Deciphering estrogen receptor alpha-driven transcription in human endometrial stromal cells via transcriptome, cistrome, and integration with chromatin landscape.

To overcome the limited responsiveness of endometrial stromal cells to E2, we activated ESR1 in THESC using a dCas9-VPR plasmid with a blasticidin resistance gene ( Figure 1A ). RT-qPCR confirmed successful transduction and expression of dCas9 in comparison to untransduced THESC; these cells will be referred to as THESC dCas9-VPR from this point forward ( Figure 1B ). Due to a reduced expression of dCas9 after 8 days without selection, we maintained THESC dCas9-VPR in media always containing blasticidin to ensure sustained expression ( Figure 1B ). To activate ESR1, we designed seven different gRNAs (ESR1–1 to ESR1–7) to target alternate ESR1 transcription start sites (TSS) annotated in the NCBI Refseq database ( Figure 1C ). RT-qPCR and western blot showed that the ESR1–3 gRNA can successfully activate ESR1 in comparison to the non-targeting control gRNA (NT gRNA) ( Figure 1D , S1 ). These cells will be referred to as THESC ESR1 and THESC NT from this point forward. We also show that dCas9-VPR and ESR1–3 gRNA can successfully activate ESR1 in myometrial cells (hTERT-HM) and a second endometrial stromal cell-line (H1644) ( Figure S2A - B ), highlighting the utility of this gRNA to active ESR1 expression across different cell-lines. To assess ESR1 functionality, we treated THESC ESR1 and THESC NT with 10 nM E2 for 24 h and measured the expression of known ESR1/E2 target genes, including the progesterone receptor gene (PGR) and Inositol polyphosphate-4-phosphatase type II (INPP4B) ( 20 ). By RT-qPCR, we observed an increase of PGR mRNA expression in response to E2, and an increase in INPP4B mRNA expression following ESR1 activation ( Figure 1E ). These results confirm that ESR1 can successfully be activated using the ESR1–3 gRNA and that the cells are ESR1/E2 responsive. To identify ESR1 E2-dependent and -independent target genes, THESC ESR1 and THESC NT were treated with 10 nM E2 or 0.01% ethanol (vehicle) for 24 h and bulk RNA-seq was conducted to observe changes in gene expression. Hierarchical Clustering and Multidimensional Scaling (MDS) plot revealed distinct clustering of THESC ESR1 and THESC NT in the presence and absence of E2, while THESC NT clustered together regardless of treatment. These results suggest that ESR1 regulates gene expression in ligand dependent and independent manners, and that ESR1 activation through CRISPRa can restore E2 responsiveness ( Figure S3A - B ). ESR1 transcriptomic regulation was separated into E2-independent and E2-dependent ESR1 and differentially expressed genes (DEGs) were identified as genes with absolute fold-change (FC) > 1.3 and adjusted p-value < 0.05 ( Figure 2A ). E2-independent DEGs were identified by comparing gene expression in vehicle-treated THESC ESR1 and THESC NT , yielding a total of 305 DEGs. Among these, ESR1 was the top activated gene (FC > 637). ( Figure 2B ; Dataset 1 ). IPA pathway analysis indicated an enrichment of pathways related to cancer, inflammation, proliferation, migration, angiogenesis, preeclampsia, and extracellular matrix organization ( Figure 2C ; Dataset 1 ). E2-dependent DEGs were identified by comparing gene expression in E2-treated and vehicle-treated THESC ESR1 , yielding a total of 369 DEGs ( Figure 2D ; Dataset 2 ). The long non-coding RNA LINC01016 was the top activated gene (FC > 37; not shown in figure because out of scale); this gene is associated with endometrial cancer ( 40 ) and breast cancer and is a direct target of ESR1 in MCF7 and T47D breast cancer cells ( 41 ). IPA revealed enrichment of pathways related to cancer, inflammation, axonal guidance, metabolism, wound healing, and Wnt/β-catenin signaling ( Figure 2E ; Dataset 2 ). To evaluate the relevance of these RNA-seq datasets in relation to in vivo human tissue, DEG lists were compared to active genes (FPKM > 1) in human endometrial tissue during the proliferative stage ( GSE132713 ), the stage when estrogen signaling is dominant ( 29 ). Overall, 72.9% and 72.1% of the E2-dependent and independent DEGs, respectively, were active in the human endometrium during the proliferative stage, supporting the significance of the genes identified using engineered THESC ( Figures 2F ; Dataset 3 ). All together, these results suggest that ESR1 regulates gene expression in both the presence and absence of E2, and triggers gene expression changes involved in endometrial stromal cell biology and function. To investigate the ESR1 cistrome, we performed ESR1 Cut&Run in THESC ESR1 treated with 10 nM E2 or vehicle for 1, 3, or 6 h ( Figure 3A ). ESR1 had the greatest number of binding peaks at the 3-h timepoint (1305), encompassing nearly all the peaks from the 1 and 6 h E2 timepoints ( Figure 3B ). At 3 h, 97% of ESR1 peaks without E2 were also present with E2 ( Figure 3C ), consistent with previous findings showing minimal positional differences in ESR1 binding in the presence and absence of E2 in the mouse uterus ( 42 ). Thus, we decided to focus on ESR1 peaks after treatment with E2 for 3 h. Most peaks (54.1%) were located distal to genes (>25 kb from the TSS) and within intergenic regions ( Figure 3D ), consistent with previous studies showing that ESR1 binds primarily to distal cis-regulatory elements ( 43 ). 58% of ESR1 peaks in THESC overlapped with ESR1 ChIP-seq peaks from proliferative-stage human endometrium ( GSE200807 ) ( 19 ) ( Figure 3E ). Differences may reflect paracrine signaling between stromal and epithelial cells in biopsies that are absent in monoculture. Motif enrichment analysis using HOMER identified the estrogen response element (ERE), the canonical ESR1 binding motif, as the most enriched motif, along with motifs that bind Jun-AP1, RUNX, and TEAD ( Figure 3F ) which have been shown to mediate ESR1 genomic binding and transcriptional regulation in breast cancer cells ( 44 – 46 ). These results support previous findings that ESR1 binds to distal cis-regulatory elements and regulates gene expression through a tethering mechanism via protein-protein interactions ( 47 ). To identify genes regulated by ESR1/E2 with nearby ESR1 binding, ESR1 peaks were annotated to all genes whose TSS was within 100kb and these genes were overlapped with E2-dependent DEGs. 134 DEGs had at least one ESR1 binding site within 100kb of their TSS ( Figure 3G ). These genes were enriched for pathways including ERa genomic signaling, inflammatory response, vasculature development, the core matrisome, and homeostasis ( Table 4 ). These results support ESR1/E2’s role as a key regulator of stromal cell homeostasis and preparation for pregnancy. Chromatin conformation capture-based methods such as Hi-C and HiChIP have emerged as useful tools to identify distal genome contact points that form promoter-enhancer interactions and topologically associated domains (TADs) ( 48 ). To investigate the 3D chromatin landscape in hESC, we performed H3K27ac HiChIP in two primary hESC cell-lines treated with vehicle or a decidualization cocktail (EPC) for 72 h. We identified 150,966 and 193,367 loops in vehicle- and EPC-treated cells, respectively ( Figure S4 ; Dataset 5 ). Differential analysis (|FC| > 2 and p-value < 0.05) identified 1,421 EPC-repressed and 2,107 EPC-activated loops ( Figure 4A ; Dataset 5 ). We focused on the differential loops that had at least one anchor at a gene’s promoter (26% of EPC-repressed and 14% of EPC-activated loops; Figure 4B ) and overlapped these genes with DEGs in response to EPC from a published RNAseq dataset ( GSE205481 ) in two matched donor hESC, and a third replicate ( Dataset 6 ) ( 28 ). There were 141 DEGs in response to EPC treatment that also had a differential promoter-contacting loop in response to EPC ( Figure 4C ). Most EPC-activated DEGs were linked to EPC-activated loops (98%), and vice versa for repressed DEGs (93%) ( Figure 4D - E ). IPA pathway analysis highlighted enrichment in pro-inflammatory pathways, prolactin signaling, and integrin interactions ( Figure 4F ). These findings suggest that chromatin architecture reorganization supports gene expression changes driving decidualization and endometrial remodeling. Estrogen signaling has been shown to be critical for decidualization, but its specific role during the differentiation of hESC remains understudied. To identify genes regulated by E2 that may be involved in the regulation of decidualization, we integrated the EPC H3K27ac HiChIP loops, ESR1 binding sites, and ESR1/E2 transcriptome to identify ESR1/E2 target genes. We then overlapped these genes with genes that are differentially expressed in response to EPC. To identify ESR1/E2 target genes, we focused on genes that have an ESR1 binding peak at the promoter of the gene (proximal binding; 89 genes), or an ESR1 binding peak that is one HiChIP loop away from the promoter (anchored binding; 1,030 genes) ( Figure 5A ; Dataset 5 ). By overlapping proximal and anchored genes with the 369 ESR1/E2 targets in engineered THESCs, we identified 72 genes regulated by ESR1/E2 that are directly or indirectly bound by ESR1 at the promoter ( Figure 5A – B ). To identify if any of these genes are involved in the process of decidualization, we compared them to genes that are differentially expressed in response to EPC. 28 genes overlapped with DEGs in response to EPC ( Figure 5C ). We identified genes involved in endometrial remodeling and decidualization, including FOXO1, that are upregulated in response to E2 and EPC, and have proximal or anchored ESR1 binding sites at the promoter ( Figure 5D - E ). These results suggest that distal ESR1 binding sites may regulate gene expression involved in decidualization through long-range chromatin loops. To investigate ESR1/E2’s regulation of genes in the endometrial stroma beyond decidualization, we identified ESR1/E2 target genes using the same approach as described, with a focus on HiChIP loops in vehicle-treated hESC. We identified 69 ESR1/E2 regulated genes that are directly or indirectly bound by ESR1 at the promoter ( Figure 6A ). Metascape pathway analysis showed an enrichment for cancer pathways, endometriomas, and endometrial neoplasms ( Figure 6B ). We identified genes that are implicated in endometrial cancer including ERRFI1 and NRIP1, as well as genes involved in other types of metastatic cancers, such as EPAS1, that are regulated by ESR1/E2 ( Figure 6C - D ). Lastly, we show that ESR1, both in the presence and absence of E2, increased cell viability in THESC through MTT assay ( Figure 6E ). We also demonstrate that in response to E2, ESR1 significantly enhanced the migration of THESC through a porous membrane, as assessed by a transwell migration assay ( Figure 6F ). These findings shed light on the key role of ESR1/E2 signaling in regulating gene expression in the endometrial stroma during decidualization and in disease. The identification of ESR1/E2 target genes enriched in cancer-related pathways suggests estrogen signaling in the stroma might be associated with disease progression, potentially through an increase in stromal proliferation and migration.

Telomerase-transformed human endometrial stromal cells, THESC ( 21 ) (ATCC, CRL-4003), were maintained in DMEM/F-12 (Gibco Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 10% Fetal Bovine Serum (FBS, Gibco Thermo Fisher Scientific) and 1% Penicillin-Streptomycin (Life Technologies, Grand Island, NY, USA), termed regular hESC media ( Table 1 ). Telomerase transformed human myometrial cells (hTERT-HM) and stromal cells (H1644) were maintained according to their respective protocols ( 22 , 23 ). Cells were cultured and grown in a 5% CO 2 and 37°C incubator. Cell media was changed every three days and cells were passaged before reaching 90% confluency. To engineer dCas9-VPR expressing hESC, the Ef1a-dCas9-VPR-Blast construct was obtained as prepackaged lentivirus from Dharmacon (CRISPRmod CRISPRa lentiviral dCas9-VPR, Dharmacon/Horizon Discovery, Lafayette, CO, USA). Cells were engineered following the manufacturers protocol with specifications detailed in the Supplemental Methods. Following transduction, cells were cultured in DEMEM/F-12 (Gibco Thermo Fisher Scientific) containing 10% FBS (Gibco Thermo Fisher Scientific) and 4 ug/mL Blasticidin (Gibco R21001 , Thermo Fisher Scientific), termed engineered hESC media, to select for cells expressing dCas9-VPR . Cells were grown up in a T175 and frozen down in aliquots. For experiments, THESC dCas9-VPR were used for experiments within 10 passages from thawing frozen stock. The media types and their compositions are summarized in Table 1 . gRNAs were designed using the CHOPCHOP ( 24 , 25 ) and CRISPick tools ( 26 ) ( Table 2 ). All gRNA expression vectors were synthesized by and acquired from VectorBuilder ( VectorBuilder.com ) and expressed neomycin resistance and GFP markers. Lentivirus was generated by transfection of the constructs and packaging vector into HEK293T cells with the support of the NIEHS Viral Vector Core Facility. Cells were transduced with gRNA lentivirus at an MOI of 12 and used for experiments within three weeks of transduction. Additional experimental details are provided in the Supplemental Methods section. ESR1 activated and control cells were maintained in engineered hESC media without hormones ( Table 1 ). Protein was isolated, quantified, and seperated as cited in Wu et al. ( 27 ). Antibodies for ESR1 (Cell Signaling Technologies (D6R2W), #13258, Danvers, MA, USA; 1:500) and GAPDH (Santa Cruz Biotech, sc-25778, Dallas, TX, USA; 1:1000) were used. Additional experimental details are provided in the Supplemental Methods section. Wells/plates were washed with PBS and media was replaced with 3 mL OptiMEM (Gibco Thermo Fisher Scientific) with 2% Charcoal Stripped Fetal Bovine Serum (csFBS, Gibco Thermo Fisher Scientific) and 1% Penicillin-Streptomycin), termed low serum media ( Table 1 ). Low serum media was supplemented with either 10nM 17β-estradiol (Sigma-Aldrich, St. Louis, MO, USA), termed E2 media, or 0.01% ethanol, termed vehicle media ( Table 1 ). Cells were collected at different time point (indicated under each protocol) for isolating RNA (RT-qPCR and RNA-seq), or conducting Cut&Run, Transwell Migration Assay, or MTT assay. The total RNA of the cells was isolated using Qiagen RNeasy RNA mini prep kit with Qiagen on-column DNase digestion following the manufacturer’s instructions. RNA was reverse transcribed into cDNA using Moloney Murine Leukemia Virus reverse transcriptase (Thermo Fisher Scientific) according to manufacturer protocol. Quantitative real time PCR was performed using SsoAdvanced Universal SYBR Green Supermix (BioRad) and primers ( Table 3 ). Each reaction was performed in technical duplicates and ΔΔCt values were calculated using 18S control amplification results to determine the relative mRNA levels per sample. Additional experimental details are provided in the Supplemental Methods section. 100,000 cells per well were seeded in 6-well plates and cultured in engineered hESC media for 24 h. Wells were washed with PBS and media was changed to 3 mL per well E2 media or vehicle media for 24 h. The total RNA of the cells was isolated using Qiagen RNeasy RNA mini prep kit with Qiagen on-column DNase digestion following the manufacturer’s instructions. Libraries were prepared and sequenced as 75 bp paired-end reads by the NIEHS Sequencing Core using the Illumina Tru-seq Stranded mRNA library kit and Illumina NovaSeq 6000 instrument. For RNA-seq analysis, adapter trimming, optical deduplication, and PCR deduplication were performed using BBMap v39.01 and the trimmed reads were subsequently mapped to the hg38 genome (GCF_000001405.40) using STAR v2.6.0c. Genes with fewer than 200 reads across all samples were excluded, and differential expression analysis was performed using edgeR v4.2.2 in R v4.4.1. Functional analysis of each gene list was performed using Ingenuity Pathway Analysis software v01.22.01 (IPA, www.ingenuity.com ) and Metascape version X0200 ( http://metascape.org ). For published RNA-seq ( GSE205481 ) ( 28 ), raw Fastq files were downloaded from GEO (P1, P2, and P3 from siNT-Pre-dec and siNT-Dec samples) and analyzed following the same protocol as aforementioned. For previously published RNA-seq ( GSE132713 ) ( 29 ), normalized counts were downloaded from GEO and filtered to include genes with average FPKM > 1 in the human endometrium proliferative samples. The ESR1 Cut&Run samples were prepared with the CUTANA ™ ChIC/CUT&RUN Kit (Version 3.5; EpiCypher, Durham, NC, USA), following the manufacturer’s instructions with modifications provided in the Supplemental Methods section. Antibodies for ESR1 (Cell Signaling Technologies, #13258, Danvers, MA, USA; 1:25) and IgG (CUTANA ™ ChIC/CUT&RUN Kit; 1:100) were used. The libraries were sequenced by the NIEHS sequencing core. CUT&RUN reads were trimmed of adapter sequences using Trim Galore! ( https://www.bioinformatics.babraham.ac.uk/projects/trim_galore/ ; v0.6.10), aligned to hg38 and E. coli K-12 MG1655 genomes using bowtie2 (v2.5.2) ( 30 ), deduplicated using Picard ( http://broadinstitute.github.io/picard ; v3.2.0), and filtered using samtools (v1.18) ( 31 ). HOMER v5.1 was used to call peaks for each sample ( 32 ) and peaks present in both replicates of a condition were used for downstream analysis. Peaks were annotated for overlaps with HiChIP loop anchors and genomic features in the NCBI RefSeq gene model as of February 06, 2025 ( 33 ). Further details regarding the analysis are available in the Supplemental Methods section. Primary endometrial stroma cells (hESCs) were isolated from endometrial biopsies obtained during the proliferative phase, under the human subject protocol number H-13062 approved by the institutional review board of Baylor College of Medicine. The biopsies were obtained from two healthy, reproductive-aged volunteers with regular menstrual cycles and no history of gynecological malignancies. All donors provided written informed consent. hESCs were maintained as cited in Li et al. (2023). To induce decidualization, hESC were treated with 0.2% ethanol (vehicle) or a mixture of 10 nM 17β-estradiol (Sigma-Aldrich), 1 μM medroxyprogesterone acetate (Sigma-Aldrich) and 100 μM dibutyral cyclic AMP (Sigma-Aldrich), in low serum media, termed EPC media ( Table 1 ), for 3 days. HiChIP libraries were prepared by Arima Genomics and sequenced by the NIEHS sequencing core. HiCUP (v0.9.2) was used to process and align HiChIP samples with Arima digest ( 34 ). Peaks were inferred from the HiChIP data using MACS2 (v2.2.9.1) ( 35 ). Loops with at least one loop anchor in a peak region were called and differential loops were identified using FitHiChIP (v11.0) ( 36 ). Loop anchors were annotated for overlapping genomic features in the NCBI RefSeq gene model as of February 06, 2025 ( 33 ). These annotations were used to identify ESR1 CUT&RUN peaks one loop away from a promoter. HiChIP visualizations were generated with Juicer and Juicebox for interaction matrices and R package Sushi for loop plots ( 37 – 39 ). Further details regarding the analysis are available in the Supplemental Methods section. 1,000 cells per well were seeded in 96-well plates and cultured in engineered hESC media for 24 h. Wells were washed with PBS and media was changed to 100 uL per well E2 media or vehicle media. Media was changed every two days and cells were incubated for a total of 6 days. Viability was assessed with a Cell Proliferation Kit I (MTT) (Roche, Basel, Switzerland) according to the manufacturer’s protocol. Briefly, 10 uL of the MTT labeling reagent (final concentration 0.5 mg/mL) was added per well and incubated for 4 h before adding 100 uL of the Solubilization buffer and incubating overnight. Absorbance at 570 nm was measured and relative cell viability was calculated by subtracting absorbance at 570 from control wells containing media only with no cells and subtracting reference wavelength at 670nm. Cells were seeded at 20% confluence in 10 cm plates and cultured in engineered hESC media until reaching 60% confluence. Plates were washed with PBS and media was changed to E2 media or vehicle media. 24 h later, 30,000 cells in 300uL of E2 or vehicle media were seeded in 8.0 um transparent PET membranes (Corning, Falcon, 353097, Corning, NY, USA) placed in 24-well plates. 700uL of regular hESC media was added to the bottom of the plates. After 48 h of culture, the cells on the top side of the membrane were wiped with a cotton swab dipped in PBS. The insert was fixed with 4% paraformadehyde in PBS for 15 min, washed with PBS two times for 5 min each, stained with 1% crystal violet (Sigma-Aldrich) for 20 min, then washed with PBS three times for 10 min each before imaging on a brightfield microscope. Intensity of staining was quantified using ImageJ. Obtained data were expressed in a mean ± standard deviation of the 2–6 biological replicates. Statistical analysis was performed using GraphPad Prism (Version 10.4.0 (621), GraphPad Software Inc.). Distribution of datapoints was assessed by Shapiro-Wilk test. If datapoints were normally distributed, statistics were performed by One-Way ANOVA Turkey test for more than two comparisons, and parametric unpaired t test for two comparisons. Significance was defined as p-value < 0.05.

In this study, we introduce a new in vitro endometrial stromal cell model with restored E2 responsiveness by activating ESR1 using CRISPRa. Integration of the ESR1/E2 transcriptome and cistrome with HiChIP data identifies its role in regulating inflammation, proliferation, and decidualization, as well as its implications in endometrial cancer, providing new insights into estrogen-mediated transcriptional regulation. This model serves as a powerful tool to study estrogen signaling in endometrial stromal biology and related pathologies.

Estrogen signaling through its receptor ESR1 is critical for successful pregnancy by driving decidualization, implantation, and uterine receptivity in the endometrium ( 49 ). Low estrogen levels are suspected to contribute to abnormal placentation in naturally conceived pregnancies, whereas an excess of estrogen may impair pregnancy development and lead to adverse outcomes ( 17 ). Moreover, its dysregulation is linked to gynecological pathologies, including endometriosis and endometrial cancer ( 50 , 51 ). Studying E2 signaling in endometrial stromal cells has been hindered by the limited E2 responsiveness of both primary and immortalized cell lines due to the silencing of ESR1 in vitro ( 20 ). In this study, we introduce a new human endometrial stromal cell model with restored ESR1 expression and E2 responsiveness using CRISPRa. Using these cells, we provide novel insights into ESR1-mediated transcriptional regulation through RNA-seq and Cut&Run assays. We also profiled the chromatin landscape in hESC and decidualized hESC through H3K27ac HiChIP. By integrating chromatin looping with ESR1 binding sites, we suggest an improved method for associating ESR1 distal binding peaks with the promoters of genes they regulate through long-range looping. We identified genes critical for decidualization and implicated in endometrial cancer that are regulated by ESR1/E2 with proximal or anchored ESR1 binding up to 500 kb away from their promoter. Over the past decades, studies in breast cancer cells have expanded ESR1/E2 actions beyond the classical model of ligand-dependent ERE binding ( 52 ), to include ligand-independent actions and binding to non-ERE genomic sites through protein-protein interactions ( 53 ). Through RNA-seq and Cut&Run assays, we show that ESR1 can bind to the genome and regulate transcription in both the presence and absence of ligand in THESC. Importantly, comparison of DEGs between control and E2-treated conditions in ESR1-activated hESCs revealed that 72% of genes overlapped with those active in human endometrial tissue during the proliferative phase, emphasizing the physiological relevance of our model. Using HOMER, we identified transcription factor motifs enriched in ESR1 binding sites in THESC, including chromatin modifiers and proteins known to regulate ESR1 tethering through protein-protein interactions, such as HOX, JUN/AP-1, RUNX, and TEAD motifs ( 44 – 46 ). Additionally, we found enrichment for zinc-finger GATA transcription factors that are implicated in endometrial function and pathology ( 54 ). These results demonstrate that ESR1 activation restores E2 responsiveness in THESC, establishing this model as a valuable tool for studying estrogen signaling in endometrial stromal cell biology. Through H3K27ac HiChIP, we also show that changes to chromatin architecture support gene expression changes that drive decidualization. Integrating differential chromatin loops with DEGs in response to EPC from published RNA-seq ( GSE205481 ) showed enrichment in inflammatory pathways, including prolactin and its downstream signaling pathway (JAK/STAT, PIP3/AKT), as well as IL-6 and IL-7 signaling ( 28 ). Blastocyst implantation, decidualization, and early pregnancy relies on a pro-inflammatory state mediated in part by cytokine release from immune cells ( 55 ). Prolactin signaling is a key marker of decidualization ( 56 ) and is indispensable for pregnancy ( 57 ), playing roles in regulating angiogenesis, glandular secretion, and immune regulation ( 58 ). Taken together, our results suggest that changes to the chromatin landscape may be involved in regulating inflammation-related gene expression changes for early pregnancy establishment and decidualization. It has become common practice to associate TF binding peaks to nearby genes, but this arbitrary approach can introduce biases, particularly for regulators like ESR1 that bind to distal cis-regulatory elements. To overcome this, we integrated H3K27ac HiChIP with ESR1 Cut&Run data in engineered THESC, revealing that chromatin looping may link distal ESR1 peaks to the promoter of ESR1/E2-regulated genes. One notable example is FOXO1 - FOXO1 is activated in response to ESR1/E2 signaling and HiChIP data suggest that ESR1 regulation of FOXO1 may occur through a long-range loop connecting the FOXO1 promoter to an ESR1 binding site located more than 500kb away from its TSS. We have previously shown that FOXO1 is a critical co-regulator of PGR during decidualization ( 59 ) and it is involved in the activation of decidual markers PRL and IGFBP1 ( 60 – 63 ). FOXO1 is also involved in PGR’s anti-proliferative effects on the endometrial stroma and epithelium, and its misregulation is implicated in endometrial cancer ( 64 , 65 ). FOXO1 expression can be induced by 8-Br-cAMP ( 60 ), BMP4 ( 66 ), and SOX4 ( 67 ) in the endometrium, and here we suggest that liganded ESR1 is also a regulator of FOXO1 in endometrial stromal cells. Taken together, these results suggest that changes to chromatin architecture may be involved in the regulation of genes for endometrial differentiation. Furthermore, we demonstrate that the practice of integrating cistromic data with chromatin looping may be a less arbitrary method to associate distal transcription factor binding sites with the promoters of genes they regulate. Through the integration of HiChIP, ESR1 Cut&Run, and RNAseq data, we also identified genes regulated by ESR1/E2 that are implicated in endometrial cancer, including genes NRIP1 (RIP140) and ERRIF1 (MIG-6). We found three distal ESR1/E2 binding sites located approximately 60kb, 200kb, and 270kb away from the NRIP1 TSS that may contact the NRIP1 promoter, as identified through HiChIP chromatin looping. NRIP1 functions as a repressor of ESR1 signaling to attenuate its expression ( 68 , 69 ) and has high rates of mutation in tumors ( 70 ). NRIP1 silencing is implicated in the progression of endometrial cancer ( 70 ), although it is suggested to act as both an activator and inhibitor of tumorigenesis in various cancers ( 71 – 73 ). In the case of ERRFI1, we identified an ESR1/E2 binding site approximately 20kb away from the ERRFI1 TSS with HiChIP chromatin looping. ERRFI1 mediates PGR’s repression on E2-induced proliferation in the endometrium and its knockdown in the uterus of mice leads to endometrial hyperplasia ( 74 ). In humans, ERRFI1 downregulation is correlated with human breast carcinoma ( 75 , 76 ), complex atypical hyperplasia, and endometrioid endometrial carcinomas ( 74 ). ESR1/E2’s regulation of ERRFI1 has previously been documented in the chicken oviduct ( 77 ). Disruption of ESR1/E2 regulation of these genes may occur in endometrial cancer, resulting in unopposed estrogen signaling that drives tumorigenesis. Although endometrial cancer originates from the epithelium, the growth of the epithelium strongly depends on paracrine signaling from the endometrial stroma. The genes identified may play a role in early signaling events during endometrial cancer development and could contribute to epithelial misregulation that drives tumorigenesis.

The endometrium consists of a luminal epithelial layer overlying a connective tissue matrix primarily comprised of endometrial stromal cells, with populations of endothelial cells and immune cells ( 1 ). The endometrium undergoes dynamic changes throughout the menstrual cycle in response to sex steroid hormones estrogen and progesterone, preparing for potential embryo implantation and supporting early pregnancy ( 2 ). Estrogen, specifically estradiol (E2), is a key regulator of endometrial function. During the proflierative phase of the menstrual cycle, E2 signaling from the stroma promotes the thickening of the endometrium by triggering proliferation of epithelial cells in a paracrine manner ( 3 , 4 ). E2 also activates the progesterone receptor gene (PGR), preparing the endometrium to respond to progesterone (P4) secreted by the corpus luteum during the secretory phase ( 5 ). A rise in P4 signaling during the secretory phase, alongside E2 signaling, creates a receptive endometrium and initiates the process of decidualization ( 6 ). During decidualization, fibroblast-like endometrial stromal cells differentiate into epithelioid-like decidual cells that create an immunotolerant environment and provide nutritional support for the implanting blastocyst prior to placentation ( 7 ). While P4 is the primary hormonal regulator of decidualization, E2 signaling is also crucial for its success, but the direct mechanisms through which E2 regulates this process remain unclear ( 8 ). E2 primarily exerts its effects in the female reproductive tract by binding to nuclear estrogen receptors, ERα and ERβ, which are encoded by the ESR1 and ESR2 genes, respectively. Both ESR1 and ESR2 are expressed in the ovaries, the endometrial stroma, and the epithelium, but ESR1 is the dominant isoform throughout the endometrium ( 9 ). Knockout of endometrial-specific ESR1 in mice leads to infertility due to defects in implantation and decidualization ( 10 – 13 ). ESR1 can regulate gene transcription by directly binding to genomic estrogen response elements (EREs) or by interacting with other transcription factors bound to DNA regions lacking EREs ( 12 ). Dysregulated E2 signaling is associated with impaired fertility and reproductive diseases including type I endometrial tumors ( 14 ), endometriosis, and polycystic ovarian syndrome ( 15 – 17 ). Endometrial cancer is the most common malignancy of the female reproductive tract and primarily affects post-menopausal women, a period characterized by unopposed estrogen signaling due to a decline in progesterone levels ( 18 ). Approximately 75% of endometrial tumors are type 1, characterized by overactivation of ESR1 expression, and are hypothesized to be estrogen-driven ( 14 ). Thus, elucidating E2/ESR1 signaling is crucial for advancing our understanding of endometrial function and developing targetted therapies for pregnancy disorders and endometrial diseases. While transcriptomic and cistromic studies of ESR1 have been extensively conducted in mice and humans using whole endometrial biopsies, the presence of both epithelial and stromal components in these samples makes it challenging to isolate the specific role of ESR1 in the stroma. Endometrial epithelial cells and organoid models have been widely used to study E2 signaling in the epithelium in vitro , as they exhibit a strong response to E2 ( 19 ). However, studying ESR1 function in the endometrial stroma remains challenging due to the low expression of ESR1 in primary and immortalized human endometrial stromal cells (hESC), which limits E2 responsiveness ( 20 ). To overcome this barrier, we engineered telomerase-immortalized hESCs (THESCs) with a CRISPR activation (CRISPRa) system to restore ESR1 expression, re-establishing E2 responsiveness in hESCs in vitro . We investigated the ligand dependent and independent ESR1 transcriptome and cistrome through bulk RNA-sequencing (RNA-seq) and Cleavage Under Targets and Release Using Nuclease (Cut&Run) assays. We also examined the chromatin landscape in hESC using H3K27ac HiChIP and demonstrated the utility of this dataset by integrating it with RNA-seq and Cut&Run to identify E2 target genes with distal ESR1 binding that loop to gene promoters. Functionally, ESR1 activation promoted cell viability and migration in response to ESR1/E2, and upregulated genes associated with endometrial cancer, including ERRFI1, NRIP1, and EPAS1. These findings reveal the ESR1-driven transcriptome in hESC, providing critical insights into ESR1’s role in fertility and endometrial pathologies.

Dataset S1 : (A) DEGs (Adjusted P-value 1.3) between ESR activated cells treated with Vehicle for 24 h versus control cells treated with Vehicle for 24 h. (B) IPA pathway analysis of DEGs. Dataset S2 : (A) DEGs (Adjusted P-value 1.3) between ESR activated cells treated with 10nM Estradiol for 24 h versus Vehicle for 24 h. (B) IPA pathway analysis of DEGs. Dataset S3 : 220 DEGs regulated by E2-independent ESR1 and 268 DEGs regulated by E2-dependent ESR1 in THESC, that are also active (FPKM > 1) in human endometrial tissue during the proliferative stage ( GSE132713 ). Dataset S4 : ESR1 ESR1 Cut&Run peaks in THESC engineered to express ESR1 and treated with E2 or vehicle (0.01% EtOH) for 1, 3, or 6 h. Columns A-C indicates peaks in bed format. Column D (tssDistance) indicates how far the peak is from the nearest transcription start site. Column E (relativeLocation) indicates the genomic feature that the peak overlaps with. Column F (overlappedPromoters) indicates the peak a gene overlaps with if it is located at the gene promoter. Column G (closestTSS) indicates the single closest gene transcription start site. Column H (genesWithTSSWithin100kb) annotates each peak to all the genes with a transcription start site that are within 100 kb of the peak. Columns I and K (Veh and EPC_annotated_loopedPromoterGenes) indicate whether the peak overlaps with an H3K27ac Hi-ChIP loop anchor that anchors to a gene promoter in (in Veh or EPC H3K27ac HiChIP). Columns J and L (Veh and EPC_annotated_loopIDs) indicates which HiChIP loop brings the peak to the gene promoter. Dataset S6 : Reanalysis of published RNAseq ( GSE205481 ) comparing gene expression in hESC treated with EPC versus vehicle ( GSM6213359 , GSM6213360 , GSM6213363 , GSM6213364 , GSM6213367 , GSM6213368 ). Dataset S5 : H3K27ac Hi-ChIP Loops in hESC treated with vehicle (0.02% EtOH) or EPC, as well as the loops that are significantly activated or repressed in response to EPC treatment in comparison to vehicle. Columns A - F indicate the loop anchors in bed format. Column G (loopID) assigns a unique ID to each loop. Columns H and I (overlappedGenes1 and 2) indicate the gene(s), if any, that the loop anchor overlaps with. Columns E and J (relativeLocation1 and 2) indicates the genomic feature(s), if any, that each loop anchor overlaps with. Figure S1: Full length uncropped western blot. Figure S2 . RT-qPCR analysis of ESR1 expression in engineered immortalized human endometrial stromal cells (H1644) and myometrial cells (hTERT-HM). Figure S3 . Unsupervised hierarchical clustering and multidimensional scaling plot of RNAseq data. Figure S4 . Raw and filtered contact matrices comparing genome-wide interactions in Veh and EPC treated hESC.