = 0.819; 95% CI: 0.713–0.925; P = 0.000).

Figure 2

qRT-PCR validation of the four selected circRNAs. (a) The results of microarray and qPCR are shown as blue and red columns. Data that coincided with the microarray results and met the statistical cut-off are marked with *P < 0.05 and †P < 0.001. (b) Effect of menstrual cycle on the two identified circRNAs. Data are expressed as fold changes relative to the values for the proliferative phase group of controls. Both the circRNAs show no significant changes between proliferative (red columns) and secretory phases (blue columns). qRT-PCR: Quantitative real-time polymerase chain reaction; qPCR: Quantitative polymerase chain reaction; circRNAs: Circular RNAs.

Figure 3

The receiver operating characteristic curve of specific circRNAs in distinguishing endometriosis. circ_0004712 and circ_0002198 were valuable biomarkers for distinguishing endometriosis, with AUC value of 0.704 (95% CI: 0.571–0.837; P = 0.008) and 0.846 (95% CI: 0.752–0.939; P < 0.001), respectively. The diagnostic power of circ_0004712 achieved notable improvement when the circ_0004712 and circ_0002198 was combined, while that of circ_0002198 was not improved (AUC = 0.82; 95% CI: 0.71–0.93; P < 0.001). AUC: Area under the curve; circRNAs: Circular RNAs; CI: Confidence interval.

The association between the two validated circRNAs and VAS or CA125. (a) The association between circ_0002198 and VAS or CA125.(b) The association between circ_0004712 and VAS or CA125. VAS: Visual analogue scale.

Click here for additional data file.

Prediction of target miRNAs and mRNAs

To explore the interactions between circRNAs and miRNAs, we predicted the target miRNAs of each circRNA. Moreover, the dominant miRNAs targeted by top 10 up- and downregulated circRNAs were ranked based on mirSVR scores [Supplementary Table 4]. Of them, miR455-3p, miR876-3p, miR661, and miR323a-5p were found to be the common targets of both circ_0004712 and circ_0002198. The details of the molecular interactions between these two circRNAs and above target miRNAs are depicted in Figure 4. To further study how the two circRNAs regulate gene expression as ceRNAs, we also predicted the related target genes according to the miRNAs targeted by circRNAs. On basis of above prediction, a total of 29 miRNAs and 62 mRNAs were recognized as downstream targets of circ_0004712 and circ_0002198. Thereafter, the circRNA-miRNA-mRNA network was constructed, which clarified the gene regulatory relationships in endometriosis [Figure 5].

Supplementary Table 4

The dominant miRNAs targeted by top 10 up- and downregulated circRNAs

circRNA	Alias	FC	P	Top five targeted miRNAs
				1	2	3	4	5
Upregulation
hsa_circRNA_104195	hsa_circ_0002198	4.63	7.600E04	miR4553p	miR8763p	miR661	miR323a5p	miR198
hsa_circRNA_405510		4.43	4.259E02	miR4307	miR4506	miR5100	miR4975p	miR68323p
hsa_circRNA_104194	hsa_circ_0004712	4.11	1.359E03	miR4553p	let7g5p	miR8763p	miR661	miR323a5p
hsa_circRNA_404646		3.33	3.341E02	miR3918	miR47265p	miR46405p	miR67625p	miR4235p
hsa_circRNA_101501	hsa_circ_0034953	3.22	4.051E02	miR146b3p	miR5065p	miR298	miR8735p	miR1855p
hsa_circRNA_406483		2.88	2.691E02	miR3353p	miR3612	miR50025p	miR47093p	miR4933p
hsa_circRNA_075503	hsa_circ_0075503	2.43	2.998E02	miR4673	miR46455p	miR12265p	miR548b3p	miR3835p
hsa_circRNA_002503	hsa_circ_0002503	2.27	4.520E02	miR67603p	miR1182	miR68375p	miR46855p	miR3723p
hsa_circRNA_026462	hsa_circ_0026462	2.25	3.844E02	miR223p	miR19083p	miR1275	miR65015p	miR68293p
hsa_circRNA_103002	hsa_circ_0004816	2.08	9.380E03	miR3305p	miR449b5p	miR449a	miR1945p	miR34c5p
Downregulation
hsa_circRNA_001062	hsa_circ_0001062	9.86	1.029E04	miR4307	miR47533p	miR68093p	miR68733p	miR607
hsa_circRNA_004183	hsa_circ_0004183	4.35	2.551E02	miR71625p	miR68753p	miR516b3p	miR516a3p	miR46873p
hsa_circRNA_083996	hsa_circ_0083996	3.64	3.337E03	miR581	miR71613p	miR6134	miR78473p	miR6780b5p
hsa_circRNA_092547	hsa_circ_0001445	3.60	3.803E04	miR67403p	miR47985p	miR507	miR12855p	miR36645p
hsa_circRNA_406544		3.42	1.569E03	miR12855p	miR3353p	miR173p	miR4422	miR6273p
hsa_circRNA_001050	hsa_circ_0001050	3.27	1.657E03	miR36535p	miR67585p	miR4325	miR44823p	miR6535p
hsa_circRNA_059914	hsa_circ_0059914	3.09	1.388E02	miR3775p	miR6086	miR47563p	miR103b	miR7673p
hsa_circRNA_101231	hsa_circ_0000467	3.08	2.611E02	miR1535p	miR3825p	miR520g3p	miR549a	miR520h
hsa_circRNA_002082	hsa_circ_0002082	2.85	3.809E03	miR5125p	miR4773	miR3611	miR47423p	miR68873p
hsa_circRNA_405210		2.83	3.729E03	miR68333p	miR4659a3p	miR68093p	miR4659b3p	miR47685p

circRNAs: Circular RNAs; FC: Fold change.

Figure 4

The molecular interactions between the two circRNAs and their dominant target miRNAs. The common and dominant target miRNAs of both circ_0004712 and circ_0002198 were miR455-3p, miR876-3p, miR661 and miR323a-5p, respectively. circRNAs: Circular RNAs.

Figure 5

The circRNA-miRNA-mRNA network. The two identified circRNAs are denoted by yellow circle nodes. The red circle nodes represent target miRNAs, and blue circle nodes are target mRNAs. circRNAs: Circular RNAs.

Enrichment analysis of circular RNA-targeted genes

GO and KEGG analysis was applied to enrich the function of circRNA-targeted genes [Figure 6]. The data indicated that the target genes of these circRNAs were mainly involved in the biological processes of creatine metabolic process, glutamine family amino acid metabolic process, and negative regulation of striated muscle cell [Figure 6a]. The main cell component which target genes participated in were mitochondrial inner membrane, organelle inner membrane, and mitochondrion, respectively [Figure 6a]. With regard to the molecular function, the circRNA-targeted genes play a role in phosphotransferase activity, SNAP receptor activity, and carboxylic ester hydrolase activity [Figure 6a]. In addition, KEGG analysis revealed that target genes might participate in arginine/proline metabolism and cytokine–cytokine receptor interaction [Figure 6b].

Figure 6

GO and KEGG analysis for the circRNA-targeted mRNAs. (a) GO enrichment for the target mRNAs, including biological processes, cell component, and molecular function. (b) Annotated pathways for the target mRNAs. GO: Gene Ontology; KEGG: Kyoto Encyclopedia of Genes and Genomes; circRNAs: Circular RNAs.

DISCUSSION

In our study, we identified 88 differentially expressed circRNAs, 11 upregulated and 77 downregulated, in eutopic endometrium of patients with endometriosis compared with those in normal endometrium. Two of four upregulated circRNAs were confirmed by qPCR following a microarray screening. The two circRNAs, circ_0004712 and circ_0002198, identified in the present study have not been reported in other diseases. However, we found that the increased level of circ_0004712 and circ_0002198 can help identify the patients with endometriosis.

To explore the role of circRNAs in endometriosis, we first performed prediction of target miRNAs for circRNAs. In the rank of top 5 target miRNAs, it was found that circ_0004712 and circ_0002198 act together to target miR455-3p, miR876-3p, miR661, and miR323a-5p, which have not been reported in endometriosis yet. Nevertheless, in preeclampsia patients, miR455-3p was significantly downregulated and was linked to the suppression of hypoxia signaling.[21] In atherosclerosis, miR876 can induce endothelial cell apoptosis.[22] Moreover, miR-661 was able to activate p53 to inhibit cell cycle progression.[23] As regards the miR-323a-5p, in patients with refractory epilepsy caused by focal cortical dysplasia, its elevated level was positively correlated with the duration of epilepsy, seizure frequency, and poor prognosis.[24]

Subsequently, we also predicted the related target genes according to the miRNAs targeted by circRNAs. On basis of above prediction, a total of 29 miRNAs and 62 mRNAs were recognized as downstream targets of circ_0004712 and circ_0002198. Among them, 4 targeted miRNAs and 2 mRNAs have been reported to be associated with endometriosis. For example, miR-503, repressed in endometriosis, induces apoptosis and cell cycle arrest and inhibits cell proliferation, angiogenesis, and contractility of ovarian endometriotic stromal cells.[25] miR-196a, overexpressed in eutopic endometrium, activates the MEK/ERK signal and represses the progesterone receptor and decidualization from women with endometriosis.[26] miR-196b, downregulated in endometriotic stromal cells, inhibits proliferation and induces apoptosis by targeting c-myc and Bcl-2 expression.[27] Hypoxia-coordinated AUF1/miR-148a was reported to destabilize DNA methyltransferase 1 mRNA during the pathogenesis of endometriosis.[28] Besides, TNFRSF6B (NM_003823) and PGC-1α (NM_002630) have been found to participate in pathogenesis of endometriosis. For instance, decoy receptor 3 (DcR3)/TNFRSF6B, a pleiotropic immunomodulator regulated by estrogen, promotes cell adhesion and enhances endometriosis development by activating focal adhesion kinase (FAK).[29] PGC-1α, highly expressed in ovarian endometrioma, promotes local estrogen biosynthesis by stimulating aromatase expression and activity.[30]

Finally, the circRNA/miRNA/mRNA network was constructed. Functional analyses and prediction of the target genes of circRNAs were carried out in eutopic and control endometrium for the first time. The two identified circRNAs were predicted to mediate arginine and proline metabolism via CKMT1A and CKMT1B. Both genes are effective modulators of ATP synthase-coupled respiration and belongs to mitochondrial creatine kinases (MtCK), which were found to be implicated in several tumors with poor prognosis.[31323334] Overexpressed MtCK may be part of a metabolic adaptation of cancer cells and could sustain high energy turnover, but would be also protective against stress situations like hypoxia and possibly protect cells from apoptosis.[35] Although endometriosis is a benign disease, it exhibits malignant-like biological behaviors, including adhesion, aggression, and angiogenesis. Consequently, we speculated that CKMT1A- and CKMT1B-mediated metabolic pathways are involved in endometriosis progression. In addition, cytokine–cytokine receptor interaction signaling pathways were predicted to have strong relationships with the target genes, such as CCL23 (ENST00000591423), CCR10 (NM_016602), and TNFRSF6B (NM_003823). CCL23 has been reported to enhance endothelial cell migration, invasion, adhesion, and angiogenesis,[3637] which were also the pathophysiologic processes involved in endometriosis. CCR10 and its receptors regulated tissue-specific migration, maintenance, and functions of immune cells. Increasing evidence also found that CCR10/ligands were frequently exploited by epithelium-originated cancer cells for their survival, proliferation, and evasion from immune surveillance.[38] Moreover, the evidence revealed that TNFRSF6B stimulated endometriosis development by activating FAK.[29]

The findings highlight the relationship between circRNAs and ovarian endometriosis, which will provide a novel biomarker for screening ovarian endometriosis and help explore the role of circRNAs in the activation of endometriosis. However, this study had several limitations. First, the sample size was small and all cases were revised AFS Stage III–IV; thus, further validation involving in larger cohorts of patients with early stages of this disease is warranted. Second, noninvasive biomarkers are more likely to evaluate the diagnostic value of circRNA in clinical applications. Further studies will be needed to assess the level of circRNA in peripheral blood or menstrual blood samples. Third, the molecular mechanism and function of the present circRNAs should be tested by associated experiments in the future.

Overall, we revealed for the first time, the circRNA expression patterns and an associated circRNA-miRNA-mRNA network between women with and without endometriosis. circ_0004712 and circ_0002198 were confirmed to be novel biomarkers in discriminating endometriosis from controls. Our findings lay the foundation for in-depth mechanistic studies on endometriosis, which can strengthen the reliability of the two identified circRNAs as diagnostic and therapeutic targets.

Supplementary information is linked to the online version of the paper on the Chinese Medical Journal website.

Financial support and sponsorship

This study was supported by grants from the National Key R&D Program of China (No. 2017YFC1001200) and National Natural Science Foundation of China (No. 81270681).

Conflicts of interest

There are no conflicts of interest.