1. Introduction
Endometriosis is a gynecological disease characterized by the presence of endometrial-like tissue outside the uterine cavity, primarily in the abdominopelvic cavity (peritoneum, ovaries, or rectovaginal septum) [
1,
2]. This functionally active ectopic tissue is sensitive to hormonal stimulation and can originate cyclic bleeding, promoting the appearance of local inflammatory reactions and triggering pain (dysmenorrhea, dyspareunia, or chronic pelvic pain), gastrointestinal disorders, or infertility, among other maladies [
3,
4].
The pathogenesis and pathophysiology of endometriosis have yet to be fully elucidated, although various theories have been proposed. Increasing evidence has been published over recent years on the implication of endocrine-disrupting chemicals (EDCs) [
5] in an increased risk of endometriosis, which has been associated with human exposure to several families of EDCs [
6,
7,
8], including parabens (PBs) and benzophenones (BPs) [
6]. PBs, which are widely used as preservatives in personal care products (PCPs), pharmaceuticals, and food [
9], include methyl- (MeP), ethyl- (EtP), propyl- (PrP), and butyl-paraben (BuP) congeners. BPs, which frequently serve as UV-filters in PCPs [
10], include benzophenone-1 (BP-1), benzophenone-3 (BP-3), and 4-hydroxybenzophenone (4-OHBP) congeners.
In vitro and in vivo studies have demonstrated the estrogen-like effects of PBs and BPs, supporting the hypothesis that they promote endometriosis, an estrogen-dependent disease [
11,
12]. EDC exposure has also been associated with inflammation and oxidative stress [
13,
14], although its role in the pathophysiology remains unknown. Five key steps in the development and progression of endometriotic lesions—cell adhesion to the peritoneum; invasiveness into the mesothelium; recruitment of inflammatory cells; angiogenesis in endometriotic tissue; and cell proliferation—have been previously described [
15]. Changes in mediators of the corresponding cell-signaling pathways have been observed in studies of endometriosis [
16,
17,
18,
19,
20,
21,
22,
23,
24,
25]. Thus, in comparison to normal endometrial tissue, ectopic tissue has evidenced upregulation of the following: integrins and claudins, key members of tight junctions [
16,
17]; metalloproteases (
MMPs), involved in extracellular matrix cleavage and therefore invasiveness [
18,
19]; proinflammatory interleukins [
20,
21]; certain angiogenesis-related genes [
22,
23]; and genes related to cell proliferation [
24,
25]. Currently, the identification of different biomarkers of endometriosis is of crucial importance due to the need for non-invasive diagnostic methods of the disease [
26]. Additionally, these biomarkers could predict whether women will respond to first-line treatment for this disease. To our best knowledge, no study has been published on the ways in which exposure to EDCs might affect these cell-signaling pathways. Following the observation by our group that exposure to PBs and BPs is associated with endometriosis risk [
6], the present study was designed to determine the expression profile of a panel of 23 genes related to five key cell-signaling pathways for endometriosis development (cell adhesion; invasion, migration and metastasis; inflammation; angiogenesis; and cell proliferation and hormone stimulation) in women with endometriosis and to explore its relationship with their exposure to PBs and BPs. These families of EDCs were selected in this study due to (i) the growing evidence of the relationship between the consumption of cosmetics and PCPs, human exposure to PBs and BPs, and adverse health effects; (ii) previous studies reporting positive associations between PB/BP exposure and endometriosis risk [
6,
8]; and (iii) the scarcity of information on PB/BP exposure-related adverse outcome pathways in endometriosis.
3. Discussion
To the best of our knowledge, this is among the very first studies to describe the expression profile of key genes involved in the main molecular processes related to endometriosis in human endometriotic tissue and its potential association with exposure to EDCs. Most of the studied genes were expressed in the majority of samples, and exposure to certain PB and BP congeners was positively associated with the expression of genes involved in cell adhesion; invasion, migration, and metastasis; inflammation; angiogenesis; and cell proliferation and hormonal stimulation, critical molecular processes for the onset and progression of this disease. This evidence should be added to our previous report addressing the associations between PB/BP exposure and expression profiles of genes related to other cell molecular processes related to endometriosis, such as the cell cycle, cell differentiation, and lipid metabolism [
27].
Adhesion of endometrial cells to the peritoneal surface is considered a primary hallmark of endometriosis in its initial stages [
28]. Exposure to various PBs was associated with adhesion-related
ITGB2 and
CLDN7 genes in the present study, in line with recent in vitro findings of enhanced adhesive capacity in human umbilical vein endothelial cells after exposure to different EDCs [
29]. Upregulation of
CLDN7 has also been reported in ovarian cancer cell lines [
30], although the evidence is contradictory, given that downregulation of
CLDN7 has been observed in human endometriotic lesions [
31] and endometrial cancer, promoting proliferation and metastasis [
32].
The adhesion of endometrial cells to the peritoneal surface is followed by a process of invasion, migration, and metastasis that involves the overexpression of metalloproteinases [
33].
MMP1 was expressed in most of the present endometriotic samples, in agreement with previous reports of
MMP overexpression in women with endometriosis [
34,
35] and more severe stages of this disease [
19,
36]. An increase in
MMP activation has been detected in ovarian cancer, enhancing cell invasiveness [
30]. In the present investigation,
MMP1 expression was positively associated with exposure to MeP, ∑PBs, and ƩBPs, which may therefore favor the invasion, migration, and metastasis of endometriotic cells. In vitro studies have linked PB exposure to increased
MMP expression and activity in dermal fibroblasts [
37] and breast cancer cells [
38]. Exposure to other EDCs also increased
MMP expression in ovarian cancer cell lines [
39].
RHOB,
FUT8, and
RRM2 genes may also participate in invasion, migration, and metastasis. The expression of RHOB has not been described in endometriosis, but upregulation of this gene has been reported in human breast tumors [
40]. The present study describes for the first time the presence of
RHOB in endometriotic tissue and its positive association with exposure to most PB congeners studied. This suggests that exposure to these EDCs may contribute to the invasiveness of endometriotic tissue in a similar manner to the observed effects of EDCs on metastasis in breast cancer.
FUT8 has not previously been studied in endometriosis tissue samples but has been implicated in the epithelial–mesenchymal transition and metastasis [
41,
42]. PrP concentrations were positively associated with
FUT8 expression, suggesting that exposure to PrP may increase endometriotic tissue
FUT8 and promote metastasis and progression to more advanced stages of disease. It is also suspected that
RRM2 upregulation may promote cell invasion and metastasis in cancer and ovarian endometriosis [
43,
44]. Its expression was increased in the present women with higher MeP exposure, which may therefore play a potential role in boosting invasiveness. Finally, although the importance of
SPRY2 in endometriosis is unknown, it plays a crucial role in the regulation of cancer cell invasion [
45,
46]. According to the present findings, MeP and ƩPBs might downregulate
SPRY2 expression and thereby favor invasiveness and metastasis in women with endometriosis.
Another hallmark of endometriosis is a complex proinflammatory microenvironment, which favors cellular adhesion and invasion processes and the vascularization and proliferation of endometriotic lesions [
47]. Upregulation of
IL1RL1,
IL6ST,
TNFRSF1B, and
NR3C1 has been described in women with endometriosis [
48,
49]. In line with the proposal that human exposure to PBs/BPs may be related to inflammation [
14,
50], the present results suggest that the upregulation of
IL1RL1,
IL6ST,
TNFRSF1B, and
NR3C1 genes associated with exposure to various congeners of PBs and BPs could contribute to the generation of an inflammatory environment in endometriotic tissue.
Angiogenesis plays an important role in the pathogenesis and pathophysiology of endometriosis, and various genes are involved in this molecular process [
51]. Increased
ANG,
VEGFA, and
sVEGFR-1 expression was previously observed in the peritoneal fluid of women with endometriosis [
23,
52]. In the present study, urinary concentrations of various PB congeners were positively associated with
ANG,
ANGPT1, and
sVEGFR-1 expression. Despite associations not reaching statistical significance, exposure to BP-1 and BP-3 was unexpectedly related to downregulated levels of VEGFA. Although they were in accordance with a previous study carried out in Ehrlich ascites carcinoma and Dalton’s lymphoma ascites cells [
53], others have found upregulation of VEGFA after exposure to other EDCs such as bisphenols in breast cancer cells [
54] and upregulation of VEGFD in endometrial cells [
55], indicating that our non-significant associations between BPs and VEGFA might be spurious. The present findings support the hypothesis that exposure to PBs could favor the development of pro-angiogenic properties via endometriotic lesions, as suggested by the observation of increased microvessel density after exposure to bisphenol A in a model of neuroblastoma [
56].
Estradiol (E
2) promotes the persistence, multiplication, and progression of endometriotic lesions, acting as a growth factor for this tissue [
57], and endometriosis is closely associated with steroid metabolism and associated pathways [
3,
58]. E
2 production requires the action of
STAR, which facilitates the entry of cytosolic cholesterol into the mitochondrion, and
CYP19A1, which is responsible for aromatization of androgens into estrogens [
59,
60]. In the present study,
STAR gene was expressed in most of the ectopic tissue samples, and exposure to BP-1 and 4-OHBP was positively associated with STAR expression, indicating that exposure to these EDCs might be related to increased E
2 production in endometriotic tissue. These results support the previous suggestion of a potential role for EDCs in the upregulation of
CYP19A1 [
61]. In contrast, a reduction in
STAR expression was observed in the women with higher concentrations of MeP. Endometriotic lesions are very sensitive not only to the production of E
2 but also to the presence of circulating estrogens. In this regard, E
2 membrane receptors (
ERα and
ERβ) have been widely reported in endometriotic tissue [
62], with some researchers describing
ERβ as the predominant receptor on the cell surface [
63]. Hence, although
ERβ gene expression was not evaluated,
ERα gene expression in endometriotic tissue was found to be positively related to exposure to most PB congeners examined, indicating that this exposure might contribute to the sensitivity of this tissue to circulating E
2. The influence of BPs or PBs on ERα expression has not previously been studied, but the present results are in line with in vivo findings of
ERα overexpression after exposure to BPA [
64].
DUSP6 gene encodes a protein that counteracts cellular proliferation and is dysregulated in various diseases, including cancer. The number of cells expressing
DUSP6 was reported to be lower in ectopic versus eutopic endometrium [
65], suggesting this gene might also play a crucial role alongside E
2 in enhancing cell proliferation in endometriotic tissue. In the present investigation, MeP and ΣPBs were associated with a lower expression of this gene in endometriotic tissue, suggesting that these EDCs might contribute to the downregulation of
DUSP6 in this tissue.
Study limitations include the small sample size, reducing the statistical power, although it is highly challenging to gather endometriotic tissue samples for investigation. The limited amount of endometriotic tissue sample available prevented us from performing protein validation analysis to confirm the reported gene expression results. Furthermore, the utilization of spot urine samples prevented consideration of the variability in daily exposure to analytes with a relatively short elimination half-life. Nevertheless, samples were all first-morning urine samples taken during hospitalization before endometriosis surgery. In addition, only two families of EDCs were measured, with no evaluation of the combined effect of PBs and BPs alongside other EDCs. Furthermore, given the reported estrogenic/antiandrogenic effect of these compounds and possible similar underlying mechanisms of action [
11,
12], it is not possible to determine whether individual differences in these compounds could be the cause of the different associations found between EDC concentrations and gene expression levels. Finally, the ability to include covariates in regression analyses was limited by the small study population. Study strengths include the measurement, for the first time in endometriotic tissue, of numerous genes involved in cellular pathways related to the pathophysiology of endometriosis. Importantly, many cases revealed consistent associations between gene expression and exposure to certain PB/BP congeners. The combined investigation of biomarkers of exposure and potential biomarkers of effect yielded evidence of different pathways for adverse outcomes in endometriosis.
4. Material and Methods
4.1. Study Population and Sample Collection
This cross-sectional study, which forms part of the hospital-based case-control EndEA study, included 33 women with endometriosis undergoing surgery at the Gynecology and Obstetrics Units of San Cecilio and Virgen de las Nieves University Hospitals (Granada, Southern Spain) from January 2018 to July 2019 [
6,
7,
27]. All cases were diagnosed with endometriosis via laparotomy or laparoscopic surgery and histological confirmation. Inclusion criteria were premenopausal status, age between 20 and 54 years, receipt of abdominal surgery, and body mass index (BMI) below 35 kg/m
2. Exclusion criteria were history of cancer (except non-melanoma skin cancer), pregnancy at study enrolment, and inability to read and sign the informed consent document. Women were staged according to the Revised American Fertility Society classification [
66]. Written informed consent was provided by all participants, and the study was approved by the Research Ethics Committee of Granada (0464-N-18).
Endometriotic tissue samples were gathered during surgery and kept in QIAazol reagent (Qiagen, Hilden, Germany) to ensure RNA stability, and first-morning spot urine samples were collected in fasting conditions (fasting time > 8 h) on the same day as the surgery and were kept in PB- and BP-free glass tubes. All samples were immediately stored at −80 °C at the Biobank of the Public Andalusian Health Care System until their analysis. The BMI of participants was calculated, and data were gathered on sociodemographic, lifestyle, and clinical variables from epidemiological and clinical questionnaires completed by participants and on surgical variables from questionnaires completed by surgeons (
Supplementary Table S2). Endometriotic tissue samples were available from 33 participants, while urine samples were obtained from only 22 of them.
4.2. RNA Isolation and Quantitative Real-Time Polymerase Chain Reaction (qRT-PCR)
Endometriotic tissue samples were maintained in QIAzol reagent (Qiagen, Hilden, Germany) during the homogenization process to ensure RNA stability. Total RNA was extracted from 30 mg of sample with the RNeasy Mini kit (Qiagen, Hilden, Germany) according to the manufacturer’s protocol. Final RNA concentration and quality (260/280 ratio) were determined using a NanoDrop 2000 (Thermo Fisher Scientific, Waltham, MA, USA). Total RNA (1000 ng) was transcribed into cDNA with the iScriptAdvanced cDNA Synthesis Kit for RT-qPCR (Bio-Rad Laboratories, Hercules, CA, USA) according to the manufacturer’s instructions.
Real-time PCR was carried out with a CFX96 Real-time PCR detection system (Bio-Rad Laboratories, Hercules, CA, USA) using SsoAdvanced SYBR
® Green Supermix (Bio-Rad Laboratories, Hercules, CA, USA). The manufacturer’s protocol was followed to measure the expression of 23 genes involved in the five cell-signaling pathways: cell adhesion (integrin beta-2 (
ITGB2) and claudin 7 (
CLDN7)); migration/invasion (matrix metallopeptidase 1 (
MMP1), matrix metallopeptidase 7 (
MMP7), fucosyltransferase 8 (
FUT8), ribonucleotide reductase M2 (
RRM2), midkine (
MDK), ras homolog gene, family, member B (
RHOB), and sprout homolog 2 (
SPRY2)); inflammation (interleukin 1 receptor, type I (
IL1R1), interleukin 1 receptor, type II (
IL1R2), interleukin 6 cytokine family signal transducer (
IL6ST), nuclear receptor subfamily 3 group C member 1 (
NR3C1), and tumor necrosis factor receptor superfamily member 1B (
TNFRSF1B)); angiogenesis (angiogenin (
ANG), angiopoietin 1 (
ANGPT1), soluble vascular endothelial growth factor receptor-1 (
sVEGFR-1), and vascular endothelial growth factor A (
VEGFA)); and cell proliferation/hormone stimulation (cytochrome P450 family 19 subfamily A member 1 (
CYP19A1), dual specificity phosphatase 6 (
DUSP6), estrogen receptor alpha (
ERα), progesterone receptor (
PGR), and steroidogenic acute regulatory protein (
STAR)). Briefly, 10 μL SsoAdvanced Universal SYBR green mix, 1 μL cDNA (250 ng), and 1 μL of each PCR primer in 5 μmol L
−1 were mixed together. Nuclease-free water was added to achieve the final volume of 20 μL. qPCR reaction conditions were the following: first, the initiation process took place, for which the temperature reaction reached a temperature of 95 °C for two minutes (1 cycle); subsequently, the denaturation process occurred through 40 cycles of 5 s at a temperature of 95 °C; next, the alignment and extension process took place, through 40 cycles of 60 s at a temperature of 65–95 °C; finally, to evaluate the reaction specificity of the assay, a cycle of 5 sec/step was carried out, at a temperature of 65–95 °C (0.5 °C increments), showing a melt curve for each of the genes in each sample analyzed. Primers used for these studies were purchased from Bio-Rad Laboratories (Hercules, CA, USA), and their assay ID, amplicon context sequence, amplicon length (bp), and chromosome location are detailed in
Supplementary Table S8. Every mRNA primer used in this study was designed and experimentally tested by Bio-Rad Laboratories, generating an amplification plot using universal RNA, conducting a melt curve analysis, and calculating the efficiency and dynamic range from a seven-point standard curve. Specificity was determined by next-generation sequencing of the amplicon.
The criteria for the selection of these genes was based on previous evidence relating exposure to EDCs with the gene expression of these genes, their participation in the different cell-signaling pathways, and their relationship with endometriosis [
67,
68].
Gene expression was analyzed via Bio-Rad CFX 96 Manager Software to determine the cycle of quantification (Ct), and this was calculated using the 2
−ΔΔCt method [
69]. All values were normalized using glyceraldehyde-3-phosphate dehydrogenase (
GAPDH) RNA expression levels, calculating the difference corresponding to the quantification cycles between the target genes and GAPDH gene: ∆Ct = Ct (gene of interest)—Ct (GAPDH). Amplification efficiencies of the 23 primers used were ranged between 91 and 104%, being similar to that of the reference gene GAPDH (97%).
4.3. Chemical Analysis
Dispersive liquid–liquid microextraction (DLLME) and ultra-high-performance liquid chromatography with tandem mass spectrometry (UHPLC-MS/MS) were performed to determine urinary concentrations of MeP, EtP, PrP, BuP, BP-1, BP-3, and 4-OHBP, as previously described [
70].
After thawing urine samples at room temperature and centrifuging at 2600× g for 10 min, 1.0 mL was taken for analysis. The total amount of these benzophenones and parabens (free and conjugated) was obtained by enzymatically treating samples with an enzyme solution of β-glucuronidase/sulfatase, pre-prepared by dissolving 10 mg of β-glucuronidase/sulfatase (3·106 U g solid−1) in 15 mL of 1M ammonium acetate/acetic acid buffer solution (pH 5.0). Enzymatically treated samples were incubated at 37 °C for 24 h. Next, 20 μL of standard replacement solution (5 mg/L of EP_13C6, 2 mg/L of BPA-D16, and 2 mg/L of BP-d10) were added, and samples were diluted with 10 mL of 10% aqueous NaCl solution (pH 2.0, adjusted with 0.5 M HCl). Samples were then mixed with a solution of 1 mL of acetone (dispersing solvent) and 0.5 mL of trichloromethane (extraction solvent), shaken manually for 30 s, and centrifuged at 4000× g for 10 min. The organic phase was then collected with care from the bottom of the glass tube using a 1 mL pipette and was placed in 2 mL glass vials. All of the extracted fluid was evaporated under a nitrogen stream, and the residue was dissolved with 100 μL of an acetonitrile/water mixture (0.1% ammonia, 70:30 [v/v]) and vortexed for 30 s. The extract was then ready for analysis via UHPLC-MS/MS. The limit of detection (LOD) was determined as the minimum detectable amount of analyte, with a signal-to-noise ratio ≥ 3. LODs were 0.05 ng/mL for BP-1, 0.06 ng/mL for BP-3 and 4-OHBP, and 0.10 ng/mL for MeP, EtP, PrP, and BuP.
4.4. Statistical Analysis
In a descriptive analysis, arithmetic means and standard deviations were calculated for continuous variables and relative frequencies for categorical variables. Urinary concentrations of PBs, BPs, and gene expression levels were summarized as arithmetic means with standard deviations and as 25, 50, and 75 percentiles. The Shapiro–Wilk test was applied to check the normality of variable distributions, and data found to be non-normally distributed underwent logarithmic transformation.
Spearman’s rank correlation coefficient was used to evaluate monotonic correlations between urinary concentrations of PBs/BPs and the expression of genes with detection/expression frequencies above 75%. Simple linear regression models were also developed to evaluate the relationship between PB/BP exposure and gene expression levels, expressing the results as β with 95% confidence intervals. In parallel, simple linear regression models were constructed after patient stratification (i) into lower or higher PB and BP exposure, i.e., below or above the median concentration of each congener; and (ii) into low, moderate, and high exposure, based on tertiles of PB and BP concentrations. Genes expressed in 25–75% of samples were considered as dichotomous variables (detected/not detected), and simple logistic regression models were used to evaluate their association with PBs and BPs (considering assumptions of each regression modality), while no analyses were performed on genes expressed in <25% of samples.
Statistical analysis was performed with SPSS Statistics 23.0 (IBM, Armonk, NY, USA) analysis, and the level of significance was p < 0.05 in all tests; however, associations with p-values between 0.05 and 0.10 are cautiously considered, given the limited sample size. Due to sample availability, descriptive analysis on sociodemographic/gynecologic characteristics and gene expression profiles were summarized from the whole cohort (n = 33), while urinary concentrations of PBs/BPs and the associations between expression profiles and exposure were accomplished with data from the subset of women with both samples collected (n = 22).