Estrogen Action in the Epithelial Cells of the Mouse Vagina Regulates Neutrophil Infiltration and Vaginal Tissue Integrity

In the female reproductive tract, the innate immune system is modulated by two sex steroid hormones, estrogen and progesterone. A cyclical wave of neutrophils in the vaginal lumen is triggered by chemokines and correlates with circulating estrogen levels. Classical estrogen signaling in the female reproductive tract is activated through estrogen receptor α (encoded by the Esr1 gene). To study the role of estrogen action in the vagina, we used a mouse model in which Esr1 was conditionally ablated from the epithelial cells (Wnt7acre/+; Esr1f/f). Histological evidence showed that in response to a physical stress, the lack of ESR1 caused the vaginal epithelium to deteriorate due to the absence of a protective cornified layer and a reduction in keratin production. In the absence of ESR1 in the vaginal epithelial tissue, we also observed an excess of neutrophil infiltration, regardless of the estrous cycle stage. The histological presence of neutrophils was found to correlate with persistent enzymatic activity in the cervical-vaginal fluid. Together, these findings suggest that ESR1 activity in the vaginal epithelial cells is required to maintain proper structural integrity of the vagina and immune response, both of which are necessary for protecting the vagina against physical damage and resetting the vaginal environment.

luteinizing hormone levels peak, restarting the cycle. Nucleated epithelial cells are the dominant cell populations observed in vaginal smears obtained from animals in proestrus 10 . Importantly, these different stages are accompanied by changes in vaginal secretory proteins such as cytokines, chemoattractants, and antimicrobial molecules 12 . E 2 signaling in the vagina induces the formation of a physical barrier by maintaining the thickness of the vaginal epithelium and increasing the secretion of antimicrobial peptides, cytokines and chemokines (which recruit and activate immune cells) [13][14][15] . The epithelial lining of the vagina is also covered by a layer of glycoprotein-containing mucus, namely mucins, that protects the vagina from infectious agents 16,17 . Vaginal epithelial cells are stimulated by E 2 signaling to produce glycogen, a substrate that is metabolized by the native microflora 18 . In postmenopausal women, a reduction in E 2 levels results in the production of fewer epithelial cells and a reduced glycogen content. With vaginal epithelial cell atrophy and a lack of lubrication, the vaginal tissue becomes vulnerable to physical damage, e.g. intercourse, and is subsequently susceptible to infection. Additionally, the composition of the vaginal microbiota often changes in postmenopausal women due to the reduction in the nutritional substrate for the microbiota 19,20 . These changes in the vaginal environment are often alleviated by treatments with a topical synthetic E 2 21 , suggesting that E 2 signaling within the vagina plays a role in the maintenance of the immune response and vaginal homeostasis. Although E 2 supplementation is the current course of treatment for most postmenopausal symptoms in women, alternative non-hormonal treatment options are often desired. A better understanding of the E 2 -regulated actions in the vaginal epithelium is critical to the development of alternative therapeutic options for postmenopausal women.
The effects of E 2 in the female reproductive tract are primarily mediated through its receptor ESR1. To assess the effects of ESR1 activity on the vaginal epithelium, Miyagawa and Iguchi utilized a keratin 5-driven Cre mouse model to genetically ablate Esr1 from the vaginal epithelial cells (Krt5 cre/+ ; Esr1 f/f ) 4 . Using this model, the authors showed a reduced thickness in the vaginal epithelium, thereby suggesting a role for ESR1 in controlling vaginal epithelial cell differentiation 4 . Although this data is informative, the changes imparted on the vaginal epithelium in the absence of ESR1 following physical stresses have never been investigated. In the absence of ESR1, physical stresses, especially those induced by mating, could exacerbate the observed effects on the vaginal epithelium. A compromised epithelial layer is likely to alter vaginal homeostasis, eliciting an observable immune response that differs from an individual with normal ESR1 function. Therefore, we hypothesized that ESR1 activity within the vaginal epithelial cells is not only required for the maintenance of vaginal tissue integrity, but is also involved in the regulation of the immune response following a physical stress.
To assess this, we generated a similar model in which ESR1 was deleted in the vaginal epithelium using the Wnt7a cre/+ (Wnt7a cre/+ ; Esr1 f/f ) 22 . Our Wnt7a cre/+ ; Esr1 f/f females ovulate normally without the need for an exogenous gonadotropins 3 . These Wnt7a cre/+ ; Esr1 f/f females are also receptive to copulation 3 , allowing us to validate the role of ESR1 in maintaining vaginal tissue integrity after physical stress in vivo. Additionally, we used this model to investigate the role of vaginal epithelial ESR1 for both neutrophil recruitment/infiltration and maintenance of the vaginal microbiota.

Results
Loss of ESR1 function in the epithelium induces vaginal tissue laceration after mating. The physiological impact of mating in animals lacking ESR1 in vaginal epithelial cells has yet to be explored. Therefore, histological evaluation was performed on vaginal tissues at 0.5 days post coitus (0.5 dpc) in Esr1 f/f and Wnt7a cre/+ ; Esr1 f/f adult (8-to 12-week-old) female mice. First, ESR1 immunohistochemical (IHC) analysis was performed to ensure ablation of ESR1 in the vaginal epithelium. ESR1 protein was expressed in all cell layers of the Esr1 f/f vaginal tract (Fig. 1). In the vaginal tract of Wnt7a cre/+ ; Esr1 f/f animals, ESR1 protein was absent from the entire epithelial cell layer, whereas expression of ESR1 in the stromal layer remained intact (Fig. 1). Next, the vaginal histoarchitecture was evaluated using hematoxylin & eosin (H&E) staining. Esr1 f/f vaginal tracts showed well-defined and distinguished epithelial layers (Fig. 1). These included a cornified outer layer (containing enucleated cells, Fig. 1, yellow arrowheads), stratified epithelial cell layers, and stromal cell layers in both the upper and lower vaginal tract. In contrast, vaginal epithelial layers within the Wnt7a cre/+ ; Esr1 f/f animals were severely disorganized (Fig. 1). Representative images showed a disrupted epithelial layer with cells infiltrating from the stroma into the vaginal lumen ( Fig. 1, black arrows).
To determine whether proliferation was disrupted in the Wnt7a cre/+ ; Esr1 f/f vaginal epithelium after mating, we performed IHC analysis with Ki67 (a proliferative marker) at 0.5 dpc. Ki67-positive cells were detected at the basal layer of the epithelium and within the lumen of the Esr1 f/f vaginal tract (Fig. 1), whereas at sites with a disrupted epithelial cell layer in the Wnt7a cre/+ ; Esr1 f/f animals, all infiltrating cells were Ki67-positive. To determine the proliferative index, we evaluated the number of Ki67-positive cells present in vaginal tissues during estrus as well as at 0.5 dpc in areas without disruption sites. The proliferative index in the epithelial layer was not different at estrus, but was significantly higher at 0.5 dpc in Wnt7a cre/+ ; Esr1 f/f compared to Esr1 f/f controls ( Fig. 2A). Because the epithelial layer contains both Ki67-positive and -negative cells in Esr1 f/f tissues (Fig. 1), the cell size of Ki67-negative cells was measured. We found that Ki67-negative cells in the epithelial cell layer were significantly smaller in Wnt7a cre/+ ; Esr1 f/f compared to Esr1 f/f controls at both estrus and 0.5 dpc (Fig. 2B).
To determine whether the keratin-rich cornified layer was also absent in Wnt7a cre/+ ; Esr1 f/f vaginal epithelium after mating, Masson's Trichome (MT) staining was performed. A clear cornified layer was observed in the Esr1 f/f mice but no cornified layer was detected in the Wnt7a cre/+ ; Esr1 f/f tissue (Fig. 1, yellow arrowheads, MT staining). Consequently, the vaginal plug was observed to be directly adhered to the basal epithelium in the Wnt7a cre/+ ; Esr1 f/f lower vaginal tract (Fig. 1B, MT staining). In some areas, the semen/copulatory plug was also observed to be lodged between the detached epithelial surface layer and the basal layer ( Supplementary Fig. S1), while in other areas, the epithelial cell layer was completely absent and the vaginal plug was adhered directly to the stromal cell layer ( Supplementary Fig. S1). Interestingly, a blue staining layer indicative of glycogen content was observed at the apical surface in only the Wnt7a cre/+ ; Esr1 f/f vaginal tissues (Fig. 1A,  (PAS) staining confirmed that a globular-shape glycoprotein (possibly mucin) was present at the apical surface of the Wnt7a cre/+ ; Esr1 f/f upper vaginal tract (Fig. 1A, black arrowhead). At estrus, intense PAS staining was also observed at the apical surface of the Wnt7a cre/+ ; Esr1 f/f vaginal tract ( Supplementary Fig. S1).

Epithelial ESR1 signaling is involved in Krt and Muc gene expression.
In Esr1 f/f control mice, the whole vaginal epithelium thickness (is composed of both epithelial cells and a cornified layer) at estrus was 77.03 ± 18.82 μm (Fig. 2C). A 10.69 μm cornified layer contributed to this vaginal epithelium thickness (Fig. 2D). After mating, the whole vaginal epithelium thickness in Esr1 f/f controls was further increased to 97.41 ± 29.40 μm with an average 15.51 μm cornified layer (Fig. 2C,D). In contrast, Wnt7a cre/+ ; Esr1 f/f animals had a significantly thinner epithelial layer at estrus (28.55 ± 8.87 μm) and at 0.5 dpc (31.34 ± 7.52 μm, Fig. 2C). Due to a lack of a cornified layer in Wnt7a cre/+ ; Esr1 f/f animals, no measurements of the cornified layer in Wnt7a cre/+ ; Esr1 f/f animals could be made. As keratinization strengthens the epithelium we next determined the expression of the keratin family of genes, specifically Krt6a, Krt6b, Krt10, Krt13, and Krt16. Only two of the tested keratin genes, Krt6a (a type 2 keratin) and Krt10 (a type 1 keratin), were expressed at significantly lower levels in the Wnt7a cre/+ ; Esr1 f/f compared to Esr1 f/f vaginal tissues (Fig. 2E, p = 0.0426 and 0.0111, respectively).
To determine whether the glycoproteins indicated by the PAS staining in Wnt7a cre/+ ; Esr1 f/f vaginal tissues were mucins, the expression of mucin genes and presence of mucin proteins were also analyzed. Muc1 and Muc4 genes were both expressed at significantly higher levels in Wnt7a cre/+ ; Esr1 f/f animals compared to Esr1 f/f controls (Fig. 2F, p = 0.026 and 0.0043 respectively). MUC1 protein was detected at the surface of the upper vaginal epithelium in Wnt7a cre/+ ; Esr1 f/f animals and was not detectable in Esr1 f/f controls (Fig. 2G). MUC1 protein was also detected at lower levels in some areas of the lower vaginal tract of Wnt7a cre/+ ; Esr1 f/f animals (Fig. 2G) compared to a non-detectable level in the controls. Mucosal MUC1 protein was present in both the uterus and endocervix of Esr1 f/f and Wnt7a cre/+ ; Esr1 f/f animals. However, the detection of MUC1 protein in Esr1 f/f animals was gradually decreased between the endocervix and ectocervix ( Supplementary Fig. S2). Increased leukocyte infiltration in the absence of ESR1 in vaginal epithelial cells. Persistent infiltration of leukocytes was one of the most dramatic phenotypes in the Wnt7a cre/+ ; Esr1 f/f vaginal tissue at 0.5 dpc. Therefore, daily vaginal smears were collected to systematically track the estrous cycle of Esr1 f/f and Wnt7a cre/+ ; Esr1 f/f animals. In Esr1 f/f animals, leukocyte levels were lowest during proestrus and estrus and highest during metestrus (Fig. 3A,B). However, leukocyte levels were present throughout the entire estrous cycle in Wnt7a cre/+ ; Esr1 f/f animals, resulting in a non-cyclic pattern of leukocyte infiltration (Fig. 3A,B).
The leukocytes present in the upper vaginal lumen appeared to be multi (or segmented)-nucleated cells, characteristic of neutrophils. In the Wnt7a cre/+ ; Esr1 f/f epithelium, infiltration of the multi-nucleated cells was present To validate whether the lack of a cornified layer alters the expression of immune genes in the vaginal epithelium of Wnt7a cre/+ ; Esr1 f/f animals, we quantified the expression of immune-related genes that were previously reported 9 . Among all tested genes, mRNA levels of Il1b and Il8 were significantly elevated (Fig. 4C, p = 0.0152 for both genes), while other immune genes remained unchanged in Wnt7a cre/+ ; Esr1 f/f compared to Esr1 f/f vaginal tissues.
To test whether there is a correlation between the bacterial populations present in the Wnt7a cre/+ ; Esr1 f/f vagina and the increasing number of leukocytes, gram staining was performed on vaginal smears. Colonies of gram-positive bacteria were observed at estrus in all Esr1 f/f smear samples (Fig. 5A, yellow arrowhead). At metestrus, smears from Esr1 f/f tissue showed neutrophil infiltration, but gram-positive bacterial colonies were not detected. In smears from diestrus and proestrus stages, we observed scarce or no bacterial colonies. In the Wnt7a cre/+ ; Esr1 f/f animals, gram-positive bacterial colonies were not observed at any stage in any of the animals (Fig. 5A). Bacterial colonies were present only in the smears from Esr1 f/f at estrus stage, but not Wnt7a cre/+ ; Esr1 f/f females (Fig. 5B). Leukocyte clusters from the vaginal smear of Wnt7a cre/+ ; Esr1 f/f females were too dense to be viewed; therefore, we did not include these areas when screening for gram-positive bacterial colonies ( Supplementary Fig. S3). Due to the fiber-like protrusion of the neutrophil nuclei in Wnt7a cre/+ ; Esr1 f/f smears, we investigated whether this protrusion was a result of neutrophil extracellular traps (NETs) 23 that eliminate surrounding bacteria. Histone 3 (H3) is a critical mediator involved in the formation of NETs 23 , so an H3 antibody was used to determine the presence of NETs in vaginal smear samples. The protrusion pattern from the nuclei region of DAPI and H3 was detected in the Wnt7a cre/+ ; Esr1 f/f , but not in Esr1 f/f smears (Fig. 5C, white arrows). Cells with a protrusion pattern in Wnt7a cre/+ ; Esr1 f/f smear samples also co-localized with Ly6G expression.

A loss of epithelial ESR1 in the vagina changes enzymatic activities in the cervical-vaginal fluid.
Matrix metalloproteinase 9 (MMP9) is produced by neutrophils that undergo NET transformation in order to induce tissue remodeling 24,25 . To test whether MMPs are produced in Wnt7a cre/+ ; Esr1 f/f cervical-vaginal fluid (CVF), enzymatic activity of CVF was determined using a gelatin zymography assay. In Esr1 f/f CVF, digestion of the gelatin gel was observed mainly at metestrus (Fig. 6A), corresponding to the leukocyte infiltration. The major bands detected were at 72 and 92 kDa, which are consistent with the size of MMP2 and MMP9, respectively (Fig. 6A). Enzymatic activity of the CVF from Esr1 f/f females was detected at lower levels during diestrus and proestrus in comparison to metestrus, and not at a detectable level at estrus (Fig. 6A). Persistent MMP digestive activity was evident in Wnt7a cre/+ ; Esr1 f/f CVF when compared to the CVF from Esr1 f/f controls (Fig. 6B). To assess the relative levels of Mmp2 and Mmp9, RT-qPCR was performed using whole vaginal tissues collected at estrus. We found that Mmp2 and Mmp9 mRNA levels were expressed at comparable levels in Wnt7a cre/+ ; Esr1 f/f and Esr1 f/f tissues (Fig. 6C). In addition to MMPs, other proteases could also induce changes in tissue integrity. In the vagina, we identified four highly expressed Klk family members in Esr1 f/f tissues. Only Klk1b5 was significantly increased in Wnt7a cre/+ ; Esr1 f/f compared to Esr1 f/f tissues (Fig. 6D, p = 0.0411).

Discussion
Our study revealed new insights regarding the role of epithelial ESR1 in maintaining tissue integrity and immune response in the lower female reproductive tract. We have summarized our findings and proposed a working model in Fig. 7. First, ESR1 in the vaginal epithelium is responsible for E 2 -induced cell hypertrophy during the estrous cycle. Without ESR1, epithelial cells become atrophic, similar to what occurs in postmenopausal women. Second, epithelial ESR1 regulates genes involved in maintaining cellular integrity (keratins) and secretions (mucins). Lack of epithelial ESR1 contributes to a loss of cornification, reduced cellular integrity, and excessive glycoprotein production. Third, epithelial ESR1 plays a vital role in immune suppression. Lacking epithelial ESR1 would result in a failure to suppress vaginal leukocytes, leading to excessive MMP activities. MMPs are proteinases that digest cellular matrix proteins, causing extracellular matrix (ECM) modification and cellular detachment. ECM breakdown may also create a feedback loop to recruit more neutrophils to clear out cellular debris. This proposed working model illustrates that epithelial ESR1 is necessary for homeostasis of the vagina and potentially facilitates the development of treatments for postmenopausal symptoms in women.
Vaginal tissues of Wnt7a cre/+ ; Esr1 f/f animals displayed sites with excessive amounts of immune cells in the different layers of the vaginal tissue and within the lumen after mating. Based on the gross morphology of the cells in the infiltrate of the Wnt7a cre/+ ; Esr1 f/f animals, a majority of the cells appear to have the characteristic of neutrophils (cells with segmented nuclei and cytoplasmic granules) and macrophages (cell with non-segmented nuclei). Ly6G-positive staining confirmed that a majority of the leukocytes were indeed neutrophils. However, further investigation is needed to identify the different immune cell types in the presence or absence of epithelial ESR1 in the vagina. Interestingly, the presence of neutrophils was not reported in the previous Krt5 cre/+ ; Esr1 f/f model 4 . At 0.5 dpc, Wnt7a cre/+ ; Esr1 f/f vaginal epithelia showed lesions with neutrophil infiltration. Some lesions were severe, resulting in a complete detachment of the vaginal epithelium. Based on these results, a lack of epithelial ESR1 results in an extremely fragile vaginal epithelium that can be easily damaged by physical stress, such as mating.  Cell atrophy contributes to vaginal dryness and sexual dysfunction in postmenopausal women 21 . Statistical analysis of the epithelial thickness in Wnt7a cre/+ ; Esr1 f/f animals indicates an overall thinning of the epithelium compared to Esr1 f/f controls. It is established that E 2 induces cell hypertrophy in the uterus 26 . In this report, a loss of epithelial ESR1 led to a significant reduction in cell size in non-proliferating epithelial cells in Wnt7a cre/+ ; Esr1 f/f at both estrus and 0.5 dpc. This finding suggests that local epithelial ESR1 signaling is necessary to induce vaginal cell hypertrophy.
Cell proliferation is essential for tissue growth. We previously reported that stromal, but not epithelial, ESR1 in the uterus is required for epithelial cell proliferation 22,27 . We speculated that vaginal epithelial cell proliferation is also dependent on stromal ESR1 signals, since the proliferation was indeed independent of epithelial ESR1 expression during estrus. However, an increased proliferation index at 0.5 dpc in Wnt7a cre/+ ; Esr1 f/f vaginal epithelium could be contributed by the Ki67-positive leukocytes in the epithelium. In Wnt7a cre/+ ; Esr1 f/f tissues during estrus, the epithelial cells at the apical side of the lumen were loosely attached to the cells underneath. At 0.5 dpc, epithelial lesions and cellular detachment were evident in Wnt7a cre/+ ; Esr1 f/f tissues. It is highly likely that differentiating cells detached prematurely before they could terminally differentiate into cornified cells, or detached due to physical stress during mating. Based on these observations, a loss of epithelial ESR1 led to a reduced number of non-proliferating cells and an increased proliferating index at 0.5 dpc.
In 1922, Long and Evans described the cyclic pattern of the rat vaginal epithelium 28 . Since then, the process of vaginal keratinization has been studied extensively and determined to be induced by E 2 . Keratin 6, 10, 13, and 16 are involved in the keratinization process [29][30][31][32][33] . In addition to the previous Krt5 cre/+ ; Esr1 f/f model, we showed that Krt6a and Krt10 were expressed at lower levels in the absence of epithelial ESR1. The expression pattern of keratin is unique to different epithelial cell types as keratins share only 30% sequence homology between different subfamily members 34 . Krt6a is a type 2 keratin, associated with pachyonychia congenita 35 . Krt10 is a type 1 keratin and a mutation in Krt10 is associated with hyperkeratosis 36,37 . Pachyonychia congenita and hyperkeratosis are rare autosomal dominant disorders, and the mutated keratin proteins impair the compaction of the stratum corneum of the skin. Formation of the intermedium filament in the skin requires the dimerization of KRT1 and KRT10 38 . Therefore, reduction in Krt6a and Krt10 expressions may contribute to the absence of the cornified layer and a loss of cellular integrity in the vaginal epithelium.
In contrast to keratinization, vaginal mucification is triggered by changes in circulating levels of P 4 and E 2 39,40 . High levels of P 4 oppose E 2 -induced action resulting in mucification -a production of glycoproteins including mucins. Mucins are the major barrier molecules in the reproductive mucosa. Two members of the mucin family, mucin 1 and 4 are expressed in human vaginal epithelium 41 . Both Muc1 and Muc4 were elevated in the Wnt7a cre/+ ; Esr1 f/f vagina. Moreover, MUC1 protein levels were increased in the upper vaginal tract in the absence of epithelial ESR1. Our findings are consistent with the previous report that suppression of ESR1 activity caused aberrant induction of MUC1 production in the mouse uterus 42 . Thus, we conclude that epithelial ESR1 in the vaginal tissue is required for normal keratinization and mucification processes. The host defense system of the lower female reproductive tract is composed of epithelial layers, the immune system, vaginal flora, low pH, and mucus in the CVF. The dynamic immune system responds to sex hormones and physical stimuli. In women, elevated E 2 during the proliferative phase suppresses mucosal immunity, creating a window of vulnerability for infection 6 . In rodents, an increased risk of infection in the lower female reproductive tract is also associated with increased levels of E 2 [43][44][45] . Our data are consistent with these findings as gram-positive bacterial colonies were present in vaginal smears during the estrus stage of control animals. At estrus, the absence of leukocytes in the vaginal lumen allows the growth of bacteria colonies during this window of vulnerability 6 . In contrast, a loss of epithelial ESR1 caused excessive leukocyte infiltration throughout all stages of the estrous cycle, resulting in a lack of gram-positive bacterial colonies. Thus, it is possible that the ovarian cycle provides a window of recovery in the female reproductive tract, allowing the tissue to grow and normal flora to colonize in an environment with minimal immunosuppression.
In vaginal smears obtained from Wnt7a cre/+ ; Esr1 f/f animals, we observed the presence of viscous material and leukocyte clusters that were similar to vaginal samples obtained from animals with a prolonged diestrus stage, where the mucus entraps neutrophils 10 . Cells within the mucus were distorted or elongated 10 . We identified the majority of the leukocyte population in the CVF from Wnt7a cre/+ ; Esr1 f/f females to be Ly6G-positive neutrophils. It is established that excessive neutrophil activity has a negative impact on tissue health, primarily due to hyperactivity of neutrophil elastase (NE) 46 , a proteinase that degrades epithelial cadherin resulting in a loss of tissue integrity 46 . In the lung, NE reduces respiratory epithelial integrity 46 . In cancer models, neutrophils enhance carcinogenesis and metastatic potential 47,48 . In blood vessels, neutrophil-extravasation increases vascular permeability 49 . In addition, NETs can also cause blood clots in the vessels 50 . However, the mechanism of how excess neutrophils affect vaginal immune system and how E 2 signaling is involved in this process are unclear. Here, our study showed that Il1β and Il8 were elevated in the absence of epithelial ESR1. This finding is interesting as Il8 is a proinflammatory cytokine produced by keratinocytes 51 and is a known neutrophil-activation factor 52,53 . Additionally increased expression of Il1β and Il8 can be triggered by a bacterial infection 54 . The elevated expression of Il1β and Il8 are consistent with their role in neutrophil activation and local inflammatory responses 55,56 .
Activated neutrophils can transform into extracellular traps (NETs) to kill pathogens using dense materials from the cell nucleus 23 . In addition, neutrophils utilize activated MMP9 to induce tissue remodeling 25 . Considering the extensive tissue damage and cell detachment in the Wnt7a cre/+ ; Esr1 f/f vaginal tissue, it is likely that this tissue damage is caused by elevated MMP activity derived from excess neutrophils. Persistent MMP2/9 activity was evident throughout all stages of the estrous cycle in the absence of epithelial ESR1. Immunofluorescent staining showed that nuclei and H3 of the neutrophils were extruding outside of the cell along with strong Ly6G signals. This histological structure of the neutrophils from the CVF of Wnt7a cre/+ ; Esr1 f/f animals matches the description of NETs. These findings suggest that MMP2/9 activity correlates with the presence of neutrophils and is likely modulated by epithelial ESR1 at a post-transcriptional level. In our previous studies, we identified roles for a different protease family, the kallikreins (KLKs), in the female reproductive tract. KLK expression is tissue-specific. In the oviduct, excessive KLK activity resulted in a lysis of the embryo leading to infertility 1 . In the uterus, a lack of KLK activity caused a semen liquefaction defect 3 . In the vagina, we found that Klk1b5 was the only KLK member with increased expression in the absence of epithelial ESR1. Therefore, Klk1b5 remains a unique candidate for our future study as we previously showed its biological function in the upper female reproductive tract.

Materials and Methods
Ethics statement. All animal handling protocols and procedures were carried out according to Washington State University (WSU) Animal Care and Use Committee guidelines and were in compliance with WSU-approved animal protocols #4702 and 4735. Studies were performed with mice that were housed in a temperature-and humidity-controlled room with access to water and food ad libitum.
Animals and experimental procedures. Adult female mice (8 to 16 weeks old) with a selective deletion of ESR1 in the epithelial cells of the female reproductive tract (Wnt7a cre/+ ; Esr1 f/f ) and their control littermates (Esr1 f/f ) were used in the experiments. Generation of experimental mice and genotyping of the animals was carried out as previously described 22 . Deletion of epithelial ESR1 in the Wnt7a cre/+ ; Esr1 f/f and Esr1 f/f experimental animals was confirmed using IHC analysis 3 . Wnt7a cre/+ ; Esr1 f/f and Esr1 f/f female mice were singly housed and bred overnight with a wild-type (WT) C57BL6/J (The Jackson Laboratory, Bar Harbor, ME) proven breeder male. If a copulatory plug was observed the next morning at 8 a.m., the female was designated as 0.5 days post coitus (0.5 dpc). At the time of tissue collection, animals were euthanized using CO 2 asphyxiation followed by cervical dislocation.

Histological procedures. Unless otherwise noted all tissue sections analyzed were obtained from four
Esr1 f/f animals and three Wnt7a cre/+ ; Esr1 f/f animals at estrus stage, and four Esr1 f/f and six Wnt7a cre/+ ; Esr1 f/f animals at 0.5 dpc. Five-micrometer (µm) paraffin sections were used in this study.
For Periodic Acid Schiff staining (PAS), sections were rehydrated as described above and oxidized in 0.5% periodic acid for 5 mins, rinsed, and placed in Schiff 's reagent (Alfa Aesar, Ward Hill, MA) for 15 mins. Sections were then washed in warm tap water for 5 mins and counterstained in Mayer's hematoxylin for 1 min.
For gram staining, vaginal smears were obtained as previously described 10 . First, smears were air-dried and heat fixed on a gentle flame for 2-3 seconds. Slides were then stained with Gram's crystal violet (1% crystal violet, 0.5% ammonium oxalate) for 1 min and color-treated with Gram's Iodine (#470301-188, Ward's science, Rochester, NY). A 50/50 mixture of 95% ethyl alcohol and acetone was applied as a decolorizer. Slides were then counterstained with 0.5% safranin, washed with tap water, and air dried before mounting with Permount (ThermoFisher Scientific Inc., Carlsbad, CA). Four animals per genotype were used for this experiment.
The immunofluorescence staining protocol was used as previously described with slight modifications 57 . Briefly, cryosections were antigen retrieved with a decloaker (BioCare Medical, Concord, CA), washed in phosphate-buffered saline (PBS) for 10 mins, and blocked with 0.1% triton, and 2% bovine serum albumin (BSA) in PBS for 1 hr at room temperature. Sections were then incubated with histone H3 antibody (1:1000, #AB46765, abcam) at 4 °C overnight. The goat anti-rabbit secondary antibody (1:100, #SA00007-2, ProteinTech, IL, USA) was applied to the sections for 90 mins at room temperature in the dark. After washing, Alexa Fluor 647 anti-mouse Ly6G (1:100, #127609, BioLegend, San Diego, CA) was applied to the sections for 90 mins at room temperature in the dark. Sections were then sealed with ProLong Gold antifade DAPI (#P36935, ThermoFisher Scientific). An Olympus DSU spinning disk confocal microscope was used to capture all fluorescent images. A Leica DMi8 microscope (Leica Microsystems Inc., Buffalo Grove, IL) was used to capture all histological images. Any modifications to images (for example, to increase brightness) were performed across the entire image in accordance with the journal's standards.

Mouse estrous cycle stage identification. Examination of vaginal smears stained with H&E were used
to determine the estrous cycle stage according to previously described procedures 10 . Vaginal smears were collected at 8:00 am, fixed in methanol (5 mins), and stained with H&E as described above. Vaginal smears with mostly cornified epithelial cells were designated as estrus, the presence of leukocytes and cornified cells were designated as metestrus, smears with very few cells were designated as diestrus, and smears in which the majority of the cells were nucleated were designated as proestrus.
Reverse Transcriptase-quantitative PCR analysis (RT-qPCR). The vaginal tract was collected at estrus or at 0.5 dpc for RT-qPCR analysis. Vaginal tissue samples were snap-frozen on dry ice upon removal and stored at −80 °C until use. Tissue samples were homogenized and total RNA was extracted using the RiboZol ME Reagent (Amresco, Solon, OH) according to the manufacturer's directions. RNA quality and quantity was determined using a NanoDrop 1000 Spectrophotometer (Thermo Scienific). Total RNA (1 μg) was reverse-transcribed using the qScript cDNA SuperMix kit according to the manufacturer's protocol (Beverly, MA). The cDNA products were diluted 1:5 in nuclease-free H 2 O. Diluted cDNA (1 μL) was used as a template for the RT-qPCR reaction using the PerfeCTa SYBR Green FastMix (QuantaBio) according to the manufacturer's instructions. PCR reactions were run and raw data was recorded on a 7500 Fast Real-Time PCR System (Applied Biosystems, ThermoFisher Scientific). Expression values in vaginal samples were calculated as fold change and normalized to eukaryotic elongation factor 2 (Eef2) expression, relative to the Esr1 f/f . The relative expression of the genes was determined with an n = 3 mice/genotype, each of which were measured in triplicate. Primer sequences for the genes analyzed are listed in Supplementary Table S1.
Cell counting, size measurement, and static analysis. Quantification of Ki67 IHC was determined using with Cell Counter Tool Plugins as previously described 58 . Three images from each tissue section were captured using the Leica Application Suite (Leica Microsystems Inc.). A total of 6 animals per genotype at estrus were used in the analysis with three-to-four consecutive sections cut per animal and used for Ki67 staining. The number of Ki67-positive cells was counted and calculated as the percentage of positive cells in the epithelium in each image as previously described 2 . Cell size was measured using FIJI with Freehands selection and measure tools. A total of 756 cells were counted from four Esr1 f/f animals at estrus stage, 1039 cells were counted from three Wnt7a cre/+ ; Esr1 f/f animals at estrus stage, 3479 cells were counted from four Esr1 f/f animals at 0.5 dpc, and 1245 cells were counted from six Wnt7a cre/+ ; Esr1 f/f animals at 0.5 dpc. It is worth noting that the analysis avoided areas of containing lesions due to its abnormal morphology and variability. Only areas where the vaginal epithelium was in contact with the vaginal plug were used in the analysis to ensure consistency in location and cell compositions, across all analyzed animals.

Cervical-Vaginal Fluid collection and zymography.
To collect the CVF, the vaginal canal was flushed with 80 μl of normal saline using P1000 pipet tip. To determine the stage of the estrous cycle, 10 μl of the CVF was aliquoted for H&E staining. The rest of the CVF was stored at −80 °C until use. A total of 6 females/genotype were used in the zymography assay. To remove cell debris, the CVF was centrifuged at 3000 × g for 2 mins and the supernatant was used for zymography. Acrylamide gelatin gel (10%) was used for the zymography assay as previously described 59 . In short, CVF samples were incubated with 2X Laemmli sample buffer (#161-0737, Bio-Rad, Hercules, CA) without reducing agents for 10 mins on ice. Then the samples were directly loaded into the gel. After separation of the band at 200 V for 90 mins, the gel was removed from the glass cassette and incubated for 1 hr at room temperature on a shaker with 2.5% Triton X-100 added to the wash buffer (50 mM Tris pH7.4, 5 mM CaCl 2 , 1 μM ZnCl 2 ) to remove sodium dodecyl sulfate. The gel was then washed with deionized water and placed in a wash buffer at 37 °C for 20 hrs. Then, the gel was stained with Coomassie stain (2.5% Coomassie G250, 30% Ethanol, 10% acetic acid) for 30 mins. To obtain clear digested bands, the gel was destained for 45 mins with 30% ethanol and 10% acetic acid. To stop the destaining process, the gel was incubated with 2% acetic acid. The visible bands on the gel were captured using a DLSR camera (Canon Rebel T3i) under room lighting, Original images were used in the figures without any editing through image processing software except cropping was performed in ImageJ. Statistical analysis. Statistical analysis was performed using GraphPad (Prism, La Jolla, CA) and all data are presented as mean ± standard error of the mean (SEM). The Mann-Whitney post-hoc test was performed on cell and genetic analyses. A two-way ANOVA was performed when data was compared between time points and genotypes unless otherwise indicated.