Murine uterine gland branching is necessary for gland function in implantation

Abstract Uterine glands are branched, tubular structures whose secretions are essential for pregnancy success. It is known that pre-implantation glandular expression of leukemia inhibitory factor (LIF) is crucial for embryo implantation; however, the contribution of uterine gland structure to gland secretions, such as LIF, is not known. Here, we use mice deficient in estrogen receptor 1 (ESR1) signaling to uncover the role of ESR1 signaling in gland branching and the role of a branched structure in LIF secretion and embryo implantation. We observed that deletion of ESR1 in neonatal uterine epithelium, stroma, and muscle using the progesterone receptor PgrCre causes a block in uterine gland development at the gland bud stage. Embryonic epithelial deletion of ESR1 using a Müllerian duct Cre line, Pax2Cre, displays gland bud elongation but a failure in gland branching. Reduction of ESR1 in adult uterine epithelium using the lactoferrin-Cre (LtfCre) displays normally branched uterine glands. Unbranched glands from Pax2Cre Esr1flox/flox uteri fail to express glandular pre-implantation Lif, preventing implantation chamber formation and embryo alignment along the uterine mesometrial–antimesometrial axis. In contrast, branched glands from LtfCre Esr1flox/flox uteri display reduced expression of ESR1 and glandular Lif resulting in delayed implantation chamber formation and embryo–uterine axes alignment but mice deliver a normal number of pups. Finally, pre-pubertal unbranched glands in control mice express Lif in the luminal epithelium but fail to express Lif in the glandular epithelium, even in the presence of estrogen. These data strongly suggest that branched glands are necessary for pre-implantation glandular Lif expression for implantation success. Our study is the first to identify a relationship between the branched structure and secretory function of uterine glands and provides a framework for understanding how uterine gland structure–function contributes to pregnancy success.


Introduction
Uterine glands are key to embryo implantation and pregnancy success.In viviparous mammals in the absence of yolk, uterine glands support the pregnancy and embryo development until the placenta forms (Kelleher et al., 2019).Uterine glands are present in humans, rodents, sheep, pigs, and horses among other studied mammals (Gray et al., 2001).Uteri of neonatal mice are devoid of glands at birth.At postnatal day (P) 5, gland buds protrude off the uterine lumen on the anti-mesometrial and lateral sides of the murine uterus and extend into the surrounding stroma (Vue et al., 2018).Uterine glands are branched in the pubertal nonpregnant and pregnancy stages (Arora et al., 2016).Although uterine glands are categorized as exocrine glands, their mechanisms of branching have not been identified unlike those of other exocrine glands (e.g.salivary glands) (Khan et al., 2022).Uterine glands facilitate the secretions of vital factors, including the key cytokine leukemia inhibitory factor (LIF) that is essential for embryo implantation (Stewart et al., 1992;Chen et al., 2000).Although the relationship between exocrine gland structure and function has been established in the case of meibomian, mammary, and salivary glands (Khan et al., 2022), this relationship in uterine glands is still unknown.
Estrogen (E2) signaling facilitates the growth and development of the uterus as suggested by various estrogen receptor (ESR) deletion models.The ESR has two isoforms, ESR1 and ESR2, which are encoded by separate genes located on different chromosomes.Both receptor isoforms are nuclear receptors and act as ligand-activated transcription factors that can alter target gene transcription (Cui et al., 2013).Female mice with a whole body ESR1 deletion are infertile (Hewitt et al., 2000).These females fail to ovulate and exhibit hypoplastic uteri that fail to display cyclic changes (Dupont et al., 2000).In contrast, females with a whole body ESR2 deletion have sub-fertility with a reduced litter size (Hewitt et al., 2000).Mice with conditional deletion of Esr1 in different compartments have been generated to highlight the celltype-specific contributions of ESR1 signaling in the uterus.Deletion of ESR1 in the neonatal epithelium, stroma, and muscle, using progesterone receptor driven Pgr Cre (where the endogenous Pgr gene is replaced by Cre recombinase gene) and in embryonic epithelium using Wnt7a Cre (where the endogenous Wnt7a gene is replaced by Cre recombinase gene) results in infertility (Winuthayanon et al., 2010;Pawar et al., 2015).Implantation has not been assessed in a Pgr Cre model and epithelial deletion of ESR1 using Wnt7a Cre results in protease-mediated embryo death in the oviduct (Winuthayanon et al., 2015).Blastocyst stage embryo transfer studies into pseudopregnant uteri of the latter mice suggest that embryo implantation is not supported in this model (Winuthayanon et al., 2010).
Gland development is variably responsive to ovarian E2 depending on the species.In the neonatal pig, gland development is both ESR1-dependent and sensitive to E2 levels (Tarleton et al., 1999(Tarleton et al., , 2003)).Unlike the pig, in sheep, ESR1-inhibition does not impact gland initiation (Carpenter et al., 2003) but reduces the number of glands, and reduces branches and coils in the uterine glands (Carpenter et al., 2003).In neonatal mice, the initiation of gland formation is thought to be ovary-, adrenal-, and steroidindependent (Ogasawara et al., 1983;Bigsby and Cunha, 1985).However, treatment with Genistein (an ESR1 agonist) during the neonatal period prevents gland development and results in implantation failure (Jefferson et al., 2020).Gland structure in Esr1deficient mice has not been assessed.
Mouse mutants with an absence of uterine glands or key glandular secretion of LIF display implantation and decidualization failure (Filant and Spencer, 2013).Examples include: micedeficient in Wnt7a (Dunlap et al., 2011); Foxa2 (Jeong et al., 2010;Kelleher et al., 2017); progesterone-induced gland knockout (Filant et al., 2012); and Lif (Stewart et al., 1992).The luminal epithelium, glandular epithelium, and stromal compartment all possess the ability to express Lif, and the coordination of LIF secretion among these compartments presumably happens in accordance with steroid hormone levels during both the mouse estrous cycle and early pregnancy (Rosario and Stewart, 2016).ESR1 signaling is critical for Lif induction.Ovulatory E2 coincides with ESR1 and Lif expression in the uterus (Shen and Leder, 1992;Song et al., 2000) and, prior to implantation, uterine glands express ESR1 and LIF (Song et al., 2000).A large dose of E2 has the effect of inducing robust Lif expression, and this response is absent in ESR1 knockout mice (Hewitt et al., 2002;Winuthayanon et al., 2010).However, the compartment in which Lif is expressed (luminal or glandular epithelium) when stimulated by a large dose of E2 remains unknown.Wnt7a Cre deletion of ESR1 results in a reduction of Lif and failed oil-mediated artificial decidualization.Exogenous Lif supplementation rescues artificial decidualization in this model (Pawar et al., 2015).This highlights the need for epithelial ESR1 signaling for pre-implantation Lif production in early pregnancy.
Studies have demonstrated that there is a link between ESR1 signaling and mammary gland branching such that both the complete abrogation of ESR1 signaling and epithelial-specific deletion of ESR1 result in significantly reduced mammary gland branching and consequently diminished lactational function (Bocchinfuso et al., 2000;Sternlicht et al., 2006;Feng et al., 2007).While the role of ESR1 signaling has been separately studied during uterine development and for pre-implantation Lif expression, a connection between gland structure and function is yet to be made.In this study in mice, we determine the relationship between E2 signaling, development of a branched gland structure, and pre-implantation glandular LIF production.We determine that neonatal stromal ESR1 signaling is key to gland elongation and pubertal ESR1 signaling is necessary for gland branching.We also uncovered that unbranched, ESR1-deficient glands fail to express Lif, whereas branched glands with much reduced levels of ESR1 can still produce enough LIF to support embryo implantation and pregnancy.Finally, we show that pre-pubertal uteri support Lif expression in the luminal epithelium but not in the unbranched glands, suggesting that gland structure (branching) is necessary for gland function (glandular LIF production and implantation).

Animals
Esr1 flox/flox mice (C57Bl6 background) were provided by Dr Pierre Chambon (Gieske et al., 2008).These mice were bred with Pgr Cre (Soyal et al., 2005) (C57Bl6 background), Pax2 Cre (Ohyama and Groves, 2004) (mixed background) (where the endogenous Pax2 gene is replaced by Cre recombinase gene), Wnt7a Cre (Winuthayanon et al., 2010) (C57Bl6 background), or Ltf Cre (Daikoku et al., 2014) (mixed background) (where the endogenous Ltf gene is replaced by Cre recombinase gene) mouse lines to generate tissue-and time-specific deletion of ESR1 (Table 1).Esr1 flox/flox females were used as controls.For pregnancy studies, females were mated with fertile CD1 males and the appearance of a vaginal plug was identified as gestational day (GD) 0 1200 h.Adult females >6 weeks were used for Pgr Cre Esr1 flox/flox (hereafter referred to as PER) and Pax2 Cre Esr1 flox/flox (hereafter referred to as XER) deletion models.In the Ltf Cre Esr1 flox/flox deletion model (hereafter referred to as LER), because Cre expression comes on at puberty (�6 weeks), adult females aged 10-12 weeks were used to allow sufficient time for Esr1 excision.For the pups study, LER virgin females were set up with CD1 males, monitored for pregnancy, and the number of live pups was recorded at birth.Wnt7a Cre Esr1 flox/flox mice (hereafter referred to as WER) (Hewitt et al., 2010) were used for analysis of uterine gland structure to compare with XER mice.These Esr1 flox/flox mice were generated independently by Dr Kenneth Korach (Hewitt et al., 2010); however, the Esr1 flox/flox mice used in our studies have exon 3 of the Esr1 gene flanked by LoxP sites and result in complete absence of the ESR1 protein.To detect implantation sites, female mice at GD4 1200 h mice were injected i.v. or retroorbitally with blue dye prior to dissection.All animals were maintained and handled according to the Michigan State University Institutional Animal Care and Use Committee guidelines.

LIF, hormone, and inhibitor treatments
For Lif rescue experiments, XER mice were injected i.p. at 1000 and 1800 h on GD3 with either PBS or 10 mg recombinant Lif (554006, BioLegend, San Diego, CA, USA) followed by blue dye injection and dissection at GD4 1200 h.Alternatively, one uterine horn of XER mice was injected intraluminally with 1 mg recombinant Lif at 1300 h on GD3.The partner horn was left untreated as a control.Mice were then dissected on GD4 1200 h following blue dye injection.For exogenous hormone treatments, 17β-estradiol (E2) (E8875, Sigma-Aldrich, St Louis, MO, USA) and progesterone (P4) (P0130, Sigma-Aldrich) were dissolved in sesame oil and three injection schemes were used.To test Lif induction by ovarian hormones in pre-pubertal mice, control P21 mice were injected s.c. with either 100 ng E2 or sesame oil at 1200 on P21 and 0900 h on P22 before dissection at 1200 h on P22.Alternatively, control P21 mice were injected s.c. with 100 ng E2 at 1200 h on P21 and P22.On P23, mice were injected with 1 mg P4 þ 6.7 ng E2 at 0900 h and subsequently dissected at 1200 h.

Cryoembedding, cryosectioning, and immunostaining
For cryoembedding, uterine horns were dissected and fixed in 4% paraformaldehyde overnight at 4 � C. The next morning, uteri were washed three times, 5 min each, with PBS and left upright in a solution of 10% sucrose overnight.Uteri were then transferred into 20% and 30% sucrose solutions, sequentially, for 2-3 h each.Finally, uteri were embedded longitudinally in Tissue-Tek OCT (45831, Andwin Scientific, Simi Valley, CA, USA) and stored at −80 � C. Tissues were cryosectioned at 7 mm, mounted on glass slides (1255015, Thermo Fisher, Waltham, MA, USA), and stored at −20 � C until immunostaining.For ESR1 immunostaining, antigen retrieval was performed by washing slides in PBS before transferring to 1× citrate buffer solution (005000, Thermo Fisher) and boiling in a beaker filled to 3 =4 in the microwave for 10 min.The slides were then washed three times for 5 min each with PBS, blocked with 2% powdered milk in PBS þ 1% Triton X-100 (PBT), and incubated with primary antibodies overnight at 4 � C. The next day, slides were washed three times for 5 min each with PBS, stained with secondary antibodies for 1 h, washed with PBS again, and sealed with a coverslip and nail polish.

Confirmation of Pax2 Cre-lineage
Pax2 lineage was confirmed by breeding the Pax2 Cre mouse line with Gt(ROSA)26Sor tm4(ACTB-tdTomato,-EGFP)Luo/J also referred to as ROSA mT/mG (007576, Jackson Labs, Bar Harbor, ME, USA) reporter mice.Uteri from P21 pups and GD3 1200 h pregnant mice were dissected and embedded for cryosectioning.Cryosections were imaged for endogenous membrane green fluorescent protein (GFP) and membrane tomato signal.Some cryosections were additionally stained with primary antibody for CD31 to identify blood vessels (see next section).

Whole-mount and tissue section immunofluorescence
Whole-mount immunofluorescence was performed as described previously (Arora et al., 2016) Uteri were dissected from mice and fixed in dimethylsulphoxide: methanol (1:4).For immunostaining, uteri were rehydrated in methanol: PBT (1% Triton X-100 in PBS) (1:1) for 15 min, washed in PBT for 15 min and incubated in blocking solution (2% powdered milk in PBT) for 1 h at room temperature.Uteri were incubated with 1:500 concentration of primary antibodies diluted in blocking solution for seven to nine nights at 4 � C.They were then washed twice for 15 min each with 1% PBT followed by three washes for 45 min each at room temperature.Uteri were then incubated with secondary antibodies at 4 � C for two or three nights, followed by one 15 min and three 45 min washes with 1% PBT and dehydration in 100% methanol for 30 min.Uteri were then bleached overnight at 4 � C in a solution of 3% H 2 O 2 in methanol.Finally, the samples were washed in 100% methanol for 1.5 h and cleared in BABB (1:2, benzyl alcohol: benzyl benzoate) (108006, B6630, Sigma-Aldrich).

In situ hybridization
In situ hybridization on uterine sections was performed using the RNAscope 2.

Confocal microscopy
Samples with whole tissue immunofluorescence, section immunofluorescence and in situ hybridization were all imaged using a Leica TCS SP8 X Confocal Laser Scanning Microscope System (Leica, Wetzlar, Germany) with white-light laser, 10× air objective (used for whole tissue) and 20× water objective (used for sections).The entire length and thickness of the uterine horn was imaged using the tile scan function with z stacks 7 μm apart.
For sections, z stacks 1.5 μm apart were used.Images were merged using Leica software LASX version 3.5.5 (Leica).

3D Reconstruction and image analysis
Image analysis was performed using commercial software Imaris v9.2.1 (Bitplane; Oxford Instruments, Abingdon, UK).The confocal image (.LIF) files were imported into the Surpass mode of Imaris.

Gland visualization
The Surface function of Imaris was used to reconstruct 3D gland surfaces based on the FOXA2 fluorescent signal.The Imaris Vantage function was used to isolate individual glands into a comprehensive gallery for visualization.

Quantitative analysis of gland length and branch numbers
The images of uterine horns were imported into Imaris v9.2.1 (Bitplane) with MATLAB (XT) module (MathWorks, Natick, MA, USA).The surface function in Imaris was used to create 3D renderings of uterine glands from the fluorescent staining by background subtraction with the diameter of the largest sphere set to 30.With files of uterine horns stained only with cytokeratin 8, glands were isolated manually from the surface by using the scissors tool to cut them away from the lumen.For gland length, the Bounding Box OOC function in Imaris was used that determines the shortest straight-line distance from the point where the gland is connected to the uterine lumen to the furthest tip.Masks of the gland surfaces were made to get a channel with uniform gland signal throughout.These channels were then imported to FIJI (ImageJ).Thresholding was used to binarize the isolated image and the 'Fill Holes' function was used to fill in any gaps to ensure a more accurate skeleton.This 3D image was then saved as a TIFF file and imported into MATLAB.The 'imread' function was used so MATLAB could read the data and a new variable was defined using the command 'uint8(Skeleton3D(imbinarize(stack))) � 200;'.This command defines the file as a binary image, calculates the 3D skeleton of a binary volume using a thinning algorithm, and increases the intensity of the signal throughout.The 'saveastiff' function was then used to save this revised file as a TIFF file, which was then added to the original Imaris file as a new channel.Filaments were created manually using that channel.The number of dendrite branch points was exported from Imaris in a Microsoft Excel (Microsoft, Redmond, WA, USA) file for statistical analysis (Khan et al., 2024).

Determination of embryo axis orientation
Embryo axis orientation was determined by identifying embryos via the Hoechst signal and using the Measurement Points module in Imaris.The embryo axis can be determined by identifying the inner cell mass (ICM) or the embryonic pole of the blastocyst and the mural trophectoderm (abembryonic pole).Proper embryo alignment is characterized by the embryonic pole facing the mesometrial side of the uterus and the abembryonic pole facing the anti-mesometrial side of the uterus.The embryo at an implantation site on GD4 was visualized using an optical XY Orthogonal Slicer or Oblique Slicer.An XZ Orthogonal Slicer was used to define the mesometrial-anti-mesometrial (M-AM) axis and was placed at the abembryonic pole of the embryo.Using the Measurement Points module, the first point was placed on the ICM on the M-AM plane.The second point was placed on the intersection of the M-AM plane and the abembryonic pole on the XY plane.The third point was placed on the intersection of the M-AM and XY planes.The value of the angle was obtained using the Statistics function of Imaris.

Quantifying the Lif signal
For quantifying the Lif signal from in situ hybridization on cryosections, the Imaris Surface function was used to construct volumetric surfaces of gland nuclei, and Lif signal.For quantitation of Lif volume per gland volume, volumetric glandular surface of the z-stack of a single cross section of uterus was constructed according to the FOXA2 signal.

Statistical analysis
To compare the numbers of implantation sites, gland length measurements of P21 treatment mice, embryo axis orientation, and Lif volume per gland volume, the Kruskal-Wallis test with Dunn's multiple comparisons was conducted.For analysis of the number of implantation sites/pups, gland length measurements of GD4 1200 h and P21 mice, and Lif volume per lumen volume, the Mann-Whitney test was applied.To assess whether the proportions were comparable in ESR1-stained sections and the number of glands with various branches, a two-proportion Z-test was used.Statistical analyses were performed using GraphPad Prism (Dotmatics; GraphPad, La Jolla, CA, USA) with advanced statistical analyses conducted using R Statistical Software (R Core Team, 2014).A P-value <0.05 was considered significant, indicating differences between comparisons.

Generation of tissue-specific ESR1 deletion mice
To assess the role of ESR1 in uterine gland structure, we generated mice with uterine-specific deletion of ESR1 in a time and compartment-specific manner (Table 1).ESR1 flox/flox mice bred with a Pgr Cre mouse line (Pgr Cre ESR1 flox/flox , hereafter referred to as PER) were generated to determine the role of neonatal epithelial, stromal, and muscle-derived ESR1 (Madhavan and Arora, 2022).To determine the role of epithelial ESR1 in gland structure, we used a Pax2 Cre mouse line to generate mice where ESR1 was deleted in the embryonic M€ ullerian duct epithelium (Pax2 Cre ESR1 flox/flox , hereafter referred to as XER).To determine the role of epithelial ESR1 in gland structure in the post-pubertal mouse, a Ltf Cre mouse line was used (Ltf Cre ESR1 flox/flox , hereafter referred to as LER).
Since this is the first application for Pax2 Cre in the uterine epithelium, we discerned the lineage of Cre expressing cells using the ROSA mT/mG (Muzumdar et al., 2007) reporter line.We observed that Pax2 lineage labeled cells contribute to both the luminal and glandular epithelium of the uterus but are absent from the muscle and stroma on P21 and gestational day (GD) 3 1200 h (Fig. 1A).We also observed that the oviduct displays partial expression of the lineage label at P21 (Supplementary Fig. S1).In addition to the luminal and glandular epithelium, the CD31þ vascular compartment also expresses the lineage label (GFP) (Fig. 1A).Immunostaining with ESR1 and the vascular marker isolectin suggested no expression of ESR1 in endothelial cells at P21, P28, proestrus stage, and GD3 1200 h (Supplementary Fig. S2).
To confirm ESR1 deletion in the glandular epithelium of LER and XER mice, immunological staining with ESR1 and glandular marker FOXA2 was performed (Fig. 1B).While XER mice displayed no expression of ESR1, 10% of LER glandular epithelial cells maintained ESR1 expression at GD3 1200 h (Fig. 1C), suggesting the Ltf Cre deletion of ESR1 is not 100% complete.Thus, LER mice used for further analysis below are not completely ESR1deficient but have reduced levels of ESR1.

Embryonic epithelial ESR1 deletion compromises gland branching
Uterine glands in mice develop in a sequential manner from a bud phase to an elongated phase in the pre-pubertal period, eventually becoming branched in adulthood.We sought to analyze gland structure in ESR1 deletion mice at different stages (Fig. 2, Table 2).At P21, a stage when the majority of glands are unbranched (Vue et al., 2018), PER and XER mice display gland buds that are comparable to controls (Fig. 2A).Diestrus, a pubertal, non-pregnant stage, displays branched glands in control uteri (Arora et al., 2016).While glands of LER mice (reduced levels of ESR1) exhibited branching similar to controls, glands of XER mice (no ESR1) displayed unbranched glands (Fig. 2B).To assess gland structure during pregnancy, we analyzed GD4 1200 h uterine glands (Fig. 2C, Supplementary Fig. S3).Control and LER pregnant uteri at GD4 1200 h displayed glands that were branched, whereas XER glands continued to be unbranched but elongated, and PER glands appeared as gland buds (Fig. 2C, Supplementary Fig. S3).Using an image segmentation algorithm (Khan et al., 2024), we quantified gland branching to support our qualitative analysis.At P21, over 90% of glands in control, PER, and XER mice are unbranched, and only 8% of control glands display 1-3 branches.Neither control nor PER or XER glands display any glands with >3 branches at this stage (Fig. 2D).At GD4 1200 h, even within controls, we observed an age-dependent effect on gland branching.For controls at 6-7 weeks, 9% of glands displayed >3 branches, and this number increased to 40% in controls aged >11 weeks (Fig. 2E).LER mice are analyzed at >11 weeks (to allow sufficient CRE-mediated ESR1 excision) and 45% of LER glands display >3 branches at GD4 1200 h.Glands in XER and PER mice failed to display >3 branches irrespective of age evaluated (Fig. 2E, age range 7-16 weeks).In addition to gland branching, we also observed that while gland length was similar in controls, PERs, and XERs at P21, gland length was greatly reduced in GD4 1200 h PERs and XERs (Supplementary Fig. S4).However, LERs displayed a gland length similar to controls at GD4 1200 h (Supplementary Fig. S4).
Since XER glands display complete loss of ESR1 (Fig. 1B), we performed an age-dependent comparison for controls and XERs to observe the dynamics of gland branching.At P28 (4 weeks of age, Fig. 2F) control and XER glands again display largely unbranched glands.At P35 (5 weeks of age, Fig. 2F), glands in controls begin to display an increased number of glands with 1-3 branches (22%) when compared to the XER glands (13%).For cycling mice, for consistency of ovarian hormone levels, we compared similar aged mice for gland branching at GD4 1200 h.We observed that at P42 (6 weeks of age, GD4 1200 h, Fig. 2F), control glands continue to display increased branch numbers when compared to XER mice.Within control mice, gland branching at GD4 1200 h for 6-week old mice and mice >11 weeks was significantly different (Fig. 2F and Supplementary Table S1) suggesting glands continue to branch with age.
To determine if gland branching defects in XERs were caused by ESR1 deletion in the embryonic epithelium and not the vasculature, we employed another M€ ullerian duct Cre Wnt7a Cre to delete ESR1 in the embryonic epithelium (Table 1).Similar to XERs, Wnt7a Cre ESR1 flox/flox mice (hereafter referred to as WERs) displayed highly reduced gland branching (Supplementary Fig. S5A  and B).In WERs, 61% of glands displayed no branching, 34% of glands displayed 1-3 branches and a small proportion displayed >3 branches (5% glands) (Supplementary Fig. S5C).In agreement with previous reports, we observed FOXA2 expression in the luminal epithelium of WERs at GD3 1200 h (Supplementary Fig. S5A) and no embryos were observed in the uterine lumen.These results suggest that while both WNT7A and PAX2 are expressed in the embryonic M€ ullerian duct epithelium (Carroll et al., 2005), there are differences in Wnt7a and Pax2 promoterdriven CRE expression that result in differential excision of Esr1 and resulting uterine gland branching.Despite this, the extent of branching in WER uterine glands was greatly reduced compared to controls (Supplementary Fig. S5C).Altogether, these observations suggest that ESR1 plays an important role in determining uterine gland elongation and branching during uterine gland development.

Branchless glands fail to support pregnancy
To determine whether gland branching defects result in functional deficits, we evaluated ESR1-deficient mice at GD3 1200 h, GD4 1200 h, GD4 1800 h, and GD5 1200 h.Since WERs show a failure of blastocyst development in the oviduct, we wanted to first confirm that embryos did indeed enter the uterine lumen in XERs.At GD3 1200 h, when embryos are expected in the uterine horn (Carroll et al., 2005), we observed that one-third of the XER mice displayed embryos in the oviduct only, one-third of the mice displayed embryos in both the oviduct and the uterus, and one-third of the mice displayed embryos in the uterus alone (Table 3).When evaluated a day later at GD4 1200 h, 4/7 XER mice displayed embryos in their uterus and the average number of embryos per mouse was lower than controls (Table 3).At GD4 1200 h, implantation sites were clearly observed in control mice using the blue dye reaction (Psychoyos, 1961) (Fig. 3A).XER mice failed to display any implantation at all, and LER mice displayed faint implantation sites indicating delayed implantation (Lufkin et al., 2023) (Fig. 3A-C).When analyzed a day later for decidual sites at GD5 1200 h, control and LER mice displayed visible deciduae whereas XER mice failed to show decidualization (Fig. 3A and  C).Since LERs exhibited delayed implantation, we assessed both embryo development at mid-gestation and the ability of LER females to deliver pups.LER females had apparently normal embryos at GD13 1200 h (Fig. 3B and D) and displayed a normal number of pups at birth (Fig. 3E).Thus, despite the loss of ESR1 in 90% of their glandular epithelium in the pre-implantation stage, LER mice can carry pregnancy to term.

Mice with branchless glands fail to form an implantation chamber and display aberrant embryo-uterine axes alignment
Previous research from our laboratory has demonstrated that the formation of a V-shaped embryo implantation chamber is necessary for the alignment of the embryonic and the uterine axes, such that the ICM of the blastocyst faces the mesometrial pole of the uterus (Madhavan et al., 2022).This event coincides with a shift of PTGS2 (prostaglandin synthase 2) expression from the luminal epithelium to the stroma underlying the implantation chamber (Madhavan et al., 2022).When we evaluated ESR1deficient uteri at GD4 1200 h, we observed that both XERs and LERs showed a failure of V-shaped implantation chamber formation; embryo-uterine axes alignment; and PTGS2 expression in the stroma under the embryo attachment site (Table 2, Fig. 4A ac 00 ).Control embryos displayed an array of axis angles with a median value of 24.30 � (Fig. 4C), while embryos of LER and XER mice were misaligned with the ICM oriented in a perpendicular direction (median angle LER 87.70 � and XER 88.6 � ) with respect to the M-AM axis (Fig. 4C).However, at GD4 1800 h, LERs began to display a small V-shaped chamber, improved alignment of the embryo-uterine axis (median angle 57.70 � ), and PTGS2 expression in the stroma underlying the chamber (Fig. 4A d-e 00 and C).Further, by GD5 1200 h, both control and LER uteri displayed comparable embryo development commensurate with the epiblast stage (Fig. 4B).We failed to observe live embryos in XER mice at GD4 1800 h (one misaligned embryo was observed in 1/10 uterine horns from five mice analyzed), suggesting that in the absence of branched glands and ESR1, attachment fails at GD4 1200 h and embryos fail to survive beyond this stage (Table 2).Uterine gland branching is key for gland function | 7

ESR1-deficient unbranched glands fail to express preimplantation Lif
Since ESR1-deficient mouse models have shown reduced LIF expression (Pawar et al., 2015) we determined Lif mRNA expression using in situ hybridization.First, we observed that in control mice at GD0 1200 h, Lif is highly expressed in the luminal compartment but not in the glandular compartment (Supplementary Fig. S6).At pre-implantation stage GD3 1200 h in controls, strong Lif expression was observed colocalizing with FOXA2-expressing uterine gland cells (Fig. 5A, Supplementary Fig. S6) and Lif expression was not observed in the luminal compartment.XER glands failed to express Lif while, surprisingly, a small amount of Lif was observed in LER glands (Fig. 5A and B).Since we observed delayed implantation in LERs, we tested for Lif expression a few hours later at GD3 1800 h.We discovered that LER glands showed an increase in Lif expression compared to GD3 1200 h (Fig. 5A and B).We quantified total Lif volume normalized for gland volume using image segmentation protocols.Control glands express a uniformly high volume of Lif normalized for gland volume at GD3 1200 h and GD3 1800 h while LER and XER mice exhibited a much lower volume of Lif at GD3 1200 h.We observed a slight but significant increase in the amount of Lif volume in LER mice at GD3 1800 h (Fig. 5B).Visually, we observed that although LER gland cells do not uniformly express Lif, a few gland cells express exaggerated amounts of Lif, sufficient to support implantation (Table 2).Since 10% of LER gland cells continue to make ESR1 (Fig. 1B), Lif expression could be occurring in residual ESR1expressing gland cells or because LERs display branched glands and branching is critical for gland function and Lif synthesis.

Lif supplementation partially rescues implantation in ESR-1 deficient uteri
We observed almost no Lif expression in XERs, therefore we assessed whether recombinant Lif administration at GD3 could rescue implantation in XER mice at GD4 1200 h (Table 4).For XER mouse uteri that did contain blastocysts, using both i.p. and intraluminal routes of LIF administration (Jeong et al., 2010;Kelleher et al., 2017), we achieved 75% and 67% implantation rescue, respectively (Fig. 5C, Supplementary Fig. S7).Altogether, this demonstrates that the critical difference between XER mice with unbranched glands and LER mice with branched glands is the presence of a threshold amount of LIF in LER mice.

Unbranched glands with E2-ESR1 signaling fail to express Lif
XER glands are unbranched but also lack ESR1 signaling.To separate the contributions of gland branching and ESR1 signaling for glandular Lif expression, we used pre-pubertal P21 mice with    unbranched glands and treated them with either E2 alone, E2 þ P4, or vehicle (sesame oil) to evaluate the effect of ovarian steroids on Lif expression.We observed that P21 glandular epithelium, luminal epithelium and stroma express high levels of ESR1 in the presence of vehicle alone whereas ESR1 levels are reduced in the epithelium in the presence of two doses of E2 (Fig. 6A, a-a 00 , b-b 00 ).On the other hand, ESR1 is highly expressed in glandular epithelium when P21 mice are treated with E2 and then P4 (Fig. 6A, c-c 00 ).We also observed that P21 glands treated with vehicle or hormones express Lif in their luminal epithelium (Fig. 6A d-d 00 , e-e 00 , f-f 00 and C).Finally, although we observed varying levels of ESR1 expression in the glandular epithelium of mice treated with vehicle or hormones, E2 treatment failed to elicit Lif expression in glandular epithelium (Fig. 6A and B).It has been suggested that glandular Lif expression occurs under P4 dominant conditions (Ma et al., 2003); therefore, we also evaluated P21 mice treated with both E2 and P4 but again did not detect any Lif expression in the glands (Fig. 6A, f-f 00 ).When we evaluated the 3D gland structure of the P21 uteri treated with different hormonal regimens (Fig. 6D), we observed that treatment with E2 alone or E2 þ P4 did not result in a change in gland length compared to vehicle treatment (Fig. 6E).Further, in all treatment groups, glands were found to be primarily unbranched (Fig. 6F) and no glands with >3 branches were observed.Since ESR1 was Uterine gland branching is key for gland function | 11 expressed in the glandular compartment, the observed lack of Lif expression at P21 cannot be related to an absence of ESR1 signaling.Thus, branchless glands, even in the presence of E2-ESR1 signaling, fail to produce Lif.

Discussion
Uterine glands are exocrine glands with a distinct structure that play vital roles in implantation and pregnancy success.Exocrine gland structure is well-known to contribute to gland function; however, such structure-function relationships for uterine glands have not been described.In our study, we discovered that developmental ESR1 signaling in different compartments of the murine uterus is necessary to build a branched gland.Further, we determine that both the branched structure of a uterine gland and ESR1 signaling are critical for gland function; specifically, preimplantation LIF production that is necessary for successful implantation.

Secretory epithelial domains in branched uterine glands
During development, mouse uterine glands bud off the uterine lumen (Branham et al., 1985) at P4. Glands elongate into the stroma, going through changes in morphology, including teardrop, elongated, and sinuous stages, until some begin branching at P21 (Vue et al., 2018).In the adult mouse, uterine glands are prominently branched (Arora et al., 2016).Human uterine glands in the functionalis and the basalis layer have also been characterized for 3D structure (Yamaguchi et al., 2021;Tempest et al., 2022).Glands in the functionalis layer are coiled and show some branching, while glands in the basalis layer show more prominent branching compared to the functionalis glands.Exocrine glands are generally classified as tubular or branched.Tubular exocrine glands do not have designated secretory cells, although tubular sweat glands have a coiled portion that holds secretions (Kurata et al., 2017).In contrast to tubular glands, branched exocrine glands have acinar end-pieces that are secretory in nature (de Paula et al., 2017).The presence of functional sub-structures in uterine glands (end-pieces, ducts, coils) has not been established.It is possible that the individual uterine gland has a ductal portion and a secretory end-piece (the branched end); however, it is also possible that individual uterine glands are end-pieces, and the uterine lumen accumulates the secretions as ducts do in other branched exocrine glands.Our study establishes that, in mice, unbranched uterine glands (Pax2 Cre ; Esr1 flox/flox , and P21) are unable to produce Lif.This supports the idea that branched portions of uterine glands may carry the secretory function.
Whether this is true for human uterine glands will be an avenue for future investigation.
ESR1 signaling and gland structure in mammary and uterine glands E2 signaling plays a key role in establishing the branching pattern of the exocrine mammary glands that are essential for lactation (Sternlicht et al., 2006;Feng et al., 2007).A whole-body deletion of Esr1 leads to mammary glands with minimal branching (Bocchinfuso et al., 2000;Sternlicht et al., 2006), while a mammary epithelium-specific deletion of Esr1 displays a loss of duct elongation and side branching (Feng et al., 2007).Thus, gland proliferation and branching are ESR1 signaling-dependent in the mammary gland.In mice with continuous estrous, the uterine glands display a cribriform structure (Stewart et al., 2022), suggesting that aberrant ESR1 signaling must impact uterine gland shape.In our study, we determined that neonatal deletion of ESR1 in the epithelial and stromal compartments resulted in gland buds, while deletion only in the epithelial compartment resulted in glands that elongate but fail to branch.Together, this suggests that stromal ESR1 signaling regulates gland bud outgrowth and elongation in a non-cell-autonomous manner while epithelial ESR1 signaling regulates uterine gland branching in a cell-autonomous manner.In the mammary gland, E2-ESR1 signaling regulates bud growth and branching morphogenesis in conjunction with Fgf10, Wnt, and heparan sulfate proteoglycan signaling (Khan et al., 2022).Signaling pathways interacting with E2-ESR1 signaling that are critical for bud elongation and branching still remain to be discovered and will be a subject of future investigations.
The gland bud phenotype in Pgr Cre Esr1 flox/flox glands is remarkably similar to gland structure observed in uteri with neonatal deletion of Foxa2 using the same Pgr Cre driver line (Fig. 4F in Matsuo et al. (2022)).Depletion of ESR1 in the adult epithelium using the Ltf Cre driver in our study did not cause any quantifiable structural gland phenotypes, suggesting that epithelial ESR1 signaling during puberty is required for gland branching.Once glands are branched, reduced E2-ESR1 signaling did not affect gland structure during estrous cycles.This is in contrast with loss of FOXA2 in the adult epithelium using the same Ltf Cre driver line, which results in loss of Lif and the gland structure appears abnormal (Fig. 4F in Matsuo et al. (2022)).We propose that ESR1 signaling and FOXA2 may work together for gland bud elongation and for Lif production; however, they may be redundant to maintain branched glands post-puberty.

Utility of Pax2 Cre for studying uterine epithelium
Both Wnt7a and Pax2 are expressed in the embryonic precursor to the uterus, the M€ ullerian duct (Douglas et al., 2012).Thus, Cre lines driven by the promoters of these genes are useful tools to Table indicates the total number of XER mice injected either intraluminally (1 μg) or intraperitoneally (10 μg) with leukemia inhibitory factor, the number of mice where embryos were observed in the uterus, the number of mice with implantation sites, and the percentage of rescue of mice with implantation sites compared to mice with embryos.The total number of embryos observed in the uterus and the total number of implanted embryos along with the corresponding percentage rescue is also indicated.ESR1, estrogen receptor 1; IS, implantation site; XER, Pax2 Cre ESR1 flox/flox (embryonic uterine epithelial deletion of ESR1 using paired box 2 (Pax2) promoter-driven CRE expression).excise genes in the uterine epithelial cells.Wnt7a Cre ESR1 flox/flox mice do not support embryo entry into the uterus as the embryos die in the oviduct, precluding analysis of embryo implantation studies.Using the Pax2 Cre mouse line, we were able to bypass the oviductal phenotype and embryos were observed in the uterus during the pre-implantation and implantation time points.This may be because we have incomplete and patchy expression of the Pax2 Cre lineage in the oviductal epithelium.Recently, Hancock et al. (2023) determined that Wnt7a Cre ESR1 flox/flox mice also display ectopic expression of gland marker FOXA2 in the uterine luminal epithelium at GD3 1200 h.In Wnt7a Cre mouse, Wnt7a coding exons are replaced by Cre, making this model heterozygous for the Wnt7a gene (Winuthayanon et al., 2010).It is unknown if the ectopic expression of FOXA2 in the Wnt7a Cre ESR1 flox/flox lumen is related to the depletion of ESR1 alone or to the additional loss of one copy of Wnt7a.WNT7A has key functions in uterine epithelium and gland specification (Dunlap et al., 2011;Goad et al., 2017) and could interfere with the ESR1 mutant analysis.We did not observe ectopic expression of FOXA2 in our Pax2 Cre ESR1 flox/flox uterine lumen, and we observed severe gland branching phenotypes in both models where ESR1 signaling is disrupted in the embryonic epithelium.This supports the idea that alternate Cre lines, such as Pax2 Cre , may be powerful tools to study the impact of loss of gene function in the embryonic uterine epithelium where Wnt7a Cre cannot be used.

E2-ESR1 signaling and Lif production
LIF is an essential gland secretion that is absolutely necessary for implantation in mice.Lif knockout mice and other genetic models that fail to produce LIF exhibit embryo implantation failure.ESR1 signaling is predicted to be upstream of Lif expression because ESR1 binding sites are observed in the coding region and in the 3 0 untranslated region of the Lif gene (Hewitt et al., 2012).In uterine biology literature, often a bolus of E2 injection results in increased levels of Lif mRNA, which is usually observed using whole tissue analysis such as quantitative PCR.Since E2-ESR1 signaling can upregulate Lif in both luminal and glandular epithelium but only glandular LIF promotes embryo implantation, it is vital that these studies be interpreted carefully and the location of Lif upregulation be determined when employing E2 treatment.We observed that ovulatory E2 only induces high Lif expression in the luminal epithelium, while in the P4-dominant pre-implantation stages, Lif is confined to the glandular epithelium.It has been shown in various mammalian species, such as mice (Ma et al., 2003), hamsters (Ding et al., 2008), marmosets (Kholkute et al., 2000), and rabbits (Yang et al., 1996), that E2 alone is unable to induce Lif expression in the glands and other factors may be necessary for LIF production.Our work suggests that, in addition to E2-ESR1 signaling, the branched structure of the uterine gland is necessary for Lif synthesis.Lif mRNA was negligible when uterine glands lacked ESR1 signaling in addition to a branched structure.Similarly, intact E2-ESR1 signaling in unbranched prepubertal glands also resulted in low Lif mRNA.Our quantitative data from both mouse models with epithelial deletion of ESR1 suggests that glands with greater than three branches are key for producing threshold levels of glandular LIF to support implantation resulting in pregnancy success.We also observed that uteri with normally branched glands but severely reduced ESR1 signaling were still able to produce threshold levels of LIF to support embryo implantation and pregnancy success.This indicates that gland functional ability (LIF secretion and implantation) is dependent on its structure, particularly whether a gland is branched or unbranched.

Uterine gland structure-function relationships
Loss of ESR1 signaling in the mammary glands results in a poorly developed branched structure and compromises the lactation function of the glands.We also show that unbranched glands compromise Lif secretion, resulting in implantation failure, and this phenotype is partially rescued when pregnancy is supplemented with LIF.It is critical to note that supplemental LIF can rescue both implantation and pregnancy, at least partially, in genetic mutants where glands are branched but simply fail to produce LIF (adult epithelial ESR1 deletion, this study, and adult epithelial FOXA2 deletion (Kelleher et al., 2017)).However, if a genetic mutant displays failure of glands to branch and fails to produce Lif, then implantation is rescued (embryonic ESR1 deletion, this study and neonatal FOXA2 deletion (Dhakal et al., 2021;Kelleher et al., 2017;Matsuo et al., 2022)) but pregnancy fails to continue post-implantation (Kelleher et al., 2017).These data support essential functions for branched uterine glands in the post-implantation phase.Reduced levels of LIF are present in the uterine luminal fluid of infertile women (Lass et al., 2001;Tawfeek et al., 2012); however, LIF is unable to rescue implantation in case of assisted reproduction in women with recurrent unexplained implantation failure (Brinsden et al., 2009;Aghajanova, 2010).Nevertheless, in addition to LIF, gland structure (branching and coiling) may be key to glandular secretions that may be critical for human implantation.
Mice with neonatal or adult epithelial deletion of FOXA2 displayed implantation failure, but the embryos apparently entered a diapause state.Embryos in our embryonic epithelial deletion of ESR1 did not survive beyond implantation stages.This could be linked to differences in the progesterone signaling in these mutant mice and will be a subject of future studies.However, even with embryos that entered diapause in FOXA2 deletion mutants, the viability of embryos that failed to implant was higher if glands were branched compared to bud-staged glands (Matsuo et al., 2022), supporting multiple roles for branched uterine glands beyond LIF secretion and embryo implantation.
From our research into the structure-function relationship of uterine glands in the context of pregnancy outcomes, we have gathered new insights regarding how uterine estrogen signaling affects gland development and structure, and how gland structure contributes to the induction of Lif and events surrounding embryo implantation.To expand the knowledge of how branched and unbranched glands differ in functional capability, further research is needed to identify the secretory factors a branched gland makes and if there are specific secretory cell types in uterine glands, similar to mammary and salivary glands (Khan et al., 2022).Additionally, it will be useful to assess the cellular differentiation events that define how glands bud off the main luminal epithelial tube, the steroid hormone-dependent and independent signaling mechanisms, and the changes in the extracellular matrix that guide gland bud elongation and branching.Further, post-pubertal gland development, especially in the context of regeneration in the post menses and the post-partum uterus, remain understudied.More research is therefore needed to characterize how cyclicity and pregnancy remodel uterine glands as well as how new gland buds may arise in adulthood.Gland structure-function dynamics in humans are challenging as it will be necessary to assess both the basalis and the functionalis layers of endometrial glands.Therefore, the mouse can provide novel insights into gland structure-function relationships relevant to early pregnancy in mammals.

Figure 3 .
Figure 3. Implantation and pregnancy phenotypes of mice with uterine epithelial estrogen receptor 1 deletion.(A) Dissected uteri from control, LER, and XER mice at GD4 1200 h, GD4 1800 h, and GD5 1200 h.Uteri at GD4 1200 h and GD4 1800 h were injected with blue dye prior to dissection for visualization of implantation sites.Asterisks indicate implantation/decidual sites.XER mice exhibit failed implantation while LER mice exhibit delayed implantation.(B) Dissected uteri from control and LER mice at GD13 1200 h showing that LER mice display pregnancy progression comparable to controls.(C and D) Quantification of the number of implantation sites at GD4 1200-1800 h and GD5 1200 h in control, LER, and XER mice (C) and at GD13 1200 h in control and LER mice (D).Each dot represents one uterus analyzed.Median values shown.Data analyzed using Mann-Whitney test.(E) Quantification of number of live pups born in first pregnancies of control and LER mice.Each dot represents one uterus analyzed.Median values shown.Data analyzed using Mann-Whitney test.( � ) ¼ P < 0.05.(ns) ¼ P > 0.05.ESR1, estrogen receptor 1; GD, gestational day; LER, Ltf Cre ESR1 flox/flox (adult uterine epithelial deletion of ESR1 using lactoferrin promoter driven CRE expression); XER, Pax2 Cre ESR1 flox/flox (embryonic uterine epithelial deletion of ESR1 using paired box 2 (PAX2) promoter driven CRE expression).

Figure 5 .
Figure 5. Epithelial estrogen receptor 1 deletion mice display defects in leukemia inhibitory factor expression. (A) Uterine sections of control, LER, and XER mice at GD3 1200 h and GD3 1800 h with staining for Hoechst (nuclei, white), FOXA2 (gland marker, green), and Lif mRNA (red).Lif expression in LER glands is high in a subset of cells (b-b 00 , g-g 00 ) and lower in others (c-c 00 , g-g 00 ).sLif expression increases from GD3 1200 h to GD3 1800 h in LER glands while XER glands express undetectable levels of Lif at both time points.Scale bar, 100 μm.(B) Quantitative analysis of the amount of Lif per gland volume (normalized units) in control, XER, and LER mice at GD3 1200 h and GD3 1800 h.n ¼ 9 regions each from three mice per genotype per stage were analyzed.Data in (B) analyzed using Kruskal-Wallis test with Dunn's multiple comparisons.( � ) ¼ P < 0.05.(C) Uterine horn of XER mouse intraluminally injected with 1 μg recombinant Lif into the right horn on GD3 1200 h and dissected on GD4 1200 h following blue dye injection.Asterisks indicate implantation rescue sites, and the total number of embryos found in each horn is shown.ESR1, estrogen receptor 1; LER, Ltf Cre ESR1 flox/flox (adult uterine epithelial deletion of ESR1 using lactoferrin promoter driven CRE expression); XER, Pax2 Cre ESR1 flox/flox (embryonic uterine epithelial deletion of ESR1 using paired box 2 (PAX2) promoter-driven CRE expression); FOXA2: forkhead box A2; LIF, leukemia inhibitory factor; GD, gestational day; em, embryo.

Figure 6 .
Figure 6.Pre-pubertal, unbranched glands fail to produce leukemia inhibitory factor even when estrogen receptor 1 signaling is intact.(A) Uterine sections of P21 control mice injected with either oil, E2, or E2 þ P4 and stained for Hoechst (nuclei, white), FOXA2 (gland marker, green), ESR1 (a-c, red), and Lif mRNA (d-f, red).Scale bar, 40 μm (a-c).Scale bar, 80 μm (d-f).n ¼ 9 regions each from three mice per treatment.Quantitative analysis of (B) the amount of Lif normalized for gland volume and (C) the amount of Lif normalized for lumen volume.n ¼ 9 regions each from three mice per treatment.(B and C) Data analyzed using Mann-Whitney test.( � ) ¼ P < 0.05.Yellow arrowheads point towards the luminal epithelial cells and white arrows points towards the glandular epithelial cells (D) Representative 3D reconstructions of P21 control uterine glands with various treatments.n ¼ 3 mice per treatment group.Scale bar, 100 μm.Quantitative analyses of (E) average gland length measurements (each dot represents one mouse) and (F) percentage of glands with 0, 1-3, and >3 branches in P21 control mice with various treatments.Hormonal treatment does not affect gland length or gland branching.85-450 glands analyzed per mouse.(E) Data analyzed using Mann-Whitney test.( � ) ¼ P < 0.05.Two-proportion Z-test determined that the differences in percentage of gland branches between oil, E2, and E2 þ P4 treatment is not statistically significant.FOXA2, forkhead box A2; ESR1, estrogen receptor LIF, leukemia inhibitory factor; P, postnatal day; GD, gestational day; E2, estrogen; P4, progesterone.

Table 1 .
Genotypes of the estrogen receptor 1 murine deletion models and abbreviations used in the study.

Table 3 .
Embryo location in mice with embryonic epithelial deletion of estrogen receptor 1.

Table 4 .
Implantation rescue in mice with embryonic epithelial deletion of estrogen receptor 1 supplemented with leukemia inhibitory factor.