Analysis of differences in the transcriptomic profiles of eutopic and ectopic endometriums in women with ovarian endometriosis

Ethical approval

Our study was approved by the institutional ethics committee of Shanghai First Maternity and Infant Hospital (KS16107). All of the subjects included in our study signed an informed consent form before research recruitment.

Tissue samples

We recruited six women with endometriosis and two women without endometriosis. All subjects had undergone laparoscopic and hysteroscopic surgery during the 8th–11th day of their menstruation, which is the early follicular phase, at Shanghai First Maternity and Infant Hospital in Shanghai, China. Six women with endometriosis were suspected to have ovarian endometriosis and infertility. Among those six women, four of them did not have dysmenorrhea, and the dysmenorrhea visual analogue scale (VAS) of the other two subjects was in the range of 1–2. Two women used as controls were suspected to have tubal infertility with an ovarian follicular cyst more than 4 cm in diameter. Both did not have dysmenorrhea. Ectopic endometrial tissues were obtained by laparoscopy, and eutopic endometrial tissues were achieved by dilation and curettage (D&C). Patients who had any history of acute inflammatory diseases, systemic autoimmune disorders, malignant tumors, or hormone therapies within six months were excluded. Endometriosis was diagnosed by visual confirmation under laparoscopy and pathological analysis. The absence of endometriosis was confirmed by diagnostic laparoscopy.

Cell samples

All the cryopreserved tissues were minced and incubated with collagenase (Sigma-Aldrich, C2674-100MG) at 37 °C for 10 min, followed by filtration. Endometrial tissues were digested by collagenase/hyaluronidase (Sigma-Aldrich, C2674-100MG/H3506-100MG) at 37 °C for 2 h. After digestion, the endometrial stroma cells and epithelial cells were dispersed. Dozens of cells were mixed and randomly placed into tubes as one group, and 29 groups from eight patients were collected from those tissue samples. Among them, 12 groups were ectopic endometrial cells, 14 groups were eutopic endometrial cells from the patients with ovarian endometriosis, and 3 groups were eutopic endometrial cells from the controls (Table S1).

Library construction and RNA sequencing

We used the Smart-seq2 protocol with minor modifications due to the low number of cells in each group, which mainly changed the reverse transcriptase for all samples (Picelli et al., 2014). In brief, cells were picked into a lysis buffer to release total RNA and were converted into cDNA using the Oligo-dT30VN primer and Superscript III reverse transcriptase (Invitrogen, 18080044) and further pre-amplified using KAPA HiFi HotStart ReadyMix (Kapa Biosystems, KK2601); AMPure XP beads (Beckman Coulter, A63881) were used to purify the cDNA products and the subsequent tagmentation reaction was performed using Nextera Tagmentation (Illumina, FC-131-1096). The concentration of the cDNA library was adjusted to obtain a total of 5 nM and was sequenced on Illumina HiSeq2500 sequencers in the PE150 model.

Processing of RNA sequencing data

Raw RNA sequencing reads were trimmed to remove the sequencing adaptor, low-quality reads, and bases using Trimmomatic (V0.33) (Bolger, Lohse & Usadel, 2014). After trimming and quality control, clean reads were aligned to the human genome (GRCh38) using STAR (version 2.7.1a) (Dobin et al., 2013). Uniquely mapped reads were counted with HTSeq and the gene expression level was quantified using the read count.

Weighted gene co-expression network analysis (WGCNA)

The genes with a mean read count of less than 20 were removed. The read counts of these genes were normalized using the function varianceStablizingTransformation in the DESeq2 package (V1.22.2). Normalized data from each sample in the eutopic endometrial cell group, ectopic endometrial cell group, and control group were used for weighted gene co-expression network analysis (WGCNA) (Langfelder & Horvath, 2008). According to the results from the pickSoftThreshold function, the power was set to 4 to find different gene modules using the blockwiseModules function. Eight gene modules were obtained. The gene co-expression networks of the turquoise and blue modules were extracted using MCODE (version 1.5.1) in Cytoscape (V3.7.2) (Shannon et al., 2003). GO (Gene ontology) enrichment analysis of the hub genes in these two networks was performed using DAVID (version 6.8) (Huang et al., 2007).

Differential expression analysis

Differential expression analysis among the eutopic endometrial cell group, ectopic endometrial cell group, and control group was performed using the DESeq2 package (V1.22.2). Genes with mean read counts value of less than one were filtered to remove low-expressed features. The differentially expressed genes (DEGs) were calculated using the DESeq function. DEGs were identified with a p-value of <0.05 and abs(log2FoldChange) >2. GO (Gene ontology) enrichment analysis of DEGs was performed using DAVID (version 6.8).

Validation with qRT-PCR

Significantly differentially expressed genes were validated by qRT-PCR using the QuantStudioTM 3 Real-Time PCR System (Applied Biosystems). The primers used in this study are shown in Table S2. All of the genes were normalized to β-ACTIN. qRT-PCR was performed in a total reaction volume of 10 µl, including 5 µl 2x TB green premix EX Taq II (Takara, RR820A), 0.4 µl forward primer (10 µmol/L), 0.4 µl reverse primer (10 µmol/L), 3.2 µl DNase-free water, and 1 µl cDNA. qPCR was achieved by heating for two minutes at 95 °C, 40 cycles of 95 °C (5 s), and 60 °C (30 s). We detected for each gene three times, independently.

Statistical analysis

Our results were analyzed using the ΔΔCT method. An analysis of variance (ANOVA) or t test was conducted using GraphPad Prism 6 to determine the significance among three groups or between two groups, respectively. p < 0.05 was considered to be statistically significant.

Gene co-expression network analysis of eutopic and ectopic endometrial cells

We normalized the read counts of 12 eutopic endometrial cell samples and 14 ectopic endometrial cell samples from women with endometriosis, as well as 3 control endometrial cell samples from healthy women without endometriosis using DESeq2. The genes with a mean read counts value less than 20 were removed and 6,440 genes remained. WGCNA analysis was used and eight gene modules were found related to these three groups (Fig. 1A), including the turquoise module (3,837 genes), blue module (616 genes), brown module (336 genes), yellow module (314 genes), green module (198 genes), red module (150 genes), black module (45 genes), and grey module (944 genes). The eutopic cell group was found to be closely related to the turquoise module according to the correlation analysis (Fig. 1B); the hub genes were enriched in cell adhesion, platelet aggregation, and tumor occurrence by Gene ontology (GO) enrichment analysis. GO terms included cell–cell adhesion, platelet aggregation, embryo implantation, movement of cell or subcellular component, vascular endothelial growth factor receptor signaling pathway, etc. (Figs. 1C and 1D). These genes may play an important role in the pathogenesis of endometriosis. The ectopic endometrial cell group was mainly related to the blue module (Fig. 1B) and the genes were enriched in cell migration, epithelial cell maturation, gap junction assembly, as well as platelet aggregation, etc. (Figs. 1E and 1F), suggesting that genes including PIK3CB, SORBS2, CARMIL1, PTK7, and CEACAM1 may promote cell migration and the formation of eutopic endometrial cells. Genes related to platelet aggregation were found in the blue module, including RAP2B, TLN1, and PIK3CB. These genes cause platelets to aggregate in endometriotic lesions and play critical roles in the development and progression of endometriosis.

Differential expression analysis among eutopic endometrial cells, ectopic endometrial cells, and control endometrial cells

To explore the mechanism of endometriosis, differential expression analysis was performed among the eutopic endometrial cell group, the ectopic endometrial cell group, and the control group using DESeq2. Our results showed that there were 688 genes significantly up-regulated and 298 genes down-regulated in the eutopic endometrial cell group compared with the control group (Fig. 2A). qRT-PCR was used to validate the differentially expressed genes (DEGs) and the results for the down-regulated expression in the eutopic endometrial cell group suggested the defect of ARNTL in the eutopic endometrial cell group may lead to an increased risk of infertility in patients with endometriosis (Fig. 2B, Fig. S1A). MMP11 was significantly up-regulated for the up-regulated expression genes, suggesting that endometrial cells in patients with endometriosis present a mechanical phenotype of tissue remodeling. The increased expression of MMP11 may be the cause of endometrium hyperplasia (Fig. 2C, Fig. S1B).

In addition, there were 155 genes significantly up-regulated and eight genes down-regulated in the ectopic endometrial cell group compared with the control group (Fig. 2D). PIWIL2 was observed, showing an up-regulated expression in ectopic endometrial cells (Fig. 2E, Fig. S1C), suggesting that ectopic endometrial cells were closely related to cell adhesion and deformation. FLT1 plays a very important role in cell migration, chemotaxis, cell survival, angiogenesis regulation, embryonic vasculature development, and tumor cell invasion. FLT1 and SCN11A were up-regulated and GNLY was down-regulated, suggesting that endometriosis is associated with an inflammatory response (Figs. 2F–2H and Figs. S1D–S1F).

Our analysis of differential expression between the ectopic endometrial cell group and the eutopic endometrial cell group showed that there were 7,450 genes up-regulated and 1,327 genes down-regulated in the ectopic endometrial cell group (Fig. 2I). MUC19 was significantly up-regulated in the ectopic endometrial cell group, which indicates that adhesion is stronger in ectopic endometrial cells than in eutopic endometrial cells (Fig. 2J, Fig. S1G). In addition, the numbers of DEGs between ectopic endometrial cells and eutopic endometrial cells were much higher than other groups of DEGs, and many genes were significantly up-regulated in ectopic endometrial cells, regardless of their comparison with eutopic endometrial cells or control cells (Fig. 2K).

The malignant transformation tendency of endometriosis

Ovarian endometriosis tissue is known as a benign cyst but the risk of malignant transformation exists. In our analysis, we also saw a considerable number of genes associated with tumorigenesis in endometriosis cells. The results of WGCNA showed that the turquoise and blue modules, which are highly related to the eutopic endometrial cell group and the ectopic endometrial cell group, all contained tumor related genes, such as UBC, UBA52, RPS27A, PIK3CB, IRS1, and CEACAM1 (Figs. 1C and 1E). Whereas, in the differentially expressed gene analysis, the tumor related genes TEX41, POLQ, and FLT1 were highly expressed in the ectopic endometrial cell group of endometriosis, and the expression of tumor suppressor gene PLCD1 was down-regulated when compared with the control group (Figs. 3A–3F, 2F and S1D). These results suggested that the endometrium has tumor-like characteristics, such as implantation, metastasis, and recurrence. Notably, in the ectopic endometrial cell group of endometriosis, tumor suppressor genes such as OSR2 were down-regulated compared with the eutopic endometrial cell group (Figs. 3G and 3H).