{"paper_id":"e8f87eb4-13c0-437e-87b0-501b73b11d00","body_text":"1\nVol.:(0123456789)Scientific Reports |        (2021) 11:20870  | https://doi.org/10.1038/s41598-021-00299-2\nwww.nature.com/scientificreports\nProteomic analysis of peritoneal \nfluid identified COMP and TGFBI \nas new candidate biomarkers \nfor endometriosis\nV. Janša1, T. Klančič2, M. Pušić2, M. Klein3, E. Vrtačnik Bokal1,4, H. Ban Frangež1,4* & \nT. Lanišnik Rižner2*\nEndometriosis is a common non-malignant gynecological disease that significantly compromises \nfertility and quality of life of the majority of patients. The gold standard for diagnosis is visual \ninspection of the pelvic organs by surgical laparoscopy and there are no biomarkers that would \nallow non-invasive diagnosis. The pathogenesis of endometriosis is not completely understood, \nthus analysis of peritoneal fluid might contribute in this respect. Our prospective case–control \nstudy included 58 patients undergoing laparoscopy due to infertility, 32 patients with peritoneal \nendometriosis (cases) and 26 patients with unexplained primary infertility (controls). Discovery \nproteomics using antibody microarrays that covered 1360 proteins identified 16 proteins with \ndifferent levels in cases versus the control patients. The validation using an ELISA approach confirmed \nsignificant differences in the levels of cartilage oligomeric matrix protein (COMP) and transforming \ngrowth factor-β-induced protein ig-h3 (TGFBI) and nonsignificant differences in angiotensinogen \n(AGT). A classification model based on a linear support vector machine revealed AUC of > 0.83, \nsensitivity of 0.81 and specificity of 1.00. Differentially expressed proteins represent candidates for \ndiagnostic and prognostic biomarkers or drug targets. Our findings have brought new knowledge that \nwill be helpful in the understanding of the pathophysiology of endometriosis and warrant further \nstudies in blood samples.\nEndometriosis is a common non-malignant gynecological disease with an estimated prevalence of 10% world -\nwide, which increases to 50% for women with infertility or chronic  pain1. There are multiple theories including \nretrograde menstruation, venous dissemination, lymphatic dissemination, congenital Mullerian remnants, coe-\nlomic metaplasia, and engrafting of bone-marrow stem  cells2. Despite major efforts, endometriosis still remains \na disease that is not completely understood and that has a poorly defined etiology and a complex pathogenesis. \nTo date, its pathogenesis is known to involve degradation of the extracellular matrix, aberrant apoptosis, angio-\ngenesis, enhanced cell adhesion, cell proliferation, increased oxidative stress, inflammation processes, a disturbed \nimmune system, and  more3–6. As a result of these pathophysiological processes, endometrial cells survive and \nproliferate at ectopic sites, which can evoke chronic pelvic  inflammation7.\nEndometriosis significantly compromises the quality of life in the majority of women, and it is a major \ncofactor in infertility. The gold standard for diagnosis is visual inspection of the pelvic organs by surgical lapa -\nroscopy. Due to the non-specific symptoms and surgery-associated risks, it can be over 10 years before women \nare diagnosed and correctly  treated8.\nThe peritoneum, retroperitoneum, ovaries, bowel, and other sites represents a ‘local environment’ for endo-\nmetrial lesions. Ectopic endometrial cells evoke local inflammation, which is mediated by immune cells and their \npro-inflammatory  products4,5,9. The analysis of peritoneal fluid, that bathes the endometriotic lesions might be \na key to a better understanding of  endometriosis10. Revealing of the underlying pathophysiology of endome-\ntriosis at the molecular level would be of great clinical importance. As peritoneal fluid has a complex role in the \nOPEN\n1Department of Obstetrics and Gynecology, University Medical Centre Ljubljana, Šlajmerjeva 3, 1000 Ljubljana, \nSlovenia. 2Faculty of Medicine, Institute of Biochemistry, University of Ljubljana, Vrazov trg 2, 1000 Ljubljana, \nSlovenia. 3Sciomics GmbH, Karl-Landsteiner -Straße 6, 69151 Neckargemünd, Germany. 4Faculty of Medicine, \nUniversity of Ljubljana, Vrazov trg 2, 1000 Ljubljana, Slovenia.  *email: helena.ban.frangez@gmail.com ; \ntea.lanisnik-rizner@mf.uni-lj.si\n\n2\nVol:.(1234567890)Scientific Reports |        (2021) 11:20870  | https://doi.org/10.1038/s41598-021-00299-2\nwww.nature.com/scientificreports/\netiopathogenesis of endometriosis, it should serve as a source of new diagnostic or predictive biomarkers, which \nmight also contribute to the identification of new drug targets. The surface of the peritoneal cavity is large, and \nit allows passive dialysis of substances between the peritoneal fluid and the blood plasma, where the diffusion \nrates decrease with molecular  weight11–13. Studies of peritoneal fluid might thus contribute to the identification \nof blood biomarkers for non-invasive diagnosis of endometriosis.\nWe used a proteomic approach with a high-content antibody microarray that covers 1360 proteins to search \nfor proteins with significantly different levels in peritoneal fluid samples from women with endometriosis (cases) \ncompared to women with unexplained primary infertility (controls). The aim of the study was to identify poten-\ntial diagnostic and predictive biomarkers or drug targets and to contribute to a better understanding of the \npathophysiology of endometriosis.\nMaterial and methods\nThis study was designed as a prospective case–control study. It consisted of a discovery phase (proteomic study: \nantibody array analysis) and a validation phase (enzyme-linked immunosorbent assay [ELISA] validation).\nPatient selection. The discovery proteomic analysis included a total of 12 women with primary infertil-\nity who were stratified according to laparoscopy and histological confirmation (Table 1, Fig. 1). The case group \nincluded six women with endometriosis and primary infertility, and the control group included six women with \nunexplained primary infertility. The validation study with ELISA included 46 women with primary infertility \n(Table 2, Fig. 1), where the case group included 26 women with endometriosis, and the control group included \n20 women with unexplained primary infertility. All of the patients had laparoscopy carried out due to infertility \nand the diagnosis was confirmed histologically. All of the women had a body mass index (BMI) in the nor -\nmal range, with a regular menstrual cycle (21–35 days). The partner semen analyses were normal for all of the \nwomen included. The further inclusion criteria included: no previous pelvic surgery, no known pelvic inflamma-\ntory disease, and ultrasound examination showed no pathology (controls) other than endometriosis (cases). The \nexclusion criteria included patients who had undergone hormonal therapy in the last year, those with irregular \nmenstrual cycles, and patients with autoimmune diseases, malignant or suspected malignant diseases, previous \npelvic inflammatory disease, and leiomyoma uteri or polycystic ovaries. None of the patients had undergone \nprevious pelvic surgery.\nSample and data collection. All of the women who met the inclusion criteria were additionally evalu-\nated. They filled out a questionnaire on their health history, stress levels, use of medications and types of pain \n(dysmenorrhea, dyspareunia, chronic pain), using a validated visual analogue scale. Stratification was carried \nout based on the results of the laparoscopy and histological confirmation: the case group only included women \nwith endometriosis and no other pathology, and the control group only included women without any pathology \nat laparoscopy.\nPeritoneal fluid samples were collected during the laparoscopy, before any intra-abdominal procedures were \ncarried out. The pneumoperitoneum was reached using a Veress needle at the umbilicus, and peritoneal fluid \nTable 1.  Clinical characteristics of the 12 patients included in the discovery phase.\nParameter Units Detail Controls Cases p-value\nTotal patient numbers n – 6 6 –\nMean age (mean ± SD) years – 29.6 ± 2.8 28.1 ± 3.1  > 0.05\nMean body mass index (mean ± SD) kg/m2 – 23.8 ± 3.1 23.8 ± 1.7  > 0.05\nMenstrual phase n (%)\nFollicular 6 (100) 6 (100)  > 0.05\nLuteal 0 (0) 0 (0)\nOral contraceptives (last 3 months) n (%)\nNo 6 (100) 6 (100)\nYe s 0 (0) 0 (0)\nHormonal therapy (last 3 months) n (%)\nNo 6 (100) 6 (100)\nYe s 0 (0) 0 (0)\nMedications (last 1 week) n (%)\nNo 6 (100) 6 (0)\nYe s 0 (0) 0 (0)\nSmoking status n (%)\nNon-smoker 6 (100) 6 (100)\nSmoker 0 (0) 0 (0)\nOccasional smoker 0 (0) 0 (0)\nFormer smoker 0 (0) 0 (0)\nEndometriosis n (%) Ovarian plus peritoneal 0 (0) 6 (100)\nRevised American Society for Reproductive Medicine score n (%)\nI 0 (0) 0 (0)\nII 0 (0) 0 (0)\nIII 0 (0) 6 (100)\nIV 0 (0) 0 (0)\n\n3\nVol.:(0123456789)Scientific Reports |        (2021) 11:20870  | https://doi.org/10.1038/s41598-021-00299-2\nwww.nature.com/scientificreports/\nFigure 1.  Flowchart of patient recruitment.\nTable 2.  Clinical characteristics of the 46 patients included in the validation phase. *Fisher’s exact test. **Chi-\nsquare test for trend.\nParameter Units Detail Controls Cases p-value\nTotal patient numbers n – 20 26 –\nMean age (mean ± SD) years – 30.5 ± 4.4 29.4 ± 3.7 0.369\nMean body mass index (mean ± SD) kg/m2 – 21.8 ± 1.7 22. 7 ± 3.7 0.754\nMenstrual phase n (%)\nFollicular 7 (35) 15 (58)\n0.149*\nLuteal 13 (65) 11 (42)\nOral contraceptives (last 3 months) n (%)\nNo 19 (95) 26 (100)\n0.4348*\nYe s 1 (5) 0 (0)\nHormonal therapy (last 3 months) n (%)\nNo 19 (95) 26 (100)\n0.4348*\nYe s 1 (5) 0 (0)\nMedications (last 1 week) n (%)\nNo 17 (85) 21 (80.8)\n > 0.999*\nYe s 3(15) 5 (19.2)\nSmoking status n (%)\nNon-smoker 14 (70) 19 (73.1)\n0.520**\nSmoker 4 (20) 3 (11.5)\nOccasional smoker 2 (10) 0 (0)\nFormer smoker 0 (0) 4 (15.4)\nType of endometriosis n (%)\nPeritoneal 0 (0) 5 (19.2)\nOvarian 0 (0) 8 (30.8)\nOvarian plus peritoneal 0 (0) 10 (38.4)\nDeep infiltrating 0 (0) 3 (11.4)\nRevised American Society for Reproductive Medicine score n (%)\nI 0 (0) 7 (26.9)\nII 0 (0) 0 (0)\nIII 0 (0) 17 (65.4)\nIV 0 (0) 2 (7.7)\n\n4\nVol:.(1234567890)Scientific Reports |        (2021) 11:20870  | https://doi.org/10.1038/s41598-021-00299-2\nwww.nature.com/scientificreports/\nwas aspirated from the Douglas space using a 2-mm needle. The samples of 3 mL peritoneal fluid were collected \nin 12-mL plastic tubes (Greiner, Monroe, North Carolina, USA). The tubes were stored at 4 °C. The samples \nwere centrifuged within an hour of collection, at 900×g for 5 min at 4 °C. The samples were then aliquoted and \nstored at −  80 °C. They were used for in-house validation using ELISA assays, or were transported in dry ice to \nSciomics GmbH for the proteomics analysis.\nAntibody microarray analysis. The mean bulk protein concentrations of the peritoneal fluid were not \nsignificantly different between cases and controls, as 37.92 ± 5.05 mg/mL and 37.53 ± 2.15 mg/mL, respectively \n(p > 0.05). Twelve samples were labelled for 1 h with scioDyes 1 and then 2 (Sciomics GmbH, Heidelberg, Ger -\nmany). Excess dye was removed, and the buffer was exchanged to phosphate-buffered saline (PBS) using size \nexclusion chromoatography. A reference sample was established by pooling identical volumes of all scioDye 2 \nlabelled samples. The 12 scioDye 1 labelled samples were then analyzed in a dual-color approach with a reference-\nbased design using scioDiscover antibody microarrays (Sciomics GmbH, Heidelberg, Germany) that targeted \n1360 different proteins with 1830  antibodies14. Each antibody was measured on the array by four technical rep-\nlicate spots. The arrays were blocked with scioBlock (Sciomics GmbH, Heidelberg, Germany) on a microarray \nhybridisation station (Hybstation 4800; Tecan GmbH, Grödig, Austria), and then incubated for 3 h with identi-\ncal volumes of sample and reference. The slides were then thoroughly washed (1× PBSTT; 0.1× PBS; water) and \ndried with nitrogen. Slide scanning was conducted using a microarray scanner (Powerscanner; Tecan, GmbH, \nGrödig, Austria) with identical instrument laser power and photomultiplier settings.\nStatistical analyses of protein microarray data. Spot segmentation was performed with the Gene-\nPix Pro 6.0 software (Molecular Devices, Union City, CA, USA). The raw data acquired were analyzed using \nthe linear models for microarray data (limma) package of R-Bioconductor after up-loading the median signal \nintensities. For normalization, a specialized invariant Lowess method was  applied15. For analysis of the samples, \na one-factorial linear model was fitted with limma which provided two-sided t-tests or F-tests based on moder-\nated statistics. All of the p values presented are adjusted for multiple testing by controlling the false discovery \nrate according to Benjamini and Hochberg. The proteins were defined as differential for ǀlog2FCǀ > 0.5 and an \nadjusted p < 0.05. Differences in the protein levels between the different samples or sample groups are presented \nas log-fold changes (logFC) calculated as base 2. In such studies, comparing samples (cases) versus  controls, \nlogFC = 1 means that the sample group had on average a  21 = twofold higher signal than the control; logFC =  − 1 \nmeans  2−1  = 1/2 of the signal in the sample than the control. Analyses for protein–protein interactions and gene \nontology (GO) were performed using String (http:// string- db. org)16. GO by String was used to classify differen-\ntially expressed proteins according to functional enrichment.\nEnzyme-linked immunosorbent assay validation and statistical analysis. The analysis of the sam-\nples was performed using commercially available enzyme-linked immunosorbent assay (ELISA) kits, according \nto the manufacturer instructions. The following ELISA kits were used: transforming growth factor-β-induced \nprotein ig-h3 (TGFBI; MyBioSource, San Diego, CA, USA; Catalogue No. #MBS177286; Lot No. #7481574715), \ncartilage oligomeric matrix protein/thrombospondin 5 (COMP; Merck Millipore, Saint Louis, MO, USA; Cata-\nlogue No. #1764; Lot No. #0102F2396), and angiotensinogen (AGT; Merck Millipore, Saint Louis, MO, USA; \nCatalogue No. #RAB1021; Lot No. #9217F2027).\nThe ROUT method was used for the outlier analysis. If outliers were identified, they were not included in the \nanalysis. The dataset without outliers was tested for normality using Shaphiro–Wilk tests. To compare groups, \nif the data were normal (p > 0.05), parametric unpaired t-tests were used; if the data were not normal (p < 0.05), \nnon-parametric Mann–Whitney tests were used. Statistical analysis was performed using GraphPad Prism 8. \nThe level of significance was set at p < 0.05.\nThe linear support vector machine (SVM) classification model using selected proteins as features was trained \nand tested using stratified fivefold cross validation. Hyperparameters were selected using fivefold cross validation \nseparately on the training sets of each split. The ROC curve and AUC calculation were based on the test sample \npredictions of each respective split.\nEthics approval and consent to participate. The study was conducted with the approval of the Medical \nEthics Committee of the Republic of Slovenia (No 0120-049/2016-4) and the research was performed in accord-\nance with relevant regulations. The informed consent was obtained from all of the participants before their inclu-\nsion in the study. The research was carried out according to The Code of Ethics of the World Medical Association \n(Declaration of Helsinki). Trial registration number: NCT04591548.\nResults\nCharacteristics of patients with endometriosis and control patients. The clinical characteristics \nof the patients with endometriosis (cases) and the control patients (controls) are presented in Table  1 for the \ndiscovery phase and in Table 2 for the validation phase. There were no significant differences between the case \nand control groups for either phase and for any of the characteristics examined here (Tables 1, 2, Supplementary \nTable S1).\nDiscovery phase. The mean age of the patients in the endometriosis group was 28.1 ± 3.1 years, and for the con-\ntrol group, 29.6 ± 2.8 years. All of these women had a BMI in the normal range (cases: 23.8 ± 1.7 kg/m2; controls: \n23.8 ± 3.1 kg/m2) and a regular menstrual cycle (21–35 days). All of the samples were collected in the follicular \n\n5\nVol.:(0123456789)Scientific Reports |        (2021) 11:20870  | https://doi.org/10.1038/s41598-021-00299-2\nwww.nature.com/scientificreports/\nphase of their menstrual cycle. All of the cases had endometriosis stage III, according to the Revised American \nSociety for Reproductive Medicine classification of  endometriosis17.\nValidation phase. The mean age of the patients in the validation endometriosis group was 29.4 ± 3.5 years, and \nfor the control group, 30.5 ± 4.3 years. All of these women had a BMI in the normal range (endometriosis group: \n22.7 ± 3.7 kg/m2; control group: 21.8 ± 1.7 kg/m2), and a regular menstrual cycle (21–35 days). In the endome-\ntriosis group, 58% of samples were collected in the follicular phase of the menstrual cycle, and in the control \ngroup, 35%. The remaining samples were collected in the luteal phase of the menstrual cycle. There were no sig-\nnificant differences between the validation case and control groups for any of the characteristics examined here \n(Table 2). According to the laparoscopy and histological verification 19% had peritoneal endometriosis, 31% \novarian endometriosis, 38% combined ovarian endometriosis with peritoneal lesions, and 12% deep infiltrat-\ning endometriosis. Seventy-three percent of the validation cases had at least stage III disease, according to the \nRevised American Society for Reproductive Medicine Classification of  Endometriosis17.\nDiscovery of 16 proteins with significantly different levels in peritoneal fluid allows separa -\ntion of patients with endometriosis from controls. The antibody microarray based analysis of 1360 \nproteins identified 16 proteins with significantly higher levels in the peritoneal fluid in cases (patients with endo-\nmetriosis) versus controls (Table 3). All 16 of these proteins showed > 1.5-fold differences in their levels in the \nperitoneal fluid for the cases versus controls. The six proteins; angiotensinogen (AGT); proinflammatory calcium \nbinding protein S1000A8/9 (S10A8/9); scavenger receptor cysteine-rich type 1 protein M130 (C163A); trans-\nforming growth factor-β-induced protein ig-h3 (TGFBI); epidermal growth factor receptor (EGFR) and tissue \ninhibitor of metalloproteinase 1 (TIMP1) showed the strongest differences with fold changes > 2 (log FC > 1). To \nthe best of our knowledge, AGT, TGFBI, cartilage oligomeric matrix protein/thrombospondin 5 (COMP) and \nangiopoietin-4 (ANGP4) have not previously been associated with endometriosis.\nThe results of the statistical analysis of these protein array data are presented as volcano plots in Fig.  2. For \nthese plots, the proteins with significantly higher levels in the peritoneal fluid of cases versus  controls are posi-\ntioned on the right side, above the red line that indicates the significance level of adjusted p value < 0.05.\nHierarchical clustering of the array data filtered for these 16 differential proteins nicely separated patients \nwith endometriosis from control patients (Fig. 3). Only the AD003 control sample clustered together with endo-\nmetriosis samples, although there were no apparent reasons for this.\nValidation confirms higher levels of COMP and TGFBI in peritoneal fluid from patients with \nendometriosis versus controls. Three proteins that had not been associated with endometriosis previ-\nously were selected for validation using ELISA. This validation confirmed that the levels of the COMP and \nTGFBI proteins in the peritoneal fluid of cases versus  controls were consistent with the microarray proteomic \ndiscovery study (Fig.  4). These COMP and TGFBI levels were significantly higher, at 1.7-fold and 1.3-fold for \ncases versus controls, respectively (p < 0.0005, for both). The levels of AGT were also 1.9-fold higher in the endo-\nmetriosis patients versus the control patients (p = 0.0199).\nFurther receiver operating characteristic (ROC) analysis revealed that COMP and TGFBI have very good \ndiagnostic characteristics, with areas under the curve of 0.78 and 0.84, respectively (Fig.  5). With the cut-off  \nTable 3.  Proteins with different levels in the peritoneal fluid from endometriosis patients versus control \npatients. The proteins given in bold were selected for the validation study. *p values adjusted for multiple \ntesting according to Bonferoni and Hochberg.\nProtein Protein abbreviation logFC adjustedp-value* Uniprot identifier\nAngiotensinogen AGT, ANGT 1.70 3.1e−03 P01019\nProinflammatory calcium binding protein S100A8/9 S10A8/9 1.66 2.8e−04 P05109\nScavenger receptor cysteine-rich type 1 protein M130 C163A 1.43 5.2e−07 Q86VB7\nTransforming growth factor-β-induced protein ig-h3 TGFBI, BGH3 1.23 3.8e−03 Q15582\nEpidermal growth factor receptor EGFR, HER1 1.13 1.1e−02 P00533\nTissue inhibitor of metalloproteinase 1 TIMP1 1.08 2.8e−04 P01033\nLumican LUM 0.95 1.3e−03 P51884\nCartilage oligomeric matrix protein/thrombospondin 5 COMP 0.84 6.3e−04 P49747\nα-2-Antiplasmin A2AP , SERPINF2 0.77 4.8e−02 P08697\nPhospholipid hydroperoxide glutathione peroxidase GPX4 0.72 3.2e−02 P36969\nInsulin-like growth factor-binding protein 4 IBP4, IGFBP4 0.68 4.8e−02 P22692\nHepatocyte growth factor activator HGFA 0.68 3.2e−02 Q04756\nMatrix metalloproteinase-2 MMP2 0.66 8.2e−03 P08253\nDickkopf-related protein 3 DKK3 0.60 2.7e−03 Q9UBP4\nCellular tumour antigen p53 P53 0.59 1.1e−02 P04637\nAngiopoietin-4 ANGP4 0.53 4.7e−02 Q9Y264\n\n6\nVol:.(1234567890)Scientific Reports |        (2021) 11:20870  | https://doi.org/10.1038/s41598-021-00299-2\nwww.nature.com/scientificreports/\npoint selected nearest to the top left-most corner of the ROC  curve18, as a cut-off of 1047 ng/mL, TGFBI showed \nsensitivity of 88.5% and specificity of 70.0%, and with a cut-off of 180 ng/mL, COMP sensitivity was 95.0% and \nspecificity was 54.3%. AGT showed only weak diagnostic potential, with an area under the curve of 0.67. An \nadditional classification model based on a linear support vector machine (SVM) using all three proteins was \ngenerated, yielding an AUC of > 0.83, as well as sensitivity of 0.81 and a specificity of 1.00 at the optimal clas-\nsification cutoff. An investigation of the sample distributions (Fig. 5B) reveals that, in terms of TGFBI values, at \nthe chosen cutoff there is low separation between control and endometriosis samples, meaning a classifier based \non TGFBI could potentially be sensitive to slight measurement variations. Comparatively, the margin between \nsample groups is more robust when applying the SVM classifier.\nProtein–protein interactions and gene ontology analysis identify direct and indirect interac -\ntions between these proteins that show differential levels in peritoneal fluid. GO analysis of \ndifferentially abundant proteins revealed that among molecular functions; signaling receptor binding, enzyme \nbinding, protein-containing complex binding, protease binding, collagen binding and endopeptidase inhibi-\ntor activity were most enriched (Supplementary Table S2). KEGG pathways related to the differential proteins \ninclude Proteoglycans in cancer, PI3K-Akt, HIF-1 and MAPK signaling pathways, Bladder cancer, Endocrine \nresistance, Human papilloma virus infection, Pathways in cancer and others (Supplementary Table S3). The pro-\ntein–protein interaction analysis using the STRING  database16 revealed several direct and indirect interactions \nbetween the biomarker candidates (Fig. 6). Most of the proteins identified (i.e., MMP2, TIMP1, COMP , EGFR, \nAGT, ANGP4, TGFBI, HGFA, SERPINF2, GPX4, IGFBP4) are located in the extracellular region and are mainly \ninvolved in extracellular matrix re-modulation, negative regulation of apoptosis, inflammatory responses and \nresponses to stress.\n−2 −1 012\nlog− fold change (logFC)\nadj. p value\n10 0\n10 −1\n10 −2\n10 −3\n10 −4\n10 −5\n10 −6\nHigher abundance in\nendometriosis samples\nHigher abundance in\ncontrol samples\nS100A8\nAHSG\nTIMP1\nS100A8\nCD163\nEGFR\nTIMP1\nLUM\nSERPINF2\nGPX4\nIGFBP4\nHGFAC\nMMP2\nDKK3\nTP53\nANGPT4\nAGTTGFBI\nCOMP\nFigure 2.  Volcano plot of the protein array data to visualise the adjusted p values and the corresponding log-\nfold changes (logFC). Horizontal red line, adjusted p = 0.05; vertical lines, logFC cut-offs (IlogFCI > 0.5). Proteins \nwith positive logFC had higher levels in the peritoneal fluid of the cases versus controls; and vice versa for \nproteins with negative logFC.\n\n7\nVol.:(0123456789)Scientific Reports |        (2021) 11:20870  | https://doi.org/10.1038/s41598-021-00299-2\nwww.nature.com/scientificreports/\nFigure 3.  Heatmap displaying the relative expression of proteins identified as differential. Values were centered \nand scaled by proteins (A). Array value differences between individual endometriosis samples and the average \nof control samples for the selected differential proteins (B). Rhombs indicate sample group means. Whiskers \nindicate one standard deviation.\n\n8\nVol:.(1234567890)Scientific Reports |        (2021) 11:20870  | https://doi.org/10.1038/s41598-021-00299-2\nwww.nature.com/scientificreports/\nDiscussion\nBiomarker discovery in the field of endometriosis has so far focused mainly on biomarkers from  blood4,6,9,19–21. \nThus, studies on peritoneal fluid biomarkers remain  rare4,7, and to the best of our knowledge, to date there have \nnot been any proteomic antibody microarray analyses of peritoneal fluid in population of women with endome-\ntriosis. For diagnostic procedures, peritoneal fluid samples could not replace blood samples since the collection \nis more invasive and thus risky. However, studies of peritoneal fluid might contribute to the identification of \nblood biomarkers for non-invasive diagnosis of endometriosis considering the surface of the peritoneal cav-\nity is large, and it allows passive dialysis of substances between the blood plasma and the peritoneal  fluid11–13. \nDorien et al. studied plasma samples from patients with endometriosis and a control group using an antibody \nmultiplex array approach, although they concluded that discovery and verification of potential markers is chal-\nlenging using this method, mainly due to issues of  reproducibility22. The only protein identified as a potential \nbiomarker was interleukin-3122.\nIdentification of potential biomarkers in peritoneal fluid, which represents the local environment of endome-\ntriotic lesions, might represent the first step towards identification of clinically important biomarkers for non-\ninvasive diagnostics from other body fluids (e.g., peripheral blood, urine, saliva). Peritoneal fluid undoubtedly \nhas a role in the etiopathogenesis of endometriotic lesions, and thus proteomic analysis might also provide new \nknowledge about the basic mechanisms of this disease.\nThe antibody microarray analysis that was used in the present study identified 16 proteins that showed dif-\nferences, whereby their levels were all increased in peritoneal fluid from patients with endometriosis (cases) \ncompared to the control patients. Six of these proteins showed the greatest increases in their levels, which \nwere twofold to fourfold higher for cases versus  controls: AGT, S10A8/9, C163A, TGFBI, EGFR and TIMP1. \nFigure 4.  Validation of transforming growth factor-β–induced protein ig-h3 (TGFBI) (A), cartilage oligomeric \nmatrix protein/thrombospondin 5 (COMP) (B) and angiotensinogen (AGT) (C)  levels in peritoneal fluid \nof endometriosis patients and control patients. 3D Scatterplot (D) shows distribution of TGFBI, cartilage \noligomeric matrix protein/thrombospondin 5 (COMP) and angiotensinogen (AGT) levels across samples with \nmeasurements for all three proteins.\n\n9\nVol.:(0123456789)Scientific Reports |        (2021) 11:20870  | https://doi.org/10.1038/s41598-021-00299-2\nwww.nature.com/scientificreports/\nAccording to our review of the literature here, COMP , AGT, TGFBI and ANGP4 have not yet been associated with \nendometriosis, while S100A8, C163A, EGFR and TIMP1  have23–34. However, EGFR has not yet been studied in \nperitoneal fluid, and for C163A, there is as yet no explanation of how it might be involved in the pathogenesis of \nthis disease, with no data available on its potential value as a  biomarker25,26. The other nine of these 16 proteins \nalso showed statistically significant increased levels in the peritoneal fluid of the cases, which were ≥ 1.5-fold \nhigher than for the controls.\nFor three candidate biomarkers, COMP , TGFBI and AGT, with increased levels in the peritoneal fluid no \nearlier reports in the context of endometriosis were found. Therefore, these proteins were selected for validation \nby ELISA. This validation confirmed the significant increases for COMP and TGFBI while the increase for AGT \ndid not reach statistical significance.\nCOMP is a glycoprotein that is mainly localized to the extracellular matrix of cartilage, synovium, ligaments \nand  tendons35. Increased levels of COMP have been associated with fibrogenesis in systemic sclerosis, skin \nkeloids, vascular atherosclerosis, lung fibrosis, rheumatoid arthritis, osteoarthritis, pseudoachondroplasia, acute \ntrauma and systemic lupus  erythematosus 36. In-vitro, COMP modulates pathological collagen-I deposition, \ndespite up-regulation of matrix  metalloproteinases36. We hypothesize that COMP induces collagen deposition \nand participates in extracellular matrix remodeling, and might thus contribute to the pathophysiology of intra-\nperitoneal adhesions in endometriosis. Significantly higher levels of COMP in patients with endometriosis indi-\ncate that COMP has a role in pathogenesis of endometriosis and can serve as a potential biomarker or drug target.\nTGFBI has been associated with a range of diseases, which include nephropathy, atherosclerosis, rheumatoid \narthritis, corneal disorders and malignant  diseases37. In malignant diseases, TGFBI appears to have either tumor-\nsuppressing or tumor-promoting roles, with reports that suggest that TGFBI can mediate cancer cell invasion and \nmetastasis, and can enhance cancer cell  extravasation38–41. In ovarian cancer, TGFBI can enhance cell adhesion, \nmotility and  invasion38. The loss of TGFBI in cancer cells has a pro-tumorigenic role, while its overexpression in \nperitoneal cells aids the metastatic  process38. As suggested for ovarian cancer, TGFBI might have a similar role \nin the development of endometriotic peritoneal implants.\nThese significantly higher levels of TGFBI in the peritoneal fluid from women with endometriosis in the \npresent study suggest a role for TGFBI in the pathogenesis of endometriosis. This is the first report that has \nassociated TGFBI with endometriosis, and we show the great potential for TGFBI as a diagnostic tool, and even \nFigure 5.  ROC curves assessing the diagnostic profiles of ig-h3 (TGFBI), thrombospondin 5 (COMP) and \nangiotensinogen (AGT) and a linear SVM model using all three features. Dots represent the decision thresholds \nyielding the most promising classifiers for the TGFBI and SVM predictors (A). One-dimensional sample \ndistributions. For the single feature distributions, values were [0,1]-transformed. For the SVM distribution, \ncalculated class probabilities for the class \"case\" were plotted (B). Blue lines mark samples for each group which \nare closest to the respective optimal decision thresholds of (A).\n\n10\nVol:.(1234567890)Scientific Reports |        (2021) 11:20870  | https://doi.org/10.1038/s41598-021-00299-2\nwww.nature.com/scientificreports/\nmore importantly, as a predictive biomarker for endometriosis. These data call for further validation studies, \nespecially in patients with peritoneal endometriosis who are in great need for novel diagnostic options.\nAGT is a component of the renin–angiotensin system. AGT is released from the liver and is cleaved in \nthe peripheral blood by renin, to form angiotensin I. Angiotensin I is then converted into angiotensin II by \nangiotensin-converting enzyme. Angiotensin II is the most active component of the renin–angiotensin system, \nand it acts through interactions with two major receptors: the angiotensin II type 1 and type 2 receptors. One of \nthe effects of activation of these receptors is stimulation of the synthesis of vascular endothelial growth factors, \nwhich can directly induce the formation of new blood  vessels42. To date, only one study has reported any associa-\ntion of AGT with endometriosis, whereby Kowalczyńska et al. suggested a role for AGT M235T polymorphism \nin endometriosis. However, they did not provide any evidence for association of AGT with the development or \nclinical course of endometriosis, and did not indicate any prognostic value for AGT 43. Although much remains to \nbe learned, ANGT might have a role in the development and survival of endometriotic lesions in the peritoneal \ncavity, due to its biological function in angiogenesis.\nThe other proteins that were identified as showing specific changes in their levels in peritoneal fluid in \nendometriosis in our antibody array discovery analysis were MMP2, TIMP1, EGFR, ANGP4, C163A, HGFA, \nS10A8/9, LUM, A2AP/SERPINF2, GPX4, IBP4/IGFBP4, DKK3 and P53. These have not (yet) been validated as \nhaving important roles in endometriosis.\nHowever, the expression of matrix metalloproteinases (MMPs) and the tissue inhibitors of metalloproteinases \n(TIMPs) have been shown to be involved in the pathogenesis and pathophysiology of  endometriosis44–49. These \nproteins are at the forefront of extracellular matrix remodeling, and so changes in their levels might contribute \nto fibrosis and adhesions. A balance of these proteins is required for normal follicular development, ovulation, \nembryo implantation and further  embryogenesis50,51. TIMP1 is secreted by endometriotic lesions, and it regulates \ncell differentiation, migration and death. Indeed, it might be part of the mechanism that causes endometriosis-\nassociated  infertility52. Significantly higher levels of TIMP-1 were found in peritoneal fluid of patients with \nendometriosis as compared to control  women28. As shown in animal models, excessive TIMP1 was deleterious to \novulation and embryo  development52. In the present study, the patients with endometriosis showed higher levels \nFigure 6.  Analysis of protein–protein interactions from the STRING database for association networks. These \nrevealed several direct (thick lines) and indirect (thin lines) interactions of the biomarker candidates. See Table 3 \nfor protein name abbreviations.\n\n11\nVol.:(0123456789)Scientific Reports |        (2021) 11:20870  | https://doi.org/10.1038/s41598-021-00299-2\nwww.nature.com/scientificreports/\nof TIMP1 and MMP2 in the peritoneal fluid, as compared to the controls. Protein–protein interaction analyses of \nthe identified proteins have also confirmed the central position of MMP2 and TIMP1 in protein–protein interac-\ntion networks. Furthermore, TIMP1 has already been described as a protein that is secreted by endometriotic \n lesions53, and the present study supports its great value as a potential biomarker.\nEGFR has important roles in signal-transduction, and it mediates a variety of cellular processes, including \ncell survival, proliferation, migration and angiogenesis, and inhibition of  apoptosis30,33. It has a role in the cycli-\ncal growth of the endometrium, and might have a key role in the pathogenesis of endometriosis. Ejskjaer et al. \nhave reported on different cyclical mRNA levels of EGF receptors and their ligands in eutopic endometrium \nfrom patients with endometriosis compared to endometrium from healthy  individuals34, and EGF receptors in \nendometrium were significantly up-regulated in endometrial cancer and endometrial hyperplasia, compared \nto healthy menopausal  endometrium54,55. EGFR is important also for female reproductive functions. Ding et al. \nreported higher levels of embryonic EGF and their receptors when mouse oocytes and embryos were cultured in \nmedia with peritoneal fluid obtained from women with mild endometriosis. Here, the fertilization of oocytes and \nthe development potential of embryos were  decreased56. In endometriosis most studies have compared EGFR/\nHER1 expression in eutopic and ectopic endometrium, and have reported contradictory data: no significant \n differences29–31, and lower  levels32 or higher EGFR protein and mRNA levels in eutopic and ectopic endometrium \nversus healthy  endometrium33,34. In serum samples Matalliotakis et al. found no significant differences in the \nlevels of soluble EGFR in patients with endometriosis versus  those without  endometriosis57. To the best of our \nknowledge, EGFR has not been studied in peritoneal fluid of women with endometriosis, where the higher levels \nin the present study suggest that it is association with subfertility.\nOur antibody microarray proteomic analyses of the peritoneal fluid of patients with endometriosis also iden-\ntified higher levels of angiopoetin 4 (ANGP4), a protein that is involved in angiogenesis. We have so far been \nunable to find any report of the detection of angiopoietin in peritoneal fluid, and also no reports about detection \nof angiopoietin in other body fluids in endometriosis patients. Angiopoietins have crucial roles through their \npromotion of pericyte recruitment and vascular  branching58, and they might be involved in the pathogenesis of \nendometriosis, and thus might also represent potential biomarkers.\nC163A has previously been investigated in serum and peritoneal fluid of women with endometriosis, but \nthe data have been  inconclusive25–27. C163A is a membrane receptor that is only expressed by monocytes and \nmacrophages. The importance of peritoneal macrophages in the development of endometriosis is well  known59–61. \nC163 has been reported to be related to the binding of hemoglobin:haptoglobin  complexes62. C163A is regulated \nby other cytokines, where IL-6 and IL-10 have been shown to induce the expression of both its messenger RNA \nand the C163A surface receptor  protein63. The extracellular (soluble) portion of this C163A (known as sC163) is \nshed from the cell surface when macrophages are stimulated by inflammatory  cytokines64. The biological function \nof sC163 remains unknown to date, although it has been suggested to be a marker for monocyte/ macrophage \nactivity in diverse inflammatory  diseases64,65. There has only been one study so far that has examined sC163 in \nperitoneal fluid of women with endometriosis, and it reported no significant differences in comparison with \nthe case group of patients who underwent laparoscopy due to infertility or elective tubal  sterilization27. As we \nfound higher levels of C163A in the peritoneal fluid of these women with endometriosis versus the controls, this \nindicates that C163 has some diagnostic potential.\nHepatocyte growth factor (HGF) is produced by endometrial stromal cells, and it promotes cell proliferation \nand migration, and lumen formation of endometrial epithelial  cells66,67. Y oshida et al. showed significantly higher \nlevels of HGF in peritoneal fluid of patients with endometriosis compared with patients without  endometriosis67. \nHere, we found higher levels of one of its activators, hepatocyte growth factor activator (HGFA), which is con-\nsistent with the published literature on a role for the HGF system in the etiopathogenesis of  endometriosis67.\nEndometriosis is considered to be a chronic inflammatory disease, and previous studies have shown that \ninflammatory processes are involved in its pathogenesis and are associated with its characteristic symptoms. The \npro-inflammatory calcium binding protein S100A8 has already been studied in patients with endometriosis, in \nsamples of peritoneal fluid and cervical  mucus23,24. S100A8 predominantly acts as a heterodimer with S100A9, \nand is thus named calprotectin. Overexpression at sites of inflammation has been well established for S100A8, and \nalso elevated serum levels have been reported for a variety of inflammatory  diseases68–70. S100A8 is released by \nphagocytes and is a potent chemoattractant for neutrophils and monocytes both in vitro and in vivo71. Our report \nhere of higher levels of S100A8 in peritoneal fluid of women with endometriosis thus supports the published \n data23. A previous study has also reported higher protein levels for S100A8 in peritoneal fluid from patients with \ndeep endometriosis, as compared to patients with superficial  lesions23. There is no literature on S100A8 in the \nperipheral blood of patients with endometriosis. As S100A8 is known to be involved in inflammatory processes, \nit is probably not specific to endometriosis, and subsequently, its diagnostic value will be limited.\nConclusion\nTo the best of our knowledge, the present study is the first that has used antibody arrays for the identification of \ndifferential levels of proteins in peritoneal fluid from patients with endometriosis. We defined 16 proteins with \nsignificantly increased levels in this peritoneal fluid, which are mainly related to fibrinogenesis, extracellular \nremodeling, pathogenesis of inflammation, induction of a dysfunctional immune system, and angiogenesis. This \nstudy also reports the first time that the proteins COMP , AGT, TGFBI and ANGP4 have been associated with \nendometriosis. For COMP and TGFBI, validation by ELISA confirmed the proteomic array data obtained here. \nOur findings have brought new knowledge that will contribute to better understanding of the pathophysiology of \nendometriosis. COMP and TGFBI thus represent potential biomarkers, which therefore warrant further studies \nalso in blood samples, which are currently in progress.\n\n12\nVol:.(1234567890)Scientific Reports |        (2021) 11:20870  | https://doi.org/10.1038/s41598-021-00299-2\nwww.nature.com/scientificreports/\nData availability\nThe datasets used and/or analyzed during the current study are available from the corresponding author on \nreasonable request.\nReceived: 22 May 2021; Accepted: 30 September 2021\nReferences\n 1. Rogers, P . A. et al. Defining future directions for endometriosis research: Workshop report from the 2011 World Congress of \nEndometriosis in Montpellier, France. Reprod. Sci. 20(5), 483–499 (2013).\n 2. Sampson, J. A. Peritoneal endometriosis due to the menstrual dissemination of endometrial tissue into the peritoneal cavity. Am. \nJ. Obstet. Gynecol. 14(4), 422–469 (1927).\n 3. Dunselman, G. A. et al. ESHRE guideline: Management of women with endometriosis. Hum. Reprod. 29(3), 400–412 (2014).\n 4. Rižner, T. L. Diagnostic potential of peritoneal fluid biomarkers of endometriosis. Expert Rev. Mol. Diagn. 15(4), 557–580 (2015).\n 5. Burney, R. O. & Giudice, L. C. Pathogenesis and pathophysiology of endometriosis. Fertil. Steril. 98(3), 511–519 (2012).\n 6. Coutinho, L. M., Ferreira, M. C., Rocha, A. L. L., Carneiro, M. M. & Reis, F . M. New biomarkers in endometriosis. Adv. Clin. Chem. \n89, 59–77 (2019).\n 7. Agic, A. et al. Is endometriosis associated with systemic subclinical inflammation? Gynecol. Obstet. Investig. 62(3), 139–147 (2006).\n 8. Rogers, P . A. et al. Research priorities for endometriosis. Reprod. Sci. 24(2), 202–226 (2017).\n 9. Rižner, T. L. Noninvasive biomarkers of endometriosis: Myth or reality? Expert Rev. Mol. Diagn. 14(3), 365–385 (2014).\n 10. Vid, J., Josko, O., Eda, V . B., Lanišnik, T. R. & Helena, B. F . Biomarkers of endometriosis: How far have we come and where are we \ngoing? Slov. Med. J. 90, 1–10 (2021).\n 11. Koninckx, P . R., Kennedy, S. H. & Barlow, D. H. Endometriotic disease: The role of peritoneal fluid. Hum. Reprod. Update 4(5), \n741–751 (1998).\n 12. Y oung, V . J., Brown, J. K., Saunders, P . T. & Horne, A. W . The role of the peritoneum in the pathogenesis of endometriosis. Hum. \nReprod. Update 19(5), 558–569 (2013).\n 13. Bedaiwy, M. A. & Falcone, T. Peritoneal fluid environment in endometriosis. Clinicopathological implications. Minerva Ginecol. \n55(4), 333–345 (2003).\n 14. Schröder, C. et al. Dual-color proteomic profiling of complex samples with a microarray of 810 cancer-related antibodies. Mol. \nCell Proteomics 9(6), 1271–1280 (2010).\n 15. Sill, M., Schröder, C., Hoheisel, J. D., Benner, A. & Zucknick, M. Assessment and optimisation of normalisation methods for dual-\ncolour antibody microarrays. BMC Bioinform. 11, 556 (2010).\n 16. Szklarczyk, D. et al. STRING v11: Protein-protein association networks with increased coverage, supporting functional discovery \nin genome-wide experimental datasets. Nucleic Acids Res. 47(D1), D607–D613 (2019).\n 17. American Society for Reproductive. Revised American Society for Reproductive Medicine classification of endometriosis: 1996. \nFertil. Steril. 67(5), 817–821 (1997).\n 18. Hoo, Z. H., Candlish, J. & Teare, D. What is an ROC curve? Emerg. Med. J. 34, 357–359 (2017).\n 19. Nisenblat, V . et al. Blood biomarkers for the non-invasive diagnosis of endometriosis. Cochrane Database Syst. Rev. 2016(5), \nCD012179 (2016).\n 20. Anastasiu, C. V . et al. Biomarkers for the noninvasive diagnosis of endometriosis: State of the art and future perspectives. Int. J. \nMol. Sci. 21(5), 1750 (2020).\n 21. Knific, T. et al. Multiplex analysis of 40 cytokines do not allow separation between endometriosis patients and controls. Sci. Rep.  \n9(1), 16738 (2019).\n 22. Waelkens, E., Vanhie, A., Peterse, D., Fassbender, A. & D’Hooghe, T. The use of antibody arrays in the discovery of new plasma \nbiomarkers for endometriosis. Reprod. Sci. 27(2), 751–762 (2020).\n 23. Ferrero, S. et al. Proteomic analysis of peritoneal fluid in women with endometriosis. J. Proteome Res. 6(9), 3402–3411 (2007).\n 24. Grande, G. et al. Cervical mucus proteome in endometriosis. Clin. Proteomics 14, 7 (2017).\n 25. Itoh, F . et al. Possible involvement of signal transducer and activator of transcription-3 in cell-cell interactions of peritoneal mac-\nrophages and endometrial stromal cells in human endometriosis. Fertil. Steril. 99(6), 1705–1713 (2013).\n 26. Jee, B. C., Suh, C. S., Kim, S. H. & Moon, S. Y . Serum soluble CD163 and interleukin-6 levels in women with ovarian endometrio-\nmas. Gynecol. Obstet. Investig. 66(1), 47–52 (2008).\n 27. Bersinger, N. A. et al. PAPP-A and osteoprotegerin, together with interleukin-8 and RANTES, are elevated in the peritoneal fluid \nof women with endometriosis. Am. J. Obstet. Gynecol. 195(1), 103–108 (2006).\n 28. Marí-Alexandre, J. et al. Micro-RNA profile and proteins in peritoneal fluid from women with endometriosis: Their relationship \nwith sterility. Fertil. Steril. 109(4), 675–684 (2018).\n 29. Prentice, A. et al. Epidermal growth factor receptor expression in normal endometrium and endometriosis: An immunohisto-\nchemical study. Br. J. Obstet. Gynaecol. 99(5), 395–398 (1992).\n 30. Scotti, S., Regidor, P . A., Schindler, A. E. & Winterhager, E. Reduced proliferation and cell adhesion in endometriosis. Mol. Hum. \nReprod. 6(7), 610–617 (2000).\n 31. Uzan, C. et al. Status of HER1 and HER2 in peritoneal, ovarian and colorectal endometriosis and ovarian endometrioid adeno -\ncarcinoma. Virchows Arch. 454(5), 525–529 (2009).\n 32. Huang, J. C. & Y eh, J. Quantitative analysis of epidermal growth factor receptor gene expression in endometriosis. J. Clin. Endo-\ncrinol. Metab. 79(4), 1097–1101 (1994).\n 33. Bergqvist, A., Borg, A. & Ljungberg, O. Protooncogenes in endometriotic and endometrial tissue. Ann. N. Y. Acad. Sci. 626, 276–283 \n(1991).\n 34. Ejskjaer, K. et al. Expression of the epidermal growth factor system in eutopic endometrium from women with endometriosis \ndiffers from that in endometrium from healthy women. Gynecol. Obstet. Investig. 67(2), 118–126 (2009).\n 35. Hedbom, E. et al. Cartilage matrix proteins. An acidic oligomeric protein (COMP) detected only in cartilage. J. Biol. Chem. 267(9), \n6132–6136 (1992).\n 36. Magdaleno, F . et al. Cartilage oligomeric matrix protein participates in the pathogenesis of liver fibrosis. J. Hepatol. 65(5), 963–971 \n(2016).\n 37. Ween, M. P ., Oehler, M. K. & Ricciardelli, C. Transforming growth factor-beta-induced protein (TGFBI)/(βig-H3): A matrix protein \nwith dual functions in ovarian cancer. Int. J. Mol. Sci. 13(8), 10461–10477 (2012).\n 38. Ween, M. P . et al. Transforming growth factor-beta-induced protein secreted by peritoneal cells increases the metastatic potential \nof ovarian cancer cells. Int. J. Cancer 128(7), 1570–1584 (2011).\n 39. Tang, J. et al. BetaIg-h3 is involved in the HAb18G/CD147-mediated metastasis process in human hepatoma cells. Exp. Biol. Med. \n(Maywood) 232(3), 344–352 (2007).\n 40. Tang, J., Wu, Y . M., Zhao, P ., Jiang, J. L. & Chen, Z. N. Betaig-h3 interacts with alpha3beta1 integrin to promote adhesion and \nmigration of human hepatoma cells. Exp. Biol. Med. (Maywood) 234(1), 35–39 (2009).\n\n13\nVol.:(0123456789)Scientific Reports |        (2021) 11:20870  | https://doi.org/10.1038/s41598-021-00299-2\nwww.nature.com/scientificreports/\n 41. Ma, C. et al. Extracellular matrix protein betaig-h3/TGFBI promotes metastasis of colon cancer by enhancing cell extravasation. \nGenes Dev. 22(3), 308–321 (2008).\n 42. Khakoo, A. Y ., Sidman, R. L., Pasqualini, R. & Arap, W . Does the renin-angiotensin system participate in regulation of human \nvasculogenesis and angiogenesis? Cancer Res. 68(22), 9112–9115 (2008).\n 43. Kowalczyńska, L. J. et al. Is M235T polymorphism of the angiotensinogen gene involved in the development of endometriosis? \nGinekol. Pol. 88(11), 585–590 (2017).\n 44. Stilley, J. A., Woods-Marshall, R., Sutovsky, M., Sutovsky, P . & Sharpe-Timms, K. L. Reduced fecundity in female rats with surgically \ninduced endometriosis and in their daughters: A potential role for tissue inhibitors of metalloproteinase 1. Biol. Reprod.  80(4), \n649–656 (2009).\n 45. Chung, H. W . et al. Matrix metalloproteinase-9 and tissue inhibitor of metalloproteinase-3 mRNA expression in ectopic and eutopic \nendometrium in women with endometriosis: A rationale for endometriotic invasiveness. Fertil. Steril. 75(1), 152–159 (2001).\n 46. Cox, K. E., Piva, M. & Sharpe-Timms, K. L. Differential regulation of matrix metalloproteinase-3 gene expression in endometriotic \nlesions compared with endometrium. Biol. Reprod. 65(4), 1297–1303 (2001).\n 47. Zhou, H. E. & Nothnick, W . B. The relevancy of the matrix metalloproteinase system to the pathophysiology of endometriosis. \nFront. Biosci. 10, 569–575 (2005).\n 48. Sharpe-Timms, K. L. et al. Partial purification and amino acid sequence analysis of endometriosis protein-II (ENDO-II) reveals \nhomology with tissue inhibitor of metalloproteinases-1 (TIMP-1). J. Clin. Endocrinol. Metab. 80(12), 3784–3787 (1995).\n 49. Osteen, K. G., Y eaman, G. R. & Bruner-Tran, K. L. Matrix metalloproteinases and endometriosis. Semin. Reprod. Med. 21(2), \n155–164 (2003).\n 50. Nothnick, W . B., Soloway, P . & Curry, T. E. Jr. Assessment of the role of tissue inhibitor of metalloproteinase-1 (TIMP-1) during \nthe periovulatory period in female mice lacking a functional TIMP-1 gene. Biol. Reprod.  56(5), 1181–1188 (1997).\n 51. Curry, T. E. Jr. & Osteen, K. G. Cyclic changes in the matrix metalloproteinase system in the ovary and uterus. Biol. Reprod. 64(5), \n1285–1296 (2001).\n 52. Stilley, J. A. et al. Neutralizing TIMP1 restores fecundity in a rat model of endometriosis and treating control rats with TIMP1 \ncauses anomalies in ovarian function and embryo development. Biol. Reprod. 83(2), 185–194 (2010).\n 53. Luddi, A. et al. Expression of matrix metalloproteinases and their inhibitors in endometrium: High levels in endometriotic lesions. \nInt. J. Mol Sci. 21(8), 2840 (2020).\n 54. Niikura, H., Sasano, H., Kaga, K., Sato, S. & Y ajima, A. Expression of epidermal growth factor family proteins and epidermal growth \nfactor receptor in human endometrium. Hum. Pathol. 27(3), 282–289 (1996).\n 55. Ejskjaer, K. et al. Expression of the epidermal growth factor system in endometrioid endometrial cancer. Gynecol. Oncol. 104(1), \n158–167 (2007).\n 56. Ding, G. L. et al. Attenuated oocyte fertilization and embryo development associated with altered growth factor/signal transduc-\ntion induced by endometriotic peritoneal fluid. Fertil. Steril. 93(8), 2538–2544 (2010).\n 57. Matalliotakis, I. M. et al.  Expression of serum human leukocyte antigen and growth factor levels in a Greek family with familial \nendometriosis. J. Soc. Gynecol. Investig. 10(2), 118–121 (2003).\n 58. Krikun, G. Endometriosis, angiogenesis and tissue factor. Scientifica (Cairo) 2012, 306830 (2012).\n 59. Santanam, N., Murphy, A. A. & Parthasarathy, S. Macrophages, oxidation, and endometriosis. Ann. N. Y. Acad. Sci. 955, 183–198 \n(2002).\n 60. Sturlese, E. et al. Dysregulation of the Fas/FasL system in mononuclear cells recovered from peritoneal fluid of women with \nendometriosis. J. Reprod. Immunol. 92(1–2), 74–81 (2011).\n 61. Laganà, A. S. et al. Evaluation of M1 and M2 macrophages in ovarian endometriomas from women affected by endometriosis at \ndifferent stages of the disease. Gynecol. Endocrinol. 36(5), 441–444 (2020).\n 62. Fabriek, B. O., Dijkstra, C. D. & van den Berg, T. K. The macrophage scavenger receptor CD163. Immunobiology 210(2–4), 153–160 \n(2005).\n 63. Graversen, J. H., Madsen, M. & Moestrup, S. K. CD163: A signal receptor scavenging haptoglobin-hemoglobin complexes from \nplasma. Int. J. Biochem. Cell Biol. 34(4), 309–314 (2002).\n 64. Møller, H. J., Peterslund, N. A., Graversen, J. H. & Moestrup, S. K. Identification of the hemoglobin scavenger receptor/CD163 as \na natural soluble protein in plasma. Blood 99(1), 378–380 (2002).\n 65. Moestrup, S. K. & Møller, H. J. CD163: A regulated hemoglobin scavenger receptor with a role in the anti-inflammatory response. \nAnn. Med. 36(5), 347–354 (2004).\n 66. Zhang, X. Hepatocyte growth factor system in the mouse uterus: Variation across the estrous cycle and regulation by 17-beta-\nestradiol and progesterone. Biol. Reprod. 82(6), 1037–1048 (2010).\n 67. Y oshida, S. et al. Hepatocyte growth factor/Met system promotes endometrial and endometriotic stromal cell invasion via autocrine \nand paracrine pathways. J. Clin. Endocrinol. Metab. 89(2), 823–832 (2004).\n 68. Odink, K. et al. Two calcium-binding proteins in infiltrate macrophages of rheumatoid arthritis. Nature 330(6143), 80–82 (1987).\n 69. Zwadlo, G., Brüggen, J., Gerhards, G., Schlegel, R. & Sorg, C. Two calcium-binding proteins associated with specific stages of \nmyeloid cell differentiation are expressed by subsets of macrophages in inflammatory tissues. Clin. Exp. Immunol. 72(3), 510–515 \n(1988).\n 70. Foell, D., Frosch, M., Sorg, C. & Roth, J. Phagocyte-specific calcium-binding S100 proteins as clinical laboratory markers of inflam-\nmation. Clin. Chim. Acta 344(1–2), 37–51 (2004).\n 71. Devery, J. M., King, N. J. & Geczy, C. L. Acute inflammatory activity of the S100 protein CP-10 Activation of neutrophils in vivo \nand in vitro. J. Immunol. 152(4), 1888–1897 (1994).\nAcknowledgements\nThe authors thank their study participants, who kindly donated their samples and time. The authors thank the \npersonnel of the Department of Obstetrics and Gynecology, University Medical Centre Ljubljana, Ljubljana, \nSlovenia, especially Mrs. Tanja Lončar. The authors also thank Mrs. Vera Troha Poljančič and Prof. Dr. Joško \nOsredkar at the University Medical Centre Ljubljana, Clinical Institute of Clinical Chemistry and Biochemistry, \nDr. Tamara Knific at the Institute of Biochemistry, Faculty of Medicine, University of Ljubljana, for processing \nthe samples and Dr. Chris Berrie for critical reading of the manuscript.\nAuthor contributions\nV .J., T.L.R. and H.B.F .: conception and design of study, acquisition of data, analysis of data and approval of the \nfinal version. M.K.: data analysis. V .J.: drafting the manuscript. T.K.: carried out ELISA experiments, T.K. and \nM. P . analyzed the data. T.L.R., H.B.F . and E.B.V .: revising the manuscript critically for intellectual content. V .J., \nH.B.F . and T.L.R. initiated the project and were responsible for the study. All authors have approved the final \nversion of the manuscript.\n\n14\nVol:.(1234567890)Scientific Reports |        (2021) 11:20870  | https://doi.org/10.1038/s41598-021-00299-2\nwww.nature.com/scientificreports/\nFunding\nThe preparation of this manuscript was supported by a Grant J3-1755 from the Slovenian Research Agency to \nT.L.R.\nCompeting interests \nThe authors declare no competing interests.\nAdditional information\nSupplementary Information The online version contains supplementary material available at https:// doi. org/ \n10. 1038/ s41598- 021- 00299-2.\nCorrespondence and requests for materials should be addressed to H.B.F . or T.L.R.\nReprints and permissions information is available at www.nature.com/reprints.\nPublisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and \ninstitutional affiliations.\nOpen Access  This article is licensed under a Creative Commons Attribution 4.0 International \nLicense, which permits use, sharing, adaptation, distribution and reproduction in any medium or \nformat, as long as you give appropriate credit to the original author(s) and the source, provide a link to the \nCreative Commons licence, and indicate if changes were made. The images or other third party material in this \narticle are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the \nmaterial. If material is not included in the article’s Creative Commons licence and your intended use is not \npermitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from \nthe copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.\n© The Author(s) 2021","source_license":"CC0","license_restricted":false}