The role of genetics in estrogen responses: a critical piece of an intricate puzzle

Abstract

The estrogens are female sex hormones that are involved in a variety of physiological processes, including reproductive development and function, wound healing, and bone growth. They are mainly known for their roles in reproductive tissues—specifically, 17β-estradiol (E₂), the primary estrogen, which is secreted by the ovaries and induces cellular proliferation and growth of the uterus and mammary glands. In addition to the role of estrogens in promoting tissue growth and development during normal physiological states, they have a well-established role in determining susceptibility to disease, particularly cancer, in reproductive tissues. The responsiveness of various tissues to estrogen is genetically controlled, with marked quantitative variation observed across multiple species, including humans. This variation presents both researchers and clinicians with a veritable physiological puzzle, the pieces of which—many of them unknown—are complex and difficult to fit together. Although genetics is known to play a major role in determining sensitivity to estrogens, there are other factors, including parent of origin and the maternal environment, that are intimately linked to heritable phenotypes but do not represent genotype, per se. The objectives of this review article were to summarize the current knowledge of the role of genotype, and uterine and neonatal environments, in phenotypic variation in the response to estrogens; to discuss recent findings and the potential mechanisms involved; and to highlight exciting research opportunities for the future.—Wall, E. H., Hewitt, S. C., Case, L. K, Lin, C.-Y., Korach, K. S., Teuscher, C. The role of genetics in estrogen responses: a critical piece of an intricate puzzle.

Estrogens are now known to have important roles in many tissues, including adipose tissue, bone, brain, skeletal muscle, skin, and vasculature (1). Therefore, any processes that alter the sensitivity to estrogens, many of which are dictated by genetic factors, can lead to marked variation in the physiology of many tissues during both normal development and disease. An improved understanding of the factors involved could present opportunities for optimizing the development and function of various tissues (i.e., fertility, lactation, and skeletal development) and also could help to prevent and treat diseases.

UTERUS AND VAGINA

The physiological response of the uterus to 17β-estradiol (E₂), which has been well characterized, consists of early and late phases. The early rapid phase, which occurs within 4 h of E₂ stimulation, is characterized by changes in gene transcription, a marked increase in vascular permeability, and water imbibition (2, 3). The late phase, which occurs 18 to 30 h after E₂ stimulus, is characterized by an influx of leukocytes into the uterine stroma, changes in transcription of late-phase genes, and an increase in epithelial cell proliferation and differentiation (3, 4). Together, these early and late phases comprise the classic uterotropic response to E₂ (3).

It has long been established that genetics plays a role in the responsiveness of the uterus to estrogens. Early studies demonstrated that the uterotropic response to E₂ is genetically controlled, with marked variation in tissue growth or regression observed, depending on the strain of mouse studied (5). Similar observations have been made for the vagina, with a wide range of epithelial responses observed across mouse strains (6, 7). More recently, research has shown that the infiltration of leukocytes, particularly eosinophils, into the uterine stroma is also genetically controlled (8).

To determine the degree to which genotype influences uterine responsiveness to E₂ across several strains of mice, uterine peroxidase activity was used as a quantitative trait variable, a measurable phenotype regulated by multiple genes, for assessing phenotypic variation in the immature ovariectomized (OVX) mouse uterotropic assay (9). Among the 6 different inbred strains studied, there was a marked effect of strain as well as a continuous distribution of uterine peroxidase activity (Fig. 1), indicative of polygenic inheritance (10, 11). More specifically, the observed phenotype is the result of the interaction between many genes. This assay showed C57BL6/J (B6) to be a high responder to E₂, whereas C3H/HeJ (C3H) was shown to be the lowest. Results obtained with immature mice were confirmed in adult OVX mice (9).

Figure 1.

Genetic control of uterine growth in response to E₂. The increase in uterine peroxidase activity in sexually immature inbred strains of mice at 24 h after 3 daily injections of E₂. (See ref. 9 for additional information).

Genetic mapping experiments using inbred strains of mice that are high responders (B6) or low responders (C3H) to E₂ have led to the identification of quantitative trait loci (QTLs), which are stretches of DNA containing or linked to the genes that underlie a quantitative trait (12), controlling quantitative variation in uterine growth and eosinophil infiltration (8, 13). Specifically, E₂-induced uterine growth is determined by QTLs on chromosomes 5 (Estq2) and 11 (Estq3), whereas the number of infiltrating eosinophils is controlled by QTLs on chromosomes 4 (Estq1) and 10 (Estq4) and an interactor on chromosome 16 influencing both traits. Interestingly, none of the identified QTLs has been associated with estrogen receptors (ERs) or any known estrogen signaling pathways.

To understand the strain differences in uterine E₂ sensitivity at the cellular level, we identified proliferating and apoptotic epithelial cells in uterine tissue 72 h after E₂ treatment and found that in all strains, treatment with E₂ elicited a marked increase in cell proliferation (9). However, in C3H and B6 × C3H F₁ (B6C3) hybrid mice, the uterine epithelia were undergoing apoptosis, whereas no apoptosis was detected in the B6 mice (ref. 9 and Fig. 2A, B). Both induction of cellular proliferation and prevention of apoptosis are necessary for a complete uterine epithelial response to E₂ (14). These results indicate that ability to limit apoptosis immediately after the proliferative response to E₂ is impaired in C3H and B6C3 mice, consistent with the low-responder phenotype.

Figure 2.

Apoptosis in uterine epithelium 72 h after treatment with E₂. A) Localization of TUNEL-positive cells (brown, TUNEL-positive; blue, hematoxylin). B) Quantity of CC3 expression in uterine epithelia of B6, C3H, and B6C3 mice. (See ref. 9 for additional information).

In agreement with previous observations on genetic control of the response of the vagina to estrogens (6, 7), ongoing experiments have revealed that the degree of vaginal engorgement following treatment with E₂ is also genetically controlled. Interestingly, the response to E₂ by the vagina appears to be the opposite of that of the uterus: the C3H strain is the high responder (high level of vaginal engorgement), and the B6 strain is the low responder (Fig. 3A, B).

Figure 3.

Genetic control of the change in vaginal area in response to E₂. A) Images depicting vaginal area of vehicle- and E₂-treated B6 and C3H mice. B) Quantification of vaginal area 10 d after vehicle or E₂ treatment. Letters above bars indicate differences in the vaginal area between strains; P < 0.05. There was also a strain-by-treatment interaction; P < 0.01 (unpublished results).

MAMMARY GLAND

In the mammary gland, E₂ promotes the proliferation of epithelial cells, as well as the branching and elongation of mammary ducts (15). Like the uterus (5), phenotypic variation in the responsiveness of the adult mammary gland to E₂, during both normal and pathological conditions, including tumor formation, was reported long ago (6, 16, 17). In addition, genetic control of mammary development (18), response to hormones (19), and ductal morphology (20, 27) have been observed. Blair et al. (17) observed marked variation in mammary alveolar development induced by treatment with E₂ and progesterone in several strains of male mice, including B6 and C3H. Many of the reports investigating the genetic control of the mammary response to E₂ were published before 1999 (6, 16, 17), prior to the eradication of mouse mammary tumor virus (MMTV) in inbred mice (Peter Kelmenson, The Jackson Laboratory, Bar Harbor, ME, USA, personal communication, June 4, 2013). MMTV is a β-retrovirus that is transmitted through the milk of lactating female mice to their offspring. The major cellular targets for MMTV are mammary epithelial cells, and infection can lead to the formation of mammary tumors during adulthood in susceptible strains of mice. Therefore, many of the studies investigating the strain differences in the response of the mammary gland to E₂ that were published before 1999 were either confounded by the presence of MMTV or were designed to investigate the mammary response to E₂ in the context of MMTV (7, 16, 17, 22–26). More recent reports have confirmed that, even in the absence of MMTV, the response of the mammary gland to E₂ and progesterone (19) and to environmental estrogens (27, 28) is genetically controlled.

In rats, extensive genetic mapping experiments have been conducted to determine the QTLs and individual genes that control the sensitivity of the mammary gland to E₂. Loci have been identified that regulate mammary cell proliferation and differentiation (29), mammary immune function (30), and susceptibility to E₂-induced mammary cancer (31, 32). Some of these QTLs have been shown to have relevance in humans, making them interesting targets for prediction and prevention of human mammary disease (29, 33). Interestingly, genetic mapping experiments have revealed that the genetic determinants of estrogen sensitivity are tissue specific, both in regard to the genes involved and the associated phenotype (34).

Paradoxically, mouse strains observed to be high uterine responders to E₂ are often low mammary gland responders (6, 17). To determine whether B6 and C3H mice also possess this inverse relationship, we conducted an experiment comparing the uterine vs. mammary responses to E₂. Our observations revealed that the mammary response to E₂, like the uterotropic response, is also genetically controlled (35). Moreover, in agreement with previous reports on other strains of mice (6, 17), we observed that the uterine response to E₂ correlates negatively with the E₂-induced increase in mammary ductal length (35). Specifically, B6 mice are high uterine responders and low responders for mammary ductal growth, whereas the reverse is true of the C3H strain (Fig. 4). In contrast, B6 was a high responder for mammary ductal side branching (Fig. 4). Similar inverse relationships in E₂ sensitivity have been observed in other species. For example, the ACI rat is susceptible to E₂-induced mammary and pituitary tumors, but it is resistant to E₂-induced uterine infection (31, 36). Interestingly, we observed a positive correlation between the E₂-induced increase in uterine weight and ductal side branching (35). Although the functional implications of this relationship are unclear, it warrants further investigation, since genetically controlled differences in the effect of E₂ on mammary morphology could influence both the function of the mammary gland and susceptibility to disease. It is plausible that the E₂-induced signaling cascades responsible for the uterotropic response are the same ones responsible for E₂-induced mammary ductal side branching, but distinct from those associated with E₂-induced mammary ductal growth. The existence of such shared or distinct signaling mechanisms, however, remains to be established.

Figure 4.

Genetic control of mammary ductal growth and side branching in response to E₂. Mammary ductal length and side branching were measured 3 d after 2 daily injections of E₂. Letters above bars indicate differences in the ductal length or total branches between strains; P < 0.05. (See ref. 35 for additional information).

In agreement with our observation that there are distinctions in the response of mammary tissue to E₂ in C3H vs. B6 mice, Singh et al. (16) reported that relative to C3H mice, B6 mice require longer hormonal priming in vivo for mammary tissue to respond to hormones in vitro. The factors underlying strain-specific mammary responsiveness to E₂ are unclear; however, it has been suggested that interactions between the tissue and the host environment dictate hormone responsiveness (20, 21). As was described for the uterus, the cellular bases for differences in mammary responsiveness to E₂ appear to involve differences in E₂-induced cell proliferation and apoptosis. Our recent experiments comparing B6 and C3H mice have revealed an increase in both mammary epithelial cell apoptosis and cell proliferation in E₂-treated B6 mice compared to C3H (ref. 35 and Fig. 5). Therefore, E₂-treated B6 mice appear to be undergoing more extensive mammary remodeling or mammary epithelial cell turnover, and this may be the reason for the differences in mammary ductal morphology since, in mammary gland, morphogenesis involves both cell proliferation and apoptosis (15).

Figure 5.

Proliferation and apoptosis in mammary epithelium in response to estrogen. Quantification of Ki-67 and CC3 in mammary epithelia of B6 and C3H mice 72 h after treatment with E₂. Letters above the bars indicate differences in the Ki-67- or CC3-positive cells between strains; P < 0.05. (See ref. 35 for additional information).

OTHER TISSUES

In additional to the genetic control of the response of reproductive tissues to E₂, quantitative variations in the responsiveness of a variety of other tissues to E₂, including bone and skin, have been observed (5, 17, 37–40). Understanding the genetic contribution to this variation is of particular interest because both bone loss and delayed wound healing are significant problems for postmenopausal women and for the elderly in both sexes.

Bone

The role of genetics in determining the response of bone to estrogens has been extensively studied due to the known decrease in bone mineral density that occurs in postmenopausal women (41). During and after menopause, there is a gradual decline in estrogen production by the ovaries that leads to an increase in bone turnover, decreased bone mineral density, and increased risk of bone fracture (41). Similarly, in aging men, a decrease in bioavailable estrogen is a major predictor of bone loss (41). Notably, however, there is a large amount of variation in the rate and extent of bone loss, such that some individuals are much more susceptible than others (42). Family and twin studies have revealed that genetics factors explain a large portion of this variance, and experiments using female OVX mice have confirmed these observations. Specifically, it has been reported that the magnitude, timing, and location of bone loss after ovariectomy are genetically determined (39, 40). In addition, it has been observed that genetic background contributes to variations in the rate and extent of fracture healing in inbred mice (43). These studies lay the groundwork for future experiments designed to identify the specific gene families that control the response of bone to estrogen. Once identified, such genes have the potential to be used as therapeutic targets during both menopause- and age-associated changes in bone density.

Skin

Estrogens are known to play a role in the physiology of the skin, and together with insulin-like growth factor-I, estrogens prevent inflammation and stimulate wound healing (44). Similar to bone, the structure and integrity of the skin is known to decrease after menopause and with advanced aging, and this deterioration can be partly prevented by estrogen replacement therapy (44). Although we are unaware of any studies that directly assessed the contribution of genetic background on the response of skin to estrogens, there are at least two reports on the variation in wound healing in different strains of mice. In one study, genetic background had a striking effect on wound healing in mice, such that MRL/MpJ mice underwent more extensive wound healing and tissue regeneration than did B6 and BALB/c mice (45). More recently, these observations were confirmed, and it was also revealed that genetic background controls not just the degree to which wound healing occurs, but also the nature of wound healing (regeneration of the tissue vs. wound closure and scar formation; ref. 46). Because delayed wound healing is a very common issue for the elderly and for many postmenopausal women, understanding the specific genetic factors that regulate skin physiology could provide opportunities for preventing skin damage and accelerating wound healing after injury.

Adipose tissue

Estrogens are known to play an integral role in the metabolic function of many tissues, including adipose tissue. The activation of ERs by E₂ suppresses lipogenesis, enhances energy expenditure, and improves insulin sensitivity (47). In both males and females, the role of estrogens in energy homeostasis becomes critical during the aging process, as secretion of gonad hormones decreases, resulting in increased risk of obesity and type 2 diabetes (47). It is well established that susceptibility to obesity and type 2 diabetes is genetically controlled (48, 49); therefore, an understanding of the gene by estrogen interactions that regulate the disruption of energy homeostasis is critical for both prevention and treatment of metabolic diseases.

MOLECULAR BASIS FOR DIFFERENCES IN ESTROGEN SENSITIVITY

The observed genetically linked differences in estrogen responses logically lead to questions of what differs between the mouse strains that accounts for these strain-sensitivity outcomes. Understanding the mechanisms governing estrogen-regulated responses as they currently stand may indicate features that underlie strain differences.

Mechanism of ER–mediated response

Estrogen elicits responses in tissues via the ER, a nuclear receptor family member (50). There are two ERs: ERα, which is expressed in female reproductive tissues, including the uterus and mammary glands, and ERβ, predominantly found in ovarian granulosa cells (51, 52). Isolation and cloning of the ERα protein and cDNA revealed aspects of its molecular functions, as ERα is composed of 6 domains (A–F), each encoding regions needed for the various functions of ERα (Fig. 6). These include the DNA-binding domain (DBD; C domain), which specifically interacts with estrogen responsive element (ERE) DNA motifs (GGTCAnnnTGACC), thus targeting ERα to responsive genes (51–54). ERα also binds with high affinity and specificity to natural and synthetic estrogens via its ligand-binding domain (LBD; E domain). These two key functions govern the specificity of estrogen regulation, including activation by hormones and interaction with chromatin to modulate transcription of estrogen target genes. The classic paradigm of ER-mediated responses involves the diffusion of lipophilic estrogen into cells and interaction with nuclear localized ER with high affinity (Kd<1 nM for E₂), leading to activation of transcription focused at target genes via specific interaction with ERE. ER can also tether, in a nonclassic manner, to other transcription factors such as AP-1, Sp1, and NF-κB, to indirectly influence gene expression (55–57). Two mouse models have been developed that impair ERα–ERE interaction by mutating the DBD to restrict ERα-mediated functions to tethered signaling. Although the first (Nerki or KIKO) model seemed to indicate that the tethered signaling that remained in this mouse model mediated uterine responses (58, 59), the second (EAAE) model had phenotypes that resembled ERα-null models (60, 61). The differences in the two models were demonstrated to result from remaining DNA binding activity of the Nerki/KIKO ERα mutant (61). Therefore, it appears that in the absence of ERE binding activity, nonclassic signaling is not sufficient to support estrogen responses. In addition to its nuclear functions, ER also exerts rapid nongenomic effects through interactions with cell membrane–associated growth factor receptors and components of signal transduction cascades in the cytoplasm (61). Recently, 2 mouse models that have mutations in ERα that prevent membrane responses have been developed (63, 64). The resulting phenotypes differ, but indicate roles for this mechanism.

Figure 6.

ER structure and molecular mechanism. A) Domain structure of ER. From the amino terminus (N) to the carboxy terminus (C), the ER contains 6 domains: A, B, C, D, E, and F. The A/B domain contains one of the transcriptional AFs (AF1). The C domain contains the DBD. Domain E binds to estrogen ligands, with critical helix 12 (H12) mediating interactions with transcriptional coregulators. This ligand-dependent activity defines AF2. B) Mechanism of ER-mediated transcriptional regulation. Access of ER to ERE motifs is facilitated by pioneer factors such as FoxA1. E₂ binds to the ER, which leads to recruitment of coactivators and chromatin remodelers. Increased histone acetylation makes chromatin more accessible and allows RNA polymerase II (RNA Pol II) to access target gene transcription start sites (TSS).

The transcriptional activity of ER is modulated by two transcriptional activation functions (AFs), one in the A/B domain (AF1), described as hormone independent, and the second in the LBD (AF2), shown to depend on estrogen binding (52, 54). These AF domains orchestrate interactions with transcription factors, especially members of the steroid receptor coactivator (SRC) family (SRC1, -2, and -3; ref. 65). To interact with EREs and regulate transcription, the ER must reach DNA that is packaged into chromatin. It has become clear that the ER acts as a nexus on which the “machinery” that enables chromatin remodeling and modulation of transcription is assembled in a manner that is induced by estrogen (66, 67). The structure of the LBD is composed of a 3-layer α-helical sandwich to create a hydrophobic ligand binding pocket. Before binding estrogen, the ER is conformed such that interaction with coactivator molecules is not favored. Binding of estrogen leads to rearrangement of the α helices, creating an interface for SRC interaction with helix 12 of the LBD (68). SRC functions together with massive multimeric complexes that facilitate access to chromatin. The activities of the complex dynamically coordinate the steps of transcription, including initiation, elongation, termination, and clearing or turnover of the transcriptional modulators. Regulation of access to chromatin is facilitated or impeded in part by modification of histones, the proteins that package DNA. Acetylation of histone tails is mediated by histone acetyl transferases (HATs), which increase access, and deacetylation by histone deacetylases (HDACs), which increase chromatin compaction. SRC itself has HAT activity, and ER–SRC interacts with other HATs and HDACs (66, 69, 70).

Chromatin immunoprecipitation (ChIP) combined with microarray (ChIP-on-chip) or next-generation sequencing (ChIP-seq) technologies allowed for comprehensive analysis of all the sites of interaction between ERα and DNA in a cell or tissue. These types of studies have indicated that most ERα binding occurs not at gene promoters, but often in regions >10⁵ bp distal from transcribed genes (71, 72). It was through these genome-wide ChIP studies that the role of FoxA1 as a pioneer factor, facilitating ERα access to chromatin in breast cancer cells, was discovered (73–75). Although estrogen greatly increases ERα recruitment to chromatin and is needed to regulate transcription of target genes, ERα ChIP-seq analysis of mouse uterine tissue revealed that some ERα is prebound to DNA in the absence of estrogen, with the number of sites per gene increased with hormone treatment (76).

Understanding the complexity of the mechanism of estrogen response in tissues highlights the many aspects involved in estrogen signaling that could underlie strain-specific differences. For example, studies with SRC1-null mice indicate decreased hormone sensitivity (77, 78). Thus, strain-specific differences in SRC levels may affect estrogen response. Similarly, strain-specific variations in other cofactors and chromatin remodelers may affect estrogen responses. Studies of mice engineered to express a mutated form of ERα and impaired in their AF1 region or mutated in the AF2 region indicate that loss of AF1-mediated ERα signaling results in decreased uterine response to estrogen (79, 80). Thus, strain differences in the AF1 region of the ERα or in molecules that interact with AF1 may affect estrogen responses as well.

Transcriptional profiles underlying uterine estrogen response

Each phase of the uterotropic response is associated with distinct transcriptional signatures, implicating unique sets of differentially expressed genes in each of the physiological effects of E₂ (81). Given that the uterotropic response is genetically controlled (8, 13), we hypothesized that the transcriptional response of the uterus to E₂ is similarly genetically controlled. Eight-week-old female ovariectomized B6, C3H, and B6C3 F₁ hybrid mice were treated with exogenous E₂, and the uteri were collected at 2 and 24 h after treatment. Microarray analysis revealed the existence of both common and distinct E₂-regulated transcriptional signatures at both time points (9). Pathway analysis of strain-specific E₂-responsive genes showed that cell death pathways were significantly enhanced in C3H but not B6 mice at 2 and 24 h after treatment with E_2. Consistent with our observations of increased uterine epithelial cell apoptosis in C3H mice (Fig. 2), we found that expression of cleaved caspase-3 was increased in C3H mice, and estrogen was significantly less effective at inducing the apoptosis inhibitor Birc1a (Naip) in C3H than in B6 (9). Uterine epithelial cell expression of estrogen receptor 1 (ESR-1), which is dispensable for eliciting an E₂-regulated uterine proliferative responsiveness but critical for inhibition of apoptosis (82), was not different across the strains (9).

As was seen for the uterus, observations in mammary tissue revealed that the differences in the response to E₂ in each strain are not the result of differences in the activation or inhibition of a common set of regulatory molecules (35). Rather, these findings suggest that E₂ elicited the activation or inhibition of strain-specific regulatory molecules that may contribute to the observed differences in mammary ductal length and side branching between the strains. For example, signal transducer and activation of transcription 3 (STAT-3) mRNA, which is known to be associated with mammary involution and remodeling (83), was induced in B6 mice coincident only with an increase in mammary epithelial cell apoptosis (35). In contrast, epidermal growth factor RNA, which is known to be a promoter of mammary ductal growth (84), was selectively increased in C3H mice (35), consistent with increased E₂-induced ductal length compared with that of B6. In agreement with the observation of increased mammary epithelial cell proliferation and ductal side branching in B6 vs. C3H mice, transforming growth factor β-3 transcript, which has a known role in mammary ductal branching morphogenesis (85), was selectively up-regulated in B6 mice (35).

Notably, the findings of our microarray experiments in both uterine and mammary tissue have revealed that although overall E₂ elicits very similar transcriptional profiles in both strains, for the transcripts that exhibit strain-selective activity, the quantitative differences in uterine responsiveness to E₂ are not simply due to differences in the magnitude of expression of similar genes. Rather, E₂ also induces distinct transcriptional programs in each of the strains. To gain insight into the differences that exist between the strains before E₂ treatment, we combined microarray analysis on uterine tissue from untreated mice with our QTL mapping data (8, 13) and determined that, among the genes differentially expressed in untreated mice, 84 reside within our previously identified QTLs (9). In addition, the genes exhibited different inheritance patterns; in some cases, B6C3 segregated with B6 and in other cases segregated with C3H. We propose that these genes, because they are differentially expressed in the strains at baseline, are positional candidates that may control the observed differences in uterine responsiveness to E₂. Pathway analysis (Ingenuity Systems, Redwood, CA, USA) revealed that most enriched functions associated with the 84 genes include Runx1. The Runx-related family of transcription factors comprises at least 3 transcriptional regulators known to be involved in several cellular processes, including cell proliferation and differentiation (86), and a role for Runx1 has been suggested in the development of endometrial cancer (87, 88). We identified Runx1 as a positional candidate on chromosome 16 that epistatically interacts with other QTLs that control uterine responsiveness to E₂ (8, 13). Interestingly, we observed that uterine expression of Runx1 mRNA was increased 2 h after E₂ treatment, and uterine expression of Runx1 was greater in B6 compared to C3H mice (9). In contrast, mammary expression of Runx1 was greater in C3H vs. B6 mice (35); therefore, the genetic control of Runx1 expression is reciprocal across uterine tissues from the 2 strains. This reciprocal relationship has clear functional relevance: expression of Runx1 correlated positively with both the E₂-induced increase in uterine weight (r=+0.97; P<0.0001) and the increase in mammary ductal length (r=+0.40; P<0.04), whereas it correlated negatively with E₂-inuced ductal side branching (r=−0.63; P<0.001) (35).

It has been proposed that Runx1 is a potentiator of E₂-induced nonclassic ESR-1 signaling (independent of EREs) in the uterus by acting as a tethering factor (89). Interestingly, however, the reverse has been observed for E₂-induced cellular proliferation in the mammary gland, where Runx1 has been proposed to be an antagonist of E₂ and a tumor-suppressor gene (90). Because these observations support a positive relationship between Runx1 expression and E₂-induced mammary ductal growth, and a negative relationship between Runx1 expression and E₂-induced ductal side branching, the role of Runx1 during mammary ductal growth remains unclear. Confirmation or exclusion of Runx1 as the gene controlling this response, or a shared gene controlling uterine and mammary gland responsiveness to E₂, necessitates a physical mapping–based forward genetic approach in congenic strains of mice.

There are clearly many potential areas, both at the transcriptional and protein level, wherein even slight changes in mRNA or protein expression, as well as the expression of various gene isoforms, could be linked to genetically controlled differences in E₂ responses. Research in this area has revealed the complexity of E₂ signaling in the context of strain-specific phenotypes, and it is anticipated that additional studies will continue to identify factors underlying genetic variation in E₂ responses at the molecular level.

ROLE OF ENVIRONMENT IN DETERMINING SENSITIVITY TO ESTROGENS

There are factors that are intimately linked to heritable phenotypes for estrogen responses but do not represent genotype, per se. The most noteworthy of these factors are the uterine and maternal and neonatal environments and the exposure to environmental estrogens during critical periods of development.

Uterine environment

The intrauterine environment is known to influence the subsequent responsiveness of adult tissues to estrogens via intrauterine position (IUP). The IUP effects are caused by differences in steroid hormone transfer between embryos as a consequence of their positioning in the uterine horns (91, 92). In litter-bearing species, fetuses are organized side by side in the uterine horns (91), and those not located at either end of the horns will be positioned between 2 females (0M), 2 males (2M), or 1 female and 1 male (1M) (91). Extensive IUP studies in mice have demonstrated profound effects of androgen exposure on developing female fetuses, influencing numerous phenotypes such as adult social behavior (93, 94), reproductive organ development (95, 96), and susceptibility to endocrine disruption (97). Notably, the degree to which androgen exposure influences the masculinization of females is genetically controlled as evidenced by the lack of an IUP effect noted for certain strains of inbred mice (98).

Although most IUP studies have been conducted in rodent models, the effects of the intrauterine environment have been described in non-litter-bearing species capable of carrying >1 fetus, including cows, sheep, goats, and humans (99). Interestingly, in cattle carrying opposite-sex dizygotic twins, the female is almost always born as a freemartin, with a female appearance but a chimeric genotype (carrying both XX and XY chromosomes) and no functioning ovaries (100, 101). The freemartin phenotype is the result of endocrine disruption, mainly associated with very high levels of anti-Müllerian hormone in the female owing to the presence of the male in the uterus (102). Opposite-sex dizygotic twin studies have also provided compelling evidence of IUP's masculinizing effects in humans. Females with male twins tend to display an increase in masculine traits, such as sensation seeking (103), masculinized left hand finger-length ratio (104), aggression level (105), and brain size (106).

A few studies have also described the effects of IUP on male fetuses due to intrauterine E₂ exposure. Adult 0M males exhibit prostates that are enlarged compared with those of adult 2M males. Furthermore, 0M male embryos have elevated levels of blood E₂, which is believed to be the result of E₂ exposure from the neighboring female fetuses (103, 107). These data provided seminal evidence of an effect of low-dose exposure to estrogens on in utero embryo development that persist into adulthood. It has now been established that neonatal exposure to estrogens can elicit epigenetic modifications via changes in methylation status that lead to altered gene expression with functional and developmental consequences (108–111). Exposure to hormones can manifest not only via IUP, but also by crossing of molecules through the placenta or through the ingestion of milk during the neonatal period.

To further explore the role of in utero E₂ effects on embryo development, pregnant mice were exposed to varying doses of either E₂ or the synthetic estrogen diethylstilbestrol, and the prostates of adult male mice were analyzed for enlargement and androgen receptor expression (112). The results revealed that small elevations in E₂ exposure in utero led to an increase in the enlargement and in the expression of the androgen receptor in the prostate. Interestingly, higher doses of E₂ in utero did not lead to an increased exacerbation of these phenotypes in adult male mice, thus generating a nonmonotonic dose–response curve, such that low doses of E₂ were stimulatory, and high doses were inhibitory (112). The effects of low-dose hormone exposure on development have been corroborated by human twin studies that indicate that female twins can influence their co-twins, including behavior (113) and susceptibility to breast (114, 115) and testicular cancers (115). Taken together, these data lead to the question of whether the many known endocrine-disrupting chemicals that are present at environmentally relevant doses and are deemed safe by the regulatory community may in fact yield adverse effects when exposure occurs during development.

Maternal environment

Although it is clear that the responsiveness of the mammary gland and uterus to E₂ is genetically controlled, it is unclear what role, if any, the maternal environment plays in genetic variations in E₂ responses. Variations in the maternal environment and the quality of maternal care are known to influence the responsiveness of reproductive tissues to E₂ (116, 117), E₂ signaling in the brain (118), and susceptibility to many diseases, including diabetes (119) and autoimmune disease (120). Indeed, in humans, epidemiological data indicate that the source of nutrients for newborns (breastfeeding vs. formula feeding) has long-term effects on susceptibility to many diseases, including breast cancer, diabetes, obesity, and asthma (121, 122).

Factors from the maternal environment are delivered to the neonate through the milk (lactocrine signaling, ref. 123). The molecular events associated with lactocrine signaling to the neonate, and subsequent uterine development, are well-established in pigs (117, 124). In addition, nonobese diabetic mice have different risks of developing autoimmune type 1 diabetes depending on the strain of their foster mothers, representing a clear example of gene-by-maternal postnatal environment effects influencing adult-onset autoimmune disease (119). Although it is clear that lactocrine signaling exists across multiple species, including humans (121, 122), it is unclear how ingestion of milk permanently alters gene expression and subsequent function in reproductive tissues and other organs. Because the maternal environment and lactocrine signaling can program only neonatal tissues during a very small window of time (116), it is likely that an epigenetic mechanism is responsible for the long-term effects (118).

Another important factor that can influence E₂ responses is the parent of origin (POO). More specifically, the genotypes of each parent may uniquely contribute to the phenotype of the offspring. Experiments designed to investigate POO effects do not usually distinguish between the POO and the maternal environment (cross-fostering is necessary to make this distinction), but they can answer important questions about the nature of the inheritance pattern of the phenotype. We recently tested the hypothesis that the POO would influence the response of the uterus to E₂. Parental B6 and C3H strains were studied, along with B6C3 and C3B6 F1 hybrids. (In both cases, the first strain listed is the strain of the mother; the second is father.) Our preliminary findings revealed that the F1 hybrid strains have a similar uterine response to E₂ and are C3H-like; therefore, the POO does not appear to influence this response (Fig. 7). These observations are consistent with those of Trentin (7), who observed no POO effect on the vaginal response to estrogen, and they warrant additional experimentation using cross-fostering experiments to distinguish between genetic regulation and contributions of the maternal environment in regulating estrogen responsiveness.

Figure 7.

Effect of POO on the uterine response to E₂ in female mice. Letters above bars indicate differences in the E₂-induced increase in uterine weight between strains; P < 0.05 (unpublished results).

Environmental estrogens

Environmental estrogens are molecules found in the environment that mimic the actions of E₂, presumably by binding to ERs. Exposure to these molecules during critical phases of development, especially during physiological transitions, such as puberty, can have marked effects on reproductive tissue development and function, as well as the physiology of nonreproductive tissues. Consistent with lactocrine signaling and the importance of the maternal environment, it has been reported that exposure to endocrine disrupters through ingestion of milk from treated dams elicits uterotropic responses in prepubertal rats (125); increases ovarian apoptosis (126) and neuroendocrine development in adult rats (127); permanently influences expression of ERα and -β across multiple tissues in both prepubertal and adult rats (125); and increases susceptibility to mammary cancer in adult rats (128).

The genetic control of the responsiveness to environmental estrogen disruptors has been documented. For example, treatment of juvenile male mice with E₂ resulted in decreased testis weight, and this response varied markedly across mouse strains (28, 129). Genetic background also controlled the degree to which E₂ influenced spermatogenesis, which complete disruption in some strains and no effect in others (28). In addition, the development of male reproductive organ abnormalities after neonatal exposure to tamoxifen is genetically determined (130). Moreover, recent findings showed that the effects of developmental exposure to bisphenol A on adult testes gene expression are genetically controlled (131). These findings highlight the importance of using multiple genetic backgrounds when investigating the impact of environmental estrogens on tissue function and when determining optimum dosages for therapeutic estrogens.

CONCLUSIONS

It is now accepted that estrogens have a vast array of functions across multiple tissues and that genetic factors play a highly significant role in the variation in tissue responses to estrogens. Indeed, marked variation in estrogen responses, based on differences in genetic background, have been documented for many tissues including uterus, vagina, and mammary gland, as well as skin, bone, and others. Understanding the factors involved in phenotypic variation in response to estrogens, as well as the regulation of these factors, could provide critical insight into the genetic variation in other highly relevant E₂-regulated processes in humans, including bone loss in postmenopausal women (132, 133), premature ovarian failure (134), fertility (135, 136) and libido (135), success rate of fertility treatments (137, 138), and sensitivity to environmental endocrine disrupters (28, 129).

As clinicians and researchers continue to strive to piece together this physiological puzzle, several exciting research opportunities are evident. The first is that, although it has been clearly established that genetics explain a large part of the variation in estrogen responses, we still do not know the specific genes involved. Therefore, genetic mapping experiments, designed to positionally clone the genes underlying the QTLs controlling tissue responsiveness to E₂ and determining their functional role in modulating tissue sensitivity to E₂ across various physiological states and genetic backgrounds, are needed. The orthologous gene candidates can then be used in other species, including humans, to identify global regulators of E₂ responses. Such analyses are central to understanding inheritance patterns of disease susceptibility and providing insight into how individual genetic variation influences responses to treatment of E₂-dependent diseases and sensitivity to hormonal agents and therapeutics.

The second research opportunity is that, although the role of the maternal environment is clear, the true genetic contribution vs. that of lactocrine signaling is still unclear. Therefore, it is important to distinguish between POO effects that manifest through a genetic mechanism and gestational environment vs. maternal environment effects that occur during the neonatal prepubertal period and how each of these effects contributes to variation in estrogen responses during adulthood. Determining these distinctions could lead to clearly defining a role for these factors in the responsiveness of various adult tissues to E₂, and identifying therapeutic targets for preventing and treating disease.

Finally, more work is needed to clearly identify the critical windows wherein exposure to natural or environmental estrogens can permanently alter tissue physiology, the mechanisms by which this occurs, what genetic factors render an individual sensitive or resistant to exposure, and how these processes can be manipulated to optimize development and prevent disease. As we strive to understand the key players and how to manipulate them, we can hope to achieve more effective disease prevention, improved therapeutics for disease and reproductive dysfunction, and a more refined approach to hormone therapies based on physiological state and genetic background.