Signaling via retinoic acid receptors mediates decidual angiogenesis in mice and human stromal cell decidualization.

A.V.B., Q.Z., R.A., and N.C.D. designed the study. C.‐A.S., P.T., N.M., R.L., Q.Z., and T.W. conducted experiments. A.C., E.K.C., C.‐A.S., P.T., R.A., N.M., R.L., Q.Z., and X.Z. analyzed data. A.C., A.V.B., E.K.C., C.‐A.S., Q.Z., R.A., and N.C.D. wrote the manuscript.

We determined the spatiotemporal expression of genes encoding transport and synthesis proteins, enzymes, and receptors involved in RA signaling (Figure  1 ). Using RT‐qPCR, we compared mRNA expression in E0.5 uteri and decidual sites at E5.5 and E7.5. We first confirmed changes in mRNA expression of prolactin family 8, subfamily a, member 2 ( Prl8a2 ), a marker of decidualization. As expected, mRNA expression of Prl8a2 increased from fertilization at E0.5 to E5.5; from E5.5. to E7.5; and from E0.5 to E7.5 (Figure  1B ). Next, we investigated gene expression changes in the RA synthesis and signaling pathway. 49 From E0.5 to E5.5, mRNA expression of Lrat , the enzyme that promotes retinyl ester formation from retinol, increased 5.3‐fold. Transporter protein RBP4 carries retinol from the liver to peripheral tissues. mRNA levels of Rbp4 were similar at E0.5 and E5.5 but increased 9.8‐fold from E5.5 to E7.5 and 10.6‐fold from E0.5 to E7.5. Coincident with increased stromal cell decidualization and decidual angiogenesis from E5.5 to E7.5, mRNA levels of Aldh1a3 , one of the genes encoding a RA‐synthesizing enzyme that converts retinaldehyde to RA, increased 24.3‐fold. Two intracellular proteins, FABP5 and CRABP2, transport RA to its nuclear receptors and are thought to promote cellular differentiation and apoptosis, respectively. 23 mRNA expression of Fabp5 increased 3.1‐fold from E5.5 to E7.5, while expression of Crabp2 decreased 5.6‐fold from E0.5 to E7.5. mRNA expression of RA receptors Ppard , Rara , and Rarb increased significantly with decidualization, while mRNA levels of receptor Rarg were not altered with decidualization. From E0.5 to E5.5, mRNA levels of Ppard increased 1.8‐fold and Rara increased 2.2‐fold. From E0.5 to E7.5, mRNA levels of Rara increased 2.4‐fold and Rarb increased 15.8‐fold. mRNA expression of CCAAT/enhancer‐binding protein beta ( Cebpb ) and Connexin 43 ( Gja1 ), two downstream targets of RA signaling that have been implicated in stromal cell decidualization and angiogenesis, 50 , 51 , 52 increased 2.3‐fold and 9‐fold, respectively. Together, these data (Figure  1C ) show distinct temporal expression of components of the RA pathway and support the existence of active RA signaling in early mouse pregnancy. Spatial transcriptomics technology was recently utilized to identify uterine microenvironments and determine the uterine transcriptome at E7.5. 53 We analyzed this dataset and localized expression of RA pathway components to five distinct microenvironments within the decidual site (Figure  2A ). As previously described, in this type of analysis, bubbles in the same uterine microenvironment are labeled with one color and gene expression in the gastrulation stage conceptus (C) is identified. 53 Three microenvironments in the mesometrial region, the mesometrial decidua (MD), vascular sinus folds (VSF), and fetal‐maternal interface (FMI) are identical to those previously delineated. Gene signatures can sub‐divide the anti‐mesometrial region into the primary decidual zone (PDZ), transitional decidual zone (TDZ), and secondary decidual zone (SDZ), and as expected, the prominently expressed genes in the PDZ, TDZ, and SDZ are different than the those expressed in the adjacent undifferentiated stroma. 53 For our analyses, we combined gene expression in the PDZ, TDZ, and SDZ into one decidual zone (DZ) microenvironment within the anti‐mesometrial region. We then determined mRNA expression of genes encoding proteins involved in decidualization (Figure  2B ) and genes associated with retinol transport and RA synthesis and signaling as highlighted in Figure  1 (Figure  2C ). Decidualization markers show high levels of expression throughout the mesometrial region. There was highest mRNA expression of Hoxa10 , Hoxa11 , and Wnt4 in the MD; Hand2 in the VSF; and Bmp2 and Wnt4 in the FMI. mRNA expression of Prl8a2 was highest in the VSZ in the mesometrial region and the DZ in the anti‐mesometrial region. mRNA expression of Ptgs2 was high in the conceptus and FMI and MD in the mesometrial region. The retinol transport gene, Rbp4 , was expressed in the VSZ of the mesometrial region and DZ of the anti‐mesometrial region. mRNA expression of RA synthesizing retinol dehydrogenase (Rdh) and aldehyde dehydrogenase (Aldh) enzymes was prominent in the conceptus. Rdh10 was also expressed in the MD of the mesometrial region and the DZ of the anti‐mesometrial region, while Rdh11, Rdh12 , and Aldh1a1 were also expressed in the DZ. Although mRNA expression of both Crabp2 and Fabp5 was localized to the FMI and DZ of the anti‐mesometrial region, the percent of cells expressing these RA‐binding proteins differed, possibly reflecting unique roles. In the DZ, >75% of cells expressed Fabp5 and <25% of the cells expressed Crabp2 . Spatial transcriptomics analysis of RA pathway components in the E7.5 decidual site. (A) Spatial map of the gastrulation stage conceptus (C) and four uterine microenvironments, the mesometrial decidua (MD), vascular sinus zone (VSZ), fetal‐maternal interface (FMI) in the decidual site. Bubbles in the same uterine microenvironment are labeled with one color. (B) mRNA expression of genes involved in decidualization in the MD, VSZ, FMI, C, and DZ. (C) mRNA expression of RA pathway components within MD, VSZ, FMI, C, and DZ. The color of the bubble is associated with the level of gene expression. The size of the bubble indicates the percent of cells expressing the gene. We uncovered unique, non‐overlapping expression domains of the RA receptors. Ppard mRNA expression was highest in the FMI of the mesometrial region and DZ of the anti‐mesometrial region, whereas expression of Rara , Rarb , and Rarg was highest in the MD of the mesometrial region and the conceptus. 25%–50% of cells in those microenvironments expressed Ppard , Rara , and Rarg , while <25% percent of cells expressed Rarb . mRNA expression of downstream targets of RA signaling, Cebpb and Gja1 , was detected throughout, but was highest in the FMI, the area of placental invasion with direct contact between fetal cells and the maternal decidua. To investigate the requirement for RAR signaling during the peri‐implantation period for progression of pregnancy, pregnant dams BMS493 were administered, an inverse pan‐RAR agonist shown to prevent retinoic acid‐induced differentiation. 42 , 54 After mating and detection of a copulation plug, increasing concentrations of BMS493 were administered via daily oral gavage starting at E2.5 prior to embryo implantation (Figure  3A ) or at E4.5 just after embryo implantation (Figure  3B,C ) and ending on E6.5. Pregnancies and uterine phenotypes were assessed at E7.5 and E9.5 (Figure  3D–G ). Systemic administration of BMS493 disrupts pregnancy. (A–C) Experimental design for administration of BMS493 via oral gavage. Black arrows indicate days of BMS493 administration, black arrowheads indicate day of embryo implantation, and red arrows indicate day of euthanasia. (D) Number of implantation sites at E7.5 after DMSO (control, n  = 4 dams) or increasing concentrations of BMS493 (15 μg/g, 30 μg/g, 60 μg/g body weight with n  = 3 dams per concentration) was administered from E2.5 to E6.5. (E) Number of implantation sites at E7.5 after DMSO ( n  = 6 dams) or increasing concentrations of BMS493 ( n  = 3–7 dams) was administered from E4.5 to E6.5. (F) Number of implantation sites at E9.5 after DMSO ( n  = 4 dams) or 60 μg/g BMS493 ( n  = 4 dams) was administered from E4.5 to E6.5. (G) Uterine horns at E9.5 after treatment with DMSO or 60 μg/g BMS493 from E4.5 to E6.5. (H) Representative images of decidual sites at E7.5 after treatment with DMSO or 60 μg/g BMS493. (I) Representative images of H&E‐stained sections of implantation sites from DMSO‐treated (i) and 60 μg/g BMS493‐treated dams with (ii) and without (iii) an embryo. (J) Quantification of embryo size in two implantation sites per dam for DMSO ( n  = 4) and 60 μg/g BMS493 ( n  = 4) treated dams. Each dam is represented by a unique shape. (K) mRNA expression of Cyp26a1 , Cyp26b1 and Rarb in the liver samples at E7.5 after treatment with DMSO or 60 μg/g body weight BMS493 from E2.5 to E6.5. (L) mRNA expression of Cyp26a1 , Cyp26b1 and Rarb in decidual sites at E7.5 after treatment with DMSO or 60 μg/g body weight BMS493 from E4.5 to E6.5. Gene expression was determined by RT‐qPCR normalized to levels of 18 s. The average transcript level after BMS493 treatment was set to 1. Data are presented as median + IQR. Statistical difference was determined by Mann–Whitney U test with * p  < .05. AM, anti‐mesometrial; CTL, control; e, embryo; ns, not significant. Starting at E2.5 and ending at E6.5, dams received BMS493 at concentrations of 15, 30, and 60 μg/g body weight or the corresponding volume of DMSO. At the lower concentrations of 15 and 30 μg/g BMS493, one of three dams was pregnant at E7.5, with 7 and 8 implantation sites, respectively (Figure  3D ). At the highest concentration of 60 μg/g BMS493, no implantation sites were detected, indicating the potential for a dose effect. In contrast, dams that received the DMSO vehicle control had consistently higher numbers of implantation sites at E7.5, averaging 8.8 across the dams (Figure  3D ). Given the requirement for ovarian‐derived hormones to maintain early pregnancy prior to placenta formation, we examined the ovaries to determine if exposure to BMS493 during the peri‐implantation period impairs ovulation and corpora lutea formation. Ovaries harvested at E6.5 were embedded in OCT, sectioned, and stained with H&E or endothelial cell marker (EC), CD31. We identified multiple corpora lutea, reflecting maturation of ovarian follicles and ovulation of multiple oocytes, in both BMS493‐treated and DMSO control dams (Figure  S1 ). When dams received BMS493 starting at E4.5 and ending at E6.5 at concentrations of 15, 30 and 60 μg/g or the corresponding volume of DMSO, the number of implantation sites at E7.5 ranged from 7.6 to 9.3 per dam for all treatments (Figure  3E ). When dams that received 60 μg/g body weight BMS493 starting at E4.5 and ending at E6.5 were assessed at E9.5, no implantation sites were observed (Figure  3F ). Uteri appeared dark, consistent with hemorrhage and pregnancy loss (Figure  3G ). To investigate the impact of post‐implantation systemic administration of BMS493 on decidual sites at E7.5, we removed the myometrium. We found that E7.5 decidual sites were of similar size in dams that received DMSO or 60 μg/g BMS493 (Figure  3H ). To assess embryo development, implantation sites were embedded in wax, sectioned, and stained with H&E (Figure  3I ). Embryo size was determined as the total number of H&E‐stained sections containing embryo in each implantation site. Two implantation sites per dam (Figure  3J , same symbol) were assessed in 4 control and 4 BMS493‐treated dams. An embryo was identified and scored in 12/12 implantation sites from DMSO controls (Figure  3Ii,J ). Treatment with BMS493 resulted in 7/12 implantation sites with embryo (Figure  3Iii,J ) and 5/12 implantation sites devoid of embryo (Figure  3Iiii, J ). When embryos were present, embryo size was similar in BMS493‐treated and DMSO control dams (Figure  3J ). RT‐qPCR on liver samples revealed a significant decrease in expression of Cyp26a1 , Cyp26b1 , and Rarb after treatment with all concentrations of BMS493 starting at E2.5 or E4.5 (Figure  3K ; data not shown for 15 and 30 μg/g BMS493) consistent with systemic disruption of RAR signaling. 43 Downregulation of Cyp26a1 and Rarb in decidual sites (Figure  3L ) confirmed BMS493‐induced disruption of RA signaling locally at maternal–fetal interface. Taken together, these data show that both pre‐ and post‐implantation inhibition of RAR signaling results in pregnancy loss. Abnormal decidual vasculature can result in poor placenta formation and function, fetal growth restriction and/or pregnancy loss prior to placenta formation. 45 , 55 , 56 We and others have interrogated the role of the vascular endothelial growth factor (VEGF), Notch and angiopoietin (Angpt) signaling in decidual angiogenesis. 45 , 46 , 57 , 58 Herein, we characterized the spatial expression of components of the VEGF, Notch, and Angpt signaling pathways in E7.5 decidual sites (Figure  4A ). The Vegfa ligand and VEGF family receptors, Flt1 ( Vegfr1 ), Kdr ( Vegfr2 ), Flt4 ( Vegfr3 ) were expressed in the vascular sinus zone (VSZ) and fetal maternal interface (FMI), while placental growth factor ( Pgf ) was expressed in the decidual zone (DZ). Notch pathway ligands delta‐like ligand 4 ( Dll4 ) and Jagged1 ( Jag1 ) were expressed in the DZ. Receptors Notch1 and Notch4 were expressed in the VSZ, FMI, and conceptus. Based on the bubble charts, <50% of the cells expressed these VEGF and Notch family components. Angpt receptors Tie1 and Tek were also expressed in 75% of the cells expressed Angpt2 in the FMI. We then sought to identify BMS493‐induced changes at E7.5 that may precede pregnancy loss by E9.5. We determined how treatment with BMS493 impacts mRNA expression of signaling partners Vegfa and Vegfr2 ; Dll4 and Notch4 ; and Angpt2 and Tek in decidual sites (Figure  4B ). Treatment with BMS493 significantly decreased mRNA levels of Notch4 and Vegfr2 . mRNA levels of Angpt2, Vegfa, Dll4 , and Tek were similar in BMS493 and DMSO control decidual sites. To determine if BMS493‐induced changes in gene expression affected decidual angiogenesis, wax embedded implantation sites were sectioned. Those that contained an embryo were stained for EC marker CD34, 59 and the percentage of CD34+ ECs in two implantation sites per dam was calculated (Figure  4C–H ). For this analysis, we considered both the total area of the decidual site and the anti‐mesometrial region (Figure  4D ). The total area of the decidual site and the size of the anti‐mesometrial region were similar in BMS493 and DMSO‐treated dams (Figure  4D,E ). To ensure that differences in CD34 expression were not due to differences in the relative size of the AMR, we also considered the ratio of the area of the AMR to the area of the decidua and found that these were similar in BMS493 and DMSO‐treated dams (Figure  4G ). Treatment with BMS493 significantly decreased the percentage of CD34+ ECs in the decidual site and in the AMR (Figure  4F,H ), demonstrating that decidual angiogenesis requires RAR signaling. Taken together, these data suggest that disruption of the post‐implantation decidual vasculature is the primary cause of BMS493‐induced pregnancy failure by mid‐gestation. BMS493 mediated disruption of RAR signaling alters angiogenic gene expression and decidual vasculature. (A) E7.5 spatial transcriptomics analysis of angiogenic genes in the gastrulation stage conceptus and uterine microenvironments, the mesometrial decidua (MD), vascular sinus zone (VSZ), fetal‐maternal interface (FMI) and decidualization zone (DZ). The color of the bubble is associated with the level of gene expression. The size of the bubble indicates the percent of cells expressing the gene. (B) mRNA expression of Angpt2 , Tek , Dll4 , Notch4 , Vegfa , and Vegfr2 in E7.5 decidual sites after treatment with DMSO or 60 μg/g BMS493 from E4.5 to E6.5. Gene expression was determined by RT‐qPCR normalized to levels of 18 s. The average transcript level after DMSO treatment was set to 1. (C–H) Expression of CD34 in E7.5 implantation sites was analyzed by immunohistochemistry. (C) Representative images after DMSO ( n  = 4 dams) or 60 μg/g BMS493 ( n  = 4 dams) was administered from E4.5 to E6.5. (D) Schematic to illustrate how the decidua and decidual vessels were analyzed. The total area of the decidua (highlighted yellow) was calculated as the total area of the decidual site minus the area of the implantation chamber. The anti‐mesometrial region (highlighted blue) excludes the implantation chamber and vascular sinus folds. The area of decidua (E), %CD34+ endothelial cells in the decidua (F), relative area of the anti‐mesometrial region (G) and %CD34+ endothelial cells in the anti‐mesometrial region were determined in two implantation sites per dam. Each dam is represented by a unique shape. Data in B and E–H are presented as median + IQR. Asterisks denote statistical significance, which was determined by Mann Whitney U test. * p  < .05, *** p  < .001. AM, anti‐mesometrial; AMR, anti‐mesometrial region; CTL, control; e, embryo; ic, implantation chamber; ns, not significant. Spiral arterioles arise from branches of the uterine arteries, the primary vessels supplying the uterus, and provide blood flow to the endometrium. 1 After embryo implantation, angiogenesis and vascular remodeling in the endometrium result in the formation of the vascular networks that comprise the decidual vasculature. To address the possibility that BMS493 directly affects the uterine vasculature, with the disrupted decidual vasculature as a downstream consequence, we assessed the impact of BMS493 on the vessel architecture of the uterus during early pregnancy. To reduce the contribution of varying levels of estrogen and/or progesterone on the vasculature, we initiated inhibition of RA signaling in hormonally synchronized mice at E0.5, immediately after the high estrogen estrus phase and successful mating. Starting at E0.5 and ending at E2.5, mice received 60 μg/g BMS493 or DMSO control each day. At E3.5, uteri were removed and then subjected to whole tissue analysis of CD31 expression. We observed no overt differences in the distribution of the vessels at the mesometrial or anti‐mesometrial poles or in the subepithelial space (Figure  S2 ). With these data, we conclude that BMS493 does not impact the vessel architecture of the uterus, and that the abnormal vascularization at the decidual sites is not due to the systemic action of BMS493 on uterine vasculature. Stromal cell decidualization in humans occurs with every ovulatory menstrual cycle, most frequently in the absence of embryo implantation. To investigate the role of RA signaling in human endometrium and human endometrial stromal cells (HESCs), we collected endometrial tissue in the proliferative and mid‐secretory phases of the menstrual cycles from women with proven fertility. Baseline demographic characteristics (Table  1 ) of proliferative and mid‐secretory phase subjects were similar, with ages of 26.5 ± 2.5 and 28.3 ± 5.3 years and BMI of 28.7 ± 4.8 and 28.9 ± 8.3 kg/m 2 , respectively. On the day of biopsy, serum E2 levels were similar (proliferative: 139.2 ± 90.7 vs. mid‐secretory: 116.7 ± 29.3 pg/mL, p  = .65), and as expected, serum P4 levels were significantly lower in the proliferative as compared to the mid‐secretory phase (0.4 ± 0.3 ng/mL vs. 10.8 ± 4.5, p  = .004). Histological dating of endometrial glands and stroma confirmed that exposure to post‐ovulatory ovarian hormones induced endometrial remodeling in vivo (Figure  5A,B ). We isolated pHESCs from four proliferative phase samples (PR1‐PR4) and decidualized them in vitro for six days with a combination of E2/P4 and cAMP (referred to as EPC differentiation media) (Figure  5C–F ). The significant rise in IGFBP1 and PRL mRNA levels was consistent with the epithelioid‐like morphology of day 6 decidualized primary HESCs (dpHESCs) (Figure  5D–F ). Demographic and clinical characteristics of study subjects. Note : All mid‐secretory phase biopsies well collected 9 days after detection of the urinary LH surge. Abbreviations: BMI, body mass index; E2, estradiol; P4, progesterone. Expression of RA pathway components in human endometrium and primary HESCs. Representative H&E images of proliferative (A) and mid‐secretory (B) phase human endometrium. (C) mRNA expression of RA pathway genes in the human endometrium during the proliferative and mid‐secretory phases ( n  = 4 subjects in each phase). pHESCs were isolated from the proliferative phase of four subjects denoted as PR1, PR2, PR3, PR4, and maintained in basal media. (D) Representative image of pHESCs on day 0 (D0). pHESCs were exposed to differentiation media containing estradiol (E2), progesterone (P4), and cAMP (EPC) for 6 days. (E) Representative image of dpHESCs on day 6 (D6) after exposure to EPC differentiation media. mRNA expression of IGFBP1 (F) and PRL (G) in day 0 pHESCs and day 6 dpHESCs. (H) mRNA expression of RA pathway genes in day 0 pHESCs and day 6 dpHESCs. Gene expression was determined by RT‐qPCR normalized to levels of 18 s. The average transcript level in proliferative phase endometrium and day 0 pHESCs was set to 1. Data are presented as median + IQR. Statistical difference was determined by Mann Whitney U test. * p  < .05, ** p  < .01, **** p  < .0001. Scale bars A and B = 50 μm. Scale bars D and E = 200 μm. dpHESCs, decidualized primary HESCs; MS, mid‐secretory; ns, not significant; pHESCs, primary HESCs; PROL, proliferative; spA, spiral artery. We then utilized RT‐qPCR to compare expression of RA‐related genes in the proliferative and mid‐secretory phase endometrium and in non‐decidualized (day 0) and decidualized (day 6) primary HESCs (Figure  5G,H ). We found that levels of RBP4 and FABP5 mRNA were increased while levels of CRABP2 mRNA were decreased in mid‐secretory as compared to proliferative phase endometrium. A similar pattern of mRNA expression was observed with increasing decidualization in early mouse pregnancy (Figure  1 ). After in vitro decidualization of pHESCs, mRNA expression of RBP4 increased 380‐fold, which is similar to the 273‐fold increase in expression in the mid‐secretory endometrium. In contrast to what we observed in vivo, FABP5 mRNA levels decreased in dpHESCs and CRABP2 mRNA levels varied. mRNA expression of ALDH1A3 , one of the enzymes converting retinaldehyde to RA that increased with decidualization in mouse pregnancy, increased in the mid‐secretory endometrium but decreased in dpHESCs. mRNA expression of genes encoding nuclear receptors, PPARD and RARA , was decreased in mid‐secretory as compared to proliferative phase endometrium. In contrast, with in vitro decidualization of pHESCs, expression of RARA mRNA increased and PPARD mRNA levels remained unchanged. CEBPB was recently described as an effector of the RA pathway in immortalized HESCs. 50 mRNA levels of CEBPB were similar in proliferative and mid‐secretory endometrium but significantly increased with in vitro decidualization of pHESCs. Taken together, the mouse and human data show conserved expression patterns for many RA‐related genes in the uterus/decidua of early mouse pregnancy and human endometrium during ovulatory menstrual cycles. Differences in gene expression in the human endometrium and pHESCs are likely due to inherent differences in these samples; biopsies of proliferative and mid‐secretory phase endometrium contain multiple cell types, including stromal fibroblasts, epithelial, immune, and endothelial, whereas pHESCs are stromal fibroblasts cultured in vitro. RAR signaling is essential for decidualization of immortalized HESCs 26 , 50 but, to our knowledge, the role of RARA in pHESCs and dpHESCs derived from proliferative phase endometrium has not been reported. We sought to determine whether RA signaling via RARA is necessary for the initiation of pHESC decidualization (Figure  6A ), the maintenance of the decidualized state of dpHESCs (Figure  6B ), or both. To determine whether RA signaling via RARA is required for initiation of decidualization, we transfected pHESCs with small interfering (si) RNA specific for RARA or non‐target (NT) siRNA and then exposed the transfected cells to EPC differentiation medium to induce decidualization (Figure  6A ). After exposure to EPC differentiation medium for 6 days, mRNA expression of RARA, IGFBP1 , and PRL was significantly reduced in pHESCs transfected with RARA siRNA (Figure  6E–G ), showing that reduction of RA signaling in pHESCs via silencing RARA impairs the initiation of decidualization. To interrogate the requirement for RA signaling via RARA for maintenance of the decidualized state, pHESCs were decidualized by exposure to EPC differentiation medium for 6 days. On day 6, dpHESCs were transfected with NT control siRNA or RARA siRNA, followed by a 6‐h recovery and then 6 additional days of exposure to EPC differentiation medium (Figure  6B ). In the NT siRNA controls, expression of IGFBP1 and PRL on days 6, 9, and 12 was similar (Figure  6C,D ). After 12 days of exposure to EPC differentiation medium, mRNA expression of RARA was significantly reduced, consistent with inhibition of RA signaling in dpHESCs (Figure  6H ). mRNA expression of IGFBP1 and PRL varied among the four dpHESCs that had been transfected with RARA siRNA. Compared to controls, expression of IGFBP1 and PRL in dpHESCs with RARA silenced was lower in cells derived from PR1 but higher in cells derived from PR3 and PR4 (Figure  6I,J ), demonstrating a variable response of dpHESCs to RARA silencing. RARA knockdown inhibits initiation but not maintenance of decidualization in primary HESCs. (A) Experimental design to assess initiation of decidualization in pHESCs transfected with RARA siRNA. (B) Experimental design to assess maintenance of decidualization in dpHESCs transfected with RARA siRNA. mRNA levels of IGFBP1 (C) and PRL (D) measured on day 0 (D0) pHESCs isolated from the proliferative phase of four subjects denoted as PR1, PR2, PR3, PR4), day 6 (D6) dpHESCs, day 9 (D9) (3 days after dpHESCs were transfected with non‐target siRNA), and day 12 (D12) (6 days after dpHESCs were transfected with non‐target siRNA. RARA (E), IGFBP1 (F), and PRL (G) mRNA levels in pHESCs transfected with non‐target or RARA siRNA and then exposed to EPC differentiation media for 6 days. RARA (H), IGFBP1 (I), and PRL (J) mRNA levels in dpHESCs transfected with non‐target or RARA siRNA and then exposed to EPC differentiation media for an additional 6 days. Gene expression was analyzed via RT‐qPCR normalized to levels of 18 s. Data are presented as median + IQR. Statistical difference was determined by Mann–Whitney U tests or Kruskal–Wallis tests with post hoc multiple comparisons tests. ** p  < .01, *** p  < .001, **** p  < .0001. dpHESCs, decidualized primary HESCs; pHESCs, primary HESCs; ns, not significant. Angiogenesis and vascular remodeling are integral to the preparation of the human endometrium for embryo implantation. 1 To begin to understand how RA signaling could impact vascular development and differentiation in the human endometrium, we first determined mRNA expression of key angiogenic factors, VEGFA and ANGPT2 in human endometrium and pHESCs. mRNA levels of VEGFA were consistently higher in mid‐secretory versus proliferative phase endometrial samples (Figure  7A ), whereas levels of ANGPT2 were similar or higher in the mid‐secretory versus proliferative phase (Figure  7B ). Expression of VEGFA and ANGPT2 varied among the day 6 dHESCs, with higher mRNA levels of both VEGFA and ANGPT2 in cells derived from PR1 and PR2 and lower mRNA levels in cells derived from PR3 and PR4 compared to non‐decidualized day 0 pHESCs (Figure  7C,D ). When expression of RARA was silenced in pHESCs, decidualization was inhibited (Figure  6E–G ). With reduced RAR signaling, we did not observe a consistent change in expression of VEGFA or ANGPT2 among the four pHESCs (Figure  7E,F ), suggesting heterogeneity among pHESCs that was also observed in the dpHESCs derived from them. RARA knockdown in primary HESCs does not alter VEGFA and ANGPT2 levels. VEGFA (A) and ANGPT2 (B) mRNA expression in proliferative and mid‐secretory phase human endometrium ( n  = 4 subjects in each phase). VEGFA (C) and ANGPT2 (D) in day 0 (D0) pHESCs isolated from the proliferative phase of four subjects denoted as PR1, PR2, PR3, PR4 and day 6 (D6) dpHESCs. VEGFA (E), and ANGPT2 (F) mRNA levels in pHESCs transfected with non‐target or RARA siRNA and then exposed to EPC differentiation media for 6 days. Gene expression was determined by RT‐qPCR normalized to levels of 18 s. Data are presented as median + IQR. Statistical difference was determined by Mann–Whitney U test. * p  < .05, ** p  < .01. dpHESCs, decidualized primary HESCs; pHESCs, primary HESCs; MS, mid‐secretory; PROL, proliferative.

An important metabolic regulator in endometrium, RA is integral to the establishment and maintenance of normal mammalian pregnancy 26 , 36 , 50 , 60 , 61 , 62 , 63 ; however, the unique requirements for RA signaling via its receptors, RARs and PPARD, have not been fully elucidated. In the current report, we show that systemic disruption of RA/RAR signaling during the peri‐implantation period leads to infertility in mice, and that RA/RAR signaling is required for the initiation, but not maintenance of primary HESC decidualization in vitro. To investigate RA/RAR signaling in mice, we created a pharmacologic model in which RA/RAR signaling was attenuated with daily administration of pan RAR receptor antagonist BMS493 (60 μg/g of body weight) to pregnant dams. When BMS493 was initiated before implantation (E2.5), the lack of pregnancies observed at E7.5 could be due to failed embryo implantation or pregnancy loss after implantation. When BMS493 was initiated post‐implantation (E4.5), there was a reduction in the decidual vasculature prior to placentation (E7.5) and pregnancy loss before mid‐gestation (E9.5). To investigate RA/RAR signaling in human endometrial stromal cells, we transfected siRNA specific for RARA into pHESCs and dpHESCs derived from four proliferative phase endometrial biopsies. Suppressing RARA expression levels in pHESCs inhibited their decidualization. In contrast, when RARA expression was suppressed in dpHESCs, decidualization was maintained. Both RARs and PPARD are expressed in the mouse decidua, in pHESCs and in dpHECs, but, as previously observed, 26 , 50 , 60 RA/PPARD signaling does not compensate for loss of RA/RAR signaling in vivo or in vitro. From conception to the start of placentation, development and differentiation of multiple uterine cell types, including epithelial, endothelial and stromal, and vascular remodeling are necessary for the progression of pregnancy. The RA signaling pathway, a pathway known to be required for pregnancy success, 26 , 64 is involved in many of these processes; however, there are few reports describing the spatiotemporal expression of RA pathway components in the pregnant rodent uterus prior to placentation. 25 , 65 , 66 , 67 , 68 We determined the spatio‐temporal expression of genes involved in RA synthesis and signaling in the pregnant mouse uterus, prior to placenta formation. From E0.5 to E7.5, expression of Rara , Ppard , and Cebpb increased 2‐fold, Rbp4 , Rarb , and Gja1 increased 10‐20‐fold, and Rdh10 decreased 2‐fold. After embryo implantation, the 48‐h period from E5.5 and E7.5 is characterized by stromal cell proliferation and decidualization, sprouting angiogenesis in the decidua and vascular remodeling. 1 , 69 , 70 During this period, expression of Rbp4 , Fabp5 , and Aldh1a3 increased 2‐10‐fold. Our in‐silico analysis of a published spatial transcriptomic dataset of mouse E7.5 implantation sites 53 demonstrated that many of these genes are expressed in a spatially restricted manner, with highest expression of RA target genes, Gja1 and Cebpb , at the fetal‐maternal interface. Interestingly, as uncovered in our analysis, the fetal‐maternal interface is the region in which genes that regulate stromal cell decidualization, including Bmp2 and Wnt4 , are most abundantly expressed. To our knowledge, we are the first to use spatial transcriptomics technology to identify areas of the pregnant mouse uterus expressing RA synthesis genes. Consistent with prior reports, our findings show tightly regulated expression of genes associated with RA synthesis and signaling in early mouse pregnancy. Moreover, the increased expression of RA synthesis genes and RA target genes suggests that increased endometrial RA signaling is integral to early pregnancy progression. In our study, we initiated pharmacologic attenuation of RA/RAR signaling at two timepoints in early pregnancy, pre‐ and post‐implantation. A prior report showed that endometrial receptivity and decidualization were impaired in mice with conditional expression of dominant negative RARA that disrupted RA signaling in the female reproductive tract. 26 In this genetic mouse model of reduced RA signaling, ovulation and fertilization occurred normally. We also found that disruption of RA/RAR signaling prior to embryo implantation did not interfere with ovulation but compromised fertility. Although our in‐silico analysis of the published spatial transcriptomic dataset showed expression of multiple RA synthesis genes and RA receptors in the conceptus at E7.5, it should be noted that some studies report onset of RAR expression in E4.5 blastocysts, 27 , 71 while others report the onset of RAR and RXR expression at E6.5. 71 , 72 With post‐implantation initiation of RA/RAR signaling attenuation, pregnancies progressed to E7.5 but ended in miscarriage by E9.5. At E7.5 we found normal numbers of implantation sites, with appropriately sized deciduae. Surprisingly, H&E analysis of the fetal‐maternal interface revealed deciduae without embryos as well as deciduae with embryos of variable sizes, which is similar to dams with intact RA/RAR signaling. In mice, pregnancy failure before mid‐gestation occurs because of abnormal decidualization or improper formation of the placenta. Given our long‐standing interest in identifying signaling pathways that mediate decidual angiogenesis, coupled with the involvement of RA signaling in angiogenesis, 73 , 74 , 75 , 76 , 77 , 78 we hypothesized that disrupted RA/RAR signaling alters decidual angiogenesis and the decidual vasculature, and this would be a major contributor to pregnancy loss. Maintenance of the vasculature during the post‐implantation, pre‐placentation period of early mouse pregnancy is an absolute requirement for normal pregnancy progression. 1 , 76 We and others have shown that VEGF and Notch signaling differentially regulate decidual angiogenesis and vascular remodeling, 45 , 79 , 80 , 81 and that reduced signaling causes pregnancy loss prior to placenta formation. Consistent with the requirement for VEGF and Notch, spatial transcriptomic analysis showed expression of VEGF and Notch pathway components throughout the decidual site, with highest expression in the vascular sinus zone, fetal‐maternal interface, and decidual zone. When attenuation of RA/RAR signaling was initiated post‐implantation, reduction of the decidual vasculature, assessed as the proportion of CD34+ cells, was accompanied by a 2–3‐fold reduction in the expression of Notch4 and Vegfr2 , but no change in expression of ligands Dll4 and Vegfa at E7.5. From E4.5 to E7.5, expression of endothelial cell‐associated genes is upregulated, and sprouting and intussusceptive angiogenesis within the decidua contribute to formation of a new capillary network. 1 Thus, our data show that attenuated RA/RAR signaling in decidual sites decreases expression of angiogenic receptors on the endothelium, which disrupts decidual angiogenesis and vascular remodeling and results in pregnancy failure. Aberrations in RA signaling have been implicated in human reproductive pathologies, including endometriosis, recurrent implantation failure, and preeclampsia, that are associated with defective decidualization. 7 , 50 , 61 , 63 , 68 , 82 To understand how disrupted RA signaling might contribute to these clinical conditions, we sought to examine RA synthesis genes and RA/RAR signaling in human endometrial tissues. In our study, proliferative and mid‐secretory phase endometrial biopsies were collected from a small but homogeneous group of Black and Hispanic women. Primary HESCs were isolated from proliferative phase biopsies, cultured, and decidualized; both endometrial tissue and pHESCs were then used to study RA gene expression in non‐decidualized and decidualized states. Progesterone is an essential regulator of stromal cell decidualization, which in vivo, is influenced by complex interactions of transcription factors, cytokines, and cell–cell communications. Decidualization in vitro is induced by estradiol, progesterone, and cAMP. While we observed many similarities in gene expression between decidualized mid‐secretory endometrium and dpHESCs, notable differences in patterns of expression (e.g., ALDH1A3 and RARA ) may be attributed to inherent differences between isolated, cultured endometrial stromal cells and whole tissue. The endometrium is a complex multicellular tissue that undergoes dynamic remodeling throughout the menstrual cycle, 1 , 83 whereas HESCs are one of the cell types within that endometrium. Importantly, a recent single‐cell RNA sequencing study identified and described distinct cell clusters within pHESCs, noting differences between those with gene signatures of mature, proliferative, or active fibroblasts and demonstrating that isolated pHESCs do not represent a homogenous group of cells. Heterogeneity among HESCs likely explains differences in gene expression we observed among the four pHESCs in our study and contributes to differences in gene expression between endometrial tissue and HESCs. Nevertheless, expression of RBP4 , the only specific transport protein for vitamin A, was upregulated following decidualization in all the conditions studied (early mouse pregnancy, human endometrium, and pHESCs), consistent with prior reports that RA uptake and metabolism are involved in decidualization both in vivo and in vitro. In addition to characterizing expression of RA‐synthesizing components in human endometrial samples, we designed experiments to test the requirement of RA/RAR signaling in initiation and in maintenance of decidualization. Prior studies have shown that attenuated RA signaling inhibited decidualization in primary and immortalized HESCs 26 , 33 , 60 , 61 ; however, to our knowledge, prior reports have not addressed the loss of RA/RAR in dpHESCs. We observed a requirement for RA/RAR signaling in initiation, but not the maintenance of decidualization. Our findings support the idea that decidualized HESCs are not simply modified HESCs; they are a distinct cell type resulting from terminal differentiation and the genetic reprogramming of HESCs. 84 Our interest in different molecular programs for initiation and maintenance of decidualization stems from our interest in the clinical conditions of failed implantation versus pregnancy loss in the first trimester, prior to placentation. Identifying distinct requirements for RA/RAR signaling for initiation and maintenance of decidualization increases our interest in investigating RA pathway components as potential therapeutics for infertility and/or pregnancy loss. Our study contributes two novel insights to the existing body of literature on the role of RA signaling in female fertility; RA/RAR signaling is required for maintenance of the decidual vasculature that supports early pregnancy in mice, and distinct RA signaling is required for initiation and maintenance of pHESC decidualization in vitro.

Molecular mechanisms that regulate remodeling of the endometrium to support embryo implantation and progression of early pregnancy remain incompletely defined. In mice, detection of a copulation plug, designated embryonic day (E) 0.5, is the start of pregnancy. The blastocyst stage embryo attaches to the luminal epithelium and implantation, which occurs by E4.5, initiates pro‐pregnancy endometrial remodeling, including stromal cell decidualization, altered immune responses, and angiogenesis. By E5.5, a new, rudimentary vascular network has formed. Extensive post‐implantation decidual angiogenesis and vascular remodeling which occur from E5.5 to E7.5 support pregnancy progression and serve as the scaffold for placentation. 1 , 2 , 3 In humans, endometrial remodeling in preparation for pregnancy proceeds in both non‐conception and conception cycles. In the absence of an embryo, exposure to ovarian‐derived factors including estradiol (E2) and progesterone (P4), initiates stromal cell decidualization as well as functional and phenotypic changes in epithelial, endothelial, and immune cells to create an endometrium which will be receptive to embryo implantation within the mid‐secretory phase of the menstrual cycle. 1 , 4 , 5 Successful embryo implantation then induces further differentiation and remodeling resulting in the decidua that can fully support progression of pregnancy prior to formation of the placenta. 5 In both mice and humans, aberrant endometrial differentiation may manifest as infertility due to implantation failure or as a spectrum of pregnancy disorders, including recurrent miscarriage and fetal growth restriction, that arise from the adverse “ripple effects” of embryo implantation in the setting of defective decidualization. 6 , 7 Retinoids are a group of compounds comprising retinol (ROH) or vitamin A and its metabolites, including retinoic acid (RA). Vitamin A is an essential fat‐soluble micronutrient that is crucial for diverse biological functions, including embryo development and cellular growth and differentiation in adult organs. 8 , 9 During pregnancy, deficiency or excess of vitamin A is well‐known to be associated with the adverse sequela of teratogenesis. 8 , 9 , 10 , 11 , 12 Lecithin: retinol acyltransferase (LRAT) promotes retinyl ester formation and storage of retinoids (Figure  1A ). Within cells ROH is esterified and stored as retinyl esters while in circulation ROH is bound to its sole specific transport protein retinol‐binding protein 4 (RBP4). 13 At the surface of target cells, the ROH/RBP4 complex binds signaling receptor and transporter of retinol 6 (STRA6), a receptor that transfers ROH from RBP4 to RBP1, the intracellular carrier for ROH. 13 , 14 In cells, ROH is oxidized to retinaldehyde by retinol dehydrogenases (RDHs), including RDH10, and then retinaldehyde is oxidized to RA by the action of aldehyde dehydrogenases ALDH1A‐3. 15 , 16 RA is the most bioactive retinoid; it regulates the transcription of over 500 genes 17 , 18 via binding to its classical and non‐classical nuclear receptors, retinoic acid receptors (RARs) and peroxisome proliferator‐activated receptor D (PPARD), respectively, which heterodimerize with retinoid X receptor (RXR) to form a functional receptor unit. 19 , 20 , 21 RA is delivered to its nuclear receptors by cellular retinoic acid binding protein 2 (CRABP2) and fatty acid binding protein 5 (FABP5). 14 Studies have shown that CRABP2 delivers RA to RAR to induce cellular processes like cell arrest and apoptosis while FABP5 delivers RA to PPARD to induce cell survival and differentiation. 22 , 23 Expression of RA pathway components in early mouse pregnancy. Gene expression determined by RT‐qPCR normalized to levels of 18 s. The average transcript level at embryonic day (E) 0.5 uteri was set to 1. (A) Schematic of the retinoid metabolism pathway. (B) Relative mRNA expression of Prl8a2 in E0.5 uteri and deciduae at E5.5 and E7.5. (C) Relative mRNA expression of RA pathway genes in E0.5 uteri and deciduae at E5.5 and E7.5. RNA was extracted from 4 to 8 unique biological samples, each represented by a symbol. Asterisks denote statistical significance, which was determined by Kruskal–Wallis tests with post hoc multiple comparisons tests. Data are presented as median + IQR. * p  < .05, ** p  < 0.01. Previous studies highlight the involvement of RA signaling at the maternal‐fetal interface, in both early embryo development and endometrial differentiation. In mice, multiple RA pathway components are expressed in the uterus during the peri‐implantation period. 24 , 25 When disruption of RA signaling with expression of dominant negative RARA was restricted to the female reproductive tract, endometrial receptivity and decidualization were impaired. 26 Although there are conflicting reports as to whether exposure to inappropriate levels of RA has a negative effect on embryo implantation, 27 , 28 the requirement for RA in early embryo development is well accepted. 29 , 30 Depending on the timing and severity, maternal RA deficiency has been associated with implantation failure or embryonic death in mice. Similarly, in humans, RA receptors ( RAR , RXR , and PPARD ) are highly expressed throughout the endometrium 31 and the RA signaling network is spatially localized to the stromal and epithelial compartments. 31 Several genes associated with RA synthesis, signaling and degradation (e.g. ALDH1A1‐3 , RARA , RARB , RARG ) are expressed in both the endometrial epithelium and stromal fibroblasts while some, including CYP26A1 , are almost exclusively expressed in the epithelium. 31 Moreover, there is accumulating evidence that tightly regulated RA signaling is required for normal endometrial differentiation. While recent studies support a role for RA signaling via its non‐classical receptor PPARD in driving human endometrial stromal cell (HESC) decidualization in vitro, 32 , 33 there are conflicting reports as to whether RA signaling via the classical receptor RARA inhibits or is required for in vitro decidualization of HESCs. 26 , 33 , 34 , 35 Given that many molecules in the RA signaling pathway, especially the RARs and PPARD, are major drug targets for clinical conditions including neurodegenerative disease and cancers, 36 , 37 , 38 , 39 , 40 further investigation of the role of RA signaling in endometrial remodeling and early pregnancy progression could lay the foundation for developing RA‐based therapeutic approaches for implantation failure and/or defective decidualization. In the current study, we utilized mouse models, primary human cells, and pharmacological tools to develop a mechanistic understanding of how RA promotes embryo implantation and pregnancy success. We inhibited RA/RARA signaling in primary (p) HESCs with silencing (si) RNA and administered BMS493, an inverse pan‐RAR agonist that prevents RA‐induced differentiation, to pregnant mice. 41 , 42 We found that RA/RARA signaling is required for the initiation but not maintenance of decidualization of pHESCs. Our in vivo model revealed that disruption of RAR signaling prior to embryo implantation results in implantation failure, whereas disruption of RAR signaling after embryo implantation results in pregnancy failure by mid‐gestation, with decreased expression of angiogenic genes and reduction of the post‐implantation decidual vasculature. Herein, we show the unique expression of RA synthesis genes and receptors in mouse and human endometrial tissues; the distinct requirement for RA/RAR signaling in non‐decidualized versus decidualized pHESCs; and the requirement for functional RAR signaling for maintenance of the decidual vasculature.

The authors have declared that no conflict of interest exists.

Our study exclusively examined female mice because pregnancy is only relevant in females. C57BL/6J wildtype female and male mice (Jackson Laboratories) were maintained under standard conditions. Female mice, 8–10 weeks of age and weighing ≥20 g, were mated with stud males, and noon on the day a mating plug was observed was designated as E0.5. Mice were euthanized on E0.5, E3.5, E5.5, E7.5, and E9.5. Ovaries were collected and embedded in Tissue‐Tek® O.C.T. Compound (Sakura Fine Technical). At E0.5 and E3.5, uteri were collected. At E5.5 and E7.5, individual decidual sites (implantation sites with myometrium removed) were isolated and placed in RNAprotect (Qiagen) for RT‐qPCR analyses. At E7.5, individual implantation sites were isolated and processed for analysis of embryo size and vasculature. Beginning on E2.5 or E4.5, pregnant C57BL/6J wildtype mice were administered the pan RAR receptor inverse agonist BMS493 42 (Sigma B6688—25 mg) dissolved in DMSO and diluted in corn oil at 15 μg/g, 30 μg/g or 60 μg/g body weight or vehicle (equivalent volume of DMSO in corn oil) via daily oral gavage until E6.5. Dams were euthanized on E7.5 or E9.5. Livers were collected. Total RNA was isolated from the livers and tested using RT‐qPCR for expression of Cyp26a1, Cyp26b1, Rarb , genes known to be indicators of RA signaling activity. 43 Beginning on E0.5, C57BL/6J wildtype mice were administered BMS493 dissolved in DMSO and diluted in corn oil at 60 μg/g body weight or vehicle (equivalent volume of DMSO in corn oil) via daily oral gavage until E2.5. Uteri were removed on E3.5, stained for whole tissue immunofluorescence and analyzed as described previously. 44 Briefly, uteri were fixed in DMSO: methanol (1:4) after dissection and stored at −20°C. For immunostaining, samples were rehydrated in methanol: PBST (PBS, 1% triton) (1:1) solution followed by washing in 100% PBST. Samples were blocked in a PBST solution (PBS, 1% triton, 2% powdered milk) at room temperature and then incubated with primary antibody Armenian‐hamster anti‐CD31 (1:200, DSHB, AB_2161039) in the blocking solution for seven nights at 4°C. Samples were then washed with PBST solution and incubated with secondary antibody goat anti‐Armenian hamster IgG 647 (1:500, A78967, Invitrogen) and Hoechst (1:1000, B2261, Sigma Aldrich) for three nights at 4°C. Samples were again washed with PBST solution and then incubated at 4°C overnight with 3% H 2 O 2 solution prepared in methanol. Following the overnight incubation, the samples were dehydrated in 100% methanol at room temperature and cleared overnight with benzyl alcohol: benzyl benzoate (1:2) (108006, B6630, Sigma‐Aldrich). Samples were imaged using a Leica SP8 TCS white light laser confocal microscope (Leica, Wetzlar, Germany) utilizing a 10× air objective and a 7.0 μm Z stack. 44 For image analysis, Imaris v9.2.1 (Bitplane; Oxford Instruments, Abingdon, UK) commercial software was used to analyze the files post confocal imaging (.LIF format). 35 μm thick XY slices and 14.4 μm thick XZ slices were captured using the snapshot function. Implantation sites were fixed in 4% paraformaldehyde (PFA) in phosphate buffered saline (PBS), embedded in wax, sectioned at 7 μm and stained with hematoxylin and eosin (H&E). Interembryonic regions and central parts of the decidua were identified by H&E staining. For each implantation site, the number of sections having an embryo was counted. Immunostaining was performed, as previously described. 45 , 46 To identify the vasculature in the implantation site, rabbit anti‐mouse CD34 primary antibody (ab81289, Abcam, Cambridge, UK) and goat anti‐rabbit biotinylated secondary antibodies (Vector Laboratories, Burlingame, CA) were used. To identify vasculature in the ovary, rat anti‐mouse CD31 primary antibody (ab81289, Abcam, Cambridge, UK) and goat anti‐rabbit biotinylated secondary antibodies (Vector Laboratories, Burlingame, CA) were used in conjunction with Vectastain Standard ABC Elite kit (Vector). The color reaction was performed with DAB Substrate 3,3′‐diaminobenzidine in the presence of peroxidase (HRP) enzyme. The slides were imaged using a Keyence BZ‐X710 microscope (Keyence Corp, Elmwood Park, NJ) and analyzed using ImageJ Fiji (ImageJ2, Version 2.9.0/1.53t). Imaged sections spanning at least 14 μm centered around the core of the decidua were selected. Measurements from 2 to 4 sections per decidual site were averaged to determine the area of the decidua (μm 2 ), area of the anti‐mesometrial region (μm 2 ), and amount of CD34 expression. For each pregnant dam, two decidual sites were scored. Spatial transcriptomics data were downloaded from GEO ( https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE181169 ). The matrix data were imported using Seurat R package (v4.1.1). Gene counts normalization was done using SCtransform function in Seurat. Cells were clustered using the Seurat FindClusters function, and clusters were identified as specific cell types using canonical cell type markers. Data visualization was performed using DotPlot and VlnPlot function from Seurat. Endometrial biopsies and blood samples were obtained from women who were recruited from the Ambulatory Care Clinics at University Hospital. Subjects were of reproductive age, with regular menstrual cycles (between 21 and 35 days), no significant medical or reproductive history, and had not used hormonal treatment in the 3 months prior to recruitment. All subjects had prior pregnancies, with uncomplicated full‐term live births. Subjects were confirmed to be ovulatory with urine luteinizing hormone (LH) testing and a serum progesterone level >3 ng/mL in the secretory phase in a cycle prior to enrollment. Endometrial biopsies were obtained using the Endocell ® disposable endometrial cell sampler (Wallach Surgical Devices, Trumbull, CT, United States). Four biopsies were obtained in proliferative phase between cycle days 10–13 and four in the mid‐secretory phase 8–9 days after detection of the LH surge with urine ovulation predictor kits. A venous blood draw was performed on the day of endometrial biopsy to measure serum hormone levels. Endometrial tissue was washed with cold PBS, finely minced, and enzymatically digested in 10 mL prewarmed RPMI medium with 3% FBS and 1% pen/strep with 1 mg/mL collagenase A (Roche, ref.:11088793001), and 0.1 mg/mL DNase I (Roche, ref.:10104159001). The digest was shaken at 250 rpm 37°C for 15 min. All tissue digests were passed through an 18G needle and 40 μm cell strainers. The digests were centrifuged, and the cell pellet was resuspended with 1 mL 1× PBS. Red blood cells (RBCs) were removed with RBC lysis buffer (ThermoFisher, cat. #: 00‐4333‐57) following the manufacturer's protocol. Endometrial cells were cultured in phenol red‐free Dulbecco's modified Eagle's (DMEM)/F‐12 medium containing 10% charcoal‐stripped fetal bovine serum (FBS) (ThermoFisher Scientific Cat # A3382101), 1% penicillin (pen) and streptomycin (strep), and 10 ng/mL bFGF (Sigma GF003). This medium is called basal medium (BM). Endometrial cells were passaged twice, and the resulting isolated stromal fibroblasts were frozen as pHESCs. 47 , 48 To induce decidualization, pHESCs were cultured in phenol red‐free Dulbecco's modified Eagle's (DMEM) /F‐12 medium containing 2% charcoal‐stripped FBS, 1% penicillin/streptomycin and final concentrations of 0.01 μM estradiol (Sigma E8875), 1 μM medroxyprogesterone acetate (Sigma M1629), and 500 μM N6,2'‐O‐dibutyryladenosine 3′:5′‐cyclic monophosphate (db‐cAMP) (Sigma Cat # D0627). This estradiol, progesterone, and c‐AMP (EPC) differentiation medium (DM) was changed every 48 h for 6 or 12 days. Decidualization was confirmed by visualizing morphological changes in cell shape from fibroblastic to epithelioid and increased mRNA expression of IGFBP1 and PRL (Figure  5C,D ). siRNA transfection was performed according to the manufacturer's instructions. pHESCs were transfected with siRNA specific for human RARA (Dharmacon Research, Inc., L‐003437‐00‐0005) or human non‐target (NT) siRNA (Dharmacon Research, Inc., D‐001810‐10‐05) as a control. siRNA stock solution was diluted to 60 nM/well with Opti‐MEM and incubated for 5 min at room temperature. 30× lipofectamine RNAiMAX reagent (Invitrogen, #13778) stock solution was diluted to 1× with Opti‐MEM and incubated for 5 min at room temperature. 150 μL of diluted lipofectamine and 100 μL of siRNA per well were combined and incubated for 20 min at room temperature. Cells were washed with PBS and replaced with 1 mL transfection medium (phenol red‐free DMEM/F12 containing 2% charcoal‐stripped FBS) to each well of a 6‐well plate. 250 μL siRNA‐lipofectamine combined solution was added to each well of the 6‐well plate by pipette dropping, diffuse coverage was ensured by gentle rocking, and cells were incubated for 6 h at 37°C in a 5% CO₂ incubator. After 6 h, the medium was exchanged for basal medium (phenol red‐free DMEM/F‐12 medium containing 10% charcoal‐stripped FBS and 1% pen/strep). RNA was extracted from the uteri at E0.5, from individual decidua with myometrium removed at E5.5 and E7.5, and hESCs using the Rneasy Mini Kit (Qiagen). The RNA was reverse transcribed into cDNA and relative gene expression was measured using qPCR with QuantiNova SYBR Green PCR Kit (Qiagen). Relative mRNA expression levels were quantified using the 2 −ΔΔCT method and expressed as fold change normalized to 18 s expression. Primers are listed in the Table  S1 . Normality was determined using the Shapiro–Wilk test. For non‐parametric data, data were compared using the Mann–Whitney U test or Kruskal Wallis with post hoc multiple comparisons tests, and data are expressed as medians with interquartile range (IQR). For normally distributed data, data were compared using unpaired t ‐tests and are expressed as mean ± SD. Statistical analyses were performed using Prism Version 9.0 (GraphPad Prism Software, Inc., La Jolla, CA). Statistical significance was defined as * p  < .05, ** p  < .01, *** p  < .001, **** p  < .0001. Animal studies were approved by the Institutional Animal Care and Use Committee at Rutgers New Jersey Medical School. Written informed consent for the review of medical records, collection of blood, and endometrial sampling was obtained from all human subjects under study #Pro2018002041 which was approved by the Rutgers New Jersey Medical School Institutional Review Board.

Figure S1. . Figure S2. . Table S1. .