{"paper_id":"080efaf6-dbb8-46ca-9560-0900d29494b6","body_text":"Title \nPreconceptional immunomodulation partially corrects pregnancy abnormalities induced by \nendometriosis in a m ouse model, with normalization of transcriptional alterations observed in the \ndeveloping fetal-maternal interface at the single cell level.  \nAuthors: \nKheira BOUZID 1, Roxane BARTKOWSKI 1, Alix SILVERT 3, Fabiana MORESI 1, Camille SOUCHET 1, Marine \nTHOMAS1, Isabelle LAGOUTTE 1, Vaarany KARUNANITHY1, Brigitte IZAC 1, Charles CHAPRON 2, Pietro \nSANTULLI2, Frédéric BATTEUX1, Céline MEHATS1, Louis MARCELLIN2, Ludivine DORIDOT1 \n1. Université Paris Cité, CNRS, Inserm, Institut Cochin, F-75014 Paris, France \n2. Department of Gynecology Obstetrics II and Reproductive Medicine, AP-HP, Hôpital \nUniversitaire Paris Centre, Centre Hospitalier Universitaire Cochin, F-75014 Paris, France \n3. Université Paris Cité, iPOP-UP \nAbstract \nEndometriosis is a gynecological disease that affects approximately 10% of women worldwide and has \na significant impact on patients' lives. Women with endometriosis attempting to conceive may \nexperience infertility and have a higher risk of obstetric complications. They are also more likely to \nmiscarry and have placental defects during pregnancy.  \nThe impact of endometriosis on pregnancy remains unclear. Defects in implantation and placentation \nmay be present, but these processes are difficult to study in human subjects. To address this issue, we \ncreated a mouse model of endometriosis and examined pregnancy using the CBAxDBA crossbreed, \nwhich is commonly employed to study pregnancy immunology . We examined the early and late \ngestational stages using this model.  \nIn our model, we confirmed that the presence of endometriosis -like lesions resulted in fewer \nimplantations and a higher proportion of fetal resorption, which is analogous to miscarriage. We also \nused an endometriosis model in which the immune system was preconceptionally trained to induce \nimmune tolerance. We observed that endometriotic lesions were smaller in mice with \nimmunomodulation, and gestational complications were corrected. At early gestation (E9.5), fetal-\nmaternal interfaces  were retrieved to asse ss the potential impact of endometriosis an d \nimmunomodulation of placental development using single-cell RNA-sequencing. Our data show that \nmost changes induced by endometriosis occur in decidual stromal cells, which originate from the \nendometrium. In these cells, we observed a consistent upregulation of Gata4, a transcription factor \npreviously found to be elevated in the endometrium of women with endometriosis. We also observed \ndownregulation of Prap1, which is an indicator of uterine receptivity and successful implantation i n \nmice. The transcriptomi c changes induced by endometriosis were  partially reversed when immune \ntolerance was induced in mice. In immune cells, endometriosis leads to an inflammation-associated \nsignature. Additionally, the interferon gamma response is reduced in NK cells, which has been shown \nto be crucial for spiral artery remodelling and decidual integrity, which may be involved in the observed \nincreased resorption rate.  \nOverall, our findings provide new insights into endometriosis-associated infertility and pregnancy \ncomplications in a mouse model that can be partially corrected by immune training. These results \nsuggest that targeting the immune system should be considered in future studies to improve \npregnancy outcomes in patients with endometriosis.  \n.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 2, 2026. ; https://doi.org/10.64898/2026.01.02.695166doi: bioRxiv preprint \n\nIntroduction \nEndometriosis is a complex gynecological disease that affects approximately 10% of women of \nreproductive age. It is characterized by the presence of endometrium-like tissue in ectopic regions1.  It \nis associated with pain, infertility, and a higher risk of pregnancy complications, significantly impacting \npatients’ lives. \nSeveral studies have assessed the association between endometriosis and  infertility and adverse \npregnancy outcomes 2,3.  In vitro fertilization studies suggest that women with severe endometriosis \nhave complications related to oocyte and embryo quality and difficult implantation 4. In addition , \nwomen with endometriosis have an increased risk of miscarriage (odds ratio of 1.81 in spontaneous \nconception5) and other pregnancy complications, such as preeclampsia and placenta previa. At the end \nof pregnancy, a lower birth weight  is found in babies born to endometriotic mothers 6–8. These \ncomplications suggest that o ogenesis, implantation, and placentation processes  may be defective in \nendometriosis; however, the underlying mechanisms remain to be elucidated.   \nSeveral teams have developed mouse models of endometriosis, which are obtained either by grafting \npieces of uterine horns on different tissues within the peritoneal cavity of the mouse by laparotomy9–\n11, or by cutting uterine horns/endometrium  into small pieces and injecting them into the peritoneal \ncavity through an opening on the abdominal wall12–14. These studies have shown fertility and gestation \ncomplications in these models, with delayed coitus 14, a lower pregnancy rate9,10,12, a reduced number \nof embryos12, a higher resorption rate 11–13, and a lower pup weight 13. These models can be used to \nstudy the underlying mechanisms of pregnancy outcomes in patients with endometriosis. In our team, \nwe use a mouse model of endometriosis that we obtain by grafting uterine horn fragments on the \nperitoneum, which allows us to follow the lesion size using ultrasound. We have already shown in this \nmodel that peritoneal macrophages present inflammation induced by the presence of endometriosis \nlesions15. We can use this model to study the key players in the development of gestation, involving \ndecidual and immune cells in the maternal part and trophoblast cells in the fetal part.  \nAt the maternal -fetal interface  early during gestation , most immune cells are innate immune cells \n(decidual Natural Killer cells – dNK and macrophages). They participate in fetal immune tolerance and \nimplantation-placentation through the remodelling of the local tissue 16. Women with endometriosis \nshow altered immune cell phenotypes, such as low cytotoxicity in peritoneal NK cells17 and peritoneal \nmacrophages with reduced phagocytic capacity and are prone to secrete pro-inflammatory \ncytokines18. Poor pregnancy outcomes in patients may be due to impaired immune cell function and \nrecruitment at the maternal-fetal interface.  \nInterestingly, recent research has highlighted the concept of «  trained immunity  », where innate \nimmune cells can develop a form of memory through exposure to bacterial by -products3, influencing \ntheir function in the context of inflammation19 and endometriosis15. Indeed, we recently showed that \nLPSlow-trained immunomodulated macrophages could attenuate endometriosis  in a mouse model  \nobtained by grafting uterine fragments onto the peritoneal wall . Furthermore, coculture of human \nendometriotic cells with human LPS low-trained macrophages could reduce the fibro -inflammatory \nphenotype by inducing the downregulation of genes involved in cell adhesion and fibrosis 15. LPSlow in \nvivo immune training can induce epigenetic and metabolic changes in innate immune cells , with a \nconsequential change in gene expression, leading to an anti -inflammatory prone phenotype, mainly \ndemonstrated in macrophages 19, but similar innate memory mechanisms are described in NK cells 20. \nThis prompted us to investigate the potential benefits of LPSlow immunomodulation, specifically in the \ncontext of endometriosis -associated pregnancy complications. We used the abortion-prone CBA/J × \nDBA/2N mouse model, a well-established model for studying the general immunology of pregnancy21.  \n.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 2, 2026. ; https://doi.org/10.64898/2026.01.02.695166doi: bioRxiv preprint \n\nWe first validated that we could recapitulate endometriosis-associated pregnancy complications in the \nCBA/J × DBA/2 mouse model. To determine how endometriosis affects the maternal -fetal interface, \nwe generated and analyzed the transcriptome at the single-cell level of whole developing mouse \nplacentas and isolated immune cells from these developing placentas with or without endometriosis. \nWe demonstrated the  beneficial effect of pre -conceptional LPS low immunomodulation (IMM) on \nendometriosis-associated pregnancy complications in mice. By studying the transcriptome within the \nmaternal-fetal interface, this study aimed to elucidate the mechanisms underlying endometriosis -\nassociated pregnancy complications and the beneficial effects of pre -conceptional LPSlow \nimmunomodulation. \nMaterial and methods \nMice  \nSeven-to eight -week-old CBA/J female mice were purchased from Janvier Laboratory and housed \nunder standard conditions (ad libitum food and water, 12h light -dark cycle) in the animal care facility \nof Institut Cochin.  \nEndometriosis model \nThe endometriosis model consisted of a syngeneic graft of uterine horns to generate endometriosis-\nlike lesions. In details, donor mice were sacrificed by cervical dislocation , and the uterine horns were \nsurgically extracted and transferred to a Petri dish containing PBS. The uterine horns were opened \nlongitudinally with micro scissors and 3 to 5 -mm-length samples were prepared for grafting on the \ninternal face of the peritoneum of the recipient mice (each horn fragment was weighed). A \npreoperative gavage of all donor mice with 100µg/kg/day of 17β-estradiol was performed for two days \nbefore sacrifice (to synchronize estral cycles). Recipient mice were anesthetized using isoflurane. An \nincision was made on the ventral midline, and one or two donor horn fragments were sutured onto \nthe parietal peritoneum with two 7/0 polypropylene stitches (Prolen®, Ethicon, Somerville, NJ , USA). \nIn all mice, tissue samples were sutured at identical positions in the abdominal wall to ensure that the \nhost tissue sites exhibited comparable vascularization. The incision was then sutured with a 6/0 nylon \nthread.  \nImmune training and endometriosis \nThree independent experiments were performed using 5 –10 mice per group using the previously \ndescribed endometriosis model. The control group underwent a sham surgery with a midline incision \nand peritoneal stitches. For the endometriosis group, where we evaluated in vivo LPSlow innate immune \ntraining (IMM-EDT), mice were administered daily peritoneal injections of low doses (0.1 mg/kg) of LPS \nfor 5 days the week before endometriosis induction. The PBS-EDT group received PBS injection instead. \nLesion size was measured in the PBS -EDT and IMM -EDT groups using a high-frequency ultrasound \nimaging system (Vevo® 2100 VisualSonics; Toronto, CA) at 1 and 3 weeks after surgery.   \nEthical statements \nThe animal experiment protocol was approved with the ethical approval number DAP20 -104 and \nauthorization APAFIS#30327 by the Ministry of Higher Education and Scientific Research. \nMice pregnancy follow-up \nThree weeks after endometriosis induction, CBA/J female mice from the three groups (Sham, PBS-EDT, \nand IMM-EDT) were mated with DBA/2 male mice. The presence of a vaginal plug was considered \nembryonic day 0.5 (E0.5) . A high -frequency ultrasound imaging system (Vevo® 2100 VisualSonics; \nToronto, CA) was used to assess the number of implantation sites (between E7.5 and E10.5) and the \n.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 2, 2026. ; https://doi.org/10.64898/2026.01.02.695166doi: bioRxiv preprint \n\nnumber of alive/dead/resorbed fetuses (between E11.5 and E14.5). For ultrasound imaging, pregnant \nmice were anesthetized with isoflurane and restrained on a heated stage. The fur was removed from \nthe abdomen using hair removal agents, and pre-warmed ultrasound contact gel was applied to the \nshaved abdomen. The beating heart was detected  in the living fetuses . Dead fetuses displayed no \nbeating heart but had visible organ structures. Resorbed embryos/fetuses displayed an echogenic dot \nwith no discernible organ structure. Gestational age was confirmed by observing key developmental \nfeatures (E8.5 heart, head , and whole embryo; E9.5 amniotic membrane, yolk sac , and cerebral \nventricles; E10.5 umbilical cord, placenta, and eyes; E12.5 spine; and E13.5 face, skull bones, and ribs). \nAt the end of gestation (E18.5), the mice were sacrificed to harvest the fetuses, placentas, uterus, and \nlesion tissues . The number of fetuses, their weights, and the weights of the placentas and \nendometriosis lesions were recorded. The r esorption rate was calculated as (number of implantation \nsites at the first ultrasound - number of live pups at sacrifice) / number of implantation sites at the first \nultrasound × 100. The lesion volume was calculated using measurements obtained by ultrasound \n(length × width × width). \nStatistical analyses \nAll data from the mouse experiments were analyzed using GraphPad Prism 8.0 software ( GraphPad, \nSan Diego, CA, USA). For the comparison of two groups of data, we used the Student t -test with \nvariables following a normal distribution or the Mann-Whitney U test otherwise. Statistical significance \nwas set at p < 0.05. \nRNA sequencing  \nMouse tissues sampling \nThe three groups of CBA/J mice were obtained again as described previously and mated with DBA mice \nthree weeks after surgery. Pregnancy was assessed by the presence of a vaginal plug (E0.5 stage), and \nultrasound was performed at E7.5 -E8.5 to confirm pregnancy. Mice were sacrificed at E9.5 , and the \nfeto-maternal interfaces were retrieved from the uterus and either snap-frozen in nitrogen (for snRNA-\nseq) or digested to isolate immune cells (for CITE -seq). Fetal -maternal interfaces showing signs of \nabnormal development during ultrasound and/or dissection were not harvested to avoid non-specific \nsignals (considered as dying structures).  \nNuclei isolation and single nucleus RNA sequencing (snRNAseq) \nNuclei were isolated following a published method 22. Briefly, materno-fetal interface tissues were \nlysed for 10 min at 37°C with the lysis buffer described in the method. Samples were dounced with 10 \nstrokes and filtered with a 70µm strainer. The filtrate was centrifuged and filtered again using a 40µm \nstrainer. The filtrate was centrifuged again , and the pellet was resuspended in the staining buffer  \ndescribed in the method with specific antibodies. Nuclei from each placenta were tagged using \nBioLegend ® TotalSeq ™ antibodies. Nuclei were processed using the Single Cell 3′ Gene Expression kit \nv3 (10× Chromium, 1000076) according to the manufacturer’s instructions. In brief, 10000 -16,000 \nnuclei per sample were loaded onto Single Cell Chips B to recover as many nuclei as possible (targeting \n3,000–10,000 nuclei per sample) while limiting the potential for doublets. Using a Chromium \nController, Gel Bead -In Emulsions were generated, and samples were subsequently processed to \nisolate and amplify cDNA , and ultimately construct libraries. The quality and concentration of cDNA \nwere evaluated using an Agilent 2100 Bioanalyzer. The quality  and concentration of libraries were \nevaluated by qPCR and on an Agilent 2200 TapeStation, and libraries were sequenced on an Illumina \nNovaSeq 6000 through the Genom’IC platform of Institut Cochin.  \n.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 2, 2026. ; https://doi.org/10.64898/2026.01.02.695166doi: bioRxiv preprint \n\nImmune cell isolation, surface protein barcoding, single cell RNA and epitope sequencing \n(CITE-seq) \nPreviously snap -frozen E9.5 feto -maternal interfaces were digested using a Multi Tissue  Tissue \nDissociation Kit (Miltenyi Biotec 130-110-201) and a gentleMACS Dissociator (Miltenyi Biotec 130-134-\n029) at 37°C for 15 min. Samples were filtered using a 100µm strainer and centrifuged to obtain the \npellet. The pellet was resuspended in red blood cell lysis buffer (BioLegend # 420301) for 2 min at room \ntemperature. Samples were centrifuged, and the pellet was resuspended in RPMI 2% FCS.  Samples \nwere incubated w ith TruStain FcX PLUS Blocking Reagent (BioLegend 156603) for 10 min at 4°C. \nSamples were centrifuged, and the pellet was incubated for 30 min at 4°C with a mix of antibodies \ncontaining CD45.2 and Zombie NIR (dead/live marker) for sorting, and several epitope markers listed \nbelow. Each sample was incubated with a specific antibody tagged with a hashtag (TotalSeq B \nantimouse), which is listed below. Two wash cycles (centrifugation and addition of RPMI 2% PCF) were \nperformed for each sample. Samples were sorted using a BD FACSAria ™ III Cell Sorter keeping all the \nCD45+ and Zo mbieNIR- cells. The sorted samples were then processed using the Single Cell 3′ Gene \nExpression kit v3 (10× Chromium, #1000076) and following the same sequencing protocol as previously \ndescribed.  \nAntibody Barcode Sequence Reference (Cat #) \nB0001 (anti-mouse CD4) AACAAGACCCTTGAG 100573 \nB0002 (anti-mouse CD8a) TACCCGTAATAGCGT 100783 \nB0182 (anti-mouse CD3) GTATGTCCGCTCGAT 100257 \nB0184 (anti-mouse CD335 (NKp46)) CCCTTTCACCTCGAA 137641 \nB0114 (anti-mouse F4/80) TTAACTTCAGCCCGT 123155 \nB0013 (anti-mouse Ly-6C) AAGTCGTGAGGCATG 128053 \nB0839 (anti-mouse Ly49H) CCAGTAGGCTTATTA 144721 \nB0301 Hashtag 1 ACCCACCAGTAAGAC  155831 \nB0302 Hashtag 2 GGTCGAGAGCATTCA  155833 \nB0303 Hashtag 3 CTTGCCGCATGTCAT  155835 \nB0304 Hashtag 4 AAAGCATTCTTCACG   155837 \n \nSequences quality control and alignment \nBase calling was completed using Illumina® NovaSeq 6000 RTA v3.4.4 software, and BCL base call files \nwere converted to FASTQ files using bcl2fastq conversion software (v2.20). Using the Cell Ranger \nsoftware suite (v4.0.0) (10X Genomics®), FASTQ files were aligned to the GRCm38 mouse reference \ngenome (modification steps provided by 10X Genomics), and gene -barcode count matrices were \ngenerated for all samples. \nComputational analysis of snRNA-seq and CITE seq data  \nAll basic analyses were performed in R using Seurat 4.1.1 package included in a Snakemake pipeline \nwith the following rules (with key parameters described below): quality control, demultiplexing, \nnormalization, integration , and clustering.  Muscat package was used to perform differential gene \nexpression analysis. Scripts are available at the bkheira/doridot-sncell GitHub repository. \nQuality control & demultiplexing \nAll raw gene-barcode count matrices were converted into Seurat objects in R for initial quality control \nfiltering. Nuclei/cells were filtered using their RNA molecule count (>200 and <30000), sequenced gene \ncount (>500 and < 10000) , and percentage of mitochondrial sequenced genes (<10%). Since samples \nwere tagged with antibodies presenting a barcode (TotalSeq anti-mouse hashtags), we could separate \n.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 2, 2026. ; https://doi.org/10.64898/2026.01.02.695166doi: bioRxiv preprint \n\nthem in our analysis by reading these barcodes. An assay “HTO” was created in the Seurat Object and \ncontained barcode data. HTO assay data were normalized and demultiplexed using the “HTOdemux” \nfunction of Seurat. Doublet and negative cells for the barcodes were removed from each Seurat object. \nNormalization & integration \nSeurat objects were normalized using SCTransform. To identify the cell populations present in all \ngroups, the Seurat object of each group was integrated with the others. Integration features were \nselected using the “SelectIntegrationFeatures” function. The SCTransformed data were prepared using \nthe “PrepSCTIntegration ” function. Integration anchors were found using the \n“FindIntegrationAnchors” function, and Seurat objects were integrated using “IntegrateData”. \nClustering \nClustering was performed using the “FindNeighbors“ and “FindClusters” functions of Seurat with \ndifferent clustering resolutions . The optimum resolution was chosen using Clustree package in R.  \nUniform Manifold Approximation and Projection (UMAP) was performed using the RunUMAP function. \nDEG identification & Gene Set Enrichment Analysis \nDifferentially expressed genes were identified using Muscat 23 package in R. Muscat is a package in R \nthat allows pseudobulk analysis by pooling single-cell data in groups of unique clusters per sample. We \nfollowed the authors instructions to perform a differential analysis of the mouse data. Data were then \nfiltered : we  considered differentially expressed a gene with a p -value adjusted to multiple \ncomparisons smaller than 0.05, an absolute log2FC greater than 0.6 and a minimal expression (logCPM) \nof 5. \nWe used the FGSEA package in R using the tutorial available on the Biostatsquid blog24. GSEA was \nperformed using the Hallmark database. We added some visualization plots using the package ggplot2.  \nIllustrations \nFigures 1A and 2A were created using BioRender. BOUZID, K. (2025) https://BioRender.com/trkr2lh \n  \n.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 2, 2026. ; https://doi.org/10.64898/2026.01.02.695166doi: bioRxiv preprint \n\nResults \nEndometriosis reduces the number of embryo implantations and increases fetal \nresorption rate in the CBA/JxDBA/2 mouse model \nWe first wanted to check if the presence of endometriosis -like lesions had a deleterious effect on \npregnancy in the CBA/JxDBA/2 mouse model, as has been demonstrated in other mouse strains in the \ncontext of syngenic crosses (C57BL/69,10, Swiss Webster11 or Balb/C12). Our experimental design (Figure \n1A) was as follows: we induced endometriosis in CBA females by surgically suturing two syngeneic \nuterine fragments on the peritoneal wall, as previously described15. This model allows us to follow the \nlesions using ultrasound.  Three weeks later, the mice were mated with DBA males. Ultrasound imaging \nwas performed to count the number of implantation sites (between E6.5 and E8.5) and evaluate pup \ndevelopment or resorption at each site (between E13.5 and E15.5).  The volume of the endometriosis \nlesions was also measured. At E18.5, the mice were sacrificed, and the number of live fetuses, their \nweight, and the weight of the endometriosis lesions were recorded.  \nMice with endometriosis lesions displayed a slight but significant decrease in the number of \nimplantations (10% decrease, p=0.03, Figure 1B). A 2-fold increase in the resorption rate in the group \nwith endometriosis lesions (18.9 % vs. 36.9%, p=0.047, figure 1C) was also observed. This effect seems \nto be dose -dependent, as mice with only one endometriosis lesion showed an intermediate non-\nsignificant increase in  resorption rate  (fold change of 1.5 with 27.8% resorption rate, p=0.14, \nSupplementary Figure 1 ). Fetal weights at E18.5 were equivalent in all groups ( Figure 1D , Mann -\nWhitney test, p = 0.32).   \n \n \n \n \nFigure 1: Gestational complications in a mouse model of endometriosis \n(A) Experimental protocol: briefly, CBA female mice underwent either a sham or an endometriosis (EDT) \ninduction surgery. After 3 weeks of recovery, female mice were mated with DBA males, then gestation and \nlesion size were followed up by ultrasound at appr oximately E7.5 and E14.5. At E18.5, one day before \nparturition, the mice were euthanized, and the fetuses and placentas were retrieved. These are the combined \nresults of three similar experiments. The number of mice is indicated in the figure by n, with a precision of \nfetuses (F) and gestant mice (M), if relevant. (B) Number of implantations per mouse (C) Fetal resorption rate \nper mouse. For (B) and (C), data are presented as median ± interquartile range. (D) Fetal weight per pup at \nE18.5. *: p-value < 0.05 (Mann-Whitney U test).  \n.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 2, 2026. ; https://doi.org/10.64898/2026.01.02.695166doi: bioRxiv preprint \n\nHistological evaluation of E18.5 placentas did not reveal any obvious anomalies associated with \nendometriosis (Supplementary Figure 2).  \nPre-conceptional introduction of low-dose LPS reduces gestational complications \nassociated with endometriosis in a mouse model \nAs we previously showed that immunomodulatory (IMM by repeated exposure to low doses of LPS) \ntreatment can reduce the lesion size in endometriosis mice 15, we wanted to see if this \nimmunomodulatory treatment could also reduce the deleterious effect of endometriosis on \npregnancy. We conducted an experiment (Figure 2A) in which endometriosis induction was preceded \nby peritoneal injections of repeated low doses of LPS (IMM) or PBS.  Three weeks after surgery, the \nendometriosis mice were mated with DBA/2 males, and gestation was followed up as previously \ndescribed. We previously showed that IMM has no effect on the resorption rate in mice without \nendometriosis21.  \n \n \n \n \n \nFigure 2: Immunomodulation (IMM) treatment partially corrects gestational complications in endometriosis \nmice. (A) Experimental protocol: repeated injections of PBS or low-dose LPS were administered before inducing \nendometriosis by surgery, followed by the same protocol as shown in Figure 1A. (B) Evolution of the lesion \nvolume as observed on ultrasound. For (C), (D ) and (E), data are presented as mean ± SEM. PBS-EDT: group \nreceiving PBS and two lesions of endometriosis. IMM-EDT: group receiving LPS at low doses and with two lesions \nof endometriosis. (C) Total weight of the two lesions at the time of sacrifice. Student ’s t-test (D) Number of \nimplantations per mouse, Student ’s t -test (E) Fetal resorption per mouse, Student ’s t -test (F) Correlation \nbetween total lesion weight and resorption rate per mouse, PBS -EDT Pearson r = 0.72 (p = 0.0013), LPS -EDT \nPearson r = -0.21 (p = 0.38). (G) Fetal weights.  *: P-value < 0.05, **: P-value < 0.01.  \n.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 2, 2026. ; https://doi.org/10.64898/2026.01.02.695166doi: bioRxiv preprint \n\nThe volume of the lesions was measured during ultrasounds following gestation, and IMM treatment \nshowed a trend toward reducing the volume of the lesions (Figure 2B, p=0.15 by mixed effect analysis). \nAt the end of gestation, the lesion weight of the IMM -EDT group was significantly lower than that of \nthe PBS-EDT control group (Figure 2C, p=0.01). The number of implantations was significantly higher \nin the IMM-treated group (20% increase, Figure 2D, p = 0.048).  The fetal resorption rate was 38% in \nthe PBS-treated group and 24% in the IMM-treated group (Figure 2E, p = 0.06), which is closer to that \nobserved in mice without endometriosis (sham group in Figure 1C, 19%). The significant correlation \nbetween lesion size and resorption rate in the endometriosis group (r=0,73, p=0.0013) was lost with \nthe IMM (Figure 2F). Fetal weights at E18.5 were equivalent in all groups (Figure 2G). \nSnRNAseq reveals an increase of Gata4 and a decrease of P rap1 expression in decidual \ncells in endometriotic mice  \nSince there are a decreased mean number of implantation sites and an increased resorption rate in \nthe endometriosis group, we hypothesized that there is a possible dysfunction of the decidua and/or \nthe placenta that leads to an altered or interrupted embryo development. To validate this hypothesis, \nfeto-maternal interfaces at E9.5 without visible signs of ongoing resorption were collected from mice \nwith sham surgery or with two lesions of endometriosis, and their transcriptomes were sequenced at \nthe single-nuclei level. \nWe identified 13 cell clusters in our dataset that were distributed homogeneously among the samples \n(Figure 3A, 3B). Wt1, a known marker of decidualized cells25, was found in seven clusters (Figure 3D). \nAmong them, there are two decidual fibroblast/stromal clusters, identified by the expression of \ncollagen genes ( Col3a1) and several genes involved in the remodelling of the extracellular matrix of \nthe decidua during implantation and placentation (Sulf1, Adamtsl1)26,27 ,28. Interestingly, these decidual \nfibroblasts expressed Il15 and its receptor Il15ra, which are involved in the activation of uterine Natural \nKiller (uNK) 29,30. The other Wt1+ clusters were identified as decidual stromal cells (DSC) as they \nexpressed several known markers of decidual stromal cells at different levels ( Hand2, Pgr, and Esr1), \nsuggesting that the clusters may represent different stages of differentiation or different subtypes of \ndecidual stromal cells. All clusters’ markers are available in Supplementary Table 1. \nThe expression of the maternal gene Xist in placental samples associated with a male embryo \nsegregated the cells of embryonic origin from those of maternal origin (Figure 3C). We confirmed their \ntrophoblast origin by the expression of specific genes that mark them as trophoblast progenitors, such \nas Epcam and Ror2 expression in embryonic cells 31, 32 and trophoblast cells with Mct1/Sl16a1, \nCD147/Bsg, Glut1/Slc2a133, and Nr6a134 (Figure 3F). The expression of Pecam1 was used to identify \nendothelial cells within the placenta 35,36. Cells of hematopoietic origin were identified by Ptprc (CD45 \ngene, Figure 3E). Pericytes have been identified using the Pdgfrb gene. \n \n  \n \n \n \n \n \n.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 2, 2026. ; https://doi.org/10.64898/2026.01.02.695166doi: bioRxiv preprint \n\n \n \n \n \n \nThrough differential expression analysis, we identified several differentially expressed genes  (DEG) in \nall clusters (Supplementary Table 2). In the different clusters of decidual stromal cells, we observed \ntwo genes that were consistently differentially expressed: Gata4 was upregulated and Prap1 was \ndownregulated in the group with endometriosis -like lesions. Interestingly, these 2 genes differential \nexpression were normalized by the immunomodulatory treatment, with a decrease of Gata4 and an \nincrease in Prap1 in the IMM-EDT group compared to the PBS-EDT group.  \nFigure 3: Single -nucleus RNA sequencing of E9.5 fetomaternal interface reveals its composition and \ncharacteristics (A) UMAP of cell transcriptomes (B) Graph representing the proportion of a cell type (cluster) in \neach sample (top) and the number of cells retained for analyses after quality control filtering (bottom) (C) \nExpression of Xist gene in materno-fetal interfaces with male embryos (D) Expression of Wt1 (E) Expression of \nPtprc (CD45 gene) (F) Dotplot of cluster identifying genes. DSC: Decidual stromal cells  \n.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 2, 2026. ; https://doi.org/10.64898/2026.01.02.695166doi: bioRxiv preprint \n\n \nFor the next analysis, we grouped the clusters by big cell types (decidual stromal cells, trophoblast \ncells, endothelial cells, immune cells, and pericytes) and compared the groups of mice (sham, PBS-EDT, \nIMM-EDT) by performing differential expression analysis and gene set enrichment analysis (GSEA)  \n(Figure 4, Supplementary Table 3 ). When plotting a heatmap of the top (up to 10) DEG in each cell \nFigure 4: Differential expression analysis reveals alterations driven by endometriosis and partially corrected \nby the IMM treatment \n(A) Number of differentially expressed genes per group of cells (B) Heatmap of the top 10 differentially \nexpressed genes between the sham and endometriosis groups. Cell clusters were grouped by cell type as \nshown in Figure 3B. (C) Volcano plot showing the differentially expressed genes within the decidual stromal \ncells (D) Dotplot of Gene Set Enrichment Analysis: top 10 significantly enriched pathways from Sham vs PBS -\nEDT analysis and how they are in the PBS -EDT vs IMM -EDT in the pooled DSC group. Pathways involved in \nproliferation are surrounded by red rectangles. * Significant enrichment after adjustment.  \n.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 2, 2026. ; https://doi.org/10.64898/2026.01.02.695166doi: bioRxiv preprint \n\ngroup, we observed that the transcriptomic expression of the IMM-EDT group was often between the \nexpression of the control and endometriosis groups, indicating a partial correction of the gene \nexpression differences induced by endometriosis by IMM treatment.  We observed transcriptomic \nchanges induced by endometriosis in all the cell types ( Figure 4A, B). We then chose to focus on \ndecidual stromal cells, which are the group of cells with the most differentially expressed genes \ninduced by the presence of endo metriotic lesions ( Figure 4A ). The differential expression analysis \nshowed an overexpression of Gata4 (log2FC of 2.27 and adjusted p -value of 1.7 × 10−5) and a \ndownregulation of Prap1 (log2FC of -2.37 and adjusted p-value of 1.63 × 10−5 ) in the PBS-EDT group \ncompared to the Sham group (Figure 4C). IMM led to the correction of these expressions (Figure 4B). \nGSEA showed that pathways related to proliferation (Myc targets, E2F targets, G2M checkpoints, and \nKRAS signalling) were negatively enriched in the endometriosis group.  This downregulation was \nsignificantly corrected by IMM treatment, except for the G2M checkpoint (Figure 4D).  \nCITEseq analysis reveals transcriptomic differences in all immune cell types and an \ninflammation in endometriotic placentas \nAfter immunomodulation, it is relevant to examine the consequences o n immune cells directly.  As \nthese cells account for a small proportion of the maternal fetal interface and thus are not well \nrepresented in the whole interface snRNAseq dataset, we performed Cellular Indexing of \nTranscriptomes and Epitopes by Sequencing (CITEse q or proteotranscriptomics ) on isolated immune \ncells (CD45+) in the E9.5 materno-fetal interface. \nWe identified 11 clusters of cells in this dataset  (Figure 5A ). Cluster markers are available in \nSupplementary Table 5. All the cell types were equally represented in the samples (Figure 5B). A major \ngroup of myeloid cells was recognized by Cd68 and Itgam (CD11b) expression ( Figure 5G ) and the \nexpression of the monocyte marker Ly6C ( Figure 5D). Within this group, we identified dendritic cells \nexpressing Cd209a and a small cluster of plasmacytoid dendritic cells expressing Siglech. Neutrophils \nwere identified by the expression of calprotein genes ( S100a9 and S100a8). Macrophages, identified \nhere with the antibody F4/80 (Figure 5C), had several marker genes, such as Wwp1, Ccl8, Stab1, Gas6, \nand Fcrls. Three other clusters were identified as monocytes (all expressing Ly6C  and/or Itgam), and \nspecific genes differentiating these subcategories of monocytes were identified.  \nThe other major cell type was natural killer (NK) cells and innate lymphoid cells (ILC) identified by the \ndifferent granzyme genes (Gzmb, Gzmc, Gzme) and perforin (Prf1) and with the surface protein data \nof the CITEseq, using the NKp46 antibody ( Figure 5E). T cells were identified by the surface proteins \nCD3 (figure 5F), CD4, and CD8a, and at the transcriptomic level, they had specific expressions of Itk and \nIl7r. B cells were identified with immunoglobulin genes such as Igkc or Ighm, as well as genes like Ebf1, \nBank1, Cd79a, and Cd79b.  \nFinally, in the CD45+ cells, we also had a small cluster of endothelial cells recognized by the expression \nof known genes such as Flt1 and Egfl7.  \nWe could see in the PBS-EDT group compared to the Sham group few transcriptomic differences in all \nthe cell types ( Figure 6A). The IMM treatment induced a higher number of significant transcriptomic \ndifferences in all cell types compared to the Sham group ( Supplementary Figure 3A) and to the PBS -\nEDT group ( Figure 6A ). Indeed, immunomodulation of the immune system leads to transcriptional \nchanges37. A gene that was consistently downregulated among the cell groups  (except macrophages) \nwas Slc15a2 (Figure 6B), which encodes the peptide transporter PEPT2 (see Supplementary Table 7).  \nTo determine the pathways in which these genes are involved, we performed GSEA.  Notably, we \nobserved an upregulation of TNF α signalling in macrophages in endometriosis, which tended to be \n.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 2, 2026. ; https://doi.org/10.64898/2026.01.02.695166doi: bioRxiv preprint \n\ncorrected by IMM treatment ( Figure 6C). This upregulation is also observed in endometriosis in the  \npooled group of monocytes (Supplementary figure 3), in T and B cells, but only corrected by the IMM \nin B cells (see Supplementary Table 8).  In NK and ILC, endometriosis led to a negative enrichment of \nIFNγ and IFN α response pathways, and IMM tended to normalize this pathway ( Figure 6D ). This \nnegative enrichment of IFN γ response in endometriosis is also found in decidual stromal cells, \nendothelial cells, and trophoblasts.  \n \n  \nFigure 5: Composition of E9.5 fetomaternal interface by single cell proteotranscriptomics sequencing of \nimmune cells  \n(A) UMAP of cell transcriptomes (B) Graph representing the proportion of a cell type (cluster) in each sample \n(top) and the number of cells retained for analyses after quality control filtering  (bottom) (C) Featureplot of \nthe surface antibody F4/80 staining macrophages (D) Ly-6C (E) Nkp46 (F) CD3 (G) Dotplot of cluster identifying \ngenes and antibodies \n.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 2, 2026. ; https://doi.org/10.64898/2026.01.02.695166doi: bioRxiv preprint \n\n \n \nDiscussion \nIn this study, we first examined the impact of endometriosis-like lesions on gestation in CBA/JxDBA/2N \ncrossing, with fewer implantations and a higher proportion of fetal resorptions than in the sham group. \nOther studies have shown the consequences of endometriosis on gestation in mouse models, including \na reduction in the pregnancy rate 9,10,12 (gestant mice per group), increase in the resorption rate 11,12, \nFigure 6: Differential expression analysis in immune cells reveals alterations driven by endometriosis and \npartially corrected by the IMM treatment \n(A) Number of differentially expressed genes in each cell group. (B) Heatmap of the top 3 differentially \nexpressed genes between the sham and endometriosis groups. Cell clusters were grouped by cell types as \nshown in Figure 5A. (C) Dotplot of Gene Set Enrichment Analysis : top (up to) 10 significantly enriched \npathways from Sham vs PBS-EDT analysis and how they are in the PBS-EDT vs IMM-EDT in Macrophages and \n(D) in NK and ILC. * Significant enrichment after adjustment.  \n.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 2, 2026. ; https://doi.org/10.64898/2026.01.02.695166doi: bioRxiv preprint \n\ndecrease in litter size12, and reduced birth weight for the pups11. In our study, we found no significant \ndifference in fetal weight, which may be due to the fact that mice were sacrificed before parturition; \nhowever, other studies also showed no difference in pup weight  at E18.510. Currently, most of these \nstudies on mouse models have been used as preclinical models to test the effects of potentially \nbeneficial molecules, but the underlying mechanisms of these complications are yet to be understood. \nOur study suggests that the presence of lesions and/or inflammation induced by endometriosis leads \nto a deleterious environment for pregnancy development.  \nThis is further substantiated by the improvement in pregnancy outcomes with IMM treatment.  In \naccordance with previous findings of our team, inducing immunotolerance reduced the evolution and \nfinal size of endometriosis lesions and had a beneficial effect on pregnancy complications, with an \nincrease in the number of implantations and a decrease in the fetal resorption rate.  In the IMM \ncondition, the correlation between lesion size and fetal resorption was lost, suggesting that the \nbeneficial effect of IMM o n pregnancy may be due to a reduction in inflammation rather than a \nreduction in lesion size.  \nDue to the increased risk of miscarriage in endometriosis, there may be defects in early pregnancy. To \nstudy this, we used our mouse model again and retrieved the feto -maternal interface at E9.5, a stage \nat which the structure of the placenta is developing.  Transcriptomic sequencing has shown that \nendometriosis induces changes in all cell types of the developing placenta, even in trophoblasts, which \nare exclusively fetal.  These fetal cells are derived from oocytes exposed to chronic inflammation \ninduced by endometriosis, which may explain the consequences on trophoblastic cells. These changes \nin fetal cells of the placenta raise questions about the transmission of endometriosis consequences to \nthe fetus via epigenetic signals. Indeed, endometriosis has a genetic heritability of 50%, and until now, \nseveral GWAS have only been able to explain 5% 38 of the genetics of endometriosis.  In addition to \npotential rare variants not detected by GWAS, missing heritability could also be due to epigenetics and \nshould be studied in further research. Our model may allow for the exploration of its potential impact \non the next generation.  \nThe major cell type at E9.5 in the developing placenta is decidual stromal cells (DSC), which are derived \nfrom the decidualization of endometrial stromal cells.  We observed most transcriptomic changes in \nthe presence of endometriosis in this group of cells.  Gata4 gene expression was upregulated by \nendometriosis and normalized by IMM treatment. This upregulation of Gata4 has already been found \nin endometriosis within ectopic and eutopic endometrium in patients 39, making our mouse model \nrelevant for studying this upregulation.  The involvement of Gata4 in gestation has not been studied, \nbut its family members GATA2 and GATA3 are important  during preimplantation and early post -\nimplantation development, with knockouts of both Gata2 and Gata3 leading to placental defects40 in \ntrophoblasts.  \nDSC also showed downregulation of Prap1 in endometriosis, which was corrected by IMM treatment.  \nIn mice, Prap1 appears to be involved in establishing uterine receptivity to the embryo 41 and is an \nindicator of successful implantation 42. Therefore, its downregulation may participate in the reduced \nnumber of implantations and increased resorption observed in endometriotic mice.  Prap1 is a gene \npositively regulated by the progesterone receptor PGR in the mouse endometrium 43. Pgr expression \nin our dataset was non-significantly downregulated by endometriosis (logFC of -0.40, p -value of \n0.0044, and adjusted p-value of 0.077) and may participate in the downregulation of Prap1 expression. \nThis downregulation of Pgr, although not significant in our dataset when we adjusted the p -value for \nmultiple testing, should be further studied, and our model may be interesting to study the \nprogesterone resistance found in endometriosis in human44.  \n.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 2, 2026. ; https://doi.org/10.64898/2026.01.02.695166doi: bioRxiv preprint \n\nIn our GSEA of mouse DSCs at E9.5, most of the downregulated pathways in the endometriosis group \nwere related to proliferation. A defect in proliferation during early gestation could explain miscarriages \nor further placental pathologies found to be increased in endometriosis.  \nWe identified two clusters of decidual stromal cells that expressed extracellular matrix remodelling \ngenes and named them decidual fibroblasts. These decidual fibroblasts expressed Il15, which encodes \nIL-15, an important factor for the  differentiation and proliferation of uNK cells, allowing uNK cells to \nsupport embryo implantation, spiral artery remodelling, and immune tolerance during pregnancy 45. \nUntil now, it was known that IL -15 signalling was coming from decidual stromal cells 30, but here we \nshow that only a specific subtype of DSC has this role.  Interestingly, these cells also express Il15ra, a \ngene encoding a receptor for IL-15, suggesting the possibility of an autocrine signal within these cells.  \nWhen sorting the immune cells from the E9.5 feto -maternal interfaces, we identified a majority of \nmyeloid cells and up to 20 -25% of lymphoid cells , concordant with previous littereature 46. This is \ndifferent from human early placentas, where NK cells represent 70% of immune cells 47 and \nmacrophages represent 20 and 30%48. Previous studies have shown that less than 1% of immune cells \nare of fetal origin at E10.5 46, a later stage than that in our study; therefore, it is likely that the vast \nmajority of cells in our dataset are of maternal origin.  \nEndometriosis induced transcriptomic changes in all immune cells. When focusing on macrophages in \nendometriosis, GSEA showed an upregulation of an inflammatory pathway, TNF α via NF κB. In \nendometriosis, it has been shown that macrophages are more activated and participate in \ninflammation within the peritoneal cavity 49. Our results suggest that macrophages are inflammatory, \neven at the maternal-fetal interface. We can imagine that inflammation at an early stage of gestation \ncould lead to a poor establishment of the maternal -fetal interface and thus induce \nmiscarriages/resorptions. The upregulation of inflammatory pathways was no longer observed when \nendometriosis mice were previously trained with low doses of LPS (IMM). We have already shown that \nperitoneal macrophages in a mouse model of endometriosis become more immuno tolerant with LPS \ntraining at low doses15, and we show here that this immunotolerance is also induced in the maternal -\nfetal interface.  \nThe other major immune cells during pregnancy are NK cells. In these cells in endometriosis mice, the \nIFNγ and IFNα response pathways were downregulated, whereas the Myc target V1 pathway, which is \ninvolved in proliferation, was upregulated. uNK cells are the main source of IFN γ in the murine \nmaterno-fetal interface50. IFNγ is essential for the development of gestation in mice and participates \nin NK maturation51. Thus, NK cells, which may be less able to respond to this signal, may not mature \nwell and may disrupt the development of pregnancy.  Mice that are IFNγ-/- present a high rate of fetal \nresorptions, no spiral artery remodelling, and decidua necrosis.  They also have more uNK at the MFI, \nas if they are trying to compensate for the lack of IFN γ by having more producing cells 50. The \nupregulation of a proliferation pathway in NK cells in our dataset may be a consequence of the \ndownregulated IFNγ response. IFN-α and IFN-β are regulators of NK cell activation and their production \nof IFNγ52, and in mice that are Ifna-/-, the same phenotype as Ifng-/- is observed at the materno-fetal \ninterface, suggesting that NK cell IFN-α response is also essential for the development of the MFI. This \ndownregulated response to IFN γ was also found in the decidual cells, endothelial cells, and \ntrophoblasts of our dataset, and only the downregulation of IFN α was found in decidual cells and \ntrophoblasts. Given the importance of IFN γ signaling in vascular remodelling and maintenance of the \ndecidua, this general d ownregulation of the response may be a cause of disruption of gestational \ndevelopment.   \nOf note, B cells are among the immune cells with most differentially expressed genes in endometriosis \ncompared to sham group. GSEA showed enrichment of several pathways involved in inflammation and \n.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 2, 2026. ; https://doi.org/10.64898/2026.01.02.695166doi: bioRxiv preprint \n\nimmune response. B cells involvement in endometriosis is still incompletely defined and controversial \nwith evidence supporting increased B cell activation and auto -antibody production 53. Our mouse \nmodel could be used to further study B cells in endometriosis.  \nOur results show that a murine model is pertinent for studying gestational complications induced by \nendometriosis and reveals mechanisms affecting decidual stromal and immune cells at the beginning \nof gestation. Our findings suggest that targeting the immune system may be a therapeutic strategy to \nimprove pregnancy outcomes and reduce the risk of miscarriage.  \nAcknowledgments and funding: \nWe would like to thank the different platforms at the Institut Cochin and Université Paris Cité that \nhelped us obtain our results: GENOM’IC, CYBIO, HIST’IM, BIOINFORMAT’IC, and IPOP-UP.  \nWe would like to thank Université Paris Cité for the IDEX funding of the PlacentAtlas project (ANR-18-\nIDEX-0001).  \nK. B. received a PhD fellowship from BioSPC doctoral School at Université Paris Cité. C. M. is supported \nby ANR-20-CE14-0004. L.D. is funded by the European Union (European Research Council (ERC) starting \ngrant No 101078556). Views and opinions expressed are however those of the author(s) only and do \nnot necessarily reflect those of the European Union. Neither the European Union nor the granting \nauthority can be held responsible for them.  \n  \n.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 2, 2026. ; https://doi.org/10.64898/2026.01.02.695166doi: bioRxiv preprint \n\n1. Zhang, T., De Carolis, C., Man, G. C. W. & Wang, C. C. The link between immunity, autoimmunity \nand endometriosis: a literature update. Autoimmun. Rev. 17, 945–955 (2018). \n2. Endometriosis-associated infertility: aspects of pathophysiological mechanisms and treatment \noptions - Tanbo - 2017 - Acta Obstetricia et Gynecologica Scandinavica - Wiley Online Library. \nhttps://obgyn-onlinelibrary-wiley-com.ezproxy.u-paris.fr/doi/10.1111/aogs.13082. \n3. Lalani, S. et al. Endometriosis and adverse maternal, fetal and neonatal outcomes, a systematic \nreview and meta-analysis. Hum. Reprod. 33, 1854–1865 (2018). \n4. Brosens, I. Endometriosis and the outcome of in vitro fertilization. Fertil. Steril. 81, 1198–1200 \n(2004). \n5. Huang, Y. et al.  Miscarriage on Endometriosis and Adenomyosis in Women by Assisted \nReproductive Technology or with Spontaneous Conception: A Systematic Review and Meta -\nAnalysis. BioMed Res. Int. 2020, 4381346 (2020). \n6. Borghese, B. et al.  Low Birth Weight Is Strongly Associated with the Risk of Deep Infiltrating \nEndometriosis: Results of a 743 Case-Control Study. PLOS ONE 10, e0117387 (2015). \n7. Jenabi, E. & Fereidooni, B. The association between endometriosis and the risk of LBW: A meta -\nanalysis. J. Endometr. Pelvic Pain Disord. 10, 72–78 (2018). \n8. Gao, M., Allebeck, P., Mishra, G. D. & Koupil, I. Developmental origins of endometriosis: a Swedish \ncohort study. J Epidemiol Community Health 73, 353–359 (2019). \n9. Park, Y. et al. Oleuropein suppresses endometriosis progression and improves the fertility of mice \nwith endometriosis. J. Biomed. Sci. 29, 1–20 (2022). \n10. Bilotas, M. A. et al. Interplay between Endometriosis and Pregnancy in a Mouse Model. PLOS ONE \n10, e0124900 (2015). \n11. França, P. R. de C. et al. R-954, a bradykinin B1 receptor antagonist, as a potential therapy in a \npreclinical endometriosis model. Peptides 181, 171294 (2024). \n12. Lai, Z.-Z. et al. Protopanaxadiol improves endometriosis associated infertility and miscarriage in \nsex hormones receptors -dependent and independent manners. Int. J. Biol. Sci.  17, 1878 –1894 \n(2021). \n13. Elsherbini, M. et al.  Impact of Chronic Exposure to Endometriosis on Perinatal Outcomes: \nEstablishment of a Mouse Model. Biomedicines 10, 2627 (2022). \n14. Soni, U. K. et al.  MCP-1 promotes ILK phosphorylation at Ser246 during endometriosis \ndevelopment and affects the pregnancy outcome. Mol. Hum. Reprod. 31, gaaf004 (2025). \n15. Jeljeli, M. et al. Macrophage Immune Memory Controls Endometriosis in Mice and Humans. Cell \nRep. 33, 108325 (2020). \n16. Ander, S. E., Diamond, M. S. & Coyne, C. B. Immune responses at the maternal-fetal interface. Sci. \nImmunol. 4, eaat6114 (2019). \n17. Izumi, G. et al.  Involvement of immune cells in the pathogenesis of endometriosis. J. Obstet. \nGynaecol. Res. 44, 191–198 (2018). \n18. Chuang, P. -C., Wu, M. -H., Shoji, Y. & Tsai, S. -J. Downregulation of CD36 results in reduced \nphagocytic ability of peritoneal macrophages of women with endometriosis. J. Pathol. 219, 232–\n241 (2009). \n19. Jeljeli, M. et al. Trained immunity modulates inflammation -induced fibrosis. Nat. Commun.  10, \n5670 (2019). \n20. Mujal, A. M., Delconte, R. B. & Sun, J. C. Natural Killer Cells: From Innate to Adaptive Features. \nAnnu. Rev. Immunol. 39, 417–447 (2021). \n21. Dang, Y. et al. BCG-trained innate immunity leads to fetal growth restriction by altering immune \ncell profile in the mouse developing placenta. J. Leukoc. Biol. 111, 1009–1020 (2022). \n22. Gaublomme, J. T. et al. Nuclei multiplexing with barcoded antibodies for single-nucleus genomics. \nNat. Commun. 10, 2907 (2019). \n23. Crowell, H. L. et al.  muscat detects subpopulation -specific state transitions from multi -sample \nmulti-condition single-cell transcriptomics data. Nat. Commun. 11, 6077 (2020). \n24. Twomey, L. Easy Gene Set Enrichment Analysis in R with fgsea() - biostatsquid.com. \nhttps://biostatsquid.com/fgsea-tutorial-gsea/ (2023). \n.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 2, 2026. ; https://doi.org/10.64898/2026.01.02.695166doi: bioRxiv preprint \n\n25. Zhou, J., Rauscher, F. J. & Bondy, C. Wilms’ tumor (WT1) gene expression in rat decidual \ndifferentiation. Differentiation 54, 109–114 (1993). \n26. Lum, D. H., Tan, J., Rosen, S. D. & Werb, Z. Gene Trap Disruption of the Mouse Heparan Sulfate 6-\nO-Endosulfatase Gene, Sulf2. Mol. Cell. Biol. 27, 678–688 (2007). \n27. Kirn-Safran, C. B., D’Souza, S. S. & Carson, D. D. Heparan sulfate proteoglycans and their binding \nproteins in embryo implantation and placentation. Semin. Cell Dev. Biol. 19, 187–193 (2008). \n28. Hirohata, S. et al. Punctin, a Novel ADAMTS-like Molecule, ADAMTSL-1, in Extracellular Matrix*. J. \nBiol. Chem. 277, 12182–12189 (2002). \n29. Brighton, P. J. et al. Clearance of senescent decidual cells by uterine natural killer cells in cycling \nhuman endometrium. eLife 6, e31274 (2017). \n30. Sharma, S., Godbole, G. & Modi, D. Decidual Control of Trophoblast Invasion. Am. J. Reprod. \nImmunol. 75, 341–350 (2016). \n31. Ueno, M. et al.  c-Met-Dependent Multipotent Labyrinth Trophoblast Progenitors Establish \nPlacental Exchange Interface. Dev. Cell 27, 373–386 (2013). \n32. Marsh, B. & Blelloch, R. Single nuclei RNA-seq of mouse placental labyrinth development. eLife 9, \ne60266 (2020). \n33. Nagai, A., Takebe, K., Nio -Kobayashi, J., Takahashi -Iwanaga, H. & Iwanaga, T. Cellular Expression \nof the Monocarboxylate Transporter (MCT) Family in the Placenta of Mice. Placenta 31, 126–133 \n(2010). \n34. Mehta, D. V., Kim, Y.-S., Dixon, D. & Jetten, A. M. Characterization of the Expression of the Retinoid-\nrelated, Testis-associated Receptor (RTR) in Trophoblasts. Placenta 23, 281–287 (2002). \n35. van Mourik, J. A., Leeksma, O. C., Reinders, J. H., de Groot, P. G. & Zandbergen -Spaargaren, J. \nVascular endothelial cells synthesize a plasma membrane protein indistinguishable from the \nplatelet membrane glycoprotein IIa. J. Biol. Chem. 260, 11300–11306 (1985). \n36. Jakeman, L. B., Winer, J., Bennett, G. L., Altar, C. A. & Ferrara, N. Binding sites for vascular \nendothelial growth factor are localized on endothelial cells in adult rat tissues. J. Clin. Invest. 89, \n244–253 (1992). \n37. Netea, M. G. et al. Trained immunity: A program of innate immune memory in health and disease. \nScience 352, aaf1098 (2016). \n38. Rahmioglu, N. et al.  The genetic basis of endometriosis and comorbidity with other pain and \ninflammatory conditions. Nat. Genet. 55, 423–436 (2023). \n39. Fouquet, B., Santulli, P., Noel, J.-C. & Misrahi, M. Ovarian-like differentiation in eutopic and ectopic \nendometrioses with aberrant FSH receptor, INSL3 and GATA4/6 expression. BBA Clin. 6, 143–152 \n(2016). \n40. Home, P. et al. Genetic redundancy of GATA factors in the extraembryonic trophoblast lineage \nensures the progression of preimplantation and postimplantation mammalian development. \nDevelopment 144, 876–888 (2017). \n41. Xiong, G. -F. et al.  Estradiol-regulated proline -rich acid protein 1 is repressed by class I histone \ndeacetylase and functions in peri -implantation mouse uterus. Mol. Cell. Endocrinol.  331, 23–33 \n(2011). \n42. Diao, H., Xiao, S., Zhao, F. & Ye, X. Uterine luminal epithelium –specific proline-rich acidic protein \n1 (PRAP1) as a marker for successful embryo implantation. Fertil. Steril. 94, 2808-2811.e1 (2010). \n43. Kim, T. H. & Jeong, J.-W. Proline-Rich Acidic Protein 1 (PRAP1) is a Target of ARID1A and PGR in the \nMurine Uterus. Dev. Reprod. 23, 277–284 (2019). \n44. Zhang, P. & Wang, G. Progesterone Resistance in Endometriosis: Current Evidence and Putative \nMechanisms. Int. J. Mol. Sci. 24, 6992 (2023). \n45. Mori, M., Bogdan, A., Balassa, T., Csabai, T. & Szekeres-Bartho, J. The decidua—the maternal bed \nembracing the embryo—maintains the pregnancy. Semin. Immunopathol. 38, 635–649 (2016). \n46. Moore, A. R. et al. Gestationally dependent immune organization at the maternal-fetal interface. \nCell Rep. 41, 111651 (2022). \n.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 2, 2026. ; https://doi.org/10.64898/2026.01.02.695166doi: bioRxiv preprint \n\n47. King, A., Balendran, N., Wooding, P., Carter, N. P. & Loke, Y. W. CD3 - leukocytes present in the \nhuman uterus during early placentation: phenotypic and morphologic characterization of the \nCD56++ population. Dev. Immunol. 1, 169–190 (1991). \n48. Lessin, D. L., Hunt, J. S., King, C. R. & Wood, G. W. Antigen Expression by Cells Near the Maternal -\nFetal Interface. Am. J. Reprod. Immunol. Microbiol. 16, 1–7 (1988). \n49. Riccio, L. da G. C. et al. Immunology of endometriosis. Best Pract. Res. Clin. Obstet. Gynaecol.  50, \n39–49 (2018). \n50. Ashkar, Ali. A. & Croy, B. A. Interferon-γ Contributes to the Normalcy of Murine Pregnancy1. Biol. \nReprod. 61, 493–502 (1999). \n51. Ashkar, A. A., Di Santo, J. P. & Croy, B. A. Interferon γ Contributes to Initiation of Uterine Vascular \nModification, Decidual Integrity, and Uterine Natural Killer Cell Maturation during Normal Murine \nPregnancy. J. Exp. Med. 192, 259–270 (2000). \n52. Matikainen, S. et al. IFN-α and IL-18 synergistically enhance IFN -γ production in human NK cells: \ndifferential regulation of Stat4 activation and IFN -γ gene expression by IFN -α and IL -12. Eur. J. \nImmunol. 31, 2236–2245 (2001). \n53. Riccio, L. G. C., Baracat, E. C., Chapron, C., Batteux, F. & Abrão, M. S. The role of the B lymphocytes \nin endometriosis: A systematic review. J. Reprod. Immunol. 123, 29–34 (2017). \n \n.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 2, 2026. ; https://doi.org/10.64898/2026.01.02.695166doi: bioRxiv preprint","source_license":"CC0","license_restricted":false}