Response of adult mouse uterus to early disruption of estrogen receptor-α signaling is influenced by Krüppel-like factor 9

Abstract

Inappropriate early exposure of the hormone-responsive uterus to estrogenic compounds is associated with increased risk for adult reproductive diseases including endometrial cancers. While the dysregulation of estrogen receptor-α (ESR1) signaling is well-acknowledged to mediate early events in tumor initiation, mechanisms contributing to sustained ESR1 activity later in life and leading to induction of oncogenic pathways remain poorly understood. We previously showed that the transcription factor Krüppel-like Factor 9 (KLF9) represses ESR1 expression and activity in Ishikawa endometrial glandular epithelial cells. We hypothesized that KLF9 functions as a tumor suppressor and loss of its expression enhances ESR1 signaling. Here, we evaluated the contribution of KLF9 to early perturbations in uterine ESR1 signaling pathways elicited by the administration of synthetic estrogen diethylstilbestrol (DES) to wild-type (WT) and Klf9 null (KO) mice on postnatal days (PND) 1–5. Uterine tissues collected at PND84 were subjected to histological, immunological, and molecular analyses. Compared to WT, KO mice demonstrated larger endometrial glands and lower endometrial gland numbers; DES exposure exacerbated these differences. KLF9 loss of expression resulted in increased glandular ESR1 immunoreactivity with DES, without effects on serum estradiol levels. QPCR analyses indicated altered expression of uterine genes commonly dysregulated in endometrial cancers (Akt1, Mmp9, Slpi, Tgfβ1) and those involved in growth regulation (Fos, Myc, Tert, Syk), with loss of Klf9, alone or in concert with DES. Our data support a molecular network between KLF9 and ESR1 in the uterus and suggest that silencing of KLF9 may contribute to endometrial dysfunction initiated by aberrant estrogen action.

Introduction

The steroid hormone 17β-estradiol (E₂) regulates numerous physiological and developmental processes in target tissues by binding to its cognate receptors estrogen receptor-α (ESR1) and estrogen receptor-β (ESR2) in the presence of numerous co-regulators (Heldring et al. 2007). In the uterus, ESR1 is the predominant isoform and its presence is necessary for reproductive functions (Couse & Korach 1999; Matthews & Gustafsson 2003). Indeed, mice null for Esr1 are infertile in contrast to Esr2 KO mice (Hewitt & Korach, 2003), due in part to the failure of the uteri to properly develop (Couse et al. 1995). Conversely, the unopposed activity of uterine ESR1 promotes the development and progression of endometrial cancers in humans and animal models (Di Cristofano & Ellenson 2007). Multiple mechanisms underlie the control of ESR1 expression and activity; these include regulation at the levels of transcription involving multiple promoters (Kos et al. 2001); alternative splicing (Heldring et al. 2007); post-translational modifications via phosphorylation, acetylation, and polyubiquination (Fu et al. 2004; Masuyama & Hiramatsu 2004; Atsriku et al. 2009); and posttranscriptional silencing by micro-RNA targeting (Castellano et al. 2009). The aforementioned complex regulation of ESR1 occurring at many levels predicts that perturbations in any or all of these mechanisms will lead to reproductive dysfunctions and diseases.

Endocrine disrupting chemicals (EDCs) mimic or antagonize the effects of naturally occurring substances that elicit physiological responses in target cells (Newbold et al., 2006, 2007). Numerous studies are aimed at elucidating the mechanism(s) by which such chemicals exert potentially devastating health consequences in subjects upon environmental exposure (Diamanti-Kandarakis et al. 2009). Diethylstilbestrol (DES), a synthetic estrogen, earned its notoriety as an EDC based on the documented increased breast cancer risk in exposed mothers (Greenberg et al. 1984; Colton et al. 1993; Palmer et al. 2006) and of vaginal clear cell adenocarcinoma in daughters exposed in utero (Herbst et al. 1971). In mice, the decreased fecundity and higher incidence of structural abnormalities of the female reproductive tract with prenatal exposure to DES (Newbold et al. 1995; Kaufman et al. 2000) were recapitulated by postnatal DES administration (postnatal days 1–5), a critical window of reproductive tract differentiation (Newbold et al. 1990). The majority of the uterine defects observed with DES had been attributed to its ability to alter the expression, co-regulatory protein recruitment, and downstream signaling events of ESR1 (Newbold et al. 1995; Couse et al. 2001; Yamashita 2004). Indeed, the over-expression of ESR1 enhanced the progression of DES-induced uterine adenocarcinoma (Couse et al. 1997). While these findings implicate perturbations in ESR1 signaling to play a pivotal role in DES-mediated reproductive dysfunctions, the precise mechanisms and participating molecules remain poorly understood.

Krüppel-like factor (KLF) 9, a member of the Sp/KLF family of transcription factors (Suske et al. 2005) is a 244 amino acid protein characterized by three C-terminal zinc finger motifs that are important for binding to GC-rich sequences in DNA (Imataka et al. 1992). By using Klf9 null mice and endometrial cell lines with KLF9 knockdown in vitro, our laboratory has shown that KLF9 is a progesterone receptor (PGR) co-regulator (Zhang et al. 2002a; Zhang et al. 2003) and depending on cell context, a negative (Velarde et al. 2007) or positive (Pabona et al. 2009) regulator of ESR1 signaling. In mice, Klf9 null mutation was associated with uterine hypoplasia and reduced uterine progesterone sensitivity, leading to a subfertility phenotype (Simmen et al. 2004) and with delayed parturition due to dysregulated PGR-A expression (Zeng et al. 2008). Klf9 mRNA expression was recently shown to be attenuated in abnormal uterine conditions such as in a mouse model of endometriosis and in human leiomyoma (Rackow et al. 2008; Lee et al. 2009). These findings indicate that Klf9 plays an important role with ESR1 and PGR in uterine growth, function, and pathology.

In the present study, we evaluated the existence of a uterine ESR1/KLF9 crosstalk in the context of perturbed ESR1 signaling elicited by postnatal (days 1–5) DES exposure of WT and Klf9 null mice. We show that relative to those of WT, adult uteri of Klf9 null mice had reduced luminal epithelial height, larger endometrial glands and lower endometrial gland numbers, and that the DES effects on these indices were dependent on Klf9 expression. We further show that uterine Klf9 expression temporally lagged behind that for ESR1 (Yamashita et al., 1989; Sato et al., 1992; Newbold 1995) and was insensitive to early DES exposure. We demonstrate that while Klf9 loss of expression had no effect on basal ESR1 expression in endometrial stromal and glandular epithelial compartments of adult uteri, early DES-exposed Klf9 null mutants exhibited higher glandular ESR1 expression. Finally, we found altered expression of a subset of genes frequently dysregulated in endometrial cancers (Akt1, Mmp9, Slpi) and of pro-proliferative genes (Fos, Myc, Tert, Syk), with loss of Klf9 alone or in concert with DES exposure. These findings provide support for KLF9’s participation in endometrial ESR1-sensitive developmental events and suggest that silencing of KLF9 may contribute to the etiology of endometrial cancers initiated by aberrant ESR1 signaling.

Materials and Methods

Animals and treatments

Animal studies were conducted in compliance with protocols approved by the Institutional Animal Care and Use Committee of the University of Arkansas for Medical Sciences. C57BL/6J Klf9 WT and mutant mice were propagated as previously published (Morita et al. 2003; Simmen et al. 2004). To eliminate confounding effects of maternal phenotype to pup viability (Simmen et al. 2004), WT and mutant offspring were generated from male Klf9 (+/−) x female Klf9 (+/−) matings. At birth [postnatal day (PND) 1] until PND5, female pups were injected daily with 2 μg of diethylstilbestrol (DES) dissolved in sesame oil (vehicle) or sesame oil alone (Oil; control) (Newbold et al., 1990). At PND21, the pups were weaned and genotyped. For purposes of this study, only WT and KO (−/−) mice were compared. At PND84, body weight and stage of estrus (by vaginal smears) were determined and the females were sacrificed by CO₂ asphyxiation. For each mouse, the ovary pairs and uterine horns were weighed. One uterine horn was placed in 10% neutral buffered formalin and processed for histomorphometry, while the other horn was processed for RNA isolation. A total of 16 WT (9 and 7, respectively for Oil and DES) and 16 KO (8 and 8, respectively for Oil and DES) mice were used for tissue collection and analyses (Figure 1).

Figure 1.

Experimental design using Klf9 WT and null mutant mice according to the DES model of endometrial carcinoma. PND=postnatal day.

RNA isolation and quantitative RT-PCR (QPCR)

Total RNA was isolated from uterine tissues with TRIzol reagent (Invitrogen, Carlsbad, CA, USA) following the manufacturer’s protocol. RNA concentrations were determined using the Nanodrop ND-1000 spectrophotometer (Nanodrop, Wilmington, DE, USA). cDNA was synthesized from 1 μg of total RNA using the Taqman Reverse Transcription Reagents kit (Applied Biosystems, Foster City, CA, USA), and analyzed by QPCR using SYBR Green (Bio-Rad, Hercules, CA, USA) and the Bio-Rad MyiQ optical module and iCycler thermal cycler as previously described (Velarde et al. 2006). Intron-flanking primers were designed to eliminate the amplification of genomic DNA using the Primer Express software (Applied Biosystems) and were obtained from Integrated DNA Technologies, Inc. (Coralville, IA, USA). The primer sequences (sense and antisense, respectively) and the corresponding PCR product sizes are shown in Table 1. A standard curve was generated by serially diluting pooled cDNAs from all animals beginning with the most concentrated cDNA pool designated as 10,000 arbitrary units. Focused QPCR array (Mouse Cancer PathwayFinder^™, SABiosciences, Frederick, MD) analyses followed protocols described by the manufacturer, using cDNAs prepared from pooled RNAs (of same amount) isolated from DES-treated WT (n=5) and Klf9 mutant (n=4) mice.

Table 1.

Mouse primer sequences used for QPCR

Gene	Sense	Anti-sense	Product size
Ppia	5′-AGATGCCAGGACCTGTATGCTT-3′	5′-TGTGCCAGGGTGGTGACTTTA-3′	91 bp
Rn18s	5′-TGCTCTTAGCTGAGTGTCCCG-3′	5′-TCATGGCCTCAGTTCCGAAA-3′	153 bp
Gapdh	5′-TGTGTCCGTCGTGGATCTGA-3′	5′-GCATCGAAGGTGGAAGAGTGG-3′	173 bp
Klf13	5′-ACACAGGTGAGAGGCCTTTCG-3′	5′-AGCATGCCTGGGTGGAAGT-3′	199 bp
Slpi	5′-AAGTCCTGCGGCCTTTTACCT-3′	5′-GGCATTGTGGCTTCTCAAGCT-3′	154 bp
Pten	5′-CCCACCACAGCTAGAACTTATCAA-3′	5′-TCCGTCCCTTTCCAGCTTTAC-3′	110 bp
Myc	5′-TACTTGGAGGAGACATGGTGAATC-3′	5′-AGAAGCCGCTCCACATACAGTC-3′	102 bp
Akt1	5′-GACTGACACCAGGTATTTCGATGA-3′	5′-CTCCGCTCACTGTCCACACA-3′	102 bp
Fos	5′-TGAAGAGGAAGAGAAACGGAGAAT-3′	5′-TCCGCTTGGAGTGTATCTGTCA-3′	102 bp
Syk	5′-AGGGTCAAAGACCAATGGAAAA-3′	5′-CTGTCAATGCGGTAGTGCAATACT-3′	108 bp
Tert	5′-CCTCCAGACAGTCTGCATCAATAT-3′	5′-TCTTCCTAACACGCTGGTCAAA-3′	104 bp
Tgfb1	5′-GCCTGAGTGGCTGTCTTTTGAC-3′	5′-CACAAGAGCAGTGAGCGCTG-3′	101 bp
Mmp9	5′-TCGCGTGGATAAGGAGTTCTC-3′	5′-AACTCACACGCCAGAAGAATTTG-3′	107 bp

Immunohistochemistry, uterine morphometry, and X-gal staining

Whole uteri were fixed in 10% neutral-buffered formalin as previously described (Simmen et al., 2004). Sections (5 μm) were mounted on poly(lysine)-coated slides (Fisher Scientific, Pittsburgh, PA, USA) and used for uterine morphometry or immunohistochemistry. For morphometrical analyses, the sections were deparaffinized in xylene, rehydrated in a series of alcohols, and stained with hematoxylin and eosin following standard protocols. Luminal epithelial (LE) height and gland numbers were quantified using the Carl Zeiss Axiovert 200M microscope equipped with a Zeiss Axiocam HRc camera and Axiovision software (Carl Zeiss, Inc., Thornwood, NY, USA). Measurements were taken from 5 distinct fields (100X) per section, and data from 2–3 serial sections from each Klf9 WT and KO mouse (n=3 mice per treatment per genotype) were analyzed. For detection and quantification of ESR1-immunostaining cells, sections were treated with 3% hydrogen peroxide for 30 min to quench endogenous peroxidase activity and with Citra Plus (Biogenex, San Ramon, CA, USA) in a microwave set at power 10 for 105 sec and at power 1 for 10 min to unmask antigen. Sections were incubated in a blocking solution (VectaStain ABC kits; Vector Laboratories, Burlingame, CA, USA) for overnight to reduce non-specific staining prior to incubation for 16–24 h with rabbit anti-mouse polyclonal ESR1 antibody (1:250 dilution; sc-542, Santa Cruz Biotechnology, Santa Cruz, CA) in a humidity chamber at 4°C. Incubation with biotinylated, anti-rabbit secondary antibody (1:200 dilution; VectaStain ABC kits) was carried out for 30 min at room temperature. Sections were stained with DAB Chromogen (3,3′-diaminobenzidine, Dako, Carpinteria, CA, USA), counterstained with hematoxylin, dehydrated, cleared and coverslipped for examination under a microscope. Immunostained cells were counted from 4 randomly chosen fields (200X magnification) per slide per mouse (n=3–4 mice per genotype per treatment). KLF9 expression was quantified by X-Gal (5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside) staining since Klf9 mutant mice were generated by knock-in of the LacZ gene into the open reading frame of Klf9 exon 1 (Morita et al., 2003) and because in-house and commercially available antibodies give high background by immunohistochemistry. Tissue sections from Klf9 heterozygous (+/−) mice were prepared and processed for X-gal staining as described previously (Simmen et al. 2004; Zeng et al. 2008).

Western Blots

Western blot analyses for KLF9 protein levels in Oil (control) and DES-treated WT mouse uteri were performed as previously described (Zhang et al., 2002a). Protein extracts were prepared from TRIzol reagent (Invitrogen) following the manufacturer’s protocol. The protein extracts were dissolved in 1% SDS, diluted 1:10 in water, and quantified using the BCA Protein Assay Kit (Fisher Scientific), with bovine serum albumin diluted in 0.1% SDS as standard. Blots were probed with rabbit anti-KLF9 antibody (1:1000; generated in-house; Zhang et al., 2002a), followed by horseradish peroxidase-conjugated goat anti-rabbit IgG (1:1500; Bio-Rad) secondary antibody. Expression was detected with Amersham ECL Plus reagent (GE Healthcare, Piscataway, NJ, USA) and band intensities were quantified using Quantity One 1-D Analysis Software (version 4.6.5; Bio-Rad).

Radioimmunoassay

Approximately 500 μL of whole blood was collected by closed cardiac puncture from PND84 female Klf9 WT and KO mice. Serum was separated from whole blood by centrifugation at 4,600×g for 1 h and stored at −20°C prior to analysis. Serum E₂ levels were measured using the Ultrasensitive Estradiol Kit (Diagnostic Systems Laboratories, Webster, TX, USA) following the manufacturer’s protocol.

Data analysis

Data (mean±SEM) were analyzed for statistical significance by Two-Way ANOVA followed by All Pairwise Multiple Comparison Procedures (Tukey Test). Where applicable, Student’s t-test was performed to determine the effect of treatment within a particular genotype. All statistical analyses were performed using the SigmaStat software (Systat Software, Chicago, IL, USA) version 3.5. Differences were considered significant at P<0.05.

Results

DES-associated growth effects with Klf9 mutation

WT and KO mice administered either Oil (vehicle) or DES daily beginning at birth for 5 days (PND1-5) were examined for body weights at weaning (PND21) and at sac (PND84). At PND21, KO mice were numerically lighter than WT mice, although the difference did not reach statistical significance. With DES treatment, WT mice were significantly lighter while KO mice were unaffected, relative to respective Oil-treated controls (Fig. 2A). At PND84, WT mice were heavier with DES administration (Fig. 2B), and this was accompanied by a numerical (albeit not significant) increase in gonadal fat pad weight (Fig. 2C). Klf9 mutant mice showed no differences in body or fat pad weights with DES, relative to corresponding Oil group.

Figure 2.

Body and gonadal fat pad weights of control (Oil) and DES-treated Klf9 WT and null mutant mice at postnatal day (PND) 21 (A) and 84 (B and C). Data are mean±SEM. Means with different superscripts differed at P<0.05 by Two-Way ANOVA followed by Tukey Test.

DES-associated alterations in uterine parameters with Klf9 mutation

We have previously shown that KLF9 negatively regulates ESR1 signaling in endometrial glandular epithelial cells (Velarde et al., 2007). To investigate DES effects on the uterus with loss of Klf9 expression, we evaluated a number of uterine parameters (wet weight, LE height, gland size and numbers) which are well-recognized indicators of dysfunctional estrogenic stimulation, in PND84 WT and KO mice. Uterine wet weights relative to body weight did not differ with genotype in control- and DES-treatment groups, with both WT and KO mice showing reduced uterine weights from corresponding Oil control with DES-treatment (Fig. 3A). LE height varied as a function of Klf9 mutation, with Oil-treated KO mice showing reduced LE heights than WT mice, comparable to those of DES-treated WT (Fig. 3B). Uterine gland numbers were decreased with loss of Klf9, and further reductions were observed for both genotypes with DES-treatment (Fig. 3C). Loss of Klf9 expression was associated with a significant increase in uterine gland size (Fig. 3D), and while both genotypes showed increased gland areas with DES-treatment, the significant differences between WT and KO were maintained (Fig. 3D). Serum E₂ levels did not differ with Klf9 status among the Oil-treated groups, and DES suppressed E₂ levels only for WT (Fig. 3E). Ovarian weights normalized to body weight were unaffected by genotype or treatment (Fig. 3F).

Figure 3.

Uterine and ovarian indices of control (Oil) and DES-treated Klf9 WT and null mutant mice at PND84. Data are mean±SEM. Means with different superscripts differed at P<0.05 by Two-Way ANOVA followed by Tukey Test.

Developmental expression of uterine KLF9

Disruption of molecular and structural parameters in the mouse uterus with early DES exposure was lost with ESR1 null mutation, suggesting DES dysregulation of ESR1-dependent events as an underlying cause (Couse et al. 2001). Moreover, our previous studies suggested the existence of KLF9/ESR1 cross-talk under distinct cellular contexts (Velarde et al., 2007; Pabona et al., 2009). We determined whether KLF9 expression was manifested in the uterus during the period of DES exposure to evaluate the possibility of KLF9 contributing to later uterine phenotype with early perturbations of ESR1 signaling by DES. The temporal expression of Klf9 gene was evaluated in uterine cryosections of Klf9 heterozygous mice at PND5, 11, 21, and 28 by X-Gal staining (Fig. 4A). Uterine Klf9 expression was undetectable at PND5 and PND11, and highly visible at PND21 and PND28 (Figure 4A). Consistent with previous findings (Simmen et al. 2004), Klf9 gene was expressed predominantly in endometrial stroma, sporadically in glands and was undetectable in luminal epithelium. QPCR (Fig. 4B) and Western blot (Fig. 4C) analyses showed robust expression of both KLF9 mRNA and protein in uteri of PND84 WT mice, which were unaffected by early DES treatment.

Figure 4.

Temporal expression of uterine Klf9. (A) Uterine X-gal staining demonstrating absence of Klf9 gene expression at PND5 and PND11 and predominantly stromal Klf9 expression (arrow) at PND21 and PND28, using Klf9 (+/−) mice. Images were taken at 100X (upper panels; scale bar, 40 μm) and 400X (lower panels; scale bar, 10 μm). (B) Uterine Klf9 mRNA expression was quantified by QPCR and normalized to 18S mRNA for Oil and DES-treated Klf9 WT mice at PND84. Data (mean±SEM) are expressed as fold change relative to WT Oil treatment group (n=4–8 mice per treatment per genotype). (C) Western blot analyses for KLF9 protein levels in Oil and DES-treated Klf9 WT mice at PND84. Data (mean±SEM; n=2 independent analyses) are expressed relative to loading control, a non-specific (<17 kDa) protein detected in the same blot.

KLF9 regulation of ESR1 expression with DES-treatment

Because the effects of DES, which signals through ESR1, on estrogen-responsive uterine parameters (e.g., LE height, gland number and size) differed with Klf9 genotype at PND84 (Fig. 3), the expression of uterine ESR1 as a function of KLF9 expression status and DES treatment was assessed by immunohistochemistry (Fig. 5A). Immunoreactive ESR1 protein was detectable in all endometrial cellular compartments. Notably, glands showed the highest staining, with most cells in each gland showing strong nuclear immunoreactivity. The numbers of nuclear stromal ESR1 (quantified as % ESR1-immunopositive cells relative to the total cells counted) were similar in Oil-treated WT and KO mice, with a significant increase noted for WT but not KO with early DES exposure (Fig. 5B). In the glandular compartment, nuclear ESR1-immunopositive cells were significantly higher only in DES-treated KO mice; Oil and DES-treated WT and control KO mice showed comparable numbers of nuclear ESR1 immunopositive cells (Fig. 5B).

Figure 5.

Endometrial stromal and glandular epithelial ESR1 expression levels as a function of Klf9 genotype and DES exposure. (A) Representative images of ESR1 immunostaining in stromal and glandular epithelial cells of Oil-treated PND84 Klf9 WT and KO mice. Images are at 400X magnification; scale bar, 10μm. Control (NTC) represents tissue section from Klf9 mutant mice, in the absence of primary antibody. Arrows indicate nuclear staining. PND84 stromal (B) and glandular epithelial (C) ESR1 quantification (mean±SEM) for n=3–4 mice per genotype per treatment (Oil or DES). Means with different superscripts differed at P<0.05 by Two-Way ANOVA followed by Tukey Test.

Altered uterine gene expression in WT and Klf9 KO mice with DES treatment

To further evaluate the contribution of KLF9 to ESR1 signaling, we next analyzed the expression of genes previously shown to be regulated by KLF9 (Simmen et al., 2004) and/or DES treatment (Hong et al. 2006; Newbold et al., 2007) in the mouse uterus. Krüppel-like Factor 13 (KLF13), a member of the Sp/KLF family of transcription factors, exhibits the highest similarity in primary structure to KLF9 (Kaczynski et al. 2003). We previously showed that Klf13 expression is increased in uteri of Klf9 null mice (Simmen et al. 2004). PTEN is a tumor suppressor whose loss of expression is associated with the development of endometrial carcinoma in mice (Daikoku et al., 2008) and is considered a prognostic marker for the disease in humans (Lian et al. 2006). Secretory leukocyte protease inhibitor (SLPI) is a glandular epithelial protein whose increased expression is linked to a highly proliferative uterine state (Zhang et al. 2002b) and Klf9 expression (Simmen et al. 2008). In control (Oil-treated) mice, Klf13 mRNA expression was elevated significantly with Klf9 mutation (Fig. 6A), and DES treatment did not alter this pattern of expression with genotype. Pten mRNA expression was numerically reduced with loss of Klf9 expression in the Oil groups, albeit the decrease was not statistically significant (Fig. 6B). DES-treatment decreased Pten expression (P=0.047) in WT mice and maintained the low levels of Pten in KO mice (Fig. 6B). By contrast, uterine Slpi mRNA expression, while comparable between Oil-treated WT and Klf9 null mice, was increased by DES exposure but only in Klf9 null mice (P=0.028) (Fig. 6C).

Figure 6.

Uterine expression of selected genes as a function of Klf9 genotype and DES exposure. mRNA levels were quantified by QPCR and normalized to GAPDH (for Klf13) or 18S mRNA (for Pten and Slpi) for Oil and DES-treated Klf9 WT and null mutant mice at PND84. Data (mean±SEM) are expressed as fold change relative to WT Oil treatment group (n=4–8 mice per treatment per genotype). Means with different superscripts differed at P<0.05 by Two-Way ANOVA followed by Tukey Test.

To expand the repertoire of genes whose expression may be further disrupted by Klf9 mutation in the background of early DES exposure, we performed additional studies using the Mouse Cancer PathwayFinder^™ PCR Array. Analyses of the uterine genes altered between DES-treated WT and Klf9 mutant mice (from the 84 genes representing six biological pathways involved in transformation and tumorigenesis and present on the array) identified a total of 23 whose expression were either up- or down-regulated by at least 2-fold (data not shown). In order to validate the results from the microarrays, a subset of these up-regulated genes were examined for differential expression by QPCR using cDNAs prepared from uterine RNAs of Oil-and DES-treated WT and KO mice (Fig. 7A). Interestingly, with the exception of Fos, whose transcript levels were only altered with DES for KO mice, the majority of the genes (Akt1, Myc, Tert, Syk) demonstrated expression patterns that were increased with loss of Klf9 expression, independent of DES treatment. Mmp9 and Tgfb1 showed comparable expression patterns, with higher levels in Klf9 KO mice relative to WT and which tended to decrease with early DES exposure for KO but not WT mice (Fig. 7A).

Figure 7.

Differential expression of uterine genes identified from Focused QPCR Array. Uterine mRNA expression at PND84 (A) and PND28 (B) was quantified by QPCR and normalized to 18S mRNA for Oil and DES-treated Klf9 WT and null mice. Data (mean±SEM) are expressed as fold change relative to WT Oil treatment group (n=4–8 mice per treatment per genotype). Means with different superscripts differed at P<0.05 by Two-Way ANOVA followed by Tukey Test. For Myc (A) and Akt1 (B), additional analyses by Students t-test between bracketed groups indicated significance difference (asterisk).

Early effects of KLF9 mutations on gene expression

The gene expression data from the above analyses (Figs. 6 and 7A) identified Akt1, Fos, and Slpi as genes differentially regulated by early DES exposure in the background of Klf9 null mutation at PND84. To evaluate whether these genes may function as early markers of disrupted ESR1 signaling influenced by Klf9 loss of expression, their expression levels were evaluated in uteri of prepubertal (PND28) WT and Klf9 null mice. As shown in Fig. 7B, only the Slpi expression pattern duplicated that found at PND84. While Akt1 transcript levels were affected by Klf9 null mutation similar to that at PND84 (p=0.059 for WT Oil vs KO Oil), Akt1 response to DES exposure was different at PND28 from that at PND84. Fos expression at PND28 remained unchanged with Klf9 genotype and DES treatment, which contrasted with the findings for PND84 uteri (Fig. 7A).

In silico analysis of selected genes in endocarcinoma

To correlate the changes in gene expression identified in the present study with previously reported alterations in gene expression for endometrial tumors (Mutter et al., 2000), an online cancer data-mining resource (www.oncomine.org) was used as previously described (Simmen et al., 2009). Table 2 lists those genes that were commonly altered between the findings reported here and the published data set (Mutter et al., 2001). The changes in expression of KLF9, AKT1, MMP9, and SLPI but not TGFβ1, were in the same direction for endometrial tumors as that found in uterine tissues of Klf9 null relative to WT mice. By contrast, the uterine expression of MYC, SYK, and TERT genes, which were altered in Klf9 null relative to WT mice, did not differ in endometrial tumors from corresponding non-diseased uteri (Mutter et al. 2001).

Table 2.

A subset of differentially expressed genes in human endocarcinoma associated with KLF9 expression status¹

Gene	Endocarcinoma2	P-value3
AKT1	+	.00018
KLF9	−	.035
MMP9	+	.00019
SLPI	+	.08
TGFB1	−	.014
MYC	NC	.408
SYK	NC	.548
TERT	NC	.904

¹In silico analysis of Mutter et al. (2001)

²(+), increased with carcinoma; (−); decreased with carcinoma; NC, no change with carcinoma

³P-value relative to normal endometrium

Discussion

In this study, we established a central role for KLF9 in the response of uterine endometrial cells to early perturbations in ESR1 signaling. Using the well-characterized DES-induced model for dysregulation of ESR1 function leading to pathology (Newbold et al. 1990, 2007), we showed that loss of Klf9 expression alone altered uterine structural parameters (LE height, gland numbers, gland size) comparable to those in DES-treated WT mice and increased endometrial gland sensitivity to DES. We demonstrated that Klf9 null mutation enhanced glandular ESR1 expression with DES, and this was positively associated with increased expression of Slpi, a gene encoding an endometrial glandular secreted protein whose over-expression is implicated in the pathogenesis of many cancers, including that of the endometrium, due to its pro-proliferative actions (Zhang et al. 2002b; Chen et al. 2004). Finally, we demonstrated that unlike ESR1, whose uterine expression was found to be an early target of DES, occurring within days of DES administration (Sato et al. 1992) and which remained dysregulated later in life, KLF9 expression was undetectable in the developing uterus (PND5, PND11) and was unaffected by DES later in life. Our results suggest that a primary site of KLF9 action is the endometrial glandular compartment where Klf9 ablation promoted long-term DES effects on ESR1 and Slpi expression and on glandular morphology (gland size and number). Given that endometrial cancer originates from endometrial glands (Di Cristofano & Ellenson 2007) and consistent with our previous report of increased ESR1 expression with estradiol-17β treatment upon siKLF9 knockdown in Ishikawa (glandular epithelial) cells (Velarde et al. 2008), we propose that KLF9 constitutes an important regulator of ESR1 signaling in endometrial glands and that loss of this regulation may predispose to increased risk for endometrial cancers.

Our findings that early (postnatal) DES exposure altered adult uterine phenotype (PND84) are consistent with previously reported long-term consequences of DES-induced early developmental perturbations for this tissue in mice (Couse et al. 2001). The precise mechanism by which early disruption of ESR1 signaling leads to abnormalities in uterine function and phenotype as adults, remains poorly understood. Epigenetic changes involving modifications to chromatin structure due to aberrant histone methylation, histone acetylation, and DNA methylation, are increasingly considered to underlie the onset and progression of cancers (Ellis et al. 2009). Consistent with this, neonatal DES exposure has been recently reported to alter genomic DNA methylation due to changes in expression of uterine DNA methyltransferases (Sato et al., 2009). Our data showing the lack of co-expression of KLF9 with ESR1 (Yamashita et al. 1989; Sato et al. 1992; Newbold 1995) in uteri of mice in the first two weeks of life suggest that KLF9 mediates glandular ESR1 expression in adult uteri subsequent to and/or independent of early events arising from epigenetic modifications of ESR1. KLF9 may alter nuclear ESR1 expression through effects on ESR1 stability, nuclear localization, and/or promoter activity in the adult uteri, the latter consistent with our previous findings that reduced KLF9 expression prevented the recruitment of ESR1 to its own promoter (Velarde et al. 2008) to elicit ESR1 negative autoregulation. While our data does not address the stage in uterine development wherein loss of Klf9 expression on ESR1 signaling is manifested, we suggest that this likely occurs as early as PND28 based on the comparable patterns of uterine Slpi expression in PND28 and PND84 DES-exposed Klf9 null mice, when compared to corresponding-aged WT mice, and given that Slpi is an ESR1-induced gene (Chen et al. 2004). Further studies will address ESR1 expression in Klf9 null uteri of younger (prepubertal) mice. Interestingly, while our previous study of the human endometrial cancer cell line HEC-1A, which lack detectable ESR1 expression, demonstrated induction of SLPI expression with KLF9 over-expression (Simmen et al. 2008), results presented here indicate negative regulation of Slpi expression by KLF9 in ESR1-expressing cells with perturbed ESR1 signaling. Taken together, our data suggest that while the contribution of epigenetic mechanisms to dysregulated ESR1 expression with DES exposure (Li et al. 2003; Bromer et al. 2009) cannot be excluded, KLF9 involvement in the regulation of ESR1 and Slp1 expression is most likely a consequence of its action at the level of the ESR1 promoter.

To determine if changes in estrogen-responsive parameters were attributable to altered ovarian function, we measured both serum E₂ and ovarian weights of adult WT and Klf9 null mice with and without early DES exposure. The comparable E₂ levels in both genotypes as well as the lack of significant differences in ovarian wet weights as a function of genotype and DES administration support our previous reports of the lack of major KLF9 effects on the ovary (Simmen et al. 2004; Zeng et al. 2008). Thus, ESR1/KLF9-dependent effects noted in glandular epithelial cells is largely ovarian-independent and likely a consequence of direct uterine targeting by KLF9.

We used a combination of candidate gene and focused gene array approaches to identify genes whose expression was differentially altered in adult (PND84) WT and Klf9 mutant uteri, based on their relationships to KLF9, ESR1, and/or cancer risk. Of the uterine genes confirmed to be regulated with loss of Klf9 expression, alone or as a function of DES exposure, two major categories are proposed: one category is comprised of genes that are regulated by KLF9, independent of ESR1 (DES) signaling (Klf13, Myc, Tert, Syk, Akt1), while the second category includes genes that are influenced by KLF9/ESR1 cross-regulation (Fos, Slp1). The increased expression of pro-proliferative/cell survival, ESR1-regulated genes Akt1 (Altomare & Testa 2005; Vilfelm et al. 2006), Myc (Cole 2000), Syk (Coopman et al. 2000), and Tert (Boggess et al., 2006) with loss of Klf9 expression is consistent with the potential tumor suppressor function of KLF9, as suggested by the reported loss of KLF9 expression in endometrial tumors (Mutter et al. 2001; Simmen et al. 2008). Moreover, the coincident elevated expression of Myc and Tert mRNAs is consistent with the observation by Wu and colleagues (1999) that MYC upregulates Tert promoter activity. Fos and Slpi are estrogen-regulated genes, whose gene products exhibit pro-proliferative activities (Zhang et al. 2002b; Li et al. 2003; Chen et al. 2004), hence, their increased expression with Klf9 mutation coincident with increased ESR1 expression (with DES exposure) suggests a pathway linking the loss of KLF9-mediated ESR1 negative regulation to enhanced proliferation and provide further support to potential KLF9 tumor suppressor activity. The increased expression of Klf13 with Klf9 null mutation implicates a regulatory loop between these highly-related family members, as suggested from our previous studies (Simmen et al. 2004). Pten mutations are among the earliest identifiable aberrations in endometrial carcinoma (Lian et al. 2006); hence, our findings of a robust decrease in Pten transcript levels in DES-exposed WT uteri relative to control WT are consistent with increased incidence of endometrial carcinoma in peri-natal DES-exposed mice (Newbold et al. 2006, 2007). While we demonstrated only a trend for decreased uterine Pten expression in Klf9 null mutants without and with DES exposure relative to WT mice, Pten transcript levels for Klf9 null mutants without DES exposure were numerically comparable to those of DES-exposed WT mice, suggesting that dysregulation of Pten expression by early DES exposure is partially mimicked by Klf9 null mutation. In this regard, recent data showing Slpi upregulation in the uteri of Pten heterozygous mice which develop atypical hyperplasias and carcinomas in situ (Lian et al. 2006) and that conditional deletion of endometrial Pten resulted in the rapid development of endometrial carcinoma in mice (Daikoku et al. 2008) are consistent with our predicted linkages between expression of Klf9, Slpi and Pten and uterine tumorigenesis. Studies evaluating uterine tumor development in Klf9 null mice with and without postnatal DES exposure will be critical to addressing the link between KLF9 and uterine endocarcinoma and are ongoing in our laboratory.

In summary, our data suggest a novel role for the transcription factor KLF9 in modulating uterine responses to ESR1-dependent DES actions, the latter leading to a dysfunctional uterus. Our in vivo model superimposing Klf9 null mutation with early (postnatal) DES exposure has revealed a molecular network between KLF9 and ESR1 operative in the uterus involving key estrogen-regulated uterine morphological and molecular indices that may constitute early predictors of uterine tumor development. While direct evidence for KLF9 as a tumor suppressor awaits further studies, our results raise the intriguing possibility that the silencing of KLF9 either by genetic and/or epigenetic mechanisms, as demonstrated for many tumor suppressors and oncogenes (Di Cristofano & Ellenson, 2007), may contribute to increased susceptibility to developing endometrial cancers initiated by aberrant estrogen action.