Characterization of a profile of epigenetic alterations involved in the aetiopathogenesis of endometriosis. Validation of molecular biomarkers for diagnosis and prognosis of endometriosis

Introduction: Endometriosis is an oestrogen-dependent inflammatory disorder defined by the presence of endometrial-like tissue in ectopic locations, which limits the quality of life of affected women. This pathology affects 10% of reproductive-age women from all ethnic and social groups, although the prevalence in those patients experiencing pain, infertility or both is as high as 35%-50%, being the estimated prevalence of this condition around 176 million worldwide. Endometriosis is associated with an average diagnostic delay of 7 years, what could be partially explained by the lack of non-invasive biomarkers for diagnose, since the gold standard method is laparoscopy followed by histological confirmation of the presence of endometrial-like glands and stroma in biopsies. Although several aethiogenic theories have attempted to explain the disparity of manifestations of the disease, the most accepted theory is still the Sampson’s retrograde menstruation theory, proposed almost 100 years ago.According to Sampson’s proposal, endometrium shed during menstruation is capable of reaching the pelvic cavityby retrograde flow through the fallopian tubes, then being able to implant and proliferate. Albeit it has been documented that retrograde menstruation occurs in 90% of healthy women in reproductive age with patent fallopian tube, the fact that only a small percentage develops the disease suggests that there must be additional mechanisms that allow the migrated tissue to implant and survive. In this context, it has been proposed that endometrial and peritoneal factors could be involved in the pathophysiology of endometriosis. At tissue level, endometriotic implants require neovascularization to proliferate, invade the extracellular matrix (ECM) and establish the lesion, displaying an oncomimetic behaviour. Neovascularization is achieved by means of angiogenesis, or the process of formation of new blood vessels from pre-existing ones. Previous results from our group showed that protein levels of VEGF-A were significantly elevated in the eutopic endometrium of patients with respect to the endometrium of women without endometriosis. In addition, in the various endometriosis lesions, peritoneal implants had levels of VEGF-A significantly increased compared to ovarian endometriomas.Additionally, remodelling of ECM plays a critical important role in the establishment of endometriotic lesions. Abnormal expression of components of metalloproteinase system and proteolysis system has been reported in both the endometrium and endometriotic tissue of women affected by endometriosis, suggesting their involvement in the establishment of lesions. Thus, it seems reasonable to simultaneously study components of the angiogenic and proteolytic system in our patients. Furthermore, peritoneal fluid (PF) represents one of the most important peritoneal factors. This biofluid is defined as an ovarian exudation product originated mainly as a result of an increased vascular permeability, with cyclic variation in volume and steroid hormones. Because ectopic endometriotic lesions located in the pelvic peritoneum are in close relationship with this biofluid, their components have emerged as an important field of study. In PF, there are cell types such as erythrocytes, macrophages and endometrial cells. Endometrial cells secrete products such as glycodelin, the peritoneum substances such as CA-125 and macrophages secrete a large diversity of molecules, among which are cytokines, growth factors and angiogenic factors. Regarding angiogenesis, VEGF-A has been detected in higher concentrations in the PF of women with endometriosis (McLaren et al, 1996; Gilabert-Estellés et al, 2007, Marí-Alexandre et al, 2017b), being correlated with the disease stage (Shifren et al, 1996). It has been suggested that increased levels of VEGF-A in PF in patients with endometriosis are the consequence of combined expression by ectopic lesions and activated macrophages present in the fluid. In addition, some studies indicate that the PF of women with endometriosis induces cell proliferation in vitro, although the underlying mechanism is unknown. Epigenetics is defined as “the structural adaptation of chromosomal regions so as to register, signal or perpetuate altered activity states”. There exist four epigenetic mechanisms, whose marks are dynamic and reversible: non-coding RNAs (ncRNAs), DNA methylation, histone modifications and chromatin remodelling. microRNAs (miRNAs are small (19-22nt) non-coding RNAs that can act as post-transcriptional regulators of gene expression, reducing the expression of their target mRNAs either inhibiting its translation or promoting its degradation. miRNAs usually regulate gene expression by binding to the 3’ UTR (UnTranslatedRegion) of their target mRNAs. Importantly, several miRNAs can target a given mRNA and a single miRNA can target several mRNA, increasing the complexity of the regulatory mechanism mediated by these molecules. A growing body of literature, including papers from our research group supports the regulatory function of miRNAs in a myriad of processed involved in the pathophysiology of endometriosis, such as angiogenesis, proteolysis, extracellular matrix remodelling, inflammation, proliferation or apoptosis. Nevertheless, the lack of overlapping among different studies is noticeable (Marí-Alexandre et al., 2017). DNA methylation is a covalent modification defined by the addition of a methyl (-CH3) group at the 5’ carbon of cytosines, being by far the most widely studied epigenetic mark. Deviations in DNA methylation patterns can occur via gain (hypermethylation) or loss (hypomethylation) of methylation marks with respect to methylomes defined as normal. Whereas hypermethylation at CpG islands of gene promoters is associated with the repression of gene expression, hypomethylation has been linked to gene expression. Albeit largely studied in oncological malignancies, the role of DNA methylation in the pathophysiology of endometriosis alterations has just begun to be explored. Hypothesis and objectives: the overall hypothesis is that an aberrant epigenetic profile could modulate the molecular mechanisms underlying the aetiology and pathophysiology of endometriosis. Getting into detail, the first hypothesis is that endometrial factors, in terms of an aberrant miRNA and DNA methylation profile, are responsible for the altered survival, proliferation, angiogenesis and extracellular matrix remodelling observed in the disease. The second hypothesis is that peritoneal factors, in terms of altered composition of peritoneal fluid in patients with endometriosis, could mediate the deregulation of miRNAs in endometrial and endometriotic stromal cells, being responsible for the altered biological processes observed in this pathology. Finally, the third hypothesis is that the expression of VEGF-A in stromal cell cultures can be epigenetically modulated by means of selected miRNA mimics and of targeted DNA methylation at VEGFA promoter. With regards to the objectives, the first objective was to assess miRNA expression profiles in samples of eutopic and control endometria and ovarian endometrioma from women with endometriosis; and also, to validate the results in a higher cohort of samples, including paired peritoneal implants and rectovaginal nodules. Additionally, we aimed to evaluate the role of deregulated miRNAs on the expression of the main regulators of angiogenesis and fibrinolysis. The second objective was to validate our in vitro model in primary stromal cell cultures from endometrial samples and from ovarian endometrioma. Additionally, we wanted to evaluate the influence of peritoneal fluid on the expression of selected angiogenesis- and fibrinolysis-related miRNAs. The third objective was to assess the influence of peritoneal fluid on miRNA expression profiles in primary endometrial and endometriotic stromal cell cultures. Additionally, we wanted to validate VEGFA mRNA as a target of miR-16-5p, miR-29c-3p, miR-424-5p. The fourth objective was to define the DNA methylation profiles in endometrial samples from patients with endometriosis and from control women by employing the Illumina Infinium MethylationEPIC BeadChip. Finally, the fifth objective was to down-regulate the expression of VEGF-A in primary stromal cell cultures of patients with endometriosis by employing techniques for targeted DNA methylation (fused Zinc fingers proteins – Dnmt3a3L enzyme). Results: To allow a better understanding of the experimental results, these have been structured in 5 chapters. In Chapter 1 we investigate the implication of miRNA deregulation at tissue level in the pathophysiology of endometriosis. To this end, we established a case-control study involving specimens from 51 women with endometriosis and 32 women without the disease. miRNA expression profiles (GeneChip miRNA2.0; Affymetrix) were performed in samples of control (CNT) and eutopic (EUT)endometria and ovarian endometrioma (OMA) from women with endometriosis.After statistical treatment of microarray results, in silico analysis allowed us to select 12 differentially expressed miRNAs (namely (miR-16-5p, -29c-3p, -138-5p, 202-3p, -373-3p, -411-5p, -411-3p, -424-5p, -449b-3p, -556-3p, -636, -935), which targets were involved in angiogenesis, fibrinolysis, or that were implicated in endometriosis. These miRNAs were validated in a larger cohort of samples, including rectovaginal nodules and peritoneal implants. Additionally, levels of vascular endothelial growth factor(VEGF-A), thrombospondin-1 (TSP-1), urokinase plasminogen activator (uPA) and plasminogen activator inhibitor-1 (PAI-1) were measured by ELISA. We observed that EUT tissue showed significantly lower levels of miR-202-3p, miR-424-5p,miR-449b-3p and miR-556-3p, and higher levels of VEGF-A and uPA than CNT endometrium. However, OMA showed significantly lower expression of miR-449b-3p than endometrium from both controls and patients, and higher levels ofPAI-1 and the angiogenic inhibitor TSP-1.Asignificant inverse correlation between miR-424-5p and VEGF-A protein levels was observed in EUT, and an inverse correlation between miR-449b-3p and TSP-1 protein levels was observed in OMA. Peritonealimplants had significantly higher levels of VEGF-A than ovarian endometrioma samples. Hence, differences in miRNA levels could modulate the expression of VEGF-A and TSP-1, whichmay play an important role in the pathogenesis of endometriosis. The higher angiogenic and proteolytic activities observed in EUTfrom patients might facilitate the implantation of endometrial cells at ectopic sites. In Chapter 2 and Chapter 3 we investigate the influence of peritoneal fluid from patients with endometriosis on the deregulation of miRNAs in endometrial and endometriotic stromal cells, being responsible for the altered biological processes observed in this pathology. Specifically, in Chapter 2 we validated an in vitro model of endometriosis,previously established by our research group,in which primary stromal cell cultures from CNT (n=8), EUT (n=11) and from OMA (n=11) were exposed to PF pools from patients (EPF) and control women (CPF) in order to simulate the peritoneal microenvironment. Afterwards, we quantified the expression of six miRNAs (miR-16, -17-5p, -20a, -125a, -221, and -222) by qRT-PCR and protein and mRNA levels of VEGF-A, TSP-1, uPA and PAI-1 by ELISA and qRT-PCR, respectively. We observed that EPF and CPF pools induced a significant reduction of the levels of all studied miRNAs in endometrial and endometriotic cell cultures. Moreover, both peritoneal fluids (CPF and EPF) induced a significant increase in VEGF-A, uPA and PAI-1 protein levels in all cell cultures without significant increase in mRNA levels, suggesting a post-transcriptional mechanism of gene expression regulation. Endometrial cell cultures from patients treated with EPF showed lower expression of miRNAs and higher expression of VEGF-A protein levels than cultures from controls. All in all, this in vitro study allowed us to validate previous results from our research group, alsoindicating that EPF modulates the expression of miRNAs that could contribute to the angiogenic and proteolytic disequilibriumobserved in this disease. In Chapter 3, we broaden the scope of our interest to assess the influence of peritoneal fluid on miRNA expression profiles in primary endometrial and endometriotic stromal cell cultures. Thus, we performed miRNA expression arrays (Affymetrix, GeneChip 4.0) in cells stimulated with the experimental condition described in Chapter 2. We observed that EPF modified themiRNAexpression profile in eutopic cells. A total of 267miRNAweremodified in response to EPF compared with eutopic cellswithout PF stimulation (ØPF). 9differently expressed miRNAs (namely miR-16-5p, miR-21-5p, miR-29c-3p, miR-106b-5p, miR-130a-5p,miR-149-5p, miR-185-5p, miR-195-5p, miR-424-5p) with potential targetsinvolved in angiogenesis, proteolysis or endometriosis, were validated in further experiments. Except for miR-149-5p, all validated miRNAs showed significantly lower levels after EPF treatment in primary cell cultures from EUT in comparison with ØPF. Additionally, functional experiments were performed in primary cell cultures using mimics for miRNAs miR-16-5p, miR-29c-3p and miR-424-5p in order to assess their effect as VEGF-A expressionregulators. We observed a significant down-regulation of VEGF-A protein expression after transfection of stromal cells with miRNAs mimics for miR-16-5p, miR-29c-3p and miR-424-5p, without significantly modifying VEGFAmRNA. One step further, we performed luciferase expression assays to confirm a repressive action of miR-29c-3p through forming miRNA:VEGFA duplexes. In Chapter 4, we defined the DNA methylation profiles in endometrial samples from patients with endometriosis and from control women by employing, for the first time in endometriosis, the Infinium MethylationEPIC BeadChip(Illumina). Arrays were performed in 13EUT from women with endometriosis (III-IV stage) and 11CNT from women without the disease, paired by age and phase of menstrual cycle. Unsupervised hierarchical clustering of methylation levels of 5000 random selected CpGs showed no segregation of endometrial tissues neither by disease status (patients vs control) nor by cycle phase (proliferative, secretory or menstrual). This was also confirmed by scatter plots of β-values from patients’ and controls’ samples. To define the existence of potential specific elements of distinction by disease status we used a supervised elasticnet-penalized logistic regression model. We determined 7 CpGs (4 CpGs related to LRP5, SEMA4F, MYL12B and PIK3AP and 3CpGs not assigned to any known gene) which together provided the highest discrimination power between patient and control women’s samples. However, Δβ were rather small, regardless of the phase of the menstrual cycle considered. Due to the importance of the endometrium in the implantation process and the singularity of the tissue in terms of cyclical changes in gene expression under hormonal influence, we aimed to provide a detailed description of genome wide DNA methylation profile of this tissue. We observed a bimodal distribution of CpGs, with 23.3% of them hypomethylated(β ≤ 0.2), whereas 24.8% of the CpGs were hypermethylated (β≥0.8). Additionally, when looking at functional distribution of CpGs, we observed an enrichment of promoters in hypomethylated regions, whereas hypermethylated CpGs were enriched in gene body regions. This was also extended to the main regulators of angiogenesis (VEGF-A, TSP-1) and fibrinolysis (uPA, PAI-1), in agreement with an active state of transcription for these genes in the endometrium and a notable gene stability. This supports the potential involvement of a post-transcriptional mechanism of regulation, highlighting the importance of miRNA regulation of these proteins partially deciphered in this work (Chapter 1 and Chapter 3). In Chapter 5 we wished to down-regulate the expression of VEGFA in primary stromal cell cultures from EUT by employing breakthrough techniques for targeted DNA methylation (Zinc finger proteins and the CRISPR-dCas9 system). Firstly, we employed bisulphite sequencing to validate results from methylation arrays, confirming a hypomethylated state at VEGF-A promoter.Secondly, we performed experiments to establish passage 3 as the optimum in which experiments would be performed. Finally, we transfected primary endometrial stromal cells from EUT with plasmids encoding Dnmt3a, GFP and either a specific ZFP or a dCas9/gRNA. Although multiple transfection conditions were employed (FuGene® HD, Lipofectamine LTX and Lipofectamine 3000 transfection reagents), we were unable to maintain cells alive for 120h with a reasonable transfection efficiency. In conclusion, the combined toxicity of transfection reagents and the multiple plasmids co-transfected is not a suitable strategy for primary endometrial stromal cell cultures, albeit it has shown to be successful for easily transfectable cells (e.g. HEK293 cell lines) and resistant tumour cells (e.g. ovarian cancer SKOV3 cell lines). In conclusion, our results corroborate the importance of both endometrial and peritoneal components in the pathophysiology of endometriosis, through a deregulation of angiogenesis and proteolysis. The lack of significant differences in DNA methylation in EUT vs CNT endometrium highlights the importance of other epigenetic mechanisms, such as miRNAs, in the survival of migrated endometrial tissues and establishment of endometriotic lesions