| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Receptor Biology Section, Laboratory of Reproductive and Developmental Toxicology, National Institute of Environmental Health Sciences, National Institutes of Health, Research Triangle Park, North Carolina 27709
 |
Abstract |
|---|
 Top Abstract I. Introduction—A...II. Estrogen ReceptorsIII. Reproductive Tract...IV. Mammary GlandV. Reproductive Tract Phenotypes...VI. Neuroendocrine SystemVII. Phenotypes in Peripheral...VIII. Comparison with Human...IX. SummaryReferences |
|---|
ERKO phenotype and oncogene-induced tumorigenesis: Wnt-1/ERKO mice and aromatase deficiency |
I. Introduction—A Historical Perspective |
|---|
|
TopAbstractI. Introduction—A...II. Estrogen ReceptorsIII. Reproductive Tract...IV. Mammary GlandV. Reproductive Tract Phenotypes...VI. Neuroendocrine SystemVII. Phenotypes in Peripheral...VIII. Comparison with Human...IX. SummaryReferences |
|---|
At the time of the above statement, the laboratories of Elwood Jensen and Jack Gorski had spent 10 yr providing experimental evidence to support the concept of an intracellular "receptor" protein for steroid hormones. Their combined work had even led to a proposed model by which the interactions of the receptor and the steroid were involved in mediating the cellular effects of the hormone (1). The first of these receptors to be characterized was for the female sex hormone, 17ß-estradiol (E2) (2, 3). Since this time, similar receptors for testosterone, progesterone, glucocorticoids, thyroid hormone, vitamin-D3, and retinoids have been discovered and now form a portion of a large family of nuclear hormone receptors (4). Although significant strides have been made since, Jensen’s introductory statement in a review published more than 25 yr ago is still contemporary with the current research goals toward understanding the growing family of nuclear receptors. This is not to say that little progress has been made, which is quite to the contrary, but rather to state that although advances in technology and molecular biology have allowed for extensive insight, there is still a great deal to be learned as we move into the next century.
Our current understanding of the various roles of the steroid, thyroid, and retinoid hormones in development and normal physiology and the mechanisms by which these actions are mediated is due, in large part, to the generation of a series of reagents and tools over the past 40 yr. The first of these was the synthesis and use of tritium-labeled E2 with high specific activity, allowing for the first reports of the detection and simple characterization of an estrogen-binding component, or "estrophilin" (1, 3, 5). This protein exhibited a binding affinity for estradiol that was several fold higher compared with the other gonadal steroids and was found only in tissues previously shown to respond to estradiol in terms of growth and increased RNA synthesis (1, 3). These studies described the uptake and concentration of radiolabeled estradiol by a specific protein unique to the cells of the uterus, vagina, and pituitary and thereby disputed the current thought that estrogen action required enzymatic metabolism of the hormone (2, 3, 6). Thus, the concept of a hormone receptor protein in target tissues was initiated. Autoradiographic studies with radiolabeled ligand further demonstrated the strong association of estradiol with the nuclei of cells lining the rat uterus within 4 h after injection (7). These studies were significantly advanced by pharmacological approaches with early nonsteroidal estrogen antagonists, the first being the triphenylethylene, MER-25 (2), followed soon after by a series of similar compounds, e.g., clomiphene, nafoxidine, CI-628, and tamoxifen (1, 8). These reagents were previously known to inhibit the uterotropic effects of estradiol in a dose-dependent manner but, more importantly, were later shown to block estradiol uptake and binding in target tissues. Still, at this time the exact role that the hormone-receptor complex played in the final manifestation of the hormonal effects remained unclear. It was suggested that the receptor may simply fulfill a "transport" role to move the steroid hormone from the cytoplasm to the nucleus of the cell (1). Nonetheless, from these early studies, the definition of an estrogen target tissue now included not only the exhibition of a measurable response to the natural hormone (E2) but also one that possessed detectable levels of the estrogen receptor (ER).
Much of the early characterization of the ER relied on the use of sucrose gradient analysis of nuclear and cytoplasmic cell fractions under varied salt concentrations (3). This procedure, along with gel electrophoresis and filtration using radiolabeled estradiol, allowed for the generation of semipure fractions of the estradiol-binding protein. These preparations led to the generation of antisera specific to the ER protein, another notable step in receptor research (9). The later development of monoclonal antibodies to the ER and the immunohistochemical techniques that soon followed provided evidence of the predominantly nuclear localization of the receptor protein in target cells (10, 11). The late 1970s brought numerous reports of the tissue distribution and localization of the ER in humans and laboratory animals, confirming much of the findings of earlier steroid autoradiography studies (7, 11). The antibodies were also used to further purify large amounts of the ER from target tissues, allowing for more detailed studies of the receptor structure and function.
The 1980s witnessed the cloning and sequencing of the cDNAs for several of the steroid hormone receptors, which proved to be a seminal step toward understanding their mechanisms of action. The first cDNA to be cloned for a member of the nuclear receptor family was that for the human glucocorticoid receptor (12), followed soon after with the description of the human ER cDNA (13, 14). Since this time, the cDNAs encoding several other members of the nuclear receptor family have been described (4). The current list of isolated ER cDNAs includes those for the chicken (15), mouse (16), rat (17), Xenopus laevis (18), and rainbow trout (19). Sequence analysis of the various receptor cDNAs demonstrated a high degree of similarity and led to their inclusion in a superfamily of nuclear receptors possessing a defined motif of functional domains (4).
A vital contribution from the cloning of the nuclear receptor cDNAs was the ability to express the receptor proteins in in vitro mammalian (20) or yeast (21) cell systems. These techniques, as well as cell-free in vitro transcription (22), allowed detailed characterizations of the different functional domains of the receptor. Recombinant DNA methodologies provided for the construction of receptor cDNAs possessing precise truncations, deletions, point mutations, or additional sequences. These powerful techniques led to the in vitro expression of chimeric and mutant receptors and great advances in the dissection and mapping of the specific domains and distinct residues critical to receptor function (21, 22, 23, 24, 25). The late 1980s witnessed multiple descriptions of naturally occurring variants and mutations of several of the nuclear receptor transcripts (26), including the ER (27, 28, 29). Their identification and functional characterization have led to further insight into the mechanisms of action of specific domains of the receptor proteins as well as the receptors as a whole. Furthermore, the detection of these nuclear receptor transcript variants in vivo has allowed speculation concerning their possible roles in alternate or abnormal hormonal signaling in normal and neoplastic tissues (27).
By the 1990s, it was evident that the members of the steroid/thyroid hormone superfamily of receptors were intracellular proteins that functioned in the nucleus to regulate transcription of target genes. This growing family of receptors now includes those for the sex and adrenal steroids, thyroid hormones, retinoids, vitamin D3, and eicosinoids (reviewed in Refs. 4, 30). The inclusion of nuclear receptor-like proteins with no known ligand, termed orphan receptors, such as those for the chicken ovalbumin upstream promoter (COUP), and steroidogenic factor-1 (SF-1), has expanded the family to now include approximately 150 distinct proteins (30).
Our knowledge of the expression patterns and mechanism of action of the
nuclear receptors has led to a greater appreciation of their
involvement in normal physiology and disease. The current decade has
witnessed several advances in our understanding of the molecular
biology and overall physiological role of these proteins. Especially
significant to these efforts has been the discovery and/or development
of three distinct aspects. The first of these was the discovery of
coregulators, a second group of nuclear proteins that further modulate
the actions of unoccupied as well as ligand (agonist or
antagonist)-bound receptors (reviewed in Ref. 31). The continued
characterization of the coregulator proteins has provided some insight
toward the long sought explanations for the cell- and tissue-specific
mixed agonist/antagonist activity of certain ligands (31). The second
of these advances was the generation of crystal structures for the
ligand-binding domains of the retinoic acid receptor- (32), retinoic
X receptor- (33), thyroid receptor- (34), and ER (35, 36).
These new data continue to allow for the description of the long
speculated conformational changes that result in the receptor after
binding of the natural ligand, and how these changes may differ from
those induced by natural or synthetic agonists and antagonists.
The third development of great impact in this decade has been the use of gene-targeting technology and transgenic techniques to disrupt the genes encoding several members of the steroid/thyroid hormone receptor superfamily. This methodology has allowed for the generation of transgenic mice that lack a functional gene for a specific receptor, as well as germline passage of this mutation. Hormone resistance due to naturally occurring mutations in the genes encoding the receptors for androgen (37, 38), glucocorticoids (39), and thyroid hormones (40) had already been described in humans and laboratory animals and thereby provided great insight into the role of these hormones in development and normal physiology. However, similar mutations had not yet been reported for other members of the superfamily of nuclear receptors. The impact of the gene-targeting technique is evident in the generation of several "knockout" mice for the nuclear receptors, including mice lacking multiple forms of the retinoic acid receptors (41, 42, 43), the progesterone receptor (44), the vitamin D3 receptor (45), the ERs (46, 47), and the coregulator protein, steroid receptor coactivator-1 (48). If nothing else, the naturally occurring mutants combined with the generation of the knockout models, have confirmed the basic hypotheses put forth almost 30 yr ago, i.e., that the receptor proteins found to tightly and specifically bind a hormone within a tissue are indeed critical for mediating the biological effect of the hormone within the target cell (1).
This review will discuss one such advance originating from the
gene-targeting boom that occurred in the 1990s, the ER knockout mice.
The broad content of this review is a reflection of the current state
of research in the expanding field of estrogen action, as illustrated
by the discovery of a second ER, the ERß. A large extent of the
discussion will focus on the mice lacking the classical ER, the ER
knockout mice (ERKO). The reasons for this are 2-fold: 1) the
ERKO mouse has been available for more than 6 yr, compared with less
than 1 yr for the ßERKO, and therefore has been more thoroughly
studied and characterized; 2) the phenotypes of the ERKO mice appear
to be more broad in nature compared with those of the less well studied
ßERKO mice, although continued research may lessen this disparity.
The extent to which the ERKO models have been used in various fields of
biological research is illustrated in Table 1
. This review will discuss the
phenotypes of the ERKO mice and the multiple ways in which these models
continue to influence the field of estrogen research. Foremost, several
of the hypotheses of estrogen and ER action put forth from the efforts
of numerous investigators over the years were confirmed by our
observations in the ERKO. Furthermore, certain phenotypes have
introduced unforeseen critical roles of estrogen in some physiological
systems, such as in male fertility. Later years have witnessed the use
of the ERKO as a research tool to investigate specific biochemical
pathways and neoplasia in the absence of estrogen action. Where
applicable, we will contrast the phenotypes of the two ERKO models, as
well as compare the relevant phenotypes of knockout models for other
hormone signaling systems. And finally, we will discuss how these
models compare with the few cases of insufficient estrogen synthesis
and the single reported case of estrogen insensitivity in humans.
Before we continue, however, we wish to take this opportunity to
establish consistency in the abbreviations used to refer to the
ER-knockout (ERKO) mice in the literature. Although the terms ERKO
and ERßKO have appeared in previous reports from our own as well as
other laboratories, we propose that the abbreviations used above and
throughout the remainder of this review, i.e., ERKO and
ßERKO, be the consensus abbreviations used hereafter to refer to the
two models.
|
|
|
II. Estrogen Receptors |
|---|
|
TopAbstractI. Introduction—A...II. Estrogen ReceptorsIII. Reproductive Tract...IV. Mammary GlandV. Reproductive Tract Phenotypes...VI. Neuroendocrine SystemVII. Phenotypes in Peripheral...VIII. Comparison with Human...IX. SummaryReferences |
|---|
.
The two receptors are not isoforms of each other, but rather distinct
proteins encoded by separate genes located on different chromosomes. As
a result of this discovery, we now know that the ER shares a phenomenon
of multiple forms previously described for other members of the nuclear
hormone superfamily, including the receptors for thyroid hormone,
retinoids, mineralocorticoids, and progesterone (26, 30). A detailed
description of the structure and mechanism of action of the ERs is
beyond the scope of this review, and therefore only a brief discussion
of the relevant points will appear here. For more detailed descriptions
of the mechanism of steroid receptor-regulated gene
transcription, readers are encouraged to seek several recent reviews
(52, 53, 54, 55).
Transcription of the mouse ER gene in vivo
predominantly results in a single transcript of approximately 6.3 kb
transcribed from 9 exons. This transcript encodes a protein of 599
amino acids with an approximate molecular mass of 66 kDa (16).
The human ER is slightly shorter at 595 amino acids but exhibits a
similar molecular mass (13, 14). Whereas the human ER gene has been
mapped to chromosome 6 (56), the mouse ER gene is located on
chromosome 10 (57). The existence of multiple promoter and regulatory
regions in the 5'-untranslated sequences of the human and rat ER
have been described, but only a single open reading frame appears to
exist (58, 59, 60). Numerous reports have described the discovery and
characterization of naturally occurring variants and mutations of the
ER mRNA in normal as well as neoplastic tissues of several species
(reviewed in Refs. 27, 28, 61). Although the existence of true
protein products of these ER mRNA variants in vivo
remains controversial, their transactivational activities in in
vitro cell culture systems have been intensely described and have
furthered our understanding of the functional domains of the receptor
(27, 61).
Initial studies indicated that the rodent ERß was composed of 485
amino acids and an estimated molecular mass of 54 kDa and therefore was
slightly smaller than the ER (49, 50, 62). The majority of this
difference in size between the two ERs was due to a significantly
shorter N' terminus in the deduced ERß protein. Unlike the ER
gene, Northern blot analyses of ovarian RNA from both the rat and mouse
indicates the presence of multiple ERß transcripts (50, 63, 64).
Furthermore, open reading frames initiating up-stream from those
originally described have now been discovered in the mouse, rat,
bovine, and human ERß mRNA (65, 66, 67, 68, 69). These studies have also
provided evidence to support that the upstream start codons are the
likely initiation sites of translation and therefore suggest the
possibility of an ERß protein of 527–530 amino acids and a
calculated molecular mass of approximately 60 kDa (65, 66, 67, 69).
However, convincing Western blots from tissue extracts to indicate the
true in vivo molecular mass of the ERß have remained
difficult to produce with the antibody preparations currently
available. Similar to ER, a number of variants of the ERß mRNA
have already been described. These include a conserved insertion of 18
amino acids in a C'-terminal region of the ERß in the rat (70, 71),
human, and mouse (72), the deletion of one or more exons in these same
species (70, 72, 73), and various isoforms in the extreme C'-terminus
of the human ERß (68).
B. Mechanism of ER action
The ERs are classified as class I members of the
superfamily of nuclear hormone receptors, defined as a ligand-inducible
transcription factor (30). Early studies indicated the ER was
cytoplasmic and became localized to the nucleus only upon ligand
binding, providing the basis of the initial "two-step mechanism" of
hormone action (1, 3). However, it is now accepted that the ER is a
predominantly nuclear protein regardless of whether
or not it is complexed with ligand (74). The inactive ER exists in a
complex consisting of several heat-shock and other proteins that appear
to disassociate upon ligand binding, resulting in a
"transformation" of the receptor to an active state (75). With
continued research, the two-step mechanism model has evolved to state
that upon binding of estradiol, or an estrogenic ligand, the
transformed receptors form dimers that tightly associate with specific
consensus DNA sequences, consisting of 15-bp inverted palindromes in
the regulatory regions of target genes (52, 74, 75, 76). This complex then
interacts with basal transcription factors, coregulator proteins, and
other transcription factors to ultimately regulate transcription of the
target gene (52, 55, 76). However, in recent years, pathways of gene
activation by the steroid receptors that deviate from this classical
model have been described. These include gene activation by
ligand-bound steroid receptors without evidence of direct DNA
binding, but rather via interaction with other DNA-bound transcription
factors, such as an AP-1 complex (77, 78). In addition,
ligand-independent activation of the receptor through pathways that
alter the activity of cellular kinases and phosphatases has been
demonstrated both in vitro and in vivo (reviewed
in Ref. 55). The discovery of these pathways strongly supports the
great importance of the ER and its ability to possibly provide diverse
physiological functions even in the absence of ligand.
The ER and ERß proteins are composed of six functional domains,
labeled A–F, a signature characteristic of members of the superfamily
of steroid/thyroid hormone nuclear receptors (Fig. 1). The N'-terminal A/B domain is the
least conserved among all members and demonstrates only 17% identity
between the human ER and ERß (64). In contrast, the C domain is
the most highly conserved among the different members of the family. It
possesses two zinc fingers forming a helix-loop-helix motif and
primarily functions in tightly binding the receptor to the DNA hormone
response elements. The sequences encoding the two zinc fingers possess
97% homology between the ER and ERß genes and are located in
separate exons (exons 3 and 4) in each (50, 64, 79). The E domain, or
ligand-binding domain, confers ligand specificity to the receptor and
is moderately conserved among the members of the superfamily. The ER
and ERß proteins possess 60% conservation of the residues in the E
domain; however, each binds estradiol with nearly equal affinity and
exhibits a very similar binding profile for a large number of natural
and synthetic ligands (80). The D domain possesses signals for nuclear
localization of the receptor and exhibits approximately 30% identity
between the two human forms of ER (64). The C'-terminal F domain is
unique to the ER among the nuclear receptors for the gonadal and
adrenal hormones (6) but is not well conserved among the ERs of
different species nor between the ER and ERß, which share
approximately 18% homology (64). Studies using forms of the ER
missing the C' terminus have indicated a role for the F domain in
modulating transactivational activity of the ER when complexed with
mixed agonist/antagonist ligands, possibly via influencing coregulatory
function and/or dimerization of the receptor (81, 82).
|
(85).
The discovery of the ERß has introduced a new level of complexity to
the current model as well. To date, there exist no data indicating a
physiological response solely mediated by ERß. In contrast, the
ERKO mouse has confirmed the requirement for ER in mediating
several actions of estradiol, as will be discussed in this review.
Nevertheless, in vitro experiments from several laboratories
have indicated the possibility of cooperative activity between the two
receptors, acting in the form of heterodimers (50, 62, 65, 86). These
studies generally report a tendency of ER to form homodimers whereas
ERß prefers to heterodimerize with ER. However, Giguere et
al. report (87) that the heterodimer is the preferred state when
both mouse ERs are present. The transactivational activity of the
heterodimer when assayed in in vitro mammalian cell
transfection assays appears to lie between that of the more active
ER homodimer and the less active ERß homodimer (50, 62, 86). A
major consideration when evaluating the possible physiological
functions of an ER/ERß heterodimer is evidence of coexpression of
the two receptors in the same cell, which has not yet been definitively
reported (88). To this end, studies and reagents are only now becoming
available to directly assess this question.
Several functional characteristics of the two ERs are similar. The
residues critical to function of the AF-2 domain appear to be identical
in the mouse ER and ERß (87). Tremblay et al. (50, 89)
demonstrated that a tyrosine residue critical to the function of the
AF-2 domain was conserved in both the ER and ERß and that mutation
of this amino acid resulted in similar constitutive, ligand-independent
transactivational activity in both receptors. In contrast, the
N'-terminal AF-1 domain shows no significant regions of similarity
between the two ERs (87). However, a potential activation site of the
mitogen-activated protein (MAP)-kinase pathway previously shown for
ER is present and active in the ERß (50, 89). Additionally, when
acting on a basal promoter linked to a consensus estrogen response
element, both ER and ERß were able to recruit the coactivator
SRC-1 and were equally susceptible to inhibition by the antiestrogens
raloxifene, ICI 164,384, and EM-800 (50).
However, as studies continue, distinct differences at the molecular
level and in the transactivational capacities between ER and ERß
have been described. Two separate studies have demonstrated the
specificity of the agonist activity of 4-hydroxytamoxifen to be unique
to ER, although this appears to be highly dependent on the cell and
promoter context as well as experimental design (50, 90). Furthermore,
Paech et al. (91) reported that when interacting with
DNA-bound AP-1 transcription factors, the in vitro
transactivational activity of estrogen agonists and antagonists was
quite different depending on which form of ER was present. Whereas
antagonists, such as raloxifene, tamoxifen, and ICI 164,384, were able
to block the stimulatory activity of the ER/AP-1 complex, these same
compounds acted as potent agonists when bound to an ERß/AP-1 complex
(91). Further experimental support for the existence of distinct
structural and functional differences between ER and ERß was
recently provided by Sun et al., who showed that certain
nonsteroidal ligands were receptor selective in their binding and
agonist/antagonist activities (92).
Perhaps the most significant disparity lies in the tissue distribution
of the two receptors. Studies employing the techniques of RT-PCR and/or
ribonuclease protection assay (RPA) have indicated that ER mRNA is
predominant in the uterus, mammary gland, testis, pituitary, liver,
kidney, heart, and skeletal muscle, whereas ERß transcripts are
significantly expressed in the ovary and prostate (Fig. 2) (63, 80, 93, 94). These same studies
have indicated relatively equal levels of mRNA for the two receptors in
the epididymis, thyroid, adrenals, bone, and various regions of the
brain (80, 93, 95, 96, 97). However, as more studies are reported, several
discrepancies in the expression patterns of ER and ERß among
different species are becoming apparent (80, 93, 96). For example,
whereas ERß mRNA is easily detectable in the pituitary of the rat
(70, 98, 99), human (100), and rhesus monkey (96), levels in the
pituitary of the mouse appear low to undetectable (93). A similar
difference in expression is apparent in the mammary gland, in which
normal and neoplastic human tissue and cell lines express detectable
ERß mRNA (64, 68, 73, 101, 102), although the mammary gland of the
mouse appears to predominantly express ER (93). Furthermore, even in
those tissues expressing both ERs, there is often a distinct expression
pattern within the heterogeneous cell types composing the tissue. In
the ovary, ERß is apparently localized to the granulosa cells of
maturing follicles, whereas ER is detectable in the surrounding
thecal cells (69, 103, 104). In the prostate of the rat, expression of
ER and ERß is detectable in the stroma and epithelium,
respectively, but does not appear to be colocalized in any portion of
the tissue (49). However, through the combined use of
immunohistochemistry and in situ hybridization, Shughrue
et al. (88) have demonstrated colocalization of ER and
ERß to select regions of the rat forebrain.
|
The rationale for generating mice that possess no functional ER is multifaceted, but in its most simple terms, was founded on the classical ablation experiments of the early part of this century. In 1900, Knauer (105) described the ability of ovarian grafts to prevent uterine atrophy in the castrated guinea pig. Five years later, Marshall and Jolly (106) described the capacity of ovarian extracts to induce estrus when administered to ovariectomized dogs. Similar protocols were later elegantly employed by Jost (107) to substantiate the endocrine function of the testis and the importance of testosterone in sex determination. These studies were critical to establishing the following basic criteria required to verify an endocrine role for a particular organ or tissue: 1) removal or destruction of the synthesizing organ should result in predictable symptoms presumed to be related to the absence of the hormone; 2) administration of material prepared from the removed organ should relieve these symptoms; and 3) the hormone should be present in and extractable from both the organ and blood (108). In the spirit of these earlier studies, the later part of this century has witnessed the marriage of two relatively new methodologies to introduce a modern version of the ablation experiment. The combination of in vitro culture of mouse embryonic stem cells and targeted homologous recombination has generated a tool that allows for the precise disruption or knockout of a particular gene of study and the passage of this mutation to offspring. Although one can debate whether this new technique is more or less invasive than the previous surgical methods, it is obvious that the classic "ablation" experiment has been elevated to a molecular level. The function of a specific hormone can now be studied rather than the function of a whole endocrine organ that may produce multiple secretions. Furthermore, this new technology allows for the study of a particular cellular component, such as a receptor, that is intrinsic to one or more tissues. Such studies were previously impossible or relied on the use of chemical antagonists, which introduced their own inherent limitations. Additionally, gene targeting provides for in vivo methods to study the roles of a particular receptor throughout the life of the animal, including the early development stages. The current tools that allow for the generation of transgenic mice have already made significant contributions to our knowledge of particular genes, especially those involved in development and reproduction (reviewed in Refs. 109, 110).
At the time the ERKO mouse was envisioned, the ER was the only form
of ER known to exist. Furthermore, there were no reports of ER
mutations in normal tissue that resulted in estrogen insensitivity in
humans or laboratory animals. This was in contrast to the descriptions
of syndromes of receptor-based insensitivity to androgens (37, 38) and
thyroid hormones (40). Therefore, investigators were inclined to
conclude that mutations that resulted in insufficient estrogen
synthesis or resistance to the hormone at the level of the target organ
were lethal at the earliest developmental stages (111, 112). This view
was strengthened by reports of the detection of ER mRNA in both
human (113) and mouse (114) oocytes as well as in mouse blastocysts
(115). Accordingly, the concept of generating a mouse devoid of ER
was initially met with skepticism but was pursued as a collaborative
effort between our laboratory and that of Dr. Oliver Smithies of the
University of North Carolina at Chapel Hill. If disruption of the mouse
ER gene did prove lethal, a model to study the precise time and
locations of critical ER-mediated actions during early development
would be available. However, providing the animal was viable, an
in vivo model of estrogen insensitivity would now be
accessible for continued study. Six years after the initial description
of the ERKO mouse, we now take for granted that disruption of the
ER gene proved not to be lethal, but rather the animal develops
normally and exhibits a life span comparable to its wild-type litter
mates (46). However, as will be elaborated in detail in this review,
the adult ERKO mice exhibit several abnormalities and deficiencies,
most notable of which are the phenotypic syndromes that result in
infertility in both sexes. Since this time, additional discoveries have
been made to enhance the utility of the ERKO model. Soon after the
generation of the ERKO mouse, Smith et al. (116) reported
the first and only known case of clinical estrogen insensitivity due to
an inactivating mutation of the ER gene in a human patient. Later
came multiple descriptions of aromatase and subsequent estradiol
deficiency in humans (117, 118, 119, 120). And finally, a second collaborative
effort resulted in the successful generation of the ßERKO mice, which
also survive to adulthood and exhibit phenotypes unique from those of
the ERKO (47). However, splicing variants of the disrupted genes
that may encode for receptors with decreased functional activity have
been detected in small amounts in each of the respective ERKO models
(as discussed below) and therefore complicate the interpretation of the
nonlethality of the gene targeting. Nonetheless, we now know that a
loss of full function of any one of the two ERs is neither lethal nor
detrimental to embryonic and fetal development in both mice and humans.
Perhaps the survival of the ERKO mice, the aromatase-deficient humans,
and the ER-deficient male will prompt a renewed effort among
clinicians to suspect and investigate the possibility of estrogen
insufficiency or resistance in patients not responding to conventional
therapies.
The ERKO mice provide broad and multiple advantages to the research
efforts toward understanding the function and mechanism of estogen
action. Much of what is known about estrogen action was inferred from
in vivo studies involving castration or the administration
of ER antagonists or inhibitors of estradiol synthesis. These findings
have been complemented by the vast knowledge gained from in
vitro cell culture studies, employing chimeric and mutant versions
of the ER, varied cell types, multiple combinations of
promoter-reporter gene constructs, as well as synthetic agonists and
antagonists. However, there are distinct disadvantages to these
experimental schemes. Studies using aromatase inhibitors and/or
estrogen antagonists are complicated by several factors, including
variability of the compound to block the action of the natural hormone
or enzyme. The effectiveness of various antagonists is highly dependent
upon the animal model, the tissue or cell of study, the bioavailability
of the compound at different target tissues, and the class of
antiestrogen used (8). This dilemma is further complicated by the
discovery of the ERß, since no known ER-selective agonists or
antagonists have been characterized at an in vivo level. The
limitations of in vitro cell culture experimental approaches
are obvious and mostly based on their finite application to the whole
animal. Therefore, the ERKO models provide a unique tool to investigate
the role of the ER in the context of the whole animal, and equally
important, during the complete life span of the animal. At their most
fundamental level, the ERKO mice address the role of the ER in the
development and normal physiology of all organ systems, as well as in
carcinogenesis, toxicity, and aging. Furthermore, unlike the
"castrate" model in which several hormones are removed from the
system, the ERKO mice retain the capability to synthesize the gonadal
steroids, including the natural ER ligand, estradiol. Therefore, the
biochemical functions of estradiol and the ER can be investigated in
the presence of presumably intact pathways for the other gonadal
hormones. The presence of estrogens in the absence of ER also provides
for the possibility of discovering pathways of estrogen action that are
independent of nuclear ER, or mediated via previously unknown forms of
the receptor. Additionally, although the majority of in
vitro studies indicate that ER and ERß may have redundant
functions, their differences in tissue distribution and response to
certain ligands indicate the presence of distinct roles fulfilled by
each. The fact that the ERKO mice exhibit an unaltered pattern of
ERß mRNA expression strengthens the usefulness of this model to
dissect these potential ERß-mediated actions (93, 121). Finally,
consistent with those criteria discussed earlier for establishing an
endocrine function to an organ, the ERKO animals now provide a null
background available for transgenic reintroduction of the ER of other
species, mutated ERs, or targeted ER expression to a specific tissue or
cell type.
A detailed description of the targeting scheme employed to disrupt the
mouse ER gene can be found in the initial description of the ERKO
mice (46). As shown in Fig. 1
, a 1.8-kb insert possessing the gene for
neomycin (neo) resistance under the control of the
phosphoglycerate kinase (PGK) promoter and including a
PGK-polyadenylation signal was inserted into a NotI site in
exon 2 of an ER gene fragment subcloned from a genomic library of
129/J mouse DNA. The targeting insert was placed in a replacement 
type targeting vector (122) with the appropriate ER gene-flanking
sequences. Upon successful targeting in mouse embryonic stem cells
(129/J), the neo insert is placed approximately 270 bp
downstream of the ER translation start site and thereby inhibits
proper expression of the ER gene. Since this was the current state
of the technology, no portion of the ER gene was removed during the
targeting event. Standard protocols of clone selection and blastocyst
(C57BL/6J) injection were used to generate chimeric mice possessing the
disrupted gene, some of which demonstrated germ-line transmission of
the mutation when bred with wild-type mates (122). Southern blot and
PCR analysis of genomic DNA from mice of all three genotypes indicated
the correct targeting of the ER gene and the absence of any
heterologous recombination in other regions of the genome. Inbreeding
of mice heterozygous for the ER disruption resulted in a Mendelian
distribution of all three genotypes as well as a balanced sex ratio,
indicating that the ER is not critical to sex determination at the
level of the external genitalia (46).
The generation of mice homozygous for a disruption of the ERß gene
was similar to that described above for the ERKO and can be found in
detail in the initial description (47). A genomic clone that spanned a
15-kb region possessing the first three exons of the mouse ERß gene
was selected from a 129/SvJ mouse library. A replacement type
targeting construct was generated to include 5' and 3' homologous
sequences of 1.3 and 7.4 kb, respectively (Fig. 1). A PGK
promoter-regulated neo gene was inserted in the reverse
orientation into a PstI site in exon 3 of the ERß clone.
Therefore, correct targeting resulted in disruption of the sequences
coding for the first zinc finger of the ERß protein, a domain
critical to normal function of the receptor. Chimeric and heterozygous
offspring were generated as described above for the ERKO mice. Once
again, mice possessing the targeted disruption of the ERß gene were
identified by diagnostic Southern blotting and PCR of genomic DNA. As
with the ERKO, inbreeding of mice heterozygous for the disruption
yielded a Mendelian distribution of all three genotypes as well as a
balanced sex ratio (47).
In both knockout models, RT-PCR on RNA from target tissues indicated
the presence of multiple splicing variants of the respective ER
transcripts (47, 123). In neither case has wild-type-like mRNA
transcribed from the disrupted receptor gene been detected. The greater
proportion of the variants detected in each ERKO model possessed frame
shifts that would result in a severely truncated or mutated ER if
translated. However, in the ERKO, a single-splice variant capable of
encoding a mutant ER protein with significantly decreased
transactivational capacity in vitro was detected at very low
levels (123). A similar ERß splice variant, in which the reading
frame was preserved although coding sequences were removed, was
detected in ovaries of the ßERKO mice. This single variant, if
translated, would encode a mutant ERß lacking the first zinc finger
and therefore would be unlikely to transactivate due to an inability to
tightly associate with the chromatin structure within the regulatory
regions of target genes.
This is not the first report of targeted insertions resulting in
aberrant splicing of a disrupted gene. Mice possessing a targeted
disruption of the transforming growth factor- gene produce a
transcript in which the entire exon possessing the disrupting insert is
accurately removed via the conventional donor and acceptor splice
sites, preserving the normal reading frame of the gene (124). Studies
in three different human genes have demonstrated that point mutations
resulting in a premature stop codon can lead to complete excision of
the exon possessing the mutation (125, 126). Furthermore, Reed and
Maniatis (127) used artificial deletions and insertions within exons of
genes that normally display alternative splicing to demonstrate that
the proximity of the acceptor and donor splicing sequences to one
another plays a role in splicing mechanisms. It is possible that
insertion of sequences as large as those used in the ERKO mice may
disrupt the spatial requirements necessary for proper mRNA splicing.
Therefore, the above studies, along with our experiences, are relevant
to the practice of targeting genes by insertion of a large disrupting
sequence possessing internal stop codons.
1. Interpretation of phenotypes in receptor null mice. The use of methodologies to target and disrupt individual genes has created numerous models available for study (reviewed in Refs. 109, 110, 128). Furthermore, this impact has been felt in several facets of the biological sciences. Although it may be initially thought that a particular gene plays no role in the physiology of certain animal systems, such as reproduction or behavior, disruption of the gene and the subsequent phenotypes often prove otherwise. Therefore, transgenic and knockout technologies have spawned a number of collaborative efforts between investigators of varied disciplines that may have never occurred.
An issue that has become apparent from the numerous gene disruption studies and interdisciplinary collaborative efforts is a collection of caveats to be considered when evaluating phenotypical data from a knockout model. These have arisen mostly from the behavioral sciences (reviewed in Refs. 129, 130), but have expanding application to all areas of study provided by transgenic animals. The first of these caveats is one that may be most relevant to the steroid receptor mutant models, i.e., when studying the target tissue of an adult receptor-knockout mouse, one must realize the tissue passed through all the stages of development and "organization" in the absence of the respective receptor. Therefore, this tissue, and in essence the whole animal, cannot be assumed to be truly identical to the wild type in all aspects except for the absence of the targeted gene product. Any genetic redundancies or compensatory mechanisms that took place during development cannot be readily detected or accounted for during most experiments. Therefore, the lack of a phenotype does not necessarily discount the function of the disrupted gene in the physiology being studied. Additionally, it is difficult to distinguish between an organizational vs. activational basis for an observed deficit in a particular physiology when studying the adult mutant. For example, observed resistance to a hormone due to alterations downstream of the function of the disrupted gene may be apparent during adulthood, but may have been imprinted during development.
Other caveats of interpreting data from receptor null mice are founded in the methods used to generate and maintain a line of knockout mice. The standard protocol for generating a knockout mouse involves the incorporation of embryonic stem cells of a 129 strain of mice that carry the desired mutation into the blastocyst of a C57BL/6 strain. The resulting chimeric animals are then often back-crossed to the C57BL/6 strain once again until mice homozygous for the disruption are acquired. Therefore, early generations of knockout mice are composed of a somewhat chimeric genome, especially in the chromosomal regions closest to the targeted locus. This is of special importance in behavioral studies in light of the known variations in the sexual behavior of different strains of mice (131). However, most relevant to the ERKO, significant variations in estrogen responsiveness of the female reproductive tract among the different strains of mice are also known to exist (132, 133). A recent report by Roper et al. (134) has further defined the genetic basis for the variations that exist in the effects of estradiol on classical uterine parameters in mice. These limitations can be overcome to some degree by increasing experimental sample sizes and including parental strains as a control group in all experiments (135). In addition, the various models of hormone and steroid receptor deficiency that are now available no longer make necessary the complete interpretation of data from any one model. Therefore the data from these models, when interpreted as a whole, should prove invaluable to elucidating the roles of the different sex steroid receptors in both development and adult physiology.
|
III. Reproductive Tract Phenotypes of the Female |
|---|
|
TopAbstractI. Introduction—A...II. Estrogen ReceptorsIII. Reproductive Tract...IV. Mammary GlandV. Reproductive Tract Phenotypes...VI. Neuroendocrine SystemVII. Phenotypes in Peripheral...VIII. Comparison with Human...IX. SummaryReferences |
|---|
ERKO and ßERKO female mice exhibit a properly
differentiated female reproductive tract possessing the constituent
structures (46, 47). However, estrogen insensitivity has severely
disrupted sexual maturation of the whole reproductive tract in the
ERKO female and ovarian function in the ßERKO female. The
consequences of ER gene disruption on the individual components of the
female reproductive tract is the topic of this portion of the review.
Before we continue, we believe it is necessary to briefly reiterate
those studies carried out to verify successful targeting of the ER
gene in the ERKO. This discussion is appropriate for this portion of
the review because the majority of these experiments were performed on
uterine tissue. To determine the effectiveness of the gene targeting,
Western blots of adult ERKO uterine nuclear and cytosolic extracts
were probed with the H222 antibodies, a rat monoclonal antibody
specific to the ligand-binding domain of the human ER (10). Our
studies, as well as those of others, have demonstrated that this
antibody possesses high cross-reactivity to the mouse ER (139, 140, 141).
These assays detected no wild-type ER or any other immunoreactive
fragments unique to the ERKO uterus. Similar results were obtained
when blots were probed with the rabbit antiserum ER-21, directed toward
the 21 amino-terminal residues of the rat ER (141). However, binding
assays using 3H-E2 on ERKO uterine extracts
indicated the presence of high-affinity binding of the hormone at
levels approximately 3–9% of the wild type (123). In agreement with
these data, sucrose gradient analysis with
3H-E2 on low-salt cytosol extracts from ERKO
uteri indicated a binding factor with an 8S sedimentation value,
similar to that of the wild-type ER (123). The discovery of the
ERß, reported approximately 3 yr after the generation of the ERKO,
prompted a renewed assessment of this ERKO estrogen-binding data in
several publications. Unfortunately, in a number of these reports, the
original datum discussed above is not evaluated in full, and the
authors elude to ERß as the likely binding source in the ERKO
uteri. Certainly at the time of the initial characterization, concern
over the residual level of binding in the ERKO uteri was often mixed
with the wonder of possibly discovering an unknown ER. However, during
these studies we also demonstrated that when the H222 antibodies were
included in the sucrose gradient assays, the estradiol binding peak in
the ERKO uterine extract was shifted accordingly (123). The H222
antibodies have been shown by us, as well as by other laboratories, to
be ER specific and unable to recognize ERß by Western blot
analysis or immunohistochemistry (142). As described earlier in this
review, our RT-PCR analysis on mRNA from ERKO uteri demonstrated the
presence of a splicing variant of the disrupted ER gene that could
encode a mutant ER possessing both the ability to bind estradiol as
well the H222 epitope (123). Furthermore, we have recently shown that
ERß mRNA is undetectable in the uteri of adult wild-type as well as
ERKO mice when assayed by ribonuclease protection assay (93).
Therefore, we believe that relatively conclusive data have been
generated to indicate that the estradiol-binding factor present in
ERKO uteri is most likely not ERß.
A. Uterus
1. Uterine phenotype and estrogen insensitivity. The ER has
been detected by steroid autoradiography and immunohistochemical
methods in the ductal structures of the rodent female reproductive
tract during several stages of development, including the late fetal
and neonatal stages through puberty and adulthood (reviewed in Refs.
112, 143). Several reports describe the initial appearance of ER
immunoreactivity in the developing uterus as early as fetal day 15
(112, 143). ER immunoreactivity was first detectable in mesenchymal
cells, whereas induction in the epithelial cells occurs during the late
fetal stages and increases significantly during the neonatal period
(112, 143). The fully developed uterus is composed of many
heterogeneous cell types comprising three major anatomical
compartments, the outer myometrium, endometrial stroma, and
luminal/glandular epithelium. In the immature CD-1 mouse, ER
immunoreactivity is easily detectable in the stroma on day 1 and
continues to rise to a maximal level on day 10, whereas the appearance
of epithelial ER is delayed and reaches a peak around day 16 (144).
Other reports indicate variations in the exact timing of the appearance
of ER among different strains and species, most likely reflecting
temporal deviations in development (112, 143).
The presence of an intact estrogen-signaling system appears to coincide
with the appearance of ER. In several species, estrogen treatment of
fetal and neonatal females results in the stimulation of increased
uterine levels of nucleic acid (136, 145), protein synthesis (146),
ornithine decarboxylase (147), progesterone receptor (148), and
cellular proliferation (145, 149, 150). However, a full biological
response to estradiol in terms of maximum increases in uterine weight
is not possible in the neonatal uterus, and can be observed only after
the animal approaches weaning age (146). Furthermore, significant
differences in the uterine response to estradiol between the neonate
and sexually mature rodent are known (151). For example, estrogen
stimulates cellular proliferation in all tissues of the immature
uterus, whereas this response becomes limited to the epithelial
compartment during adulthood (151, 152). Therefore, sexual maturation
of the uterus is not simply marked by the presence of ER, but rather
the acquisition of the capacity to undergo the correct synchronized
phases of proliferation and differentiation elicited by the
ovary-derived sex steroids.
As shown in Fig. 3, the uteri of both
adult ERKO and ßERKO females possess all three definitive uterine
compartments, the myometrium, endometrial stroma, and epithelium.
However, in the ERKO, each is hypoplastic and results in whole
uterine weights that are approximately half that recorded for wild-type
littermates (46). In contrast, the uteri of adult ßERKO females
appear normal and able to undergo the cyclic changes associated with
the ovarian steroid hormones (47). Therefore, perinatal development of
the female reproductive tract in the mouse appears to be independent of
ER and ERß actions. However, estrogen responsiveness and
subsequent sexual maturity in the uterus has been ablated by disruption
of the ER gene. The ERKO endometrial stoma is characterized by a
less organized structure and hypotrophy, with a sparse distribution of
uterine glands compared with that of the wild type (153). Luminal and
glandular epithelial cells in the ERKO uterus most often appear
healthy, but are consistently cuboidal and lack the normal
"estrogenized" morphology of a tall columnar shape and basally
located nucleus (Fig. 3). This phenotype is interesting in light of the
increased levels of estradiol found in the serum of adult ERKO
females (Table 2). However, Lindzey
et al. (154) demonstrated that the concurrently elevated
ovary-derived androgens in the ERKO female (Table 2) do provide for
some maintenance of uterine weight, which can be further reduced upon
ovariectomy. We recently reported that ERß mRNA is barely detectable
in the adult mouse uterus, including those from ERKO mice (93),
making it unlikely that ERß could provide a compensatory role in the
ERKO uterus. Numerous immunohistochemical studies for ER and the
apparent loss of estrogen sensitivity in the ERKO uterus indicate
that the classical ER is the predominant form responsible for
mediating estrogen actions in the mouse uterus.
|
|
ERKO females
fail to exhibit components of both phases after estrogen treatment,
providing strong evidence for the requirement of ER in the full
response (46, 123). In brief, when treated with 40 µg E2
or diethylstilbestrol (DES) per kg body weight for three consecutive
days, wild-type mice exhibited the expected 3- to 4-fold increase in
uterine wet weight, whereas no such response was observed in the uteri
of ERKO mice (46, 157). It should be noted that this pharmacological
dose of estrogen is well beyond that required to achieve a maximum
response in the wild-type rodent. Nonetheless, estrogen-treated ERKO
uteri exhibited no apparent components of the initial phase of estrogen
effects, including water imbibition and hyperemia. Histological
analysis and [3H]thymidine incorporation assays indicated
a lack of significant cellular proliferation and DNA synthesis in uteri
from the estrogen-treated ERKO mice (123, 153). Interestingly,
although the heterozygous females possess approximately one-half the
normal complement of ER, their uterine response to estrogens is
equal to that of the wild-type females. In a similar study, wild-type
and ERKO mice treated with hydroxy-tamoxifen (1 mg/kg) produced
comparable results (157), eliciting the expected estrogenic response in
the wild-type and having no effect on the ERKO uterus. These studies
thereby confirm that the estrogen agonist activity of
hydroxy-tamoxifen, which is somewhat unique to the mouse uterus (8), is
mediated via the ER pathway.
The mitogenic and stimulatory action of estradiol in the uterus is a
complex process involving increased RNA polymerase and ribosomal
activity (158), resulting in the regulation of a plethora of genes. It
is well accepted that the ligand-bound ER complex is not directly
involved in the mediation of all responses elicited by estrogens in the
uterus, but rather serves as a stimulus for a cascade of signaling
pathways that act to amplify the estrogen action. However, certain
genes appear to be directly regulated by the ER-estradiol complex
and possess functional estrogen-responsive elements within their
regulatory regions. Two such examples are the genes encoding the
progesterone receptor (PR) (159, 160) and the secretory protein,
lactoferrin (161). In fact, the regulation of the uterine PR and
lactoferrin genes have often been used as assays for the estrogenic
activity of experimental compounds. Therefore, with a similar intent,
we used these estrogen markers to attest for estrogen insensitivity in
the uteri of the ERKO mouse. A single dose of estradiol, known to be
effective in inducing the PR and lactoferrin genes within 24 h in
uteri of wild-type mice, produced no such up-regulation in the uteri of
the ERKO mice, confirming the need for a direct action of the ER
in this mechanism (123). Interestingly, a recent report by Tibbetts
et al. (162) demonstrated that the estrogen-stimulated
increases in PR are localized to the stromal and myometrial
compartments, whereas the increases in lactoferrin are isolated to the
luminal and glandular epithelium in the mouse uterus. Therefore,
disruption of the ER gene has resulted in estrogen insensitivity in
all three anatomical compartments of the uterus. However, it must be
noted that constitutive levels of PR and lactoferrin mRNA are present
in the ERKO uteri, suggesting that these genes are also under the
influence of pathways independent of ER. A testimony to the
complexity of estrogen action in the uterus is the finding that while
estradiol up-regulates PR expression in the myometrium and stroma, it
simultaneously abolishes PR levels in the luminal epithelium (162).
This would indicate an inhibitory role of the estradiol-ER complex
on PR expression in this portion of the uterus. Speculating that this
pathway may therefore be lacking in the ERKO uterus, an
investigation as to the source of the PR mRNA in the ERKO uteri is
warranted.
2. Changes in growth factor functions. A component of the
cascade of events that lead to the obvious changes in the physiology of
the adult uterus after estrogen exposure are the auto- and paracrine
actions of polypeptide growth factors. Several members of the epidermal
growth factor family have been suggested as possible mediators of
estrogen-induced mitogenesis in the uterus. This hypothesis is based on
experiments demonstrating that estradiol up-regulates the uterine
levels of epidermal growth factor and its receptor (EGF, EGF-R) (163, 164), transforming growth factor- (165), and insulin-like growth
factor-I (IGF-1) (166). Furthermore, mice homozygous for a targeted
disruption of the EGF-R gene exhibit a hypoplastic uterus that is
significantly reduced in size (167), similar to that of the ERKO.
Experimental data indicate that treatment of ovariectomized wild-type
mice with EGF mimics the early effects of estradiol and DES in terms of
inducing modified cell morphology and increases in the levels of ER,
DNA synthesis, phosphatidylinositol turnover, PR, and lactoferrin in
the uterus (168, 169, 170). Further studies have illustrated that
cotreatment with anti-EGF antibodies was able to attenuate the uterine
response to estradiol, presumably due to inactivation of the
EGF-signaling pathway (168). In turn, cotreatment with the estrogen
antagonist ICI-164,384 was able to reduce the uterine response to EGF
(169). These in vivo studies suggest a cross-talk mechanism
between the EGF and ligand-independent ER- signaling pathways. The
results of the animal studies have been supported by numerous in
vitro experiments demonstrating ligand-independent activation of
the nuclear signaling pathway of the ER, possibly via altering the
phosphorylation pattern of the ER (reviewed in Ref. 55). The
culmination of these and several other studies has led to the proposed
model in which the mitogenic actions of estradiol in the rodent uterus
appear to be at least partially mediated by EGF; however, in turn the
mitogenic effects of EGF require the presence of ER.
Therefore, the ERKO female provides an excellent in vivo
model to study this cross-talk between the ER- and EGF-signaling
systems in the uterus. The uteri of ERKO females possess wild-type
levels of functional EGF and EGF-R (170). Nonetheless, the mitogenic
actions and induction of estrogen-responsive genes elicited by EGF in
the wild-type uterus have been ablated in the ERKO, confirming the
interaction of these two signaling systems (170). However, not all EGF
responses are lacking in the uteri of ERKO females, as this same
study demonstrated that the mechanisms for EGF-mediated up-regulation
of the c-fos gene remained intact (170). These studies have
thereby confirmed the need for functional ER for the mitogenic
actions of EGF in the uterus.
Cunha et al. has extended the use of the ERKO mouse to
investigate the intersecting roles of ER-mediated estrogen
stimulation and growth factors in the uterus through a series of tissue
recombination experiments. The observation of estrogenic effects in
wild-type uterine epithelial cells that are apparently lacking ER
has prompted numerous investigations to illustrate a role for paracrine
factors secreted by the underlying ER-positive stromal compartment,
and thereby mediating the epithelial response (143). These studies have
been advanced by methods that provide for the delicate construction of
tissue recombinants, in which uterine stoma and epithelium are
enzymatically disassociated and recombined with similar tissue from
uteri from animals of different treatments or models to ultimately
regenerate a chimeric stromal-epithelial unit (reviewed in Ref. 143).
These tissue recombinants are implanted under the kidney capsule of
ovariectomized nude mice, which are then acutely treated with estrogen
agonists or antagonists. Later removal of the recombinant grafts allows
for the evaluation of certain end points of estrogen action in each
portion of the recombinant. Cooke et al. (171) described
experiments in which wild-type (ER+) uterine stroma were recombined
with ERKO (ER-) uterine epithelium and vice versa.
The results of these studies illustrate that proliferation of the
epithelial portion of the recombinant was possible only when ER+
stroma were present and did not require ER in the epithelium (171).
Interestingly, similar recombinant experiments using tissue from the
EGF-R knockout mice illustrated that the estrogen-signaling pathways
required for stimulation of the stroma
