Transcriptomic changes in eutopic endometrium and ectopic lesions during endometriosis progression in a mouse model

Previously we reported a novel mouse endometriosis model with subfertility using Pgr cre/+ Rosa26 mTmG/+ mice( 25 ). In this model, the established peritoneal endometriotic lesions showed comparable histology to human disease. At 1M after disease establishment, the mice were fertile, but at 3M after disease establishment, the mice demonstrated severely reduced fertility due to defects of implantation and decidualization( 25 ). In order to identify the molecular mechanisms of the endometriosis progression and its associated infertility, we performed RNA-seq analysis of the eutopic endometrium and ectopic lesions from mice at 1M or 3M after endometriosis induction, plus the endometrium of age-matched sham-surgery mice, used as controls ( Fig. 1A ). We collected the sham, eutopic endometrium and ectopic lesions at GD 3.5. Hierarchy cluster analysis of all the RNA-seq samples indicated that 1M and 3M ectopic lesions had similar gene expression patterns which were quite different from the eutopic and sham endometrium ( Fig. 1B ). In comparison with the eutopic endometrium, we identified 2,487 DEGs in the 1M ectopic lesions from fertile mice and 5,185 DEGs in the 3M ectopic lesions from subfertile mice ( Table S2 , 3 ). Previously we reported that the histologic appearance of the ectopic lesions in our mouse model is very similar to human endometriotic lesions( 25 ). T score analysis( 47 ) uses the gene signature from our 1M and 3M mouse ectopic lesions and gene expression profile from the published human endometriosis datasets( 46 ) to predict the similarity of between mouse and human data. A positive T score indicates the mouse and human data had similar gene expression patterns. Since the human dataset contains the transcriptome from different endometriosis lesion types, we compared the mouse gene signature with the different types of human lesions. Here we would like to emphasize that due to the ethical issues, it is very hard to collect human endometrium during early pregnancy. We actually compared our pregnant mice with non-pregnant human. Despite of the significant variation of immune and hormone environments between pregnancy and non-pregnancy, we found that gene-expression patterns in the ectopic lesions from mice showed a positive T score with the transcriptome in human peritoneal lesions, deep-infiltrating lesions, and ovarian endometriomas, but a negative T score with the healthy and eutopic human endometrium ( Fig. 1C ) suggesting mouse ectopic lesion share similar expression pattern with several types of human ectopic lesions regardless of the pregnancy status. Furthermore, ectopic lesion at 3M consistently had higher positive T score compared to ectopic lesions at 1M suggesting 3M ectopic lesions were more similar to the lesions in human datasets ( Fig. 1C ). These results indicated that ectopic lesions from mice recapitulated the molecular changes of human endometriotic tissues. In comparison to the age matched eutopic endometrium, 2,093 DEGs were identified in both 1M and 3M ectopic lesions and, notably, differential expression were changed in the same directions, implying conserved gene networks during endometriosis progression ( Fig. 2A ). Pathway analysis of these conserved DEGs identified multiple widely accepted endometriosis-related pathways. Enhanced estrogen signaling and deceased and altered progesterone signaling (progesterone resistance) has been regarded as a hallmark of human disease, strongly impacting disease pathophysiology( 1 ). These changes in steroid receptor signaling were mirrored in our mouse model ( Fig. 2B ). Similar to human( 48 , 49 ), enhanced estrogen signaling is the result of inhibited Estrogen receptor α (ESR1) and activated Estrogen receptor β (ESR2) signaling ( Fig. 2B ). Several other human endometriosis relevant pathways( 50 – 53 ) including immune signaling associated with macrophage activation, enhanced T cell and B cell receptor signaling, angiogenesis associated with increased HIF1α, angiotensin, and endothelin signaling, fibrosis-related wound healing and myocyte hypertrophy, complement and coagulation cascade members were all increased in our mouse lesions. PI3K/MAPK kinases, which have been proposed as potential non-hormonal therapeutic targets for treating endometriosis( 54 ) which were also enhanced in mouse ectopic lesion. The hypertrophic neuron growth and pronociceptive factors in mouse ectopic lesion was consistent with the elevated innervation in human endometriotic lesions( 55 ) ( Fig. 2C , Table S4 ). Based on the conserved DEGs in 1M and 3M ectopic lesion, potential upstream signaling regulators were identified ( Fig. 2D , Table S5 ) and included pro-inflammatory cytokines such as TNF, IL33, TGFB1, and IL21; Increased CFB, Complement Factor B and complement components C3, the central component in complement activation, and the C5AR1, Complement C5a Receptor 1,all suggest increased inflammatory processes, while PI3 kinase p110 delta protein, PIK3CD is strongly associated with kinase activation; plasminogen activator, PLAU, and its receptor, PLAUR, and Matrix Metallopeptidase MMP13 are important for tissue remodeling and thus may play roles in fibrosis. Additionally, brain derived neurotrophic factor (BDNF), nitric oxide synthase, NOS2, can play important roles in promoting neurogenesis and angiogenesis associated with endometriosis pathophysiology. In order to gain insight into the change in fertility between 1M and 3M and to understand molecular change during disease progression, we identified the DEGs between 1M and 3M in the ectopic lesions. There are no obvious morphological changes from 1M to 3M after endometriosis induction between in the ectopic lesions. However, 3,093 DEGs were distinct to the 3M timepoint, indicating a significant transcriptomic difference between ectopic lesions during endometriosis progression, associated with acquired subfertility. Pathway analysis of the 3M unique DEG identified activated metabolism signaling associated with cholesterol, glucose, and estrogen; myocyte signaling related to cardiac hypertrophy, oxytocin, MAPK and stem cell pluripotency and inhibited sonic hedgehog signaling ( Fig. 2E , Table S6 ). Analysis of potential upstream regulators ( Fig. 2F , Table S7 ) identified cholesterol-steroid metabolism, luteinizing hormone receptor LHCGR, steroidogenic factor NR5A1, leptin LEP were all activated; for myocyte, muscle specific transcription factor MYOD1, MYOG, and oxytocin receptor OXTR were all stimulated; The expression of stem cell markers such as RUNX2, TP63, PAX3 were all increased. All these results suggested an endometriosis progression from 1M to 3M ectopic at molecular levels. Endometriosis is characterized by the ectopic focal inflammation( 50 ). We found that multiple immune pathways are activated in our mouse ectopic lesions compared to the eutopic endometrium at both 1M and 3M ( Fig. 2C ). A gene expression heat map further showed enriched dendritic cell, macrophage, T and B cell markers in the 1M and 3M mouse ectopic lesions compared to the eutopic endometrium ( Fig. 3A ), which suggested the abnormal accumulation of multiple immune cells in the ectopic lesions. In order to confirm the gene expression data, we performed immunohistochemistry using antibodies against CD11c( 56 ), F4/80( 57 ), CD3ε( 58 ) and CD19( 59 ) as markers for dendritic cells, macrophage, T cell and B cell, respectively. As expected, we found abundant membrane and cytoplasmic staining of CD11c and CD19 in spleen and F4/80 staining in liver, and cytoplasmic staining of CD3ε in spleen, while much lower expression of CD11c, CD19, CD3ε in liver and F4/80 in spleen which is consistent with previous publications( 60 – 66 ) ( Fig. S1 ). We found ( Fig 3B ,) that CD11c, CD3 and CD19 positive cells were rarely detected in the sham and eutopic endometrium but abundantly present in multiple regions of 1M and 3M ectopic lesions. Further, sham and eutopic endometrium contained F4/80 positive cells, but these were increased in several regions of the ectopic tissues. Therefore, the increased infiltration of immune cells was detected in the 1M and 3M ectopic lesions, paralleling prior studies in human lesions showing increased immune cell infiltration( 67 , 68 ). The molecular mechanism of endometriosis associated fertility remains largely unknown in which uterine defects play one of the important roles. Our endometriosis mouse models showed severe subfertility 3M after endometriosis induction in contrast to the normal fertility seen in the control mice and those 1M after endometriosis induction( 25 ). Further studies revealed that the reduced fertility in 3M endometriosis mice was directly caused by uterine decidualization defects( 25 ). To understand endometriosis induced pregnancy related uterine defects, we compared the transcriptome of eutopic with sham endometrium at pregnancy D3.5. The PCA plot and hierarchy cluster heatmap, we noticed all 1M sham (n=2), 1M eutopic (n=4), 3M control (n=3) and 2 of 4 3M eutopic samples were clustered together. Separately, one 3M sham, two 1M sham, and two 3M eutopic samples were spread out ( Fig. S2A , B ). The variable fertility in endometriosis mice( 25 ) and sham mice might explain the within-group transcriptomic differences. In accordance with the normal fertility at 1M and severe subfertility at 3M, there was very limited overlap between 1M and 3M DEGs, in which 92% (650/608) DEGs of 3M eutopic endometrium were only altered at 3M not 1M ( Fig. S2C , Table S8 , 9 ). Analysis of upstream regulators and pathways revealed activation of ESR1 and IL21 signaling as well as suppression of Wnt-β/catenin signaling CTNNB1, TGFβ signaling BMP4, and EGF ( Fig. S2D , Table S10 ), which play crucial roles in embryo implantation( 69 – 71 ). Since immune and EGF pathways were altered in the eutopic endometrium, we also performed qPCR to validate the reduced expression of the cytokine, IL15, the cytokine receptor Il20rb, the increased expression of immune and EGF related Serpina3n, and the EGF regulator Egfl6, in the 3M eutopic compared to sham uterus ( Fig. S2E ). In our endometriosis models, we found that the presence of endometrial tissues outside of the uterus severely disrupted uterine functions during pregnancy( 25 ). A long-standing question in endometriosis research is, how can ectopic lesions disrupt eutopic endometrial functions? We hypothesized interactions between ectopic lesion ligands and eutopic endometrial receptors may play a role. Although elevated inflammatory mediators, inflammatory cells, and growth factors ( Fig. 2 ) may play a systemic role, it is likely that higher levels of these factors in the peritoneal fluid could have an increased role in the pelvic region. Additionally, we found the steroid synthetic enzymes, especially for progesterone and estrogen, were increased in the 3M ectopic lesions which may increase the local estrogen and possibly progesterone concentrations, though progesterone is metabolized to estrogen ( Fig. 4A ). Meanwhile, the receptors of estrogen and progesterone, ESR1 and PGR, maintained high levels of expression in the eutopic uterus. Therefore, a hyper-estrogen (and possibly progesterone) environment in the eutopic endometrium may be induced by the ectopic lesions ( Fig. 4A ). Although peritoneal endometriotic lesions cannot physically contact the eutopic endometrium, the peritoneal fluid may circulate some ligand-receptor pairs between ectopic and eutopic endometrium to mediate the paracrine interactions. To quantify the ligand receptor interactions, we set the sham endometrium as the base line and calculated the expression fold changes of the ligand and receptor in the eutopic and ectopic endometrium. Eventually, we found a great variety of ligand and receptor mRNA were altered in the ectopic lesions (ligand: 220 (1M), 289 (3M); receptor: 203 (1M), 309 (3M), Fig. S3A ), while much fewer were changed in the eutopic endometrium (ligand: 15 (1M), 31 (3M); receptor: 7 (1M), 31 (3M), Fig. S3A ). Among them, more than 50% ligands and receptors were upregulated in the ectopic lesion, but less than 50% were upregulated in the eutopic endometrium ( Fig. S3B ). Because of the abundantly altered ligand-receptor expression, the interactions within the ectopic lesions were more frequently changed compared to the eutopic endometrium ( Fig. 4B ). Meanwhile, many secreted ligand including cytokines, growth factors, and lipoprotein etc. and their receptors were differentially expressed in the ectopic and eutopic endometrium, which may greatly affect the interactions between these two endometrial tissues ( Fig. 4C ).

All mice were maintained in the animal facility of the Institutional Animal Care and Use Committee of University of Missouri, and all animal experimental procedures were approved. Mice were bred on a mixed background of the mouse strains C57Bl/6 and SV129 under controlled humidity and temperature. Endometriosis was surgically induced as previously described( 25 ). Briefly, 8-weeks-old female Pgr cre/+ Rosa26 mTmG/+ , used as uterine tissue donors, were injected with 100ng 17β-estradiol (E2) for three consecutive days. One uterine horn of each donor mouse was cut into small fragments, mixed with 0.5ml 1X PBS and injected into the peritoneal cavity of a Rosa26 mTmG/+ recipient littermate. For the sham-surgery control group, 0.5ml 1X PBS, alone, was injected into the peritoneal cavity. One and three months after endometriosis induction, the endometriosis and sham mice were mated with C57BL/6 male mice. The morning of a vaginal plug was designated as day 0.5 of gestation (GD 0.5). Mice were sacrificed at GD 3.5. Rosa26 mTmG/+ mice have a double-fluorescent Cre reporter and express membrane-targeted tandem dimer Tomato (mT) prior to Cre-mediated excision, and express membrane-targeted green fluorescent proteins (mG) after excision( 40 ). While uteri and ectopic lesions with Pgr -positive uterine cells express mG in the Pgr cre/+ Rosa26 mTmG/+ mice, other tissues with only Pgr -negative cells express mT. Endometriosis-like lesions with mG in the mice were collected under a fluorescence microscope by dissection along with eutopic endometrium. Only endometrial tissues were collected from the sham mice. Samples were either fixed with 4% (vol/vol) paraformaldehyde for histology or immunohistochemistry or snap frozen and stored at −80 °C for RNA extraction. According to our previous report( 25 ), mice after 1-month (1M) endometriosis induction are fertile, while 63.6% mice after 3-month (3M) endometriosis induction were infertile. For each group, four 1M and 3M endometriosis mice along with four 1M and 3M sham (control) mice were used. Since the mice were dissected at GD3.5, the pregnancy status cannot be accurately predicted. A schematic diagram of this experiment is at Fig. 1A . The total RNA was isolated from each sample using the RNeasy Total RNA Isolation Kit (Qiagen) following the manufacture’s instructions. RNA purity and concentration were assessed using UV spectrospoy with a NanoDrop ™ instrument. All RNA samples were further analyzed by the Van Andel Institute Genomics Core to confirm sample concentration and purity (RIN > 8.0; concentration 100–200ng/ul) before RNA sequencing. Multiple ectopic lesions from the same mouse were pooled together as one sample for RNA isolation. For RNA sequencing, libraries were prepared from 500 ng of total RNA using the KAPA mRNA Hyperprep kit (Kapa Biosystems, Wilmington, MA USA). RNA was sheared to 300–400 bp. Prior to PCR amplification, cDNA fragments were ligated to IDT for Illumina TruSeq UD Indexed adapters (Illumina Inc, San Diego CA, USA). Quality and quantity of the finished libraries were assessed using a combination of Agilent DNA High Sensitivity chip (Agilent Technologies, Inc.) and QuantiFluor ® dsDNA System (Promega Corp., Madison, WI, USA). Individually indexed libraries were pooled and 50 bp, paired end sequencing was performed on an Illumina NovaSeq6000 sequencer using an S2, 100 bp sequencing kit (Illumina Inc., San Diego, CA, USA) to an average depth of 50 M reads per sample. Base calling was done by Illumina RTA3 and output of NCS was demultiplexed and converted to FastQ format with Illumina Bcl2fastq software v1.9.0. The raw reads were initially filtered by removing low quality reads (average quality scores < 20) and aligned to the mm39 genome and assembled using Star 2.7.9a( 41 ). Expression values of RNA-Seq were expressed as counts. Differentially expressed genes (DEGs) were identified by the “exact negative binomial test” in the EdgeR software package( 42 ). The DEGs between 1M eutopic and 1M ectopic tissues or between 3M eutopic and 3M ectopic tissues were set as max counts >50, fold change of >2 (up-regulated) or <−2 (down-regulated), False Discovery Rate (FDR) < 0.05. Because the great within-group variation in sham and eutopic groups, a less stringent threshold was set for the DEGs between 1M sham and 1M eutopic tissues or between 3M sham and 3M eutopic tissues: max counts >50, fold change of >1.5 (up-regulated) or <−1.5 (down-regulated), p value <0.05. A Principle Component Analysis (PCA) map were generated by plotMDS in EdgeR( 42 ). The expressions of DEGs were normalized, scaled, clustered to generate a heatmap using the ComplexHeatmap package( 43 ). The significantly enriched pathways and upstream regulators based on the DEGs were identified using Ingenuity pathway analysis (IPA, Qiagen). The dotplot was generated by Tidyverse package( 44 ). The full list of mouse ligand-receptor pairs were downloaded from CelltalkDB( 45 ) to compare with the mouse DEG lists. The signatures of mouse ectopic lesion at 1 or 3M were determined as the DEGs between ectopic lesions and eutopic endometrium with fold change of >2 (High) or <−2 (Low). The correlations of the gene signatures with the published human endometriosis dataset GSE141549 ( 46 ) were predicted by T score using the SEMIP R package ( 47 ) Total 1ug of RNA was reverse transcribed to cDNA using M-MLV reverse transcriptase (Invitrogen) according to the manufacturer’s instructions. Quantitative real time PCR was performed with SYBR green and specific primers to examine the relative expression of genes of interest. Experimental gene expression data were normalized against the constitutively expressed gene, Rpl7 ribosomal RNA. Table S1 shows the primer sequences used. Immunohistochemistry analysis was performed as previously described( 25 ). Briefly, dewaxed hydrated paraffin-embedded tissue sections were blocking by 10% normal goat serum and then the slides were incubated with F4/80 (cs-70076, 1:250), CD3ε (cs-78588, 1:200), CD11c (cs-97585, 1:300), CD19 (cs-90176, 1:800) antibody at 4°C overnight. The slides were incubated with biotinylated goat anti-rabbit secondary antibody (BA-1000, 1:500 Vector Laboratories) for 1h, Horseradish Peroxidase (HRP) antibody (434323, 1:1000, Fisher Scientific) for 1h, and the immune signals were stained by DAB substrate (SK-4100, Vector Laboratories). The slides were counter-stained with hematoxylin. The endometrial tissues from three different mice for each group were sectioned for immunohistochemistry. For eutopic and sham endometrium at 1M or 3M, one cross section from each endometrium was evaluated at high-powered fields for immune marker staining. Because our endometriosis mice developed various number of ectopic lesions( 25 ), 1–8 cross sections of ectopic lesions per mouse were evaluated at high-powered fields. Mouse liver and spleen has been used as positive and negative control. RL performed the staining and analyzed the images. Student’s t test with unequal variance was used to compare the real-time PCR and T score. A p value of <0.05 was defined, a priori, as significant. The RNA-Seq analysis data generated in this study have been deposited in the NCBI Gene Expression Omnibus database under accession code GSE240392 .

In this study, we performed transcriptomic analysis in ectopic lesions and eutopic endometrial tissues from both fertile and subfertile mice with endometriosis. We identified the positive correlation of the gene signatures between the mouse and human in ectopic lesions. Conserved gene networks were activated in all the ectopic lesions including estradiol, immune, fibrosis, and angiogenesis pathways. The interactions mediated through hormone, cytokine, and growth factor as well as their corresponding receptors were predicted between the ectopic and eutopic endometrium. EGF and WNT signaling were more suppressed in the eutopic endometrium from subfertile mice. Our results revealed that our mouse endometriosis model recapitulates the important transcriptomic changes of endometriosis progression in human ectopic lesions including the essential regulator network and intensive inter-communications between ectopic and eutopic endometrium. Our preclinical animal model for endometriosis will be invaluable to understand etiology and pathophysiology on endometriosis.

Endometriosis is a chronic inflammatory disease that involves both the peripheral immune system and endometrial tissues( 72 ). Elevated inflammation has been reported in the transcriptomic analysis of human endometriotic tissues( 10 , 15 ). Recent advancement of scRNA-seq and spatial transcriptome further emphasized the pro-inflammatory microenvironment in different types of endometriotic cells ( 16 – 20 ). Normal endometrium also contains abundant immune cells to support the uterine remodeling during menstrual cycle and pregnancy( 73 ). The dysregulated immune cells are central to the pathophysiology of endometriosis( 72 ). Macrophages and dendritic cells are increased in the ectopic lesions and the peritoneal fluid of women with endometriosis to promote the organogenesis, angiogenesis and neurogenesis of ectopic lesion and may also contribute to the infertility and pelvic pain( 72 , 74 – 77 ). There are many contradictory results about the density of different types of T cell and B cell in the eutopic and ectopic endometrial tissues, peritoneal fluid, and peripheral blood in which reduce, increase or non-change has all been reported which may affect the immune tolerance( 72 , 73 ). Our endometriosis mouse model also showed activated immune signaling and abundant expression of immune cell markers in the ectopic lesions, including genes associated with macrophage, dendritic and T cell, etc. More interestingly, we noticed the clonal immune cell accumulation in some regions of ectopic lesion suggesting the heterogeneous immune infiltration in the endometriotic tissues. The tissue heterogeneity provides one explanation for the conflicting immune results in human endometrial samples. Therefore, a systematic study using spatial transcriptome will be a better method to reveal the complex immune cell distribution and its crosstalk with adjacent microenvironment in endometriotic tissues. Endometrium is tightly regulated by estrogen and progesterone through their receptors ESRs and PGR. Disrupted steroid signaling causes multiple uterine diseases. Among them, progesterone resistance and estrogen dominance is the hallmark of endometriosis( 1 ). The reduced PGR expression, especially the undetectable PGR-B isoform, may cause the progesterone resistance( 78 , 79 ). It has been reported that ESR1 and ESR2 were both expressed in the endometrium, while a switch from ESR1 to ESR2 dominancy was observed in endometriotic tissues( 80 ). Besides, the steroid ligand, estrogen and progesterone levels are both increased in the endometriotic tissues due to the elevated expression of steroidogenesis enzymes( 81 , 82 ). The higher estrogen levels and ESR2 expression may contribute to the estrogen dominance. However, since many of these results were obtained from the human ovarian endometrioma, the confounding factor from ovarian tissues cannot be fully ruled out. In our endometriosis mouse model, the ectopic lesions has only been observed in the pelvic not the ovarian locations. But the steroidogenesis enzymes and Esr2 expression were increased while Esr1 and Pgr were decreased in the ectopic lesions confirming that these are conserved characteristics in endometriotic lesions. We used the GD3.5 mouse uterus instead of the non-pregnant uterus based on three major reasons. First, we used uterine samples from mice at GD3.5 since they have similar ovarian steroid conditions as the early secretory phase of humans. A key aspect of endometrial biology is the elaborate crosstalk that occurs between these tissue compartments during early pregnancy via paracrine factors regulated by progesterone and estrogen. Dysregulating this crosstalk often leads to early pregnancy loss( 83 – 87 ). The main differences in gene expression between normal and eutopic endometrium from women with endometriosis is predominantly in the early secretory phase( 14 , 88 ). Therefore, we used eutopic and ectopic endometrial tissues to identify gene pathways and networks that differ among fertile control and subfertile mice with endometriosis at GD 3.5. Second, the pregnancy events and hormone environments during early pregnancy are well-controlled. In mice, the ovulatory estrogen surge at GD0.5 was followed by increased progesterone levels and one nidatory estrogen at GD3.5 ( 89 , 90 ). Therefore, GD3.5 will be a good timepoint to check both progesterone and estrogen responses of the eutopic endometrium and ectopic lesions. Although the length of pregnancy is much longer in humans than mice, the hormone regulations during early pregnancy shares similar patterns between humans and mice( 39 ). Third, it is difficult to interpret the results from non-pregnant mice because non-pregnant mice continuously undergo variable estrous cycles that last 4–5 days. More importantly, mice do not have a progesterone dominance during the estrous stage that is comparable to the human secretory phase. Endometriosis disrupts coordinated progesterone responses in the endometrium. This progesterone resistance results in an inadequate antagonism of estrogen action, increased inflammation, and inadequate differentiation of the stroma and remodeling of the endometrium, all of which can lead to a non-receptive endometrium for embryo implantation( 84 ). However, the mechanism of progesterone resistance in endometriosis is not fully understood. Therefore, GD3.5 with high progesterone levels could be a better time point to study the progesterone resistance mechanism in endometriosis mouse model. One limitation of this study is there are great within-group variations in the sham and eutopic groups. We previously reported that our endometriosis mice displayed a range of fertility from normal to infertile, which is similar to that seen in women with endometriosis( 25 ). 63.6% of mice with endometriosis at 3M had implantation failure which were not correlated with the number or mass of the ectopic lesions ( 25 ). Therefore, some mice have non-receptive endometrium and others have receptive endometrium at pre-implantation stage GD3.5. We speculate that these different endometrial conditions result in variations of transcriptomics within the group in eutopic and sham endometrium. Consequently, much less DEGs were detected with statistically lower significance in the eutopic compared to the sham endometrium. While this subfertility phenotype of our animal model mimics to endometriosis-associated infertility in some women with endometriosis, there is a limitation for identification of dysregulated genes and pathways in mixture of non-receptive and receptive endometrium from mice with endometriosis. Furthermore, the identified DEGs could be indirect effects of genetic, hormonal and immune system changes associated with conception. Future studies will include an increased number of endometriosis mice, unsupervised clustering, and collecting endometrial tissues at later pregnancy stages when fertility defects are more obvious as these may improve our understanding of the heterogeneous endometrial transcriptome between fertile or infertile eutopic endometrium. The uterus is supplied with various types of neurons including sensory, postganglionic sympathetic and parasympathetic neurons( 91 ). Multiple neurotrophic and neuro growth factors, (including brain-derived neurotrophic factor (BDNF)( 92 ), neuropeptide Y (NPY)( 93 ), and nerve growth factor (NGF)( 94 )) are synthesized in situ in the endometrium. They have a role for neurogenesis as well as modulation of epithelium proliferation, muscle contraction and prostaglandin production. The nerve fibers are dense in myometrium and reach the functional layers in the fetus uterus( 95 ). However, they are confined in the basal layer of adult endometrium. Furthermore, the nerve fibers are detected in the functional layer of women with endometriosis( 96 ) or with pain symptoms( 97 ). There are more nerve fibers in ectopic lesions than peritoneum( 98 ), which may sensitize 73% women with endometriosis toward the pelvic pain( 4 ). In addition to the pain, the innervated neurons interact with immune cells, blood vessels and endometrial cells to promote the inflammation and angiogenesis in endometriotic tissues( 55 ). Right now, there are no long-term relief of endometriosis-associated pain treatment without adverse effects on fertility( 99 ). Our endometriosis mouse model is manifested with neurotrophic factors, cytokines and vascular growth factors in the ectopic lesions implying its application in improving research of endometriosis-associated pain.

Endometriosis is defined as the presence of endometrial-like tissues outside of the uterus. It affects about 10% of women at reproductive age and is associated with both impaired fertility and pelvic pain( 1 ). Up to 30–50% of women with endometriosis are infertile( 2 ) which has been attributed to the distortion of pelvic anatomy, non-receptive endometrium, compromised folliculogenesis and ovulation( 3 ). As many as 70% of women with endometriosis experience abdominopelvic pain( 4 ) secondary to elevated inflammation and altered innervation( 5 ). However, the etiological and pathophysiological mechanism underlying the development of endometriosis and its symptoms remains poorly understood. Furthermore, there are no established diagnostic biomarkers, especially for the early stages of endometriosis. A history of menstrual symptoms and chronic pelvic pain provides the basis for suspecting endometriosis. No individual serum marker has yet been shown to be both sensitive and specific for diagnosis or monitoring of endometriosis. Although several screening tools and tests have been proposed and tested, none are currently validated to accurately identify or predict individuals or populations that are most likely to have the disease. Ovarian endometrioma, adhesions and deep nodular forms of disease often require ultrasonography or magnetic resonance imaging (MRI) to detect( 1 ). Histologic verification, usually following surgical/laparoscopic visualization, can be useful in confirming diagnosis, particularly for the most common superficial lesions( 6 ). MicroRNAs play crucial roles in endometriosis and circulating MicroRNAs have the potential to diagnose the endometriosis at early stage( 7 , 8 ). Nanoparticles accumulate at the endometriotic lesions which have both diagnosis and therapy potentials( 9 ). Therefore, further research is needed to develop non-invasive tests for diagnosis of endometriosis. Current treatment includes suppression of estrogen action or surgery, both associated with substantial side effects and a high recurrence rate( 1 ). Considering the great economic, physical, and psychological burdens of endometriosis on individuals and society, studies of molecular mechanisms are imperative. A number of whole-tissue transcriptome studies have utilized DNA microarray and RNA sequencing to better understand the differences in signaling pathways between eutopic and ectopic endometrium in women with endometriosis as compared to endometrium of healthy women. Endometriosis is characterized by altered immune, angiogenic, and neurogenic processes( 1 ). Hall and coworkers analyzed paired eutopic and ectopic lesions from nine patients by microarray and identified a gene expression signature consistent with increased activity of inflammatory pathways( 10 ). Crispi et al. used microarray analysis to compare eight samples of paired eutopic and ectopic endometrial tissues together with six samples of control endometrium from healthy women in the proliferative phase. These investigators found that a significant alteration in expression of genes involved in organogenesis in the ectopic tissues( 11 ). Another study using microarray compared gene expression in the eutopic and ectopic endometrial samples of 18 fertile women with ovarian (stage 3 to 4) endometriosis( 12 ). Both proliferative and secretory stage samples were included( 12 ). These investigators found that immune, cell remodeling, neurophysiological pathways were affected in endometriotic tissues( 12 ). Another genome-wide study of gene expression used RNA-sequencing to evaluate eutopic and ectopic lesions from ten women with endometriosis and endometrium from nine healthy women( 13 ). These investigators detected evidence of altered extracellular matrix, complement, and coagulation pathways in women with endometriosis( 13 ). Microarray analysis compared eutopic endometrial tissue from 21 women with moderate to severe endometriosis to 16 women without endometriosis and identified a subset of progesterone regulated genes that are altered, suggesting progesterone resistance( 14 ). The largest microarray analysis, including 148 endometrial samples from women with or without endometriosis, demonstrated immune activation, steroid and growth factor signaling disruption in the eutopic endometrium of women with endometriosis compared to controls( 15 ). More recently, single cell RNA-sequencing and spatial transcriptomic analysis has provided further clues about cellular and spatial changes occurring in the presence of endometriosis( 16 – 20 ). Despite these efforts, perhaps due to differences in cycle phase, between-individual, between-lesion, and study design heterogeneity, no conserved endometriosis biomarker genes have yet been found. Therefore, we have focused on translational animal model with well-controlled experimental conditions that can potentially provide valuable insight into the effects of endometriosis on disease phenotype, gene expression, and molecular pathways vital to its progression( 21 ). Nonhuman primates such as the baboon and rhesus macaques have proven to be good translational animal models that share similar reproductive physiology and anatomy to humans, including menstruation and naturally occurring endometriosis ( 22 , 23 ). However, the long-term disease development process, costly animal husbandry and ethical concerns limit non-human primate usage( 21 ). Therefore, the mouse has been widely used as the preclinical model for biomedical research due to its low costs and easy manipulation( 24 ). Recent advances in the murine model of endometriosis enhance both its feasibility and translational potential. Endometriosis can be induced in the mouse by intraperitoneal injection, surgical engraftment and subcutaneous placement of donor endometrial tissues from littermate mice or women with endometriosis( 24 ). Fluorescent protein-labeled endometrium has been used to enhance feasibility of detection and quantitation of endometriotic lesions( 24 ). We recently developed a mouse endometriosis model with a fluorescence reporter using Pgr cre/+ Rosa26 mTmG/+ that seems to enhance translational potential by mimicking both the ectopic morphology and fertility defects seen in women with endometriosis( 25 ). As we observed in human endometriosis, we did not observe any morphological changes of the ectopic lesions between one month and three months after endometriosis induction. One month after endometriosis induction, the endometriosis mice show comparable embryo implantation sites at gestation day (GD) 7.5( 25 ). However, three months after endometriosis induction, 63.6% mice with endometriosis revealed implantation failure( 25 ). Interestingly, there is a time lag between lesion establishment and infertility in the mouse model, likely similar to, but much faster than that of human disease suggesting an endometriosis progression that is positively associated with infertility. In early pregnancy the endometrium undergoes changes coordinated by the ovarian hormones progesterone (P4) and estrogen (E2)( 26 – 29 ). In rodents, E2 stimulates proliferation of uterine epithelial cells on gestation day (GD) 0.5. Upon formation of the corpus luteum (GD 2.5), P4 inhibits E2-mediated proliferation of uterine epithelial cells and leads to uterine stromal cell proliferation( 30 ). On GD 3.5, an acute E2 spike in concert with elevated P4 levels further stimulates uterine stromal proliferation and induces uterine glands to produce leukemia inhibitory factor (LIF), which is critical for uterine receptivity( 31 ). Implantation begins at GD 4.5 with attachment of the blastocyst to the luminal epithelium, followed by invasion of the embryonic trophoblasts( 32 ). Upon invasion, uterine stromal cells decidualize (a cellular transformation characterized by morphological changes and development of a secretory phenotype), which creates a local environment supportive of the implanting embryo( 33 , 34 ). In humans, the endometrium is also tightly regulated by P4 and E2, including proliferation, secretion, and menstrual shedding. The endometrium becomes receptive for implantation during the mid-secretory phase, and decidualized stromal cells are seen in the late-secretory phase( 35 , 36 ). Therefore GD3.5 will be a good timepoint to check both estrogen and progesterone responses of eutopic endometrium and ectopic lesions( 37 , 38 ). More importantly, although the length of pregnancy is much longer in humans than mouse, the hormone regulations during early pregnancy shares similar patterns between human and mice( 39 ). Therefore, we undertook a transcriptomic analysis of the murine model at GD3.5 and its temporal progression allowing analysis of changes relevant to fertility and compared these transcriptomes to those obtained from women to better understand the translational potential of our mouse model and to uncover information relevant to human disease.