Loss of PRICKLE1 leads to abnormal endometrial epithelial architecture, decreased embryo implantation, and reduced fertility in mice.

The uterine luminal epithelium is the primary maternal contact for the implanting embryo, and several steps must coincide for successful embryo implantation to occur ( 1 ). These steps include precise timing of luminal epithelial structural changes ( 2 , 3 ), appropriate alterations to apical–basal polarity ( 4 , 5 ), intrauterine fluid resorption ( 6 ), and cellular communication between endometrial and uterine immune cells ( 7 , 8 ). However, how the uterus transiently transforms into this receptive state for proper embryo implantation is poorly understood. The luminal epithelium must undergo several morphological changes to prepare for an implanting embryo, and this term has been previously coined plasma membrane transformation ( 5 ). During this process, the timing of this transition and the associated alteration to the cell structure is crucial to its success ( 4 , 9–11 ). Major components of this process are specifically associated with planar cell polarity (PCP), a crucial component for proper epithelial layer formation and epithelial cell division in a specified plane of tissue ( 12 ). Some studies have investigated the chronology of these events utilizing the defective implantation of Wnt5a mutant mice ( 13 ). Moreover, recent work in uterine deletion of Van-Gogh-like 2 ( Vangl2 ) has shown the importance of PCP and associated proteins in embryo implantation chamber formation and successful implantation in mice ( 14 ). However, the role of PCP signaling in the uterine epithelial morphogenesis and its function during implantation is still widely not understood. PCP proteins, first identified in drosophila ( 15–18 ), provide vectorial information for cell organization and migration across a wide field. However, of the six main PCP proteins, PRICKLE has been largely understudied and particularly unstudied in PCP within the context of uterine physiology and the process of embryo implantation. The mouse genome contains four Prickle homologs, where Prickle1 and Prickle2 are the most closely related to drosophila Prickle gene. While the role of PRICKLE1 and PRICKLE2 has been previously studied in embryonic AB polarity ( 19 , 20 ), the early embryonic lethality of their conventional knockouts has limited our knowledge of their role in late-stage tissue development ( 21 , 22 ). PRICKLE1 exerts multifaceted roles in cellular polarization and tissue morphogenesis and has been implicated in regulating the noncanonical Wnt/PCP pathway ( 23 ). PRICKLE1 has yet to be studied in the context of the endometrial epithelium and its role in PCP maintenance within the uterus. To explore the PCP-related function of PRICKLE1 in uterine physiology and embryo implantation, we developed an endometrial epithelial conditional knockout (cKO) of Prickle1 using a Lactoferrin-iCre . Our results demonstrate that the loss of PRICKLE1 in the endometrial epithelium leads to decreased fertility and embryo implantation defect as early as gestational day (GD) 4.5. In addition, 3D imaging demonstrates aberrant luminal folding, impaired luminal closure, and altered glandular structure in mutant uteri at GD 4.5. Further analysis demonstrated decreased aquaporin-2 (AQP2) expression in mutant mouse uteri during diestrus. Moreover, we provide evidence for altered PCP via dysregulation of cellular architecture, altered E-cadherin expression and localization, and aberrant gene expression in nonpregnant and GD 3.5 mutant mice. Additionally, the loss of PRICKLE1 in endometrial epithelium shows evidence of epithelial-to-mesenchymal transition within luminal epithelial cells of nonpregnant mice, providing a connection to uterine pathologies. Finally, the loss of PRICKLE1 in endometrial epithelium shows alterations to the plane of cellular division, resulting in asymmetric cell division, increased multinucleated cells, and providing a connection to the evidence of epithelial–mesenchymal transition (EMT). PRICKLE1 involvement in the PCP pathway within the uterus and during embryo implantation unveils a complex interplay between molecular cues and cellular responses vital for successful pregnancy establishment.

Using Prickle1 f/+ ES cells, we developed Prickle1 f/f mice and utilized Lactoferrin-iCre ( 24 ) to conditionally ablate Prickle1 in the mouse endometrial epithelium (Fig. 1 A–D, Fig. S1 ). Prickle1 cKO in mouse endometrial epithelium displays alterations to luminal folding and decreased embryo implantation. A) Prickle1 f/f targeting construct. B) Uterine cross-sections of 6-month-old Prickle1 f/f Ltf +/icre cKO and control mice in diestrus stained for PRICKLE1 and DAPI. Arrows indicate areas of patchy PRICKLE1 expression (scale bar, 100 μm). C, D) Representative images showing altered morphology of uterus in 6-month-old Prickle1 f/f Ltf +/icre cKO mouse (D) compared with control (C) mouse during diestrus. E) Fertility assessment in control ( n = 14) and Prickle1 f/f Ltf +/icre cKO ( n = 23) 6-month-old mice showing decreased fertility in Prickle1 f/f Ltf +/icre cKO. F) Implantation assessment at GD 4.5 of control ( n = 8) and Prickle1 f/f Ltf +/icre cKO ( n = 13) 6-month-old mice indicated two distinct groups of mutants. ** P < 0.05, **** P < 0.0001. G–J) Representative images showing implantation sites with corresponding H&E stain in 6-month-old Prickle1 f/f Ltf +/icre cKO mouse (H and J) compared with control (G and I) mouse at GD 4.5 showing abnormal fluid accumulation at proposed embryo site of Prickle1 f/f Ltf +/icre cKO. Arrows indicate embryo implantation sites (scale bar, 100 μm). K, L) Optical z slice showing one horn of Prickle1 f/f Ltf +/icre cKO (L) and control uteri (K) with GD 4.5 embryos (asterisks) with corresponding 3D surface models showing aberrant luminal folding in the mutant with no embryos present (scale bar, 500 μm). PIR, periimplantation region; IIR, interimplantation region; IC, implantation chamber; M, mesometrial pole; AM, antimesometrial pole; O, ovary; Cx, cervix. M) Uterine cross-sections of 6-month-old Prickle1 f/f Ltf +/icre cKO and control mice in diestrus stained for AQP2 and DAPI. Arrows indicate areas of low AQP2 expression (scale bar, 50 μm). To investigate the impact that the loss of PRICKLE1 within the endometrial epithelium had on fertility, Prickle1 f/f Ltf +/icre cKO female mice were bred with C57BL6 wild type (WT) male mice and the number of live birth pups was tracked following a positive vaginal plug formation. These fertility trials indicated that Prickle1 f/f Ltf +/icre cKO mice have a reduced litter size of 5.70 pups per litter compared with 9.36 pups per litter in Prickle1 f/f control mice ( P < 0.0001) (Fig. 1 E). However, a wide range of litter sizes was observed within the cKO groups, indicating that Prickle1 f/f Ltf +/icre cKO mice may have a range in the severity of knockout. Therefore, implantation studies were performed at GD 4.5 (where the morning of positive vaginal plug formation was considered GD 0.5) following breeding with C57BL6 WT male mice to investigate the timing of proposed improper embryo implantation. These studies demonstrated that Prickle1 f/f Ltf +/icre cKO mice show two distinct mutant groups: one group with implantation site number similar to that of control mice and one group with no implantation sites present at GD 4.5 (Fig. 1 F–J), potentially resulting from the incomplete penetrance of iCre expression ( 25 ). The results showing a range of live births from Prickle1 f/f Ltf +/icre cKO females (Fig. 1 E) indicate further loss of embryos during pregnancy, in addition to the implantation defects. Past studies have examined the importance of luminal folding during periimplantation to the success of embryo implantation and decidualization ( 4 , 26 ). To further investigate how the mutant's uterine structures and luminal folding may differ from control mice, 3D imaging of full uterine horns with GD 4.5 implantation sites was performed (Fig. 1 K and L). This imaging confirmed the presence of two types of mutants where one set of Prickle1 f/f Ltf +/icre uteri have normal embryos present, normal luminal closure, and normal transverse luminal folding and decidual sites comparable to control mice, while the other set of Prickle1 f/f Ltf +/icre uteri have no embryos present, a completely open lumen, aberrant uterine folding, and no decidua. These more severe mutant mice demonstrate a superfolded luminal nature, indicating improper luminal folding and closure during periimplantation. Several studies have demonstrated the involvement of aquaporins in the luminal closure and their dynamic regulation during implantation ( 27–29 ). Immunostaining of nonpregnant mice at diestrus as well as GD 3.5 Prickle1 f/f Ltf +/icre uteri demonstrated areas of altered AQP2 expression as compared to control uteri (Fig. 1 M, Fig. S2 ). Moreover, a receptive uterus is marked by the down-regulation of Muc-1 in the luminal epithelium and the cessation of epithelial cell proliferation prior to implantation ( 30–32 ). It has been well established that the persistent expression of Muc-1 and stromal growth factors impede implantation ( 33 ). Here, we see an increase of fibroblast growth factors Fgf1 , Fgf2 , Fgf7 , Fgf9 , Fgf10 , and Fgf18 in addition to Muc-1 mRNA expression in the Prickle1 f/f Ltf +/icre cKO when compared with control (Fig. S3 ). In addition, we see altered expression of Pgr and Esr1 across all estrus cycle stages (Fig. S4 ) and increased Pgr , Esr1 , and Muc-1 at GD 4.5 (Fig. S5 ), indicating aberrant steroid hormone signaling and altered epithelial–stromal crosstalk. Further investigation of steroid hormone signaling indicated increased expression of progesterone receptor (PGR) and estrogen receptor (ER) within the stroma with decreased HAND2 expression (Fig. S6 ). Moreover, TaqMan analysis revealed increased expression of estrogen-responsive genes, Greb1 , Clca1 , and C3 , and progesterone-responsive genes, Ihh , Areg , Cyp26a1 , and Il13ra2 (Fig. S7 ). Steroid hormone assessment of cKO in diestrus was comparable to controls (Fig. S8 ). While evidence of luminal epithelial changes was seen within the Prickle1 f/f Ltf +/icre uteri, the glandular epithelium remained largely intact but showed altered gland length in 3D (Fig. 2 A–D). At GD 4.5, as observed previously, glands were tubular and branched in control uteri with a median gland length of 353.62 μm ( 13 , 34 ). While the glands of the Prickle1 f/f Ltf +/icre cKO mice uteri appeared similar to control uteri with respect to their tubular and branched nature, the glands of mutant uteri that displayed open lumens were significantly shorter in length (median = 292.87 μm) and glands of the mutant uteri that displayed closed lumens showed an increase in length (median = 440.92 μm) (Fig. 2 C and D). Prickle1 cKO in mouse endometrial epithelium do not alter glandular structure at GD 4.5. A, B) Transverse slices of 6-month-old Prickle1 f/f Ltf +/icre cKO (B) compared with control (A) mice at GD 4.5 stained for epithelial marker E-cadherin and glandular marker FOXA2 (scale bars, 500 μm [controls], 1,000 μm [ Prickle1 f/f Ltf +/icre ], 150 μm [ Prickle1 f/f Ltf +/icre ], and 1,000 μm [ Prickle1 f/f Ltf +/icre ]). C) Representative 3D reconstructions of glands of 6-month-old Prickle1 f/f Ltf +/icre cKO compared with control mice at GD 4.5 (scale bars, 300 μm). D) Quantitative analysis of average gland length measurements (each dot represents one gland). 250–470 glands analyzed per mouse. Data analyzed using Kruskal–Wallis test with Dunn's multiple comparisons. * P < 0.05. To further elucidate morphological alterations, cross-sections of Prickle1 f/f Ltf +/icre uteri were examined and revealed prominent areas with aberrant epithelial cell shape and presence of multinucleated cells (Fig. 3 A, B and E, Fig. S9 ). Moreover, immunostaining for PCP proteins showed increased overall expression of E-cadherin with increased basal expression as compared to lateral and apical expression in the control uteri across all cycle stages of estrus (Fig. 3 C and D, Fig. S10 ), providing evidence for PCP disruption ( 35 ). Furthermore, Prickle1 f/f Ltf +/icre uteri showed increased overall cytokeratin 7 (KRT7) expression (Fig. S11 ) and increased SOX9 nuclear expression (56.1%) compared with control uteri (31.4%) (Fig. 3 F–H), a marker of endometrial hyperplasia and endometriosis ( 36 , 37 ), demonstrating that the loss of PRICKLE1 in the endometrial epithelium leads to PCP disruption. In addition, these results were consistent at GD 3.5 (Fig. S2 ), indicating that this altered PCP expression is present both during the cycle and just prior to implantation. Prickle1 cKO in mouse endometrial epithelium leads to altered cellular structure and dysregulation in expression of Wnt/PCP proteins. A, B) H&E stain of 6-month-old Prickle1 f/f Ltf +/icre cKO (B) compared with control (A) in diestrus. Large arrows indicate areas of hyperproliferation and rounded epithelial shape. Small arrows indicate multinucleated cells (scale bars, 100 μm). (B inset) Callout of Prickle1 f/f Ltf +/icre cKO highlights area of multinucleated and spindle-shaped cells. Arrows indicate multinucleated and spindle-shaped cells (scale bars, 50 μm). C, D) Immunofluorescence stain of 6-month-old Prickle1 f/f Ltf +/icre cKO (C) compared with control (D) in diestrus for epithelial marker E-cadherin and DAPI indicating high basolateral and overall E-cadherin expression in mutant. Arrows indicate area of missing epithelium and high basal expression of E-cadherin (scale bars, 500 μm). E) Quantification of multinucleated cells in 6-month-old Prickle1 f/f Ltf +/icre cKO ( n = 7) compared with control ( n = 7) in diestrus. ** P < 0.005. F, G) Immunohistochemistry stain of 6-month-old Prickle1 f/f Ltf +/icre cKO (G) compared with control (F) in diestrus for SOX9. Arrows indicate areas of nuclear expression and multinucleated cells in the mutant (scale bars, 10 μm). H) Quantification of positive SOX9 nuclear staining in 6-month-old Prickle1 f/f Ltf +/icre cKO ( n = 5) compared with control ( n = 5) in diestrus. * P < 0.05. While alterations to structure and function were seen from our analysis of the uterus as a whole tissue, further understanding of individual cellular changes was pertinent. Therefore, to study the impact the loss of PRICKLE1 in the endometrial epithelium has on the whole uterus by cell type, single-cell RNA (scRNA) sequencing was performed on 6-month-old Prickle1 f/f Ltf +/icre and Prickle1 f/f mice in diestrus ( GSE272552 ( 38 )). Six-month-old mice were chosen for this experiment due to the pubertal expression of Ltf-iCre to ensure that the progression epithelial phenotype would be evident in the uteri of cKO mice. Results from Seurat scRNA sequencing data analysis ( 39 ), which passed the quality control markers before being analyzed (Fig. S12 ), identified 12 clusters in the uterus based on their expression profiles (Fig. 4 A). Cell types were determined via gene expression profiles using SingleR software ( 40 ) and expert curation. Based on conserved gene expression, clusters were identified as stromal (clusters 0, 4, and 5) and epithelial (clusters 3 and 8) (Figs. S13 and S14 ). Prickle1 cKO in mouse endometrial epithelium leads to increased expression of collagens, Wnt/PCP, and EMT markers in stromal clusters (0, 4, and 5). A) SingleR program predictions for cell clusters present in control and Prickle1 f/f Ltf +/icre mice uteri shown by a uniform manifold approximation projection (UMAP) plot. Cell clusters were given a number between 0 and 12. B, C) Velocity RNA analysis embedding stream for control (B) and Prickle1 f/f Ltf +/icre (C) showing higher magnitude velocity in epithelial cells (cluster 3) of Prickle1 f/f Ltf +/icre mice. A χ 2 test of homogeneity comparing Prickle1 f/f Ltf +/icre and Prickle1 f/f scRNA sequencing counts revealed significant differences in populations of myometrial cells, epithelial cells, stromal cells, and natural killer (NK) cells (Fig. S15 ). Of particular interest was an over 11 times decrease in the number of epithelial cells in the Prickle1 f/f Ltf +/icre group as compared to the control, indicating a loss of epithelial-identified cells within the cKO group. A corresponding increase in stromal cells (clusters 0, 4, and 5) was observed. Moreover, there was four times increase in the number of NK cells present in the cKO when compared with the control. Top principal component analysis (PCA) features for each cluster were identified and revealed increased REST target genes for clusters 0, 4, and 5 (stromal cells) ( 41 ) in addition to increased expression of various collagens ( Col1a1 , Col1a2 , Col6a1 , Col6a3 , Col6a2 , Col4a1 , and Col3a1 ) and known markers of EMT ( Serpinh1 , Ccn1 , Lum , Mmp19 , Lif , and Igfbp6 ) ( 42–47 ) within the Prickle1 f/f Ltf +/icre group compared with control (Figs. S16 and S17 ). In addition to PCP, PRICKLE1 is known to regulate the nuclear localization of REST ( 48 ). In addition, decreased expression of Cdh1 , Muc1 , and Tjp3 in epithelial cells was seen, while increased stromal expression of Dvl1 , Wnt5a , Tjp1 , Tjp2 , and Esr1 was seen (Fig. S18 ). These results indicate that prominent changes are occurring within the stromal cell population of the Prickle1 f/f Ltf +/icre cKO mice due to the loss of PRICKLE1 within the endometrial epithelium. Furthermore, the gene expression profile of the epithelial cluster of the cKO showed similarities to cancer, organismal injury and abnormalities, reproductive system disease, and cell-to-cell signaling (Tables 1 and 2 , Tables S1–S4 ), while stromal clusters of the cKO showed many of the same pathways in addition to inflammatory response, cellular movement, tissue morphology, and reproductive system development and function (Table 3 , Tables S5–S12 ). Moreover, the gene expression profile of the NK cell cluster of the cKO showed connections to cancer, inflammatory disease, cell cycle, cellular growth and proliferation, and tissue development (Tables S13–S15 ). Pathway analysis of predicted molecular and cellular functions associated with dysregulated genes in the epithelial cell population (cluster 3) of Prickle1 f/f Ltf +/icre cKO mice. Function predictions made by IPA software based off genes which were found to be dysregulated in epithelial cells of 6-month-old Prickle1 f/f Ltf +/icre mice cKO ( n = 3) in diestrus. Pathway analysis of predicted disease and disorders associated with dysregulated genes in the epithelial cell population (cluster 3) of Prickle1 f/f Ltf +/icre cKO mice. Disease predictions made by IPA software based off genes which were found to be dysregulated in epithelial cells of 6-month-old Prickle1 f/f Ltf +/icre mice cKO ( n = 3) in diestrus. Pathway analysis of predicted disease and disorders associated with dysregulated genes in the stromal cell population (cluster 4) of Prickle1 f/f Ltf +/icre cKO mice. Disease predictions made by IPA software based off genes which were found to be dysregulated in stromal cells of 6-month-old Prickle1 f/f Ltf +/icre mice cKO ( n = 3) in diestrus. In addition, a loss of cells within clusters 3 and 8 (epithelial cells) with a corresponding increase in cells within clusters 0, 4, and 5 (stromal cells) was observed within the Prickle1 f/f Ltf +/icre group as compared to the control group (Fig. 4 B and C, Fig. S15 ). Given the previous evidence of altered gene expression and EMT markers in the stroma, we hypothesized that these epithelial groups were potentially undergoing EMT and re-clustering as stroma in the Prickle1 f/f Ltf +/icre cKO uteri. RNA velocity analysis was performed to identify high-dimensional vectors to predict the future state of the cells on a pseudo timescale ( 49 ) (Fig. S19 ). Immediately, high velocity (as indicated by the higher ratio of unspliced/spliced RNA transcripts) was seen in clusters 0, 4, and 5 of the stroma of Prickle1 f/f Ltf +/icre cKO uteri as predicted, indicating large changes to the RNA transcripts of this group of cells in comparison with their control counterparts (Fig. 4 B and C, Fig. S19 ). To further demonstrate that this reclassification of cells is due to EMT, RNA velocity analysis of 120 EMT hallmark genes revealed high velocity in clusters 3 and 8 (epithelial) to clusters 0, 4, and 5 (stroma) within the Prickle1 f/f Ltf +/icre cKO group when compared with the control group (Fig. S1A–L ). These results provide evidence that Prickle1 knockout in the endometrial epithelium promotes EMT within the epithelial cells and potentially reclassifies these cells as stromal cells during scRNA sequencing analysis. To further validate the results of the RNA velocity analysis, immunostaining of EMT markers known to be up-regulated in endometrial cancers ( 45 , 50 , 51 ) was performed on Prickle1 f/f Ltf +/icre uteri cross-sections. Increased expression of ZEB1 (Fig. S20A ), TWIST1 (Fig. S20B ), SNAI2/SLUG (Fig. S20C ), and vimentin (Fig. S20D ) in both the endometrial epithelium and stroma of the cKO when compared with control uteri was observed, confirming evidence of increased EMT activity within the Prickle1 f/f Ltf +/icre cKO as previously seen by the scRNA sequencing analysis and providing further evidence for the loss of PRICKLE1 in promoting EMT with the endometrial epithelium. Further evaluation of the binuclear and multinucleated epithelial layer revealed alterations to the plane of cell division within the Prickle1 f/f Ltf +/icre uteri cross-sections with asymmetric epithelial cell division seen in the absence of PRICKLE1 (Fig. 5 A–D). Prickle1 f/f Ltf +/icre uteri contained much higher rates of asymmetric epithelial cell division in comparison with controls (Table S16 ). Moreover, higher levels of E-cadherin expression were present at basal and apical sides of the epithelium within the cKO uteri compared with controls, providing a potential mechanism for altered polarity and incomplete cell division (Fig. 5 D and E). Finally, during scRNA data analysis, Prickle1 f/f Ltf +/icre cKO samples demonstrated a statistically higher occurrence of doublets (3.4%) as compared to control samples (2.5%) ( P < 0.05) (Table S17 ), indicating a higher percentage of multinucleated cells present within the cKO samples, while assessment of apoptosis did not show any statistical difference between control and mutant (Fig. S21 ). Prickle1 cKO in mouse endometrial epithelium displays disruption in the polarity of cell division. A, B) H&E stain of 6-month-old Prickle1 f/f Ltf +/icre cKO (B) compared to control (A) in diestrus. Callouts highlight dividing cells with altered orientation in the Prickle1 f/f Ltf +/icre cKO (scale bars, 10 μm). C) Immunofluorescence stain of 6-month-old Prickle1 f/f controls in diestrus for epithelial marker E-cadherin and DAPI. Callouts highlight dividing cells with normal orientation (scale bars, 50 μm). D) Immunofluorescence stain of 6-month-old Prickle1 f/f Ltf +/icre cKO mice in diestrus for epithelial marker E-cadherin and DAPI. Callouts highlight dividing cells with altered orientation (scale bars, 50 μm). E) Longitudinal view schematic depicting normal epithelial cell division and asymmetric epithelial cell division in the absence of PRICKLE1.

The mechanisms that regulate the unicellular layer architecture of endometrial luminal epithelium, timing, and cycle of the uterine receptive state for proper embryo implantation are poorly understood. In addition, although some recent advancements in understanding PCP-related functions within the uterus have been made ( 14 ), the specific role of PCP pathway in uterine epithelial morphogenesis and embryo implantation has previously been understudied. Here, we demonstrate the importance of PRICKLE1 in uterine endometrial epithelial architecture, embryo implantation, fertility, and overall uterine physiology using a Prickle1 endometrial epithelial cKO mouse model. We observed two distinct sets of phenotypes related to fertility in PRICKLE1 mutants, with a more severe mutant phenotype exhibiting no implantation sites and a less severe mutant with implantation sites similar to the control at GD 4.5 but with lower live pups at birth. Previous work with Lactoferrin-iCre has demonstrated known variability in the expression, leading to variability excision within mutants ( 25 ). The more severe mutants also displayed aberrant luminal folding, an open lumen, and altered AQP2 expression, indicating that Prickle1 f/f Ltf +/icre cKO mice may be unable to remove fluid from the lumen properly, thus preventing luminal closure and inhibiting successful embryo implantation. AQP2 is known to regulate luminal closure ( 27–29 ). Moreover, the increased expression of fibroblast growth factors, Muc-1 , Pgr , and Esr1 (Figs. S3 and S4 ), coupled with the increased nuclear expression of ER, PGR, and decreased HAND2 in the cKO with increased estrogen and progesterone-responsive genes (Figs. S6 and S7 ), is consistent with results suggesting an impairment in the crosstalk between stroma and epithelium via Hand2 and progesterone ( 33 ). Decreased litter size even in mice that show normal implantation numbers at GD 4.5 may indicate this altered epithelial–stromal crosstalk affects maintenance of pregnancy. The role of PRICKLE1 in epithelial–stromal communication and steroid hormone signaling needs further investigation. Immunostaining of Prickle1 f/f Ltf +/icre uteri for E-cadherin and FOXA2 demonstrated glandular structure consistent with pregnancy at GD 4.5 (Fig. 2 C), even with an open lumen in the mutants that did not contain any positive embryo implantation sites. These results indicate that the inability of the lumen to fold and close properly, in addition to dysregulated cellular communication, may be responsible for the embryo implantation failure within these mice that otherwise should have been pregnant. Although the structure of the uterine glands of the mutant uteri appeared similar to the tubular and branched glands of control uteri, there were differences in the gland length that correlated with the open or closed luminal phenotype (Fig. 2 C and D). The potential role of PRICKLE1 in regulating uterine gland morphogenesis needs further characterization. Our scRNA sequencing analysis demonstrates how the loss of PRICKLE1 in the endometrial epithelium alters uterine gene expression and provides evidence for decreased number of cells with epithelial phenotype in the cKO with an increase in stromal cell numbers. Furthermore, RNA velocity analysis indicates a potential transition of epithelial cells into clusters with stromal-like gene expression. It has been well established that EMT is a contributor to several endometrial diseases, including endometriosis, adenomyosis, and endometrial cancers ( 52–54 ). Our data indicate a possible role for the loss of PRICKLE1 in EMT. In addition to the changes in epithelial and stromal cells in the cKO uteri, the scRNA data showed a significant increase in NK cells (Fig. S15C ). The overabundance of NK cells in the cKO provides a potential connection to the fertility and implantation phenotype as it has been shown that uterine NK cell population dysregulation is connected to recurrent miscarriage, fertility, and uterine disorders such as endometriosis ( 55–57 ). Spatial cues that organize a unicellular 2D sheet of epithelium in a plane orthogonal to the apical–basal polarity are regulated locally through various Wnt/PCP signaling molecules ( 58 ). Additionally, symmetric (conservative) cell division in the epithelium is essential to maintain the apical–basal polarity and proper epithelial function, including implantation ( 4 , 58 ). Our data indicating the presence of binucleated and multinucleated cells in the luminal epithelium, along with altered plane of cell division, indicate a crucial role for PRICKLE1 in maintaining conservative, symmetrical cell division in the epithelium and potentially in the maintenance of unicellular layer architecture. The presence of binucleated cells in the luminal epithelium may indicate incomplete cytokinesis or endoreduplication in cells ( 59 ). Proper cytokinesis requires the formation of a contractile ring at the site of E-cadherin expression ( 58 ). Therefore, increased basal E-cadherin expression and an altered mitotic plane of division support incomplete cytokinesis in our cKO. In this regard, regulation of expression and localization of E-cadherin by PRICKLE1 might hold the key for altered epithelial architecture and function. Furthermore, multinucleated spindle-shaped cells in the cKO epithelium may indicate that both defective cytokinesis and endoreduplication might be present ( 59 ). This observation, further strengthened by the increased number of doublets removed during scRNA data analysis in the cKO (Table S16 ), is unique compared with the loss of other PCP genes such as Vangl2 , Ror2 , and Wnt5a ( 60 , 61 ). Loss of these PCP genes has shown varying degrees of uterine epithelial subcellular changes as well as implantation defects. However, none of these genes have shown a dramatic loss of polarity of cell division or incomplete cytokinesis in the uterus. Moreover, the role of asymmetric cell division which gives rise to two daughter cells with distinct fates in EMT has been reported ( 62 ). Whether the altered plane of cell division observed in the cKO endometrium has a direct role in the dramatic increase in EMT is unclear at this point.

Mouse embryonic stem cell clones harboring floxed Prickle1 ( Prickle1tm1a(EUCOMM)Wtsi , ES cell clones with cKO potential, targeting vector HTGR03009_Z_6_E08) were acquired from EuCOMM. The ES cells were used to generate chimeric founder mice, which were mated to C57BL6 WT mice to create heterozygous Prickle1 f/+ mice. These mice were crossed with FLPo (Gt(ROSA)26Sor tm2(FLP*)Sor ) mice (JAX stock #009086) to remove FRT flanked bgal-neo sequences ( 63 ). These were bred together to obtain homozygous Prickle1 f/f mice, which were then crossed with transgenic Lactoferrin-iCre mice ( 24 ). The mice were genotyped by PCR using primers for Prickle1 (5′GGTTTCATGTGTTGAGACATTTC) (5′GTATTTCTGTGCCCTTTTTGTCGTCG) (5′TGAACTGATGGCGAGCTCAGACC) as well as primers for Ltf +/icre (5′AACTAGCACACCTGGTTGAGG) (5′CTTCTTGGGAGGCAGTGAAC) (5′CAGGTTTTGGTGCACAGTCA). All experiments involving animals were conducted in accordance with protocols reviewed and approved by the KUMC Institutional Animal Care and Use Committee (IACUC) (Animal Care and Use Protocol number: 23-02-296). Female Prickle1 f/f Ltf +/icre and Prickle1 f/f mice were mated with C57BL6 WT males to induce pregnancy. The morning of finding the vaginal plug was considered day 0.5 of pregnancy. Litter size analysis was tracked via number of pups born following positive vaginal plug formation at days 17–21. Implantation was assessed by sacrificing pregnant dams on day 3.5 or 4.5 following 100 μL intravenous injection of 1% Chicago Sky Blue 6B dye (Millipore Sigma, Cat# C8679) solution and counting the number of distinct blue bands present within the uterus. Whole-mount immunofluorescence was performed as previously described ( 13 ). GD 4.5 Prickle1 f/f and Prickle1 f/f Ltf +/icre uteri were dissected from mice and fixed in DMSO/methanol (1:4). Subsequently, they were rehydrated in a methanol/PBST (1% Triton X-100 in phosphate-buffered saline [PBS]) (1:1) solution for 15 min, followed by a phosphate-buffered saline with Tween (PBST) wash for 15 min. Uteri were then incubated in a blocking solution (2% powdered milk in PBST) for 1 h at room temperature. They were incubated with primary antibodies diluted in blocking solution for nine nights at 4 °C. Primary antibodies included rabbit anti-FOXA2 (Abcam, ab108422; 1:300) and rat anti-CDH1 (M108, Takara Biosciences, 1:500). Uteri were washed once for 15 min with 1% PBST followed by four additional PBST washes for 45 min each at room temperature. Uteri were then incubated with 1:500 Alexa Fluor 555 donkey antirabbit IgG (Invitrogen, A31572), Alexa Fluor 633 goat antirat IgG (Invitrogen, A21094), and Hoechst (Sigma-Aldrich, B2261) at 4 °C for three nights. Samples were washed once for 15 min and three more times for 45 min each with 1% PBST and then stepwise dehydrated into 100% methanol. Uteri were incubated overnight at 4 °C in a 3% H 2 O 2 solution prepared in methanol. Next day, the samples were washed twice for 15 min each and a final 1-h wash with 100% methanol at room temperature and then cleared in benzyl alcohol/benzyl benzoate (1:2, BABB) (Sigma-Aldrich, 108006, B6630) overnight. Whole tissue immunofluorescence samples were imaged using a Leica TCS SP8 X Confocal Laser Scanning Microscope System with white-light laser, 10× air objective, and 20× BABB objective. Using the tile scan function with Z stacks 7 μm apart (10×), images covering the entire length and thickness of the uterine horn were acquired. Higher-resolution images of implantation sites and regions flanking the implantation site were acquired using a 20× BABB objective. The tile scan function was used, and Z stacks 5 μm apart were acquired. Image analysis was performed using commercial software Imaris v9.2.1 (Bitplane). The confocal image (.LIF) files were imported into the Surpass mode of Imaris. To derive the lumen-only signal, the FOXA2 signal of glands was subtracted from the epithelial CDH1 signal using the channel arithmetic MATLAB-based function ( 26 ). 3D renderings were created using automated and manual mode in Surface function for lumen-only and FOXA2 or CDH1 (glands only) signal. Individual glands were then isolated and presented in a comprehensive gallery for visualization using the Imaris Vantage function. Gland length was determined using Bounding Box OOC function in Imaris that measures the shortest straight-line distance from its point of connection to the uterine lumen to the furthest tip. To compare gland length measurements between Prickle1 f/f mice and Prickle1 f/f Ltf +/icre cKO mice, Kruskal–Wallis test with Dunn's multiple comparisons was performed using R statistical software. A P -value < 0.05 was considered significant, indicating differences between comparisons. Uterine tissues were fixed in 4% paraformaldehyde and processed for paraffin embedding. Tissue sections were deparaffinized in xylene, rehydrated using a series of ethanol, and then stained with hematoxylin and eosin (H&E) or immunostained. For immunostaining, antigen retrieval was done using antigen retrieval unmasking solution (Vector Labs, Cat# H-3301) prior to staining. For immunofluorescence staining, rehydrated tissue sections were stained as previously described ( 64 ) using primary antibodies AQP2 (Novus Biologicals, Cat# NB110-74682SS, 1:100), E-CAD (Cell Signaling, Cat# 24E10, 1:100), ZEB1 (Cell Signaling, Cat# 70512, 1:400), TWIST1 (Cell Signaling, Cat# 90445, 1:100), SNAI2 (Cell Signaling, Cat# 9585, 1:100), PGR (Invitrogen, Cat# MA1-411, 1:100), and PRICKLE1 (BiCell Scientific, Cat# 50621, 1:100). Secondary antibodies were obtained from Jackson Immuno Research Laboratories for Alexa Fluor 488 and 594 (Cat# 711-545-152 and 711-585-152, 1:200). For immunohistochemistry staining, rehydrated samples were prepared using an ABC Universal PLUS Peroxidase kit (Vector Labs, Cat# PK-8200) and then with primary antibodies KRT7 (ThermoFisher Scientific, Cat# pA5-82291, 1:500), SOX9 (Cell Signaling, Cat# 82630, 1:100), vimentin (Cell Signaling, Cat# D21H3, 1:100), ER (Proteintech, Cat# 21244-1-AP, 1:800), and dHAND (Santa-Cruz Biotechnology, Cat# sc-398167, 1:500). For TUNEL apoptosis staining, rehydrated samples were prepared using a One-Step TUNEL In Situ Apoptosis kit (Elabscience, Cat# E-CK-A320) and then imaged. Total RNA was isolated from uterine tissue samples stored in RNAlater (Invitrogen, Cat# AM7020) and then biopulverized and placed into TRIzol reagent (Invitrogen, Cat# 15596026). Following TRIzol, a series of chloroform, isopropanol, and ethanol washes were used to isolate the RNA. Following quantification using a Nanodrop spectrophotometer, aliquots of RNA were reverse-transcribed using High-Capacity cDNA Reverse Transcription kit (Applied Biosystems, 4368814). TaqMan assays for Fgf1 (IDT, Mm.PT.56a.41158563), Fgf2 (IDT, Mm.PT.56a.5129235), Fgf7 (IDT, Mm.PT.56a.28834565), Fgf9 (IDT, Mm.PT.56a.5456225), Fgf10 (IDT, Mm.PT.58.11905869), Fgf18 (IDT, Mm.PT.58.14021387), Muc-1 (IDT, Mm.PT.58.15865847), Greb1 (IDT, Mm.PT.58.11217662), Clca3 (IDT, Mm.PT.58.9995580), C3 (IDT, Mm.PT.58.17325540), Ihh (IDT, Mm.PT.58.30489545), Areg (IDT, Mm.PT.58.31037760), Cyp26a1 (IDT, Mm.PT.58.10791878), and Il13ra2 (IDT, Mm.PT.58.28388240) were used to quantify gene expression differences utilizing the delta delta C(T) method with housekeeping genes Rn18s (ThermoFisher Scientific, Mm03928990_G1). Samples of serum from Prickle1 f/f Ltf +/icre ( n = 4) and Prickle1 f/f ( n = 4) mice aged 4 months were assayed by ELISA at the University of Virginia Center for Research in Reproduction, Ligand Assay and Analysis Core (Charlottesville, VA, USA). Whole uteri of Prickle1 f/f Ltf +/icre ( n = 3) and Prickle1 f/f ( n = 3) mice aged 6 months were isolated immediately following euthanasia and washed with cold 1× PBS. Uteri were digested as previously described ( 41 ), and samples were pooled following digestion. The KUMC genomics core processed and sequenced the cells as previously described ( 41 ). The Prickle1 f/f Ltf +/icre sample had 5,805 cells with 70,202 mean reads and 2,078 median genes per cell, while the Prickle1 f/f sample had 5,426 cells with 43,202 mean reads and 1,098 median genes per cell. Both samples had high sequence saturation levels (>60%), and the mapping rate of reads to the mouse genome (10 mm) was >90% for both samples. scRNA sequencing libraries were generated using the 10× Chromium Single Cell 3′ v3 chemistry (10× Genomics) and sequenced in an Illumina NovaSeq 6000 sequencing machine. The raw sequence reads were processed using the 10× Genomics Cellranger pipeline (v 6.1.1) to obtain UMI feature-barcode count matrices. Doublets were removed using DoubletFinder ( 65 ) and cells filtered to contain only those cells with at least 500 UMIs and over 250 detected genes with a genes per UMI ratio >0.8 and a mitochondrial gene ratio <20%. The resulting single-cell data were analyzed using the R software Seurat ( 66 ) (v4) as previously described ( 67 ). The analysis was performed at a 0.4 cluster resolution reasoned using the Clustree software ( 68 ) giving 12 stable clusters. The cluster cell types were identified using the SingleR software ( 40 ) and our expert curation as previously described ( 41 ). Annotations were based on the two reference datasets, ImmGenData from the Immunological Genome Project (ImmGen) ( 69 ) and MouseRNAseqData ( 70 ). The dynamic progression of transcription in the single-cell data was analyzed using the Velocyto package ( 49 ).