Epithelial estrogen receptor 1 intrinsically mediatessquamous differentiation in the mouse vaginaShinichi Miyagawaa,b and Taisen Iguchia,b,1

aOkazaki Institute for Integrative Bioscience, National Institute for Basic Biology, National Institutes of Natural Sciences, Okazaki, Aichi 444-8787, Japan; andbDepartment of Basic Biology, Faculty of Life Science, SOKENDAI (The Graduate University for Advanced Studies), Okazaki, Aichi 444-8787, Japan

Edited by Jan-Åke Gustafsson, University of Houston, Houston, Texas, and approved September 10, 2015 (received for review July 10, 2015)

Estrogen-mediated actions in female reproductive organs aretightly regulated, mainly through estrogen receptor 1 (ESR1). Themouse vaginal epithelium cyclically exhibits cell proliferation anddifferentiation in response to estrogen and provides a uniquemodel for analyzing the homeostasis of stratified squamous ep-ithelia. To address the role of ESR1-mediated tissue events dur-ing homeostasis, we analyzed mice with a vaginal epithelium-specific knockout of Esr1 driven by keratin 5-Cre (K5-Esr1KO).We show here that loss of epithelial ESR1 in the vagina resultedin aberrant epithelial cell proliferation in the suprabasal cell layersand led to failure of keratinized differentiation. Gene expressionanalysis showed that several known estrogen target genes, includ-ing erbB growth factor ligands, were not induced by estrogen inthe K5-Esr1KO mouse vagina. Organ culture experiments revealedthat the addition of erbB growth factor ligands, such as amphir-egulin, could activate keratinized differentiation in the absenceof epithelial ESR1. Thus, epithelial ESR1 integrates estrogen andgrowth factor signaling to mediate regulation of cell prolifera-tion in squamous differentiation, and our results provide newinsights into estrogen-mediated homeostasis in female repro-ductive organs.

estrogen receptor | vagina | keratinization | amphiregulin | epithelium

Estrogens play important roles in vertebrate reproductive biology,with effects principally mediated through the nuclear estrogen

receptors (ESRs). ESRs are members of the nuclear receptor su-perfamily and typically up-regulate gene transcription upon ligandbinding. Two different genes encoding the two ESR subtypes, ESR1(formerly named ERα) and ESR2 (formerly named ERβ), havebeen identified from a variety of vertebrate species. ESR1 is thepredominant subtype regulating estrogen-mediated proliferation anddifferentiation of female reproductive organs. Administration of es-trogens increases organ weights and promotes cell proliferation anddifferentiation, whereas such events were not evoked in the Esr1-deficient mice (1–3).Estrogen actions on the uterus and mammary gland have been

extensively analyzed. During the first step of estrogen action,epithelial–stromal interactions play a pivotal role in epithelialcell proliferation. In vitro tissue recombination experiments withEsr1 knockout (KO)- and wild-type-derived epithelium andstroma revealed that the effects of estrogen on uterine andmammary epithelial cell proliferation are mediated primarily viastromally expressed ESR1 (4-6). Estrogen-induced growth factorssecreted from the stroma subsequently promote epithelial cellproliferation (7–10). Experiments using epithelial cell-specific KOof Esr1 revealed that estrogen can induce proliferation of uterineepithelial cells despite the absence of epithelial ESR1 (11). Incontrast, similar experiments using a mammary epithelium-specificEsr1 KO showed that ESR1 is required in the epithelial cells formammary gland growth and ductal epithelial cell proliferation (12).It is uncertain whether these differences result from differentrequirements for epithelial ESR1 in the uterus versus the mam-mary gland or whether they result from methodological differencesin tissue recombination experiments (e.g., age of donor, hormonetreatment protocol) (13). Irrespective of these differences, there

are definite requirements for epithelial ESR1 expression in uterineepithelial functionalization (e.g., lactoferrin gene expression andprevention of cell death in the uterus), as well as in ductal mor-phogenesis, alveologenesis, and lactation in the mammary gland(11, 12).Another female reproductive organ, the mouse vagina, exhibits a

unique estrogen-responsive feature. The vaginal epithelium exhibitscyclical, estrogen-dependent cell proliferation and squamous dif-ferentiation during the estrous cycle. The vaginae of ovariectomized(OVX) mice show an atrophied epithelium with two to three celllayers, whereas estrogen administration rapidly induces epithe-lial cell proliferation in the basal layer. The suprabasal cells areno longer mitogenic but differentiate while moving up throughthe epithelium, resulting in keratinization of apical cells. The fullystratified and keratinized vaginal epithelium resembles the typi-cal stratified and keratinized epidermis found in the skin andother organs.The formation and maintenance of such epithelia relies on a

tightly balanced process of epithelial cell proliferation and squamousdifferentiation. Classical tissue recombination experiments suggestthat epithelial ESR1 in the vagina is dispensable for cell proliferationbut is indispensable for squamous differentiation (5, 14). Until now,no mouse model for tissue-specific KO of Esr1 in the vagina hasbeen developed. In this study, we used genetic analysis of ESR1function in the vagina by creating an epithelium-specific Esr1 KO toprovide new insights into the potential tissue-specific function ofESR1. We found that estrogen can induce cell proliferation in theabsence of epithelial ESR1, but that the epithelial cells showedsevere defects in keratinocyte commitment. We also found thatamphiregulin (Areg) is required downstream of epithelial ESR1for terminal differentiation in the epithelium. Thus, epithelialESR1 integrates estrogen and growth factor signaling to mediatethe switch between cell proliferation and squamous cell differ-entiation in the mouse vagina.

Significance

Estrogen regulation of tissue homeostasis, particularly the reg-ulation of cell proliferation and differentiation in female re-productive organs, is interesting both in its basic biology and inseveral disease models and cancer biology. Using an epithelialestrogen receptor 1 (Esr1) conditional knockout mouse model,we showed that vaginal epithelial cells lacking ESR1 failed toactivate growth factor signaling and remained undifferentiatedand in a proliferative state throughout the epithelium. Thus,epithelial ESR1 mediates the switch between cell proliferationand squamous cell differentiation, demonstrating unique rolesof ESR1 in the female reproductive tract.

Author contributions: S.M. designed research; S.M. performed research; S.M. analyzeddata; and S.M. and T.I. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. Email: taisen@nibb.ac.jp.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1513550112/-/DCSupplemental.

12986–12991 | PNAS | October 20, 2015 | vol. 112 | no. 42 www.pnas.org/cgi/doi/10.1073/pnas.1513550112

Dow

nloa

ded

by g

uest

on

June

29,

202

0

ResultsPhenotypes of Epithelial-Specific Esr1 KO Mice. ESR1 protein isexpressed in both the vaginal epithelium and the stroma in the wild-type mouse vagina (Fig. 1A). We generated a vaginal epithelialcell-specific Esr1 KO mouse model (K5-Esr1KO) by crossingEsr1-floxed mice with K5-Cre transgenic lines that express Crerecombinase in the vaginal epithelial cells (15). ESR1 proteinwas detected in the stroma, but not in the epithelium, demon-strating the epithelium-specificity of Esr1 loss of function in theK5-Esr1KO mouse vagina (Fig. 1B). We first tested whetherestrogen induced cell proliferation and/or differentiation in theK5-Esr1KO mouse vagina, using OVX mice to avoid any con-founding effects of the hypothalamus-pituitary-gonadal axis andto simplify analysis of estrogen effects. The vaginal epithelium ofboth 8-wk-old OVX control and K5-Esr1KO mice was composedof two to three layers of cuboidal cells (Fig. 1 C and D). 17β-Estradiol (E2) administration induced epithelial stratificationand full keratinization in control mice (Fig. 1E), whereas the K5-Esr1KO mice exhibited stratified epithelia, but not keratinizeddifferentiation (Fig. 1F). The epithelia of K5-Esr1KO micetreated with E2 were thinner than those of controls (58.4 ± 8.7 μmvs. 95.6 ± 8.4 μm; P < 0.05). These phenotypes were not aresult of an insufficiency of E2 exposure (see the following long-term E2 exposure experiment). The K5-Esr1KO female mice areinfertile, probably because of the expression of Cre recombinase inuterine epithelial cells (16).Next we used immunohistochemical staining to further charac-

terize the phenotypes of the K5-Esr1KO mouse vagina. The tem-poral and spatial expression profiles of specific keratins reflectedthe differentiation status of epithelial cells (17). The vaginal epi-thelial marker, keratin 14 (KRT14), was expressed throughout all

layers of the vaginal epithelium (Fig. 1 G and I). KRT14 wasexpressed in basal and suprabasal cells but was absent in the apicalcell layer in OVX and estrogen-treated K5-Esr1KO vaginae (Fig. 1H and J). KRT8, a marker for undifferentiated cells, was expressedin a few suprabasal cells in the control OVX mouse vagina, but itsexpression was augmented in the OVX K5-Esr1KOmice (Fig. 1L).KRT8-positive cells were persistent in the apical layer of cells, evenafter estrogen treatment in K5-EsR1KO mice (Fig. 1 M and N).The stratified and squamous differentiation marker, KRT1, wasexpressed in the differentiating suprabasal cells upon estrogen-treatment in the control mouse vagina. Intriguingly, it was notdetected in the K5-Esr1KO mice, although these animals did ex-hibit stratified epithelium (Fig. 1 O and P). Expression of theterminal differentiation marker, filaggrin (FLG), was absent in theK5-Esr1KO mice (Fig. 1 Q and R).TRP63 is a crucial regulator of squamous differentiation and is

no longer expressed once cells are committed to squamous epi-thelial lineages (18, 19). TRP63 was mainly expressed in thebasal cells and in the two to three upper layers of cells adjacentto the basement membrane in the vagina of control mice (Fig.1S) (20). Intriguingly, TRP63 expression was augmented andexpanded into the superficial layer of the K5-Esr1KO mousevagina (Fig. 1T). We infer from these results that epithelial cellsin the K5-Esr1KO mouse vagina exhibit impaired commitmenttoward squamous differentiation and, therefore, are maintainedin an undifferentiated state.

Loss of Epithelial ESR1 Impaired Cell Proliferation and Cell Death inVagina.We investigated the proliferation status of vaginal epithelialcells using BrdU incorporation. The proliferation indices were notsignificantly changed between control and K5-Esr1KO vaginaewith or without E2 treatment (Fig. 2 A, B, and E). Although cellproliferation was restricted in the basal cells of E2-treated controlmouse vagina, several BrdU-positive cells were detected in thesuprabasal cells in K5-Esr1KO mice (red arrowheads in Fig. 2B).Likewise, MKI67, another cell proliferation marker, was expressedin the basal cells in control vaginal epithelium, but it was robustlydetected in the upper layer cells of vaginal epithelium in K5-Esr1KO mice (Fig. 2 C and D). Thus, suprabasal cells in K5-Esr1KO mouse vagina show proliferative potency.Taken together, these data indicate that epithelial cells in the

K5-Esr1KO mouse vagina remain in an undifferentiated state andretain proliferative potency, even after moving out from the basalcell layer. We next tested whether prolonged E2 exposure couldenhance cell proliferation and result in a cancerous lesion. Wefound that the epithelium did not exhibit keratinization, althoughsome abnormal epithelial pathology, including down-growth intothe stroma, was observed in the K5-Esr1KO mouse vagina 2 moafter implantation of an E2 pellet (Fig. S1). In the absence of es-trogens, vaginal epithelial cells did not proliferate, and K5-Esr1KOmice exhibited atrophied vaginal epithelia, even in 2-y-old animals(Fig. S2).We also examined cell death in the K5-Esr1KO mouse vagina.

Apoptotic indices in vaginal epithelial cells as measured byTUNEL staining did not differ between controls and K5-Esr1KOmice (Fig. 2 F–H). However, the apoptotic index was augmentedin the stroma of estrogen-treated K5-Esr1KO mice comparedwith controls (Fig. 2 F–H), suggesting that epithelial ESR1contributes to maintaining stromal cell survival.

Evaluation of E2-Induced Gene Expression in K5-Esr1KO Mouse Vagina.We next evaluated expression of progesterone receptor (PGR)and CCAAT/enhancer-binding protein β (CEBPB) (14). Bothgenes were induced by E2 administration in the control mousevagina, and their proteins were localized in the epithelium(Fig. 3 A–I). Expression of these genes was not induced by E2administration in the K5-Esr1KO mouse vagina. It has beenreported that PGR-mediated signaling inhibits vaginal epithelial

Fig. 1. Effects of epithelial cell-specific ablation of Esr1 on the mouse vagina.ESR1 protein expression was evaluated by immunohistochemistry in control (A)and K5-Esr1KO mice treated with E2 (B), indicating successful deletion of Esr1specifically in the epithelial cells in the K5-Esr1KO mice. The vaginal epitheliumof 8-wk-old OVX control (C) and K5-Esr1 (D) mice. E2 treatment induces vaginalepithelial stratification and keratinized differentiation in the control (E), butfails to such differentiation in the vaginal epithelium of K5-Esr1KO mice (F).Expression pattern of cell differentiation markers in vaginal epithelium of OVX(G, H, K, and L) and E2-treated (I, J, M–T) mice. Immunohistochemical stainingfor KRT14 (G–J), KRT8 (K–N), KRT1 (O and P), FLG (Q and R), and TRP63 (S andT). In K5-Esr1KOmice, differentiating cell-specific markers KRT1 and FLG are notexpressed, even after E2 treatment (C–F). (Scale bar, 100 μm.)

Miyagawa and Iguchi PNAS | October 20, 2015 | vol. 112 | no. 42 | 12987

DEV

ELOPM

ENTA

LBIOLO

GY

Dow

nloa

ded

by g

uest

on

June

29,

202

0

cell proliferation (21) and that CEBPB is indispensable for epi-dermal cell differentiation (22). Therefore, the failure of E2 toinduce expression of these genes in the K5-Esr1KO mouse vaginamay be associated with the observed phenotypes.To further analyze genes involved in vaginal epithelial cell dif-

ferentiation, we conducted DNA microarray analysis of vaginaefrom control and K5-Esr1KO mice treated with E2 (Dataset S1).We found that several secreted growth factors were differentiallyexpressed (Fig. S3). Of those, we focused on the erbB ligand,amphiregulin (Areg), because erbB ligands have been implicated aseffectors of ESR1 in mammary gland development (23). Areg isalso involved in keratinocyte growth and is associated with psori-asis, a hyperproliferative skin disorder (24, 25). Expression of Aregwas increased after E2 administration in the control mouse vagina,but its expression was persistently low in the K5-Esr1KO vagina(Fig. 3J). AREG was exclusively expressed in the vaginal epitheliumin controls, but expression was not observed in the K5-Esr1KOmice (Fig. 3 K and L), suggesting an autocrine mechanism for ep-ithelial ESR1 signaling.

Areg Is a Mediator of Vaginal Epithelial Cell Differentiation. We nextevaluated the effects of AREG on vaginal epithelial cell differ-entiation, using an organ culture system. Epithelial cells exhibitedkeratinized cell morphology and expressed KRT1 and FLG inorgan-cultured vaginae from control mice treated with E2 (Fig. 4A–C). In contrast, neither KRT1 nor FLG was expressed in vagi-nae from K5-Esr1KO mice treated with E2 (Fig. 4 D–F). However,the addition of AREG and E2 induced KRT1 and FLG expression,together with keratinized morphology (Fig. 4 G–I). In the absenceof E2 supplementation, AREG alone could not induce KRT1 orFLG (Fig. 4 J–L), suggesting AREG could not bypass the wholeeffects of E2 in the mouse vagina. To investigate potential epistaticrelationships among Areg, Pgr, and Cebpb, immunohistochemistrywas performed using organ-cultured vaginae supplemented with E2and AREG. Both PGR and CEBPB were expressed in the controlmouse vagina (Fig. 5 A and B). PGR was not expressed, althoughCEBPB was induced, in the epithelial cells of K5-Esr1KO vaginae

(Fig. 5 C and D). These data indicate that CEBPB is likely to be adownstream target of erbB signaling, rather than a direct target ofepithelial ESR1.Organ-cultured vaginae from wild-type mice treated with E2 and

anti-AREG antibody showed epithelial keratinization and ex-pressed KRT1 and FLG (Fig. S4). These results suggest that theamphiregulin effects might be redundant. We further examined theselectivity of erbB ligands for vaginal epithelial cell differentiation,including heparin-binding EGF-like protein (HBEGF) and trans-forming growth factor α (TGFA). We also tested insulin-likegrowth factor-I (IGF1), which is assumed to be a stromal factorthat activates epithelial cell proliferation (10, 26). HBEGF inducedexpression of KRT1, but not FLG (Fig. S4). TGFA had a minimaleffect on KRT1 and FLG expression. IGF1 did not induce KRT1or FLG expression, and the epithelium was atrophied (Fig. S4).Thus, although there is some redundancy among erbB ligands,AREG is the most active inducer of squamous cell differentiationin the mouse vaginal epithelium.

DiscussionOrgan growth, development, and function are regulated by complexinteractions among cells and tissues. Female reproductive organsare excellent models to analyze such interactions because epithelialcell proliferation and differentiation can be completely controlledby estrogen. Classical tissue recombination experiments have sug-gested that epithelial cell proliferation in female reproductive organs(e.g., uterus, vagina, and mammary gland) is mediated by stromalESR1 in a paracrine manner (4–6). In contrast, genetic studies inmouse models have shown that epithelial ESR1 is required for cellproliferation in the mammary gland, but not in the uterus (11, 12).

Fig. 2. Ectopic cell proliferation in the epithelium and cell death in the stroma ofthe K5-Esr1KO vagina. BrdU and MKI67 immunostaining in control (A and C) andK5-Esr1KO (B and D) mouse vagina. Note BrdU-positive cells in the upper layer ofthe K5-Esr1KOmice (red arrowheads in B). BrdU-labeling indices were not alteredbetween control and K5-Esr1KO in the OVX and E2-treated group (E). TUNELassay indicates loss of ESR1 has not influenced cell apoptosis in the vaginal epi-thelium, but increase in vaginal stromal cell death in the E2-treated K5-Esr1KOmice (F–H). Black arrowheads in G indicate apoptotic cells in the stroma. Dottedlines indicate the basal layer of the epithelium. (Scale bar, 100 μm.)

Fig. 3. Expression patterns of cell differentiation markers in the mousevagina. Immunohistochemistry of PGR (A–D) and CEBPB (E–H). Both proteinsare detected in the suprabasal layer of differentiating vaginal epithelial cellsof the control mice treated with E2 (C and G), but not in the K5-Esr1KO mice(B and H). Expression of Pgr, Cebpb, and Areg mRNAs failed to increase inthe vagina of E2-treated K5-Esr1KO mice (I and J). AREG protein is detectedin the vaginal epithelium in the control mice treated with E2 (K), but not inthe K5-Esr1KO mice (L). (Scale bar, 100 μm.)

12988 | www.pnas.org/cgi/doi/10.1073/pnas.1513550112 Miyagawa and Iguchi

Dow

nloa

ded

by g

uest

on

June

29,

202

0

Therefore, to clarify and extend those observations, we con-ducted Esr1KO in the mouse vagina and demonstrated thatvaginal epithelial cells can proliferate in the absence of epithelialESR1 in vivo.The vaginal epithelium is a stratified keratinized squamous type

similar to that in the epidermis. Therefore, further cell and tissueinteractions are required for coordinated and fine-tuned mor-phogenetic events. Stratified squamous epithelia are maintainedby proliferation of basal keratinocytes and their subsequent co-ordinated cell cycle exit, migration into the suprabasal layers, andterminal differentiation. In wild-type mice, suprabasal cells exit thecell cycle and express the differentiation marker KRT1 and theterminal keratinized cell marker FLG in the upper layers. How-ever, in the absence of epithelial ESR1, differentiation markerswere not expressed, in accord with previous tissue recombinationexperiments (5, 14). We also found that vaginal epithelial cells inthe K5-Esr1KO mouse continue to proliferate, even after basallayers are removed, indicating that the suprabasal cells remainundifferentiated and proliferative while failing to differentiate intokeratinocytes. Therefore, we infer that epithelial ESR1 is in-dispensable for keratinocyte commitment and maintenance ofepithelial homeostasis by controlling cell differentiation. This is incontrast to the role of epithelial ESR1 in the uterus, where epi-thelial ESR1 is involved in the maintenance of epithelial cellsurvival by preventing excessive apoptosis, likely as a result ofinactivation of cell cycle progression-related genes (11, 27). Sub-sequently, loss of ESR1 results in enhanced apoptosis in theuterine epithelium (11); such apoptosis was not observed in thevaginal epithelium of K5-Esr1KO mice.These results suggest that loss of ESR1 in the vaginal epithe-

lium impaired the action of E2-mediated secondary factors me-diating squamous differentiation. We investigated known estrogentarget genes such as Pgr and Cebpb in the epithelia. Pgr regulationby estrogens has been well studied in the female reproductivetracts, using tissue recombination experiments that showed thatPgr expression is controlled by stromal ESR1 in the uterus andepithelial ESR1 in the vagina (28, 29). These results are consistent

with genetic mouse models in both organs (current study and ref.11). In general, progesterone-PGR signaling acts antagonisticallyto E2-induced effects (21, 30). Progesterone signaling that wouldordinarily inhibit E2-induced epithelial cell proliferation was im-paired in the uterine epithelium of the Esr1KO mouse (27). Thus,loss of epithelial PGR expression might contribute to dysregula-tion of the proliferative response to estrogen in the K5-Esr1KOmouse vagina.CEBPB is another estrogen-regulated transcription factor in

female reproductive organs (31, 32). CEBPB is a key regulator ofproliferation and/or differentiation in multiple tissues, includingsquamous epithelia. Epidermal ablation of CEBPB leads to increasedbasal keratinocyte proliferation and severe defects in keratinocytecommitment and differentiation, despite some redundancy withCCAAT/enhancer-binding protein α (Cebpa) (22). CEBPB also playsa crucial role in restricting TRP63 expression to basal keratinocytes(22). These phenotypes coincide with the results shown here in theK5-Esr1KOmouse vagina and support an important role for CEBPBin squamous differentiation in the vagina.ErbB signaling has long been implicated downstream of

steroid hormone signaling in female reproductive organs (8, 9,33). In the current study, we found that the erbB ligand,AREG, is expressed specifically in the epithelium, dependenton epithelial ESR1 expression. Organ culture experiments re-vealed that AREG can compensate for the loss of epithelialESR1 on squamous differentiation with respect to KRT1, FLG,and CEBPB expression. Furthermore, AREG is the most activeof the erbB ligands and growth factors we examined in rescuingthe effects of Esr1 loss of function. Therefore, AREG is likely tobe a key intrinsic factor required for squamous differentiation inthe mouse vagina. This indirect signaling mechanism, mediatedthrough secreted growth factors, enables the amplification andsystematic coordination of estrogen stimulation within the targetcells/tissues. In contrast, neither AREG nor other growth factorsexamined could alone induce epithelial cell growth and squa-mous differentiation. Hence, we propose that additional factorsdownstream of Esr1 in the epithelium and/or stroma are neces-sary for full activation of estrogen effects in the vagina and thecoordinated crosstalk between growth factors and estrogens that isessential for female reproductive organ homeostasis. Intriguingly, thestromal cells exhibit increased cell death when epithelial ESR1 islost. We hypothesize that secreted factors derived from the epithe-lium act on the stroma to protect against stromal cell death and arecurrently investigating the existence of these factors. The number of

Fig. 4. AREG as a potential factor for vaginal epithelial cell differentiation.HE staining (A, D, G, and J) and immunohistochemistry for KRT1 (B, E, H, andK) and FLG (C, F, I, and L) in organ-cultured vagina. Organ-cultured vaginaefrom control mice treated with E2 express KRT1 and FLG (B and C), but thevaginae of K5-Esr1KO mice do not (E and F). Note that addition of AREGleads to KRT1 and FLG expression (H and I). (Insets) Counterstaining withHeckest. (Scale bar, 100 μm.)

Fig. 5. AREG induces CEBPB in vagina of K5-Esr1KO mouse. Immunohis-tochemistry of PGR (A and C ) and CEBPB (B and D) in organ-cultured va-gina. Organ-cultured vaginae treated with E2 and AREG from control miceexpress PGR and CEBPB (A and B). In the K5-Esr1KO mouse vagina, additionof AREG fails to express PGR (C ) but rescues CEBPB expression (D). (Scalebar, 100 μm.)

Miyagawa and Iguchi PNAS | October 20, 2015 | vol. 112 | no. 42 | 12989

DEV

ELOPM

ENTA

LBIOLO

GY

Dow

nloa

ded

by g

uest

on

June

29,

202

0

apoptotic cells was not altered by supplementation of growth factorsexamined in the current organ culture study.Estrogen actions need to be tightly regulated, and it is

conceivable that dysregulation of estrogen signaling results inreproductive disorders, including tumorigenesis. Squamous cellcarcinoma is the most common invasive malignancy of the cervix,vagina, and vulva (34). Although much remains to be learnedabout the underlying molecular mechanisms, loss of ESR1 hasbeen associated with squamous cell carcinoma progression andinvasion (35). Although we showed dysregulated epithelial cellproliferation and differentiation in the K5-Esr1KO mouse vaginaand found that prolonged estrogen exposure induced abnormalepithelial invagination into the stroma, cancerous lesions andindications of metastatic phenotypes (e.g., effects on basementmembrane) were not observed. This suggested that loss of the ep-ithelial ESR1 signal alone is not sufficient to cause squamous cellcarcinoma. Therefore, other factors are likely to be involved incarcinogenesis and invasive phenotypes, and future studies areneeded to further clarify tissue-specific roles of ESR1 for carcino-genesis, invasion, and metastasis. A different mouse backgroundshould also be tested, because the current mouse background(C57BL6) is genetically resistant to such cancerous lesions fromestrogen treatment.In summary, our mouse model demonstrated that the vaginal

epithelium undergoes cell proliferation mediated by the stromalESR1, whereas differentiation is promoted by epithelial ESR1(Fig. 6). Loss of epithelial ESR1 leads to failure of commitmentof cells to keratinized differentiation, resulting in aberrant epithelialcell proliferation under stimulation by stromal ESR1. Therefore,ESR1 is required in the vaginal epithelium for appropriate responseto estrogens and tissue homeostasis. It is intriguing that ESR1has unique roles and differentially mediates estrogen action in theepithelia of the mammary gland, uterus, and vagina. Overall, animproved understanding of estrogen action at the tissue level willshed light on the mechanisms of estrogen-mediated homeostasisand underlying disorders in female reproductive organs such asendometriosis, endometrial hyperplasia, or cancer.

Materials and MethodsMouse and Treatment. C57BL/6J (CLEA), K5-Cre (36), Esr1-null, and Esr1-floxedmice (1) were maintained under 12 h light/12 h dark at 23–25 °C and fed labo-ratory chow (CA-1; CLEA) and tap water ad libitum. To obtain vaginal epithelialcell-specific Esr1KO mice (K5Cre/+;Esr1flox/-), K5Cre/+;Esr1+/− male were crossedwith Esr1flox/flox female mice. For control mice, Cre-negative-Esr1flox/+ siblings wereused. In most experiments, mice were OVX at 6 wk of age and killed at 8 wk ofage. For examining effects of estrogen, a single daily injection of 0.1 μg E2 (Sigma)was given to OVX mice for 3 d; the mice were killed 24 h after the last injection.

This timing is sufficient to allow induction of a stratified and fully keratinizedepithelium in the mouse vagina. For a long treatment, OVX mice wereimplanted with E2 pellets (0.01 mg per pellet, 2 mo; Innovative Research ofAmerica) into s.c. tissue for 2 mo. All procedures and protocols were ap-proved by the institutional animal care and use committee at the NationalInstitute for Basic Biology.

Histology and Immunohistochemistry. Hematoxylin/eosin stainingwas performedby a standardprocedure. The significance of differences in the thickness of epitheliabetween control and K5-Esr1KO mice was assessed by Welch’s t test. For immu-nohistochemistry, paraformaldehyde-fixed, paraffin-embedded sections were in-cubated with the following primary antibodies: ESR1, TRP63, PGR, CEBPB (SantaCruz), KRT8 (Progen), KRT1, KRT14, FLG (Covance), MKI67 (Thermo Fisher Scien-tific), and AREG (R&D Systems). The sections were stained with the Vectastain ABCKit (Vector Laboratories). Immunofluorescence analysis was performed with AlexaFluor protein-conjugated secondary antibodies (Life Technologies) and counter-stained with Hoechst 33342 (Sigma).

For BrdU immunostaining, mice were injected with BrdU (Sigma) at100 mg/kg body weight. One hour after the injection, tissues were collected.BrdU-incorporated cells were detectedwith anti-BrdU antibody (1:20, Roche).TUNEL assay for the detection of apoptotic cells was performed with the insitu apoptosis detection kit (Roche). Statistical analyses were performed usingStudent’s t test or Welch’s t test, followed by F-test; differences with P < 0.05were considered significant. Error bars represent the SEM. More than four(immunohistochemistry) or six (BrdU and TUNEL) animals were analyzed;representative pictures are shown.

Quantitative RT-PCR. Total RNA (2 μg), isolated from each group using anRNeasy kit (Qiagen), was used in RT-PCR reactions carried out with SuperScript IIIreverse transcriptase and SYBR Green Master Mix (Life Technologies), accordingto manufacturer’s instructions. PCR conditions and sequences of some primersets are given in a previous report (33). Others used are Pgr, TAT GAG AAC CCTTGA CGG TGT TG, CAG GGC CTG GCT CTC GTT A; Cebpb, CGG GTT TCG GGA CTTGAT GCA AT, CTA GAC AGT TAC ACG TGT GTT GCG. Relative RNA equivalentsfor each sample were obtained by standardization of ribosomal protein L8levels. More than three pools of samples per group were run in triplicate todetermine sample reproducibility. Error bars represent the SE, with all valuesrepresented as fold change compared with the control treatment group nor-malized to an average of 1.0. Statistical analysis was performed by ANOVAfollowed by Dunnett’s test; differences with P < 0.05 were considered significant.Error bars represent the SEM.

DNA Microarray. The quality of Cyanine3-labeled cRNA was analyzed using theAgilent 2100 Bioanalyzer (Agilent). Cyanine3-labeled cRNA was prepared usingtheQuickAmp Labeling Kit andOne-color RNA Spike-in kit (Agilent) and purifiedusing the RNeasy Micro Kit (Qiagen) following the manufacturer’s protocol. Thequality of cRNA was analyzed using the Agilent 2100 Bioanalyzer (Agilent).cRNA (1.65 μg) was hybridized to the 4 × 44K mouse gene expression microarray(G4122F), using standard Agilent protocols. After 17 h incubation with rotation,the microarrays were washed with Gene Expression Wash Buffer Kit (Agilent)according to the manufacturer’s protocols. DNAmicroarrays were scanned usinga GenePix 4000B scanner (Molecular Devices) at 5 μm resolution. The signalintensity of the spots was digitized using the microarray imager software(Combimatrix Molecular Diagnostics). Gene expression and statistical analyseswere performed using the Subio Platform (Subio Inc). Data files generated bythe Agilent Feature Extraction Software were imported into Subio Platform. Thedigitized raw data were normalized by the mean value of all spots of signalintensity in each array. Signal intensities (gProcessedSignal) were normalized by75th percentile and transformed into log ratios (base 2). We merged duplicatesof each experimental condition and applied statistical analysis on the averages.Subsequently, these data were assessed by t test (P < 0.05) to identify differ-entially expressed genes between treatment groups.

Organ Culture. Eight volumes of Cellmatrix type I-A (Nitta Gelatin) weremixedwith 1 volume 10× Dulbecco’s Modified Eagle’s Medium/Nutrient MixtureHam’s F-12 (DMEM/F12; Sigma) and 1 volume 200 mM Hepes buffer con-taining 262 mM NaHCO3 and 0.05 N NaOH was added to the mixture. Thiscold gelatin mixture (200 μL) was poured into inner millicell cell culture in-serts (Merck) placed into the well of a 24-well plate and allowed to gel at37 °C. Vaginae of mice aged 3 wk were cut and placed into Hank’s BalancedSalt Solution (Life Technologies). Tissues were washed three times in Hank’sBalanced Salt Solution and mixed with fresh 200-μL cold gelation mixture.Tissues and gelation mixture were overlaid onto a base of gelled collagen ineach cell culture insert and allowed to gel at 37 °C. Subsequently, 200 μLDMEM/F12 supplemented with 20% (vol/vol) charcoal/dextran-treated FBS

Fig. 6. A possible scheme depicting estrogen-induced cell proliferation anddifferentiation and its regulation of homeostasis in mouse vagina.

12990 | www.pnas.org/cgi/doi/10.1073/pnas.1513550112 Miyagawa and Iguchi

Dow

nloa

ded

by g

uest

on

June

29,

202

0

(HyClone) was added to each well, and tissues were cultured at 37 °C in ahumidified, 5% CO2/air atmosphere for 3 d. Recombinant AREG, HBEGF, TGFΑ(R&D systems), IGF1 (JRH Biosciences; 0.5 μg/μL), and anti-AREG antibody (5 μg/mL)were added to the medium from day 0 of culture. Tissues for each group werecollected from at least 4 mice.

ACKNOWLEDGMENTS. We thank Dr. J. Takeda (Osaka University) andPierre Chambon (Institute for Genetics and Cellular and Molecular

Biology) for providing K5-Cre and ERα-floxed mice, respectively. We alsothank Dr. I. Hirakawa and Dr. H. Miyakawa (National Institute for BasicBiology) and Dr. T. Nakajima and Dr. T. Sato (Yokohama City University) forcritical advice on the DNA microarray and organ culture. We are grateful toDr. B. Blumberg (University of California at Irvine) for his critical readings ofthe manuscript. This study was supported by a Grant-in-Aid for ScientificResearch B from the Ministry of Education, Culture, Sports, Science andTechnology of Japan and a Health Sciences Research Grant from the Min-istry of Health, Labour and Welfare, Japan.

1. Dupont S, et al. (2000) Effect of single and compound knockouts of estrogen re-ceptors alpha (ERalpha) and beta (ERbeta) on mouse reproductive phenotypes.Development 127(19):4277–4291.

2. Couse JF, Korach KS (1999) Estrogen receptor null mice: What have we learned andwhere will they lead us? Endocr Rev 20(3):358–417.

3. Lubahn DB, et al. (1993) Alteration of reproductive function but not prenatal sexualdevelopment after insertional disruption of the mouse estrogen receptor gene. ProcNatl Acad Sci USA 90(23):11162–11166.

4. Cunha GR, et al. (1997) Elucidation of a role for stromal steroid hormone receptors inmammary gland growth and development using tissue recombinants. J MammaryGland Biol Neoplasia 2(4):393–402.

5. Buchanan DL, et al. (1998) Role of stromal and epithelial estrogen receptors in vaginalepithelial proliferation, stratification, and cornification. Endocrinology 139(10):4345–4352.

6. Cooke PS, et al. (1997) Stromal estrogen receptors mediate mitogenic effects of es-tradiol on uterine epithelium. Proc Natl Acad Sci USA 94(12):6535–6540.

7. Nelson KG, Sakai Y, Eitzman B, Steed T, McLachlan J (1994) Exposure to di-ethylstilbestrol during a critical developmental period of the mouse reproductivetract leads to persistent induction of two estrogen-regulated genes. Cell GrowthDiffer 5(6):595–606.

8. Ignar-Trowbridge DM, et al. (1992) Coupling of dual signaling pathways: epidermalgrowth factor action involves the estrogen receptor. Proc Natl Acad Sci USA 89(10):4658–4662.

9. Nelson KG, Takahashi T, Bossert NL, Walmer DK, McLachlan JA (1991) Epidermalgrowth factor replaces estrogen in the stimulation of female genital-tract growth anddifferentiation. Proc Natl Acad Sci USA 88(1):21–25.

10. Ghahary A, Chakrabarti S, Murphy LJ (1990) Localization of the sites of synthesis andaction of insulin-like growth factor-I in the rat uterus. Mol Endocrinol 4(2):191–195.

11. Winuthayanon W, Hewitt SC, Orvis GD, Behringer RR, Korach KS (2010) Uterine epi-thelial estrogen receptor α is dispensable for proliferation but essential for completebiological and biochemical responses. Proc Natl Acad Sci USA 107(45):19272–19277.

12. Feng Y, Manka D, Wagner KU, Khan SA (2007) Estrogen receptor-alpha expression inthe mammary epithelium is required for ductal and alveolar morphogenesis in mice.Proc Natl Acad Sci USA 104(37):14718–14723.

13. Mueller SO, Clark JA, Myers PH, Korach KS (2002) Mammary gland development inadult mice requires epithelial and stromal estrogen receptor alpha. Endocrinology143(6):2357–2365.

14. Kurita T, Cooke PS, Cunha GR (2001) Epithelial-stromal tissue interaction in para-mesonephric (Müllerian) epithelial differentiation. Dev Biol 240(1):194–211.

15. Miyagawa S, Sato M, Sudo T, Yamada G, Iguchi T (2015) Unique roles of estrogen-dependent Pten control in epithelial cell homeostasis of mouse vagina. Oncogene34(8):1035–1043.

16. Villacorte M, et al. (2013) β-Catenin signaling regulates Foxa2 expression during en-dometrial hyperplasia formation. Oncogene 32(29):3477–3482.

17. Gimenez-Conti IB, et al. (1994) Expression of keratins in mouse vaginal epithelium.Differentiation 56(3):143–151.

18. Yang A, et al. (1999) p63 is essential for regenerative proliferation in limb, cranio-facial and epithelial development. Nature 398(6729):714–718.

19. Mills AA, et al. (1999) p63 is a p53 homologue required for limb and epidermalmorphogenesis. Nature 398(6729):708–713.

20. Kurita T, Mills AA, Cunha GR (2004) Roles of p63 in the diethylstilbestrol-inducedcervicovaginal adenosis. Development 131(7):1639–1649.

21. Yoo YA, et al. (2013) Progesterone signaling inhibits cervical carcinogenesis in mice.Am J Pathol 183(5):1679–1687.

22. Lopez RG, et al. (2009) C/EBPalpha and beta couple interfollicular keratinocyte pro-liferation arrest to commitment and terminal differentiation. Nat Cell Biol 11(10):1181–1190.

23. Ciarloni L, Mallepell S, Brisken C (2007) Amphiregulin is an essential mediator of es-trogen receptor alpha function in mammary gland development. Proc Natl Acad SciUSA 104(13):5455–5460.

24. Stoll SW, Johnson JL, Li Y, Rittié L, Elder JT (2010) Amphiregulin carboxy-terminaldomain is required for autocrine keratinocyte growth. J Invest Dermatol 130(8):2031–2040.

25. Cook PW, et al. (1992) Amphiregulin messenger RNA is elevated in psoriatic epidermisand gastrointestinal carcinomas. Cancer Res 52(11):3224–3227.

26. Miyagawa S, et al. (2004) Persistent gene expression in mouse vagina exposed neo-natally to diethylstilbestrol. J Mol Endocrinol 32(3):663–677.

27. Winuthayanon W, Hewitt SC, Korach KS (2014) Uterine epithelial cell estrogen re-ceptor alpha-dependent and -independent genomic profiles that underlie estrogenresponses in mice. Biol Reprod 91(5):110.

28. Kurita T, Lee KJ, Cooke PS, Lydon JP, Cunha GR (2000) Paracrine regulation of epi-thelial progesterone receptor and lactoferrin by progesterone in the mouse uterus.Biol Reprod 62(4):831–838.

29. Kurita T, et al. (2000) Paracrine regulation of epithelial progesterone receptor byestradiol in the mouse female reproductive tract. Biol Reprod 62(4):821–830.

30. Li Q, et al. (2011) The antiproliferative action of progesterone in uterine epithelium ismediated by Hand2. Science 331(6019):912–916.

31. Mantena SR, et al. (2006) C/EBPbeta is a critical mediator of steroid hormone-regu-lated cell proliferation and differentiation in the uterine epithelium and stroma. ProcNatl Acad Sci USA 103(6):1870–1875.

32. Robinson GW, Johnson PF, Hennighausen L, Sterneck E (1998) The C/EBPbeta tran-scription factor regulates epithelial cell proliferation and differentiation in themammary gland. Genes Dev 12(12):1907–1916.

33. Miyagawa S, Katsu Y, Watanabe H, Iguchi T (2004) Estrogen-independent activationof erbBs signaling and estrogen receptor alpha in the mouse vagina exposed neo-natally to diethylstilbestrol. Oncogene 23(2):340–349.

34. Platz CE, Benda JA (1995) Female genital tract cancer. Cancer 75(1, Suppl):270–294.35. Zhai Y, et al. (2010) Loss of estrogen receptor 1 enhances cervical cancer invasion. Am

J Pathol 177(2):884–895.36. Tarutani M, et al. (1997) Tissue-specific knockout of the mouse Pig-a gene reveals

important roles for GPI-anchored proteins in skin development. Proc Natl AcadSci USA 94(14):7400–7405.

Miyagawa and Iguchi PNAS | October 20, 2015 | vol. 112 | no. 42 | 12991

DEV

ELOPM

ENTA

LBIOLO

GY

Dow

nloa

ded

by g

uest

on

June

29,

202

0