Nr2f2 ^flox/flox (Nr2f2 ^f/f) and Pgr-Cre knockin (Pgr ^Cre/+) mice have been generated previously (10, 21). Conditional knockout mice of Nr2f2, Pgr ^Cre/+; Nr2f2 ^f/f (Nr2f2 ^d/d) were generated by crossing Pgr ^Cre/+ mice with Nr2f2 ^f/f mice. Nr2f2 ^f/f mice were used as wild type. Wild type and Nr2f2 ^d/d female mice at 2 - 3 months of age were mated with vasectomized male mice. The morning of a vaginal plug was designated as gestational day (GD) 0.5. The uterine tissues were obtained from the mice at GD 3.5. For ChIP assay experiment, normal C57BL/6 female mice at 2 - 3 months of age were mated with vasectomized male mice, and then the uterine tissues were obtained at GD 3.5. All animal experiments were performed in accordance with National Institute of Environmental Health Sciences (NIEHS) Animal Care and Use Committee guidelines and in compliance with NIEHS-approved animal protocol (ASP 2015-0012 and 2015-0023).

Total RNA was isolated from uterine tissues using TRIzol reagent (Thermo Fisher Scientific), and then RT was performed using MMLV reverse transcriptase (Thermo Fisher Scientific). Real-time PCR was performed using SsoAdvanced Universal Probes Supermix (Bio-Rad Laboratories) and Taqman probes (Thermo Fisher Scientific) in a CFX Connect Real-time PCR Detection System (Bio-Rad Laboratories). Real-time PCR was performed using the following protocol: 5 min at 95°C, 45 cycles of denaturation (10 sec at 95°C) and annealing/extension (30 sec at 60°C), and a final step of melting curve analysis. CFX Manager software (version 3.0, Bio-Rad Laboratories) was used to collect the PCR data. As an internal control, 18S rRNA endogenous control (Applied Biosystems) was used. The relative level of mRNA was calculated using the 2^-ΔΔCt method. The primers are listed in Table S1 .

Total RNA was isolated from uterine tissues using TRIzol reagent (Thermo Fisher Scientific), and then the RNA quality was assessed using the Agilent 2100 Bioanalyzer (Agilent Technologies). RNA-seq was performed by the Epigenomics and DNA Sequencing Core Laboratory at NIEHS. RNA libraries were generated using TruSeq RNA Library Prep kit v2 (Illumina), and then sequenced using NextSeq 500 system (Illumina).

The raw reads (76-bp, paired-end) were initially processed by applying a filter with an average quality score of 20. Adaptor sequences were subsequently removed from reads using cutadapt (v1.12). The processed reads were then aligned to the mouse reference genome (mm10) using the STAR aligner (v2.5.2b). The gene expression levels in each sample were determined by counting the total number of paired-end reads mapped to each gene using the featureCounts (v1.5.0-p1) function within the Subread program. Differential gene expression analysis was conducted using the DESeq2 R package (v1.12.4) with a significance threshold of absolute fold change ≥ 1.4 and a p-value < 0.05 (22).

Uterine tissues were fixed in 4% paraformaldehyde and then embedded in paraffin. Sections (5 µm) were mounted on glass slides, deparaffinized, and rehydrated. Antigen retrieval was performed using Antigen Unmasking Solution (H-3300, Vector Laboratories, Burlingame, CA) according to manufacturer’s instructions. The slides were incubated in 3% H₂O₂ in methanol to block endogenous peroxidase activity. Then, the samples were blocked using 5% normal goat (for rabbit primary antibodies) or rabbit serum (for goat primary antibody). The slides were incubated in primary antibodies overnight at 4°C. After washing in PBS, the slides were incubated in peroxidase-labeled goat anti-rabbit IgG (1:200; BA-1000, Vector Laboratories) or rabbit anti-goat IgG (1:200; BA-5000, Vector Laboratories) for 30 min. The ABC reagent (PK-6100, Vector Laboratories) was applied to slides according to manufacturer’s instructions. The signal was developed using a 3,3’-diaminobenzidine (DAB) substrate kit (SK-4105, Vector Laboratories), the nuclei were stained using hematoxylin, and the slides were permanently mounted. Information of primary antibodies is provided in Table S2 .

Uterine tissues were minced and fixed in 1% formaldehyde for 10 min and quenched with 0.125 M glycine. The samples were washed with cold PBS containing protease inhibitors, resuspended in homogenization buffer [50 mM Tris-HCl (pH 7.5), 1 mM EDTA (pH 8.0), 1% NP-40 (IGEPAL), 0.25% deoxycholic acid, and protease inhibitors], and then homogenized with a Polytron (Kinematica). The samples were pelleted by centrifugation, resuspended in membrane lysis buffer [5 mM PIPES (pH 8.0), 85 mM KCl, 0.5% NP-40 (IGEPAL), and protease inhibitors], and incubated on ice for 5 min. The samples were pelleted by centrifugation, resuspended in nuclear lysis buffer [50 mM Tris-HCl (pH 8.0), 10 mM EDTA (pH 8.0), 1% SDS and protease inhibitors], and incubated on ice for 10 min. Lysates were sonicated, and the DNA was sheared into an average length of 300-400 bp. Chromatin was diluted 1:9 with dilution buffer [50 mM Tric-HCl (pH 8.0), 2 mM EDTA (pH 8.0), 300 mM NaCl, 1% NP-40 (IGEPAL), 0.5% sodium deoxycholate, 0.1% SDS, and protease inhibitors] and then precleared with Dynabeads Protein A (Invitrogen, Carlsbad, CA). Chromatin immunoprecipitation (ChIP) was performed using the antibodies for NR2F2 (61213, Active Motif) and PGR (8757, Cell Signaling Technology). After incubation at 4°C overnight, Dynabeads Protein A was used to isolate the immune complexes. Complexes were washed in wash buffer [20 mM Tris- HCl (pH 8.0), 2 mM EDTA (pH 8.0), 300 mM NaCl, 0.1% SDS, 1% Triton X-100, and protease inhibitors] and eluted using elution buffer [100 mM NaHCO₃ and 1% SDS]. Cross-links were reversed in 200 mM NaCl by incubation overnight at 65°C, and then the samples were treated with RNase and Proteinase K. ChIP DNA was purified by phenol-chloroform extraction and isopropanol precipitation. The purified DNAs were subjected to sequencing or quantitative PCR. The ChIP DNA libraries were generated using NEXTflex Rapid DNA-Seq kit (Perkin Elmer), and then sequenced using the NextSeq 500 system (Illumina). H3K4me1 ChIP was performed by Active Motif, Inc. Whole uterine tissue of mice at GD 3.5 were flash frozen and sent to Active Motif. The tissue was fixed and then promptly sheared into small fragments before immunoprecipitation with the H3K4me1 antibody (cat # 39297, Active Motif, Carlsbad, CA). DNA was purified and amplified to generate a library and using the NextSeq 500 system (Illumina).

The raw reads (75-bp, single-end) were subjected to filtering, retaining only reads with average quality scores greater than 20. Subsequently, the reads were aligned to the mouse reference genome (mm10) using Bowtie (v1.1.2) (23), requiring unique mapping and allowing up to 2 mismatches per read (-m 1 -v 2). Duplicated reads with identical sequences were removed using Picard tools. To visualize the read coverage, BigWig files were generated from the bedgraph files of each sample using bedGraphToBigWig. These bigWig files were displayed as custom tracks on the UCSC genome browser. The initial peaks within each dataset were identified using MACS2, applying a cutoff of adjusted p-value < 0.0001 (24). To obtain a comprehensive set of peaks across all ChIP-seq datasets, the peak intervals from each dataset were merged using BedTools. The gene associated with each peak was predicted by searching the transcription start site (TSS) of nearby genes within a 50Kb range using HOMER (25).

For ChIP quantitative PCR (ChIP-qPCR), primers were designed to amplify regions of NR2F2 occupancy identified through ChIP-seq. ChIP-qPCR was performed in duplicate on specific genomic regions using SsoAdvanced Universal SYBR Green Supermix (Bio-Rad Laboratories). As a negative control, the Mouse Negative Control Primer Set 1 (Active Motif) which amplifies a fragment in a gene desert on mouse chromosome 6 was used. The resulting signals were normalized to input DNA and presented as percentage to input. The primer sequences are listed in Table S3 .

The following experiments were conducted in immortalized T-HESC cells (American Type Culture Collection, Cat. #4003TM), a model cell line of the human endometrial stroma cells (26). T-HESC cells were transduced with lentivirus that contained IGI-P0492 pHR-dCas9-NLS-VPR-mCherry (Research Resource Identifier: Addgene_102245) and guide RNA (gRNA)-green fluorescent protein (GFP)-expressing vectors. IGI-P0492 pHR-dCas9-NLS-VPR-mCherry was a gift from Jacob Corn (Addgene plasmid # 102245; http://n2t.net/addgene:102245; RRID : Addgene_102245). gRNA-GFP-expressing vectors were synthesized with the respective guide sequences listed in Table S4 by the VectorBuilder (Chicago, IL). The non-targeting controls were included on each experimental batch by following the same procedure.

The cells were infected by the mentioned lentivirus with a MOI of 4 for the IGI-P0492 pHR-dCas9-NLS-VPR-mCherry lentivirus and with a MOI of 4 for the gRNA-GFP-expressing vectors, including the non-targeting control. After infection for 72h at 37°C, the cells were sorted selecting cells positive for both GFP and mCherry fluorescence. Then, cells were harvested, RNA was extracted, complementary DNA (cDNA) was synthetized and qRT-PCR was applied to assess the effect of infection. All the RT-PCR results were normalized to the expression levels of the internal control 18S rRNA. The 18S rRNA probe is from ThermoFisher Scientific (4319413E, ThermoFisher Scientific, Waltham, MA) and primers for HAND2 are listed in Table S1 .

Cells were maintained in the Dulbecco’s modified Eagle medium (DMEM)/F12 medium (Thermo Fisher 11330032) with 10% fetal bovine serum (Thermo Fisher, #10437028) and 1× antibiotic–antimycotic (Thermo Fisher 15240062).

Statistical significance was analyzed using GraphPad Prism 8 (GraphPad Software). The differentially expressed genes generated by RNA-seq were analyzed using Ingenuity Pathway Analysis software (IPA, Qiagen Bioinformatics). The gene signature assay was performed on the Gene Set Enrichment Analysis (GSEA) (27). The correlation study between Nr2f2 ^d/d transcriptome profile and public data sets was performed using Correlation Engine software (BaseSpace, Illumina). The ChIP-seq data were analyzed using UCSC Genome Browser (https://genome.ucsc.edu/) (28) and Cistrome software (http://cistrome.org/ap/) (29). The CRISPRa validation statistic was analyzed using Friedman’s test. Data are presented as the mean ± SEM.

Gestation day 3.5 (GD3.5) uterine tissues from Nr2f2^f/f (Nr2f2^flox/flox ) and Nr2f2^d/d (Pgr^Cre/+; Nr2f2^flox/flox ) mice were subject to the RNA-seq assay to examine the impact of Nr2f2 ablation on the transcriptomic profile. A total of 1533 differentially expressed genes (DEGs) were identified ( Figure 1A , Dataset S1 ). A reduction of Nr2f2 mRNA levels in the Nr2f2^d/d uterus confirmed the ablation of the Nr2f2 gene and validated the model ( Figures 1A, B , Dataset S1 ). The mRNA abundance of major uterine transcription regulators for embryo implantation, including Hand2, Egr1, and Zbtb16 among others (17, 30, 31), is lower in Nr2f2^d/d compared with control mice ( Figures 1A, B , Dataset S1 ), which indicates that NR2F2 is upstream of these transcription factors in the genetic network. Notably, Nr2f2 ablation also leads to an increase in mRNA abundance of HAND2 downstream targets Fgf1 and Fgf18 ( Figure 1A, B , Dataset S1 ). Moreover, the Nr2f2^d/d uterine transcriptome manifests an estrogen hypersensitive phenotype, as shown by the positive correlation with expression profiles of estrogen responsive genes in the Gene Set Enrichment Analysis (GSEA) ( Figure 1C ) and in the Ingenuity Pathway Analysis (IPA) (beta-estradiol in IPA upstream regulators of Dataset S1 ). Since stromal Hand2 suppresses endometrial estrogen signaling at GD3.5 (17), the impact of Nr2f2 loss on the Hand2-Fgf pathway supports NR2F2 as an upstream regulator of Hand2 in modulating epithelial-stromal communication in the endometrium for embryo receptivity.

In addition to estrogen hypersensitivity, the Nr2f2^d/d transcriptome manifests a change in the immune baseline as indicated by the positive correlation with the GSEA curated datasets of “allograft rejection” and “innate immune system” ( Figures 1D, E ). This in vivo observation is consistent with the enrichment of inflammation-associated genes in the NR2F2 regulated transcriptome of the human endometrial stroma cells (HESC) (12). Based on NR2F2 downstream target gene lists, the Ingenuity Pathway Analysis (IPA) canonical pathway assay further reveals the impact of NR2F2 loss on immune and inflammation associated signaling pathways in both mouse and human cell models, involving not only both pro- and anti-inflammatory cytokines such as IL6, IL17, IL4, IL10, and IL13, but also mechanisms for pathological response such as “pathogen induced cytokine storm signaling pathway”, “acute phase response signaling”, and “ eicosanoid signaling” ( Figure 1F , Dataset S2 ). These findings suggest that NR2F2 loss creates a gene expression pattern permissive for immune challenges and that this phenotype is conserved between species.

Notably, expression of many extracellular matrix and cell-matrix interaction genes are altered in response to NR2F2 loss as captured under “wound healing signaling pathway” and “integrin signaling” (IPA canonical pathways of Dataset S1 ), which implicates a role of NR2F2 in regulation of the uterine matrix homeostasis that is important for sustaining a normal pregnancy (32). These observations provide insights in a molecular context for the embryo implantation failure phenotype in mutant mice (15). In summary, this RNAseq data revealed multiple mechanisms utilized by NR2F2 to sustain normal uterine function at early pregnancy.

Next we compared the transcriptomic changes between NR2F2 ablation and other major uterine regulator perturbations that are crucial for early pregnancy. Subsets of the 1533 DEGs from NR2F2 ablation are present in DEGs from mutations of PGR Activation function 1 domain (Pgr^AF1/AF1 , 309 genes) (33), Ihh (Ihh^d/d , 261 genes) (34), Sox17 (Sox17^d/d , 592 genes) (35), Foxa2 (Foxa2^d/d , 523 genes) (36), Wnt4 (Wnt4^d/d , 510 genes) (37), Errfi1 (Errfi1^d/d , 298 genes) (38), Arid1a (Arid1a^d/d , 360 genes) (39), and Sirt1 (Sirt1^d/d , 971 genes) (40) ( Figures 2A , S1 ). Most overlapped genes shared the same direction of gene expression alterations in response to perturbations (91% with Pgr^AF1/AF1 , 75% with Ihh^d/d , 77% with Sox17^d/d , 94% with Foxa2^d/d , 71% with Wnt4^d/d , 86% with Errfi1^d/d , 84% with Arid1a^d/d , and 79% with Sirt1^d/d ), indicating a positive correlation on gene signatures between NR2F2 and these major uterine regulators ( Figures 2A , S1 ). These findings provide transcriptomic evidence to support interactions between NR2F2 and other uterine regulators for gene expression programs at early pregnancy.

Consistent with the previously observed estrogen hyperactivity (15, 16), the Nr2f2^d/d transcriptome is positively correlated with the uterine estradiol gene signature, with 75% of overlapping genes sharing the same direction of changes ( Figures 2A , S1 ). Moreover, the result of the IPA upstream regulator assay reveals an increase of inferred mifepristone activities in the Nr2f2^d/d transcriptome ( Figure 2B ), suggesting a loss of progesterone signaling activities. This genome-wide evidence supports the phenotype of unopposed estrogen activities in the context of progesterone signaling loss, in addition to the marker-gene-based findings in Figure 1 . In addition, this progesterone signaling loss phenotype is seen in the GD3.5 uterine transcriptome of Pgr^AF1/AF1 , Sox17^d/d , Errfi1^d/d , Arid1a^d/d , and Sirt1^d/d mice that manifest hyperactive estrogen activities ( Figure 2B ) (33, 35, 38–40), indicating a common theme among these transcription regulators in modulating progesterone signaling activities. On the other hand, these uterine regulators also exhibit a divergent change of the immune baseline despite their shared loss of progesterone signaling activities, as demonstrated by the inferred lipopolysaccharide activities, a surrogate for the inflammation status, manifesting an increase in Nr2f2^d/d and Errfi1^d/d and a reduction in Pgr^AF1/AF1 , Sox17^d/d , Arid1a^d/d , and Sirt1^d/d uteri ( Figure 2B ). The difference on the change of the immune baseline is further reflected by the diverse enrichment pattern of cytokine signaling pathways that are under the control of these regulators ( Figure 2C ). These findings demonstrate the common and distinct biological processes between NR2F2 and other major uterine regulators in the uterus at early pregnancy.

Immunostaining results reveal multiple KI67-positive cells in the luminal epithelium and a reduction of PGR expression levels in the stroma of the Nr2f2^d/d uterus at GD3.5, in contrast to the negative KI67 signal and strong PGR expression in the corresponding compartments of the control uterus ( Figure 3 ). This observation indicates the presence of unopposed estrogen signaling activities in the mutant uterus, which is consistent with the RNAseq findings ( Figure 1C ) and previous reports (15, 16). In the mutant stroma, the HAND2 protein abundance is lower compared with the control ( Figure 3 ), in line with the changes in the uterine Hand2 mRNA levels and reflecting the de-repression of HAND2’s downstream targets Fgf1 and Fgf18 ( Figure 1B ). Collectively, the immunohistochemistry findings further support the transcriptomic results on the estrogen hypersensitive phenotype and the impact on stromal HAND2 expression.

Given the transcription factor feature of NR2F2, the NR2F2 genome occupying profile was mapped by the ChIP-seq assay in the GD3.5 uterus using two pools (Replicates A and B) of wild type mouse uteri. Replicates A and B have 14150 and 8285 NR2F2 occupied genomic intervals, respectively, and share 7425 common intervals ( Figure 4A , Dataset S3 ). Over 60% and 53% of the 7425 NR2F2 occupied intervals are located in the H3K27ac (41) and H3k4me1 positive histone marks loci, respectively ( Figure 4B ). This observation suggests that the majority of genome-binding NR2F2 is located in active, poised enhancers, or both. De novo motif enrichment analysis shows an overrepresentation of the (A/G)GGTCA sequence in both replicates of the GD3.5 uterus ( Figure 4C ). Such an enrichment is also seen in NR2F2 occupying sites that were mapped previously in female and male mesonephori (9) and in the embryonic atria (42). Results of this unbiased search of the in vivo NR2F2-binding motif for mice are consistent with previous in vitro findings (43). In addition, this sequence is also preserved in NR2F2 occupying sites in human cells (12, 44). Moreover, both replicates show overrepresentations of motifs of major uterine regulators including PGR, ESR1, and AP-1 among others ( Figure 4D , Dataset S4 ). Notably, motifs of two major uterine regulators PGR and FOXL2 are overrepresented only in the uterus in a survey of the NR2F2 cistrome across various tissues ( Dataset S4 ), while motifs for CTCF/CTCFL and TBX20 are only enriched in embryonic mesonephroi and atria, respectively ( Dataset S4 ). These results suggest a potential tissue-specific cistome microenvironment that permits interactions of NR2F2 with a unique group of transcription regulators to modulate gene expression. In summary, the similar findings between these two replicates together support the validity of NR2F2 ChIP-seq results in the uterus. These observations also provide a comprehensive map of the uterine NR2F2 cistrome and candidate interacting factors on genomic regulation at GD3.5.

Both NR2F2 uterine ChIP-seq replicates show NR2F2 occupancy at the Hand2 locus ( Figure 5A ). ChIP-qPCR validated the NR2F2-binding events in all three NR2F2-binding intervals surrounding the Hand2 gene body within a 52-kilobase stretch of genomic region ( Figures 5A, B ). In addition, NR2F2 also occupies the genomic loci of two other transcription factors Egr1 and Zbtb16 ( Figure S2 ) that both manifest reduced mRNA abundance in the Nr2f2^d/d uterus compared with control ( Dataset S1 ). The NR2F2 binding events were validated by ChIP-qPCR at 3 and 6 genomic intervals in the Egr1 and Zbtb16 loci, respectively ( Figure S2 ). These data provide evidence of NR2F2 genome occupancy at loci of Hand2, Egr1, and Zbtb16 that are NR2F2 downstream targets, suggesting a direct modulation of the transcription of these major uterine transcription regulators by NR2F2. Notably, PGR occupancy was also detected by ChIP-qPCR in the aforementioned NR2F2-binding regions proximal to the Hand2 gene body ( Figure S3 ), confirming previous observations (NCBI accession number GSE226623) and implicating a potential interaction between NR2F2 and PGR in regulation of Hand2 expression via recruitment to proximity in the genome.

Next, as a proof of principle, we performed functional assessment on one of the putative enhancers occupied by NR2F2 located at approximately 10-kilobases downstream of the Hand2 gene (Site 2 in Figure 5A ). This site was chosen based on the criteria that it has both H3K27ac and H3K4me1 histone marks with NR2F2 occupancy in both uterine ChIP-seq replicates within a 50-kb vicinity of the Hand2 gene body ( Dataset S3 ). The UCSC genome browser liftover tool found correspondence of this mouse putative enhancer in a human genomic region approximately 12-kilobase downstream of the human HAND2 transcription start site with the genome coordinates chr4:173516002-173519662 in the hg38 genome assembly ( Figure 5C ). The Multiz tool (45) further identified two highly conserved subareas in this genomic region, in which the Remap Atlas (46) shows NR2F2 and PGR occupancy that is documented based on previously published ChIP-seq data ( Figure 5C ). These observations provide evidence to support the subsequent functional assay. For testing the ability of a transcription activator to promote Hand2 gene expression from this location, two independent guide RNAs (gRNA48 and gRNA49) that target the two conserved subareas were transduced into immortalized human endometrial stromal cells (T-HESC) (47) for the recruitment of the dCas9-VPR CRISPR activator to the target region ( Figure 5C ). qRT-PCR results show an increase of the endogenous HAND2 mRNA abundance in cells that express gRNA48 and gRNA49 compared with those expressing non-targeting gRNA that serves as a negative control ( Figure 5D ). This finding together with the local NR2F2 occupancy suggest that this enhancer recruits NR2F2 and promotes HAND2 gene expression.