aP2-Cre-Mediated Inactivation of Estrogen Receptor Alpha Causes Hydrometra

In this study we describe the reproductive phenotypes of a novel mouse model in which Cre-mediated deletion of ERα is regulated by the aP2 (fatty acid binding protein 4) promoter. ERα-floxed mice were crossed with transgenic mice expressing Cre-recombinase under the control of the aP2 promoter to generate aP2-Cre/ERαflox/flox mice. As expected, ERα mRNA levels were reduced in adipose tissue, but in addition we also detected an 80% reduction of ERα levels in the hypothalamus of aP2-Cre/ERαflox/flox mice. Phenotypic analysis revealed that aP2-Cre/ERαflox/flox female mice were infertile. In line with this, aP2-Cre/ERαflox/flox female mice did not cycle and presented 3.8-fold elevated estrogen levels. That elevated estrogen levels were associated with increased estrogen signaling was evidenced by increased mRNA levels of the estrogen-regulated genes lactoferrin and aquaporin 5 in the uterus. Furthermore, aP2-Cre/ERαflox/flox female mice showed an accumulation of intra-uterine fluid, hydrometra, without overt indications for causative anatomical anomalies. However, the vagina and cervix displayed advanced keratosis with abnormal quantities of accumulating squamous epithelial cells suggesting functional obstruction by keratin plugs. Importantly, treatment of aP2-Cre/ERαflox/flox mice with the aromatase inhibitor Letrozole caused regression of the hydrometra phenotype linking increased estrogen levels to the observed phenotype. We propose that in aP2-Cre/ERαflox/flox mice, increased serum estrogen levels cause over-stimulation in the uterus and genital tracts resulting in hydrometra and vaginal obstruction.


Introduction
Estrogen receptor alpha (ERa, NR3A1) and ERb (NR3A2) are two nuclear receptors that mediate the physiological responses to estrogen [1]. They are ligand-activated transcription factors, encoded by the Esr1 and Esr2 genes, that bind to DNA and regulate transcription in response to their ligands [2]. ERa has important roles in both the regulation of male and female reproduction, and also in the control of metabolism [3]. During the estrous cycle, 17b-estradiol (E2) levels are regulated via feedback mechanisms involving the ovaries, hypothalamus and pituitary gland. At the mid-stage of the estrous cycle, E2 produced within the gonads exerts a stimulatory effect on gonadotropinreleasing hormone (GnRH) neurons in the hypothalamus. The resulting GnRH discharge stimulates the anterior pituitary gland to release luteinizing hormone (LH), which in turn triggers ovulation. At other stages of the estrous cycle, E2 exerts a negative feedback that results in the suppression of GnRH secretion, as reviewed in [4]. Estrogenic signals, via ERa, also control a number of functions in the uterus including early events like hyperemia and water imbibition, and later events such as epithelial cell proliferation and differentiation [5,6]. Mice lacking ERa, ERa 2/ 2 mice, are infertile and have atrophic uteri that do not respond to E2 [7,8,9,10,11].
In this study, floxed ERa mice were crossed with aP2-Cre transgenic mice to generate aP2-Cre/ERa flox/flox mice. Consistent with the well-described expression of the aP2 promoter in adipocytes, aP2-Cre/ERa flox/flox mice display down-regulation of the ERa transcript in both white and brown adipose tissue (WAT and BAT). However, down-regulation of the ERa transcript is also pronounced in the hypothalamus. Phenotypically, female aP2-Cre/ERa flox/flox mice are infertile and develop hydrometra (fluidfilled distended uteri). We propose that in aP2-Cre/ERa flox/flox mice, increased serum E2 levels cause over-stimulation of the uterus and genital tracts resulting in hydrometra and vaginal obstruction.

Animals
ERa flox/flox mice (B6.129X1-Esr1 tm1Gust ) [7] were bred with transgenic mice expressing the Cre enzyme under the control of the aP2/Fabp4 promoter (B6.Cg-Tg(Fabp4-Cre) 1Rev/J) [12] to generate mice with ERa deletion in fat cells. The final breeding step was performed using male aP2-Cre/ERa flox/flox and female ERa flox/flox mice. All mice analyzed in this study were on a congenic C57BL/6J genetic background. Genotyping of the ERa floxed locus was performed using PCR on DNA from ear or tail biopsies as described previously [7]. The presence of the aP2-Cre transgene was detected with primers 59-GTTTCACTATC-CAGGTTACGG and 59-GTACTCTAAGTCCAGTGATC. Mice were maintained on a 14 h light, 10 h dark cycle and given a continuous supply of food and water.

Fertility Tests
Fertility tests of female aP2-Cre/ERa flox/flox (n = 6) mice were performed using continuous mating with male partners for three months. Mating was started at six weeks of age, and the numbers of litters and litter size were recorded.

Measurement of Serum E2 Levels
E2 levels were determined using commercially available RIA kits (Siemens Medical Solutions, CA, USA), according to the manufacturer's instructions.

RNA Isolation and RT-PCR
Tissues were dissected from 6-9 week-old female mice and immediately frozen on dry ice for storage at 270uC. Total RNA was isolated from frozen tissues using Trizol reagent (Invitrogen) and then purified with RNeasy Plus Mini Kits (Qiagen) as described in [14]. cDNA was synthesized using random primers and either Superscript II (Invitrogen) or TaqManH Reverse Transcription Reagents (Life Technologies). PCR was performed using RedTaq DNA polymerase (Sigma-Aldrich) and the following primers: ERa exon2F: 59-CCCTACTACCTGGAGAACGA and ERa exon5R: 59-TGCCCACTTCGTAACACTTG [7]. ERa expression levels were assessed by semi-quantitative real-time PCR with a 7300 Real-Time PCR System (Applied Biosystems), using TaqManH Universal PCR Master Mix (Life Technologies) and a TaqManH Gene Expression Assay (Mm00433148_mH, Life Technologies), according to the manufacturer's instructions. Gene expression was normalized to 18S (TaqManH Ribosomal RNA Control Reagents, Life Technologies). The expression of ERa target genes and Cre were analyzed using Power SYBR Green Master Mix (Applied Biosystems), according to the manufacturer's instructions, and the following primers: Lactoferrin F 59-CAGCAGGATGTGATAGCCACAA, R 59-CACTGATCA-CACTTGCGCTTCT [15], Aqp5 F 59-TTGTGAAGGCAGTG-CAAGCT, R 59-CACCCCTTTCTGGGATGGT [16]. Cre expression was analyzed with: Cre F 59-GCCGCGCGAGA-TATGG and Cre R 59-AGCTTGCATGATCTGCGGTATT.

Histology
The female genital tract including ovaries, along with liver, kidney, brain, interscapular BAT, visceral (abdominal attached to ovaries and uterus) WAT, and inguinal subcutaneous WAT were collected from 12 mice that were between 6 and 12 weeks old. The tissues were fixed for 24 h in 4% neutral-buffered formaldehyde and stored in 70% ethanol prior to routine processing and embedding in paraffin blocks. Paraffin-embedded tissues were cut to 4 mm thickness, deparaffinized, rehydrated and stained with hematoxylin and eosin (H&E). The resulting slides were microscopically analyzed by a pathologist.

Treatment with Aromatase Inhibitor
Letrozole was purchased from Selleckchem and dissolved at 2 mg/ml in saline containing 0.3% hydroxymethyl cellulose. Letrozole (10 mg/kg body weight) or vehicle control was delivered to mice via daily sc injection.

Results
Female aP2-Cre/ERa Knockout Mice are Infertile and have Increased E2 Serum Levels Fertility in female aP2-Cre/ERa flox/flox mice was investigated by continuous mating with fertile males for a three-month period. Since the breeding did not result in any pups (data not shown), we concluded that female aP2-Cre/ERa flox/flox mice are infertile. Vaginal smears from 2 month-old aP2-Cre/ERa flox/flox and ERa flox/flox littermates demonstrated that while ERa flox/flox animals cycled normally, aP2-Cre/ERa flox/flox mice displayed vaginal smears compatible with constant estrus (Fig. 1A). We next analyzed serum E2 concentrations, and found elevated levels in aP2-Cre/ERa flox/flox females compared to control female mice (Fig. 1B). The E2 levels were more than three-fold higher in aP2-Cre/ERa flox/flox females compared to ERa flox/flox female mice (23.8 pg/ml versus 6.25 pg/ml, respectively), and within the same range as our analysis of E2 levels in female mice with a global knockout of ERa (data not shown).

ERa Expression in aP2-Cre/ERa Knockout Mice
Ablation of the ERa gene in aP2-Cre/ERa flox/flox mice was analyzed by RT-PCR utilizing primers flanking exon 3 of the ERa gene. This analysis generates a 364 bp product from the WT transcript and a 176 bp PCR product from the targeted allele that lacks exon 3. As expected, only the WT transcript was detected in samples from ERa flox/flox mice. A 176 bp product, corresponding to the knockout transcript, was present in WAT and BAT from aP2-Cre/ERa flox/flox mice ( Fig. 2A). We also detected mRNA from the knockout allele in hypothalamus. Only the WT transcript was detected in liver and muscle. Traces of the knockout transcripts could also be detected in the uterus and kidney, although their levels are extremely low compared to those of the WT transcript in these tissues. To quantitatively determine the efficiency of knockout in the different tissues, mRNA levels of the WT ERa transcript were assayed using real time PCR. Significant down-regulation of the ERa transcript was observed in aP2-Cre/ ERa flox/flox mice, compared to control mice, in inguinal and visceral adipose tissue, brown adipose tissue, hypothalamus, and kidney (reductions of 44%, 31%, 83%, 80% and 42%, respectively), but not in muscle or liver; the reduction in ERa expression aP2-Cre ERa Gene Targeting PLOS ONE | www.plosone.org levels in the uterus was just below significance (Fig. 2B). To determine if the reduction of ERa mRNA levels in hypothalamus was due to Cre expression we assayed Cre expression using real time PCR. Significant Cre expression was detected in hypothalamus as well as all adipose depots analyzed and to lesser extents in uterus and kidney (Fig. 2C).

Female aP2-Cre/ERa Knockout Mice Develop Hydrometra
All female aP2-Cre/ERa flox/flox mice had swollen abdomens at 8 weeks of age. Internal anatomical examination revealed that the uteri in these mice were extensively fluid-filled (Fig. 3A). This phenotype was observed in all examined knockout mice but not in any of the control littermates. The accumulated uterine fluid was in most cases a clear and watery liquid characteristic of that seen in hydrometra, but in some of the animals the fluid was cloudy, consistent with an inflammatory response likely resulting from bacterial infection (pyometra).
Histologic analysis of the genital tract revealed that all investigated aP2-Cre/ERa flox/flox mice had distended uteri, usually with watery contents (hydrometra) and thin walls, together with a vastly reduced glandular content and a thin muscular layer (atrophy) (Fig. 3B). All control mice showed normal uterus histology with well-developed muscular walls and glands.
Morphological analysis of the vagina and cervix (Fig. 3C) revealed that aP2-Cre/ERa flox/flox mice had marked epithelial keratosis with abnormal quantities of accumulating cornified squamous epithelial cells in the vaginal lumen. We speculate that the observed hydrometra might have resulted from vaginal keratin plugs which could functionally obstruct the vagina. No other anatomical abnormalities, such as imperforate vagina or cervical/ vaginal sagittal septa, were detected. Vaginal keratinization with variable luminal accumulation of cornified squamous epithelial cells was observed in 3 out of 6 normally cycling control mice.
The ovaries of aP2-Cre/ERa flox/flox mice consistently demonstrated hemorrhagic follicles when compared to control mouse ovaries (Fig. 3D). Additionally, aP2-Cre/ERa flox/flox ovaries did not show any signs of luteinisation, compared to control ovaries, which exhibited normal corpora lutea in 2 out of 3 observed specimens. Since antral follicles were abundant in all of the aP2-Cre/ERa flox/flox ovaries examined, aP2-Cre/ERa flox/flox ovaries appear to halt follicle development only at the final stage before ovulation. Indeed, the hemorrhages observed are likely the result of aberrant ovulation or follicle rupture. No marked differences in the numbers of atretic follicles were observed.

Up-regulation of Estrogen Target Genes
To identify genes in the uterus that could be involved in the development of hydrometra we analyzed the expression of the known E2 target genes, lactoferrin and aquaporin 5 [15,16]. Lactoferrin mRNA levels were almost 10-fold higher, and aquaporin 5 mRNA levels about 7-fold higher, in the uterus of aP2-Cre/ERa flox/flox mice compared to controls (Fig. 4).

Inhibition of Endogenous Estrogen Synthesis Reduces Hydrometra
To analyze if inhibition of endogenous estrogen synthesis via the aromatase enzyme would reverse hydrometra, we treated aP2-Cre/ERa flox/flox (n = 5) and ERa flox/flox (n = 5) mice with the aromatase inhibitor Letrozole or vehicle. Mice were 10 weeks old at the start of the treatment, at which stage all aP2-Cre/ERa flox/ flox mice displayed clear visual signs of hydrometra. Importantly, Letrozole treatment reversed visual signs of hydrometra in aP2-Cre/ERa flox/flox (n = 3) mice within one week of treatment. When the mice were sacrificed after 17 days of treatment, uteri appeared grossly normal in Letrozole-treated mice while the vehicle-treated mice presented a severe hydrometra phenotype (Fig. 5A, B).

Discussion
We generated a novel mouse line in which Cre-mediated deletion of ERa is regulated by the aP2 promoter with the initial aim of targeting adipocyte ERa signaling using the Cre/loxP system. During the generation of this mouse strain, we observed that aP2-Cre-driven deletion of ERa leads to infertility in female mice and an arrest of the estrous cycle with hydrometra and increased serum E2 levels. Expression studies showed that the aP2-Cre transgene directs expression of Cre to adipose tissue as expected. Additionally, the aP2-driven Cre gene is also expressed in the hypothalamus, with a concomitant reduction in ERa levels also in this tissue (Fig. 2). The expression of Cre in the hypothalamus in aP2-Cre mice is consistent with previous observations [17,18]. Estrogen action in the hypothalamus has drastic effects on the estrous cycle and on regulation of serum E2 levels [19,20,21], and it is conceivable that deletion of ERa in this brain region is the dominant cause of the severe reproductive effects observed in female aP2-Cre/ERa flox/flox mice.
We speculate that the increased E2 levels observed in female aP2-Cre/ERa flox/flox mice are related to deletion of ERa in the hypothalamus, resulting in disruption of the E2 feedback loop. Brain-specific ERa deletion using CamKIIa-Cre or nestin-Cre has previously been shown to cause elevated serum E2 levels and infertility [19,21]. Also deletion of ERa in distinct hypothalamic neurons (SF1-Cre or POMC-Cre) affects fertility [21]. The increased levels of serum E2 in aP2-Cre/ERa flox/flox mice may explain the block in the estrous cycle, since cycling E2 levels control the estrous cycle in mammals, exerting both negative and positive feedback effects [22,23].
E2 is known to influence both uterine weight and vaginal epithelial cytology, and treatment of mice with E2 stimulates both uterine weight gain and vaginal epithelial proliferation and keratinization. In global ERa knockout mice, E2 treatment does not increase either uterine weight or the abundance of cornified epithelial cells in the vagina, showing that ERa is necessary for both these processes [8]. Long term treatment of WT mice with E2 has been shown to result in hydrometra [24], and we suggest that the hydrometra observed in aP2-Cre/ERa flox/flox mice is a result of continuous E2 stimulation of the uterus, combined with severe vaginal hyperplasia and keratinization, resulting in accumulation of vast numbers of intraluminal keratinized squamous epithelial cells. In support of this, we show that treatment of aP2-Cre/ERa flox/flox mice with the aromatase inhibitor Letrozole reverses hydrometra (Fig 5). Interestingly, short-term E2 treatment of global ERb knockout mice also results in fluid-filled uteri, and it was speculated that this is a result of increased signaling by ERa due to the loss of ERb which was suggested to dampen the effects of ERa [25]. A similar uterine phenotype was described by Wintermantel et al. [19] in mice with a CamKIIa-Cre-driven neuron-specific ERa knockout. In contrast, mice with a global deletion of ERa have severely hypoplastic uteri [7,8,9,10,11], although serum E2 levels are increased [8,9,26,27,28]. In this case, the uterus cannot respond to the increased E2 levels due to lack of ERa in this organ. Although ERa expression was also reduced in the uteri of aP2-Cre/ERa mice this reduction did not achieve significance, and was presumably insufficient to block overstimulation in this organ in response to elevated E2 levels, followed by atrophy at later stages. In line with this, we still observed marked up-regulation of the known E2 target genes lactoferrin and aquaporin 5 in uteri of aP2-Cre/ERa mice (Fig. 4).
In summary, we have generated a conditional ERa knockout mouse model using aP2-Cre-driven gene deletion, and we here demonstrate that these mice develop hydrometra. Our results are consistent with a mechanism involving reduction of ERa expression in the hypothalamus, which results in disruption of E2 regulation and increased serum E2 levels, leading to a block of the estrous cycle and hyper-stimulation of the uterus. Collectively, these results underscore the roles of E2 and ERa as main players in the development of hydrometra, and also the challenges associated with the use of the aP2-Cre transgene to target adipose gene expression.