| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Yale University School of Medicine, Division of Reproductive Endocrinology, New Haven, Connecticut 06520-8063
Correspondence: Address all correspondence and requests for reprints to: Hugh S. Taylor, Yale University School of Medicine, Division of Reproductive Endocrinology, 333 Cedar Street, P.O. Box 208063, New Haven, Connecticut 06520-8063. E-mail: hugh.taylor{at}yale.edu
 |
Abstract |
|---|
 Top Abstract I. IntroductionII. HOX Gene Structure...III. Regulation of HOX...IV. Endocrine Regulation of...V. Endocrine Disruption of...VI. Targets of HOX...VII. HOXA10 Transcriptional...VIII. ConclusionReferences |
|---|
|
I. Introduction |
|---|
|
TopAbstractI. IntroductionII. HOX Gene Structure...III. Regulation of HOX...IV. Endocrine Regulation of...V. Endocrine Disruption of...VI. Targets of HOX...VII. HOXA10 Transcriptional...VIII. ConclusionReferences |
|---|
Although Hox genes have a well-characterized role in cell fate specification, little is known about their regulation or mechanism of action. Recently, members of the nuclear receptor superfamily and their cognate ligands have been shown to regulate Hox gene expression in both embryonic and adult tissues. Endocrine regulation of Hox genes during embryogenesis determines the axial expression of various homeobox genes, necessary for body axis formation. In the adult, endocrine-regulated Hox gene expression is necessary for functional differentiation.
Hox genes are ubiquitously expressed and likely regulate many developmental programs. This review discusses the role of hormone-driven Hox signal transduction pathways during both embryonic and adult development, as well as the consequences of disrupted endocrine signaling. Although homeobox genes likely activate a plethora of target genes, only those Hox targets that serve as effectors in well-defined endocrine-Hox gene signaling pathways are specifically discussed here. An understanding of the developmental roles of Hox genes, their endocrine regulation, and consequent effects on pathophysiology offers potential for development of novel targeted therapies.
|
II. HOX Gene Structure and Function |
|---|
|
TopAbstractI. IntroductionII. HOX Gene Structure...III. Regulation of HOX...IV. Endocrine Regulation of...V. Endocrine Disruption of...VI. Targets of HOX...VII. HOXA10 Transcriptional...VIII. ConclusionReferences |
|---|
Their nomenclature derives from a shared, highly conserved 183-bp sequence known as the homeobox, which encodes a 61-amino acid domain, termed the homeodomain. Structurally the homeodomain comprises a helix-turn-helix motif that binds target gene enhancers using binding sites containing the consensus core sequence 5'-TAAT-3' (5, 6). Because HOX genes encode transcription factors, such binding results in target gene activation or repression. There is a high degree of homology between vertebrate Hox genes and Drosophila homeobox genes that is based mainly on sequence similarity in the homeodomain, although many Hox genes have conserved sequences outside this region (7).
The genes of the HOM-C are organized along the chromosome in an order that parallels their expression along the anterior-posterior body axis; this relationship is known as colinearity (8). (Fig. 1
). Genes expressed at the 3' end of the cluster are expressed earlier during embryogenesis and in more rostral body segments compared with genes at the 5' end, which are expressed later and in more caudal body segments. For example, the most 3' gene in the HOM-C, labial, is expressed most anteriorly in the developing head, mainly in the hypopharyngeal organ and the mandibular lobe. It is also expressed earliest during development (9). It is likely that the clustered arrangement of Hox genes directs temporal and spatial expression and also serves as a mechanism enabling homeobox genes to convey regional body patterning.
|
HOX genes act as selector genes in that their expression within a certain body segment determines one particular pathway of development over another (3). The identity of an individual body segment is largely determined by the combination of HOX genes expressed. The segment-specific combination of HOX genes is known as the HOX code, and it is the genetic basis that determines the phenotype of nascent tissues. When a homeobox gene is mutated, the body segment where it is normally expressed typically develops characteristics of the segment immediately anterior to it, an effect known as anterior transformation. Although the anterior limit of expression of each Hox gene is clearly demarcated, posteriorly the expression levels gradually decline. The anterior limit of expression of the adjacent 5' Hox gene overlaps this fading posterior expression of its 3' precedent (Fig. 1
). When a Hox gene is mutated, the trailing expression gradient of the adjacent 3' Hox gene influences differentiation within the domain of the mutant gene so that the phenotype resembles that determined by the expressed immediately anterior 3' gene. The end result is a shift in phenotype toward an anterior structure. Occasionally, the gene posterior to the mutated gene dominates function, resulting in a posterior transformation.
In contrast to Drosophila, where mutation of a single homeotic gene results in a dramatic phenotypic transformation, in vertebrates targeted mutation in a single Hox gene, usually causes only a subtle transformation to resemble more anterior tissues (10, 11). This is because, compared with the eight genes of the Drosophila HOM-C complex, the 39 mammalian homeotic genes confer genetic duplication. Loss of function of a single mutated gene may be compensated for by a paralog (12). Functional redundancy also occurs between genes that are adjacent to each other in the same cluster and therefore have overlapping domains of expression (13). Therefore, in mammals, double and triple mutants of paralogous genes result in more severe transformation or absence of a structure normally present within the expression domain of the paralogous genes (14, 15).
B. Developmental function
Although there is a relative lack of naturally occurring murine and human Hox gene mutations, targeted disruption of HOX gene loci have revealed a consistent role in development and patterning of the body axis during embryogenesis. In Drosophila, for example, loss of function of the 3' HOM-C gene labial results in a failure of embryonic head involution, manifest as disorganization of such cranial structures as the salivary glands and the cephalopharyngeal apparatus. In adults, a posterior transformation of the dorsal head to a more thoracic identity is observed (16). Similarly, in Xenopus, misexpression of a posteriorly expressed homeobox gene (XIHbox 6) caused respecification of anteroposterior identity, by transforming head mesoderm into tail-inducing mesoderm (17). Developmental transformations are also seen in mice after targeted disruption of a Hox gene.
The endocrine regulation of HOX gene expression is best characterized in the reproductive tract, and this review will focus primarily on those tissues. The differentiation of the embryonic anlagen of the male and female reproductive tracts is gender-specific. In the male, the presence of a Y chromosome is associated with the expression of testicular determinant factors such as the sex-determining region Y gene (SRY), which directs differentiation of the fetal gonads into testes; absence of testicular-determining factors results in ovarian differentiation (18). In turn, the fetal testes secrete a Müllerian-inhibiting substance to suppress development of female internal genitalia and testosterone, which directs mesonephric duct differentiation into male genital structures such as the epididymis, vas deferens, and seminal vesicles (19). In the female, in the absence of testicular testosterone and Müllerian-inhibiting substance, the paramesonephric duct differentiates into the fallopian tube, uterus, cervix, and upper vagina. Evidence for the role of HOX genes in reproductive tract development is apparent in the phenotypes of mice deficient for the caudally expressed Hox genes Hoxa10 or Hoxa11 (20, 21).
In wild-type male mice, the epididymis is a single, coiled tubule comprised of three regions craniocaudally: the caput, corpora, and cauda. The ductus deferens arises from the cauda as a wide, straight tubule. In Hoxa10(–/–) mice, anterior transformation is manifest as cranialization of the corpora to a more caput-like phenotype, as well as of the proximal ductus deferens to a more caudal-like phenotype (22). A similar phenotype is also seen in Hoxa11 (–/–) male mice (21). Likewise, in female Hoxa10 (–/–) mice, the proximal portion of the uterus is transformed into a narrower tubular structure that morphologically resembles and transitions into the adjacent oviduct (22). As expected, deficient expression of a single mammalian Hox gene results in subtler developmental defects than are seen in Drosophila (22, 23). As discussed above, this is likely due to genetic redundancy from expression of paralogous Hox genes within the same domains of expression. However, as discussed above, in keeping with their collinear pattern of expression, the observed phenotypic defects in mice are nevertheless suggestive of anterior transformation, as seen in Drosophila.
The linear pattern of expression of the Abd-B orthologous Hox genes along the murine paramesonephric duct changes temporally with differentiation of the female reproductive tract. On embryonic d 15.5 through 19, the paramesonephric duct is morphologically undifferentiated and homogenous in appearance. Corresponding to this undifferentiated state, a uniform pattern of Hox gene expression is observed along the length of the paramesonephric duct, including the genes Hoxa9, Hoxa10, Hoxa11, and Hoxa13 (24, 25). In contrast, at postnatal wk 2 of age, corresponding to the time of maximal differentiation of the female reproductive tract, the previously uniform Hox gene expression becomes spatially restricted along the length of the paramesonephric duct. As expected, the spatial domain of expression of each of the Hox genes in this group corresponds craniocaudally with their chromosomal 3' to 5' order.
Accordingly, Hoxa9 is expressed in the rostral oviducts, but not in the more caudally differentiating uterus, cervix, or vagina. Hoxa10 is expressed in the relatively more caudal uterine epithelium, stroma, and muscle. Hoxa11 is expressed in the uterus as well as extending further caudally in the cervical glands and squamous epithelium, whereas Hoxa13 expression is caudally primarily restricted to vaginal epithelium (24).
C. Adult function
1. Endometrial receptivity.
Although the HOX genes have a well-characterized role in embryogenesis and anteroposterior axis specification, the function of their expression in the adult is still the subject of active investigation. We have proposed that HOX genes enable the retention of developmental plasticity in certain tissues that continue to undergo rapid developmental changes in the adult. The cells of adult tissues such as the female reproductive tract and hematopoietic system exhibit a cyclic turnover, where regenerating stem cells progressively achieve functional capability through terminal differentiation. In contrast to cell replacement and repair that commonly occur in many adult cell types, cyclic uterine endometrial regeneration involves simultaneous proliferation and differentiation of multiple cell lineages such as epithelium, stroma, and endothelium. Such coordinated development among the constituent cell types results in a tissue that is functionally capable of allowing embryo implantation during the narrow window of receptivity; failure of implantation results in apoptosis and degeneration, followed by another cycle of programmed development. In tissues such as the endometrium, the process of terminal differentiation is similar to cell fate specification during embryogenesis, wherein structurally and functionally independent tissues arise from undifferentiated cells. Hox genes have a well-characterized role in cell fate specification; commensurate with this property, these genes likely regulate adult functional differentiation through similar mechanisms as used in development (26, 27, 28, 29). Furthermore, in adult tissues, as in embryogenesis, HOX genes are regulated by several hormones acting via their cognate receptors. Here, we consider the function of HOX genes in the hematopoietic and female reproductive systems wherein they function as part of endocrine signal transduction pathways, mediating functional differentiation.
In the adult female reproductive tract, Hoxa10 and Hoxa11 have evolved a unique temporal pattern of expression consistent with their role in functional differentiation of the endometrium. Their temporal and spatial expression patterns in the adult are distinct from those during paramesonephric tract differentiation, discussed in the previous section (20, 22, 28). During each reproductive cycle, adult tissues such as the uterine epithelium and stroma display a well-defined pattern of functional differentiation that is necessary for successful pregnancy. Initial proliferation is followed by differentiation leading to a receptive state for embryo implantation. In the absence of implantation, however, apoptosis and degeneration of the uterine endometrium is observed, followed by the onset of a new reproductive cycle. From the phenotype of animals with sex steroid receptor loss of function mutations, it is evident that both estrogen and progesterone signaling is necessary for the physiology of the reproductive cycle (30, 31). Hox genes mediate some functions of sex steroids during each reproductive cycle.
HOXA10 and HOXA11 demonstrate a dynamic temporal pattern of expression in adult uterine epithelium through the reproductive cycle (28, 29). HOXA10 and HOXA11 mRNA are each expressed in human endometrial epithelium as well as stroma, throughout the menstrual cycle, with significantly higher levels of expression in the mid- and late secretory phases (28, 29, 32). The midsecretory phase coincides with the time of embryo implantation, histological peak differentiation, and high systemic levels of estrogen and progesterone. Elevated endometrial levels of expression of HOXA10 and HOXA11 therefore correspond with peak functional differentiation in this tissue. In the absence of pregnancy, endometrial senescence and breakdown follow this phase, accompanied by a decline in estrogen and progesterone levels, resulting in menstruation and the onset of a new reproductive cycle.
In the event of a successful pregnancy, the decidua of early pregnancy continues to express high levels of HOXA10 and HOXA11 mRNA (28, 29). Likewise, in pregnant mice, Hoxa10 mRNA expression is first detected in the endometrial epithelium on d 0.5 post coitum (p.c.). Although the initial site of endometrial Hoxa10 expression is limited to the epithelial layer, stromal expression appears on d 2.5 p.c., and by d 3.5 p.c. expression is limited to the stroma (20). As observed in humans, persistent high levels of expression of Hoxa10 mRNA are detected in the decidua. A dynamic temporal and spatial pattern of Hox/HOX gene expression in adult reproductive tissues likely parallels ongoing functional differentiation and appears to be evolutionarily conserved across vertebrate species (20, 28).
A functional consequence of aberrant adult endometrial Hoxa10 expression is evident as defective implantation, seen in Hoxa10 (–/–) mice. Female Hoxa10 (–/–) mice demonstrate uterine factor infertility due to preimplantation and implantation defects (20). Implantation site defects are common and frequently consist of hemorrhage within the site itself, as well as in the adjacent uterine lumen. Also observed are small implantation sites with disorganized embryos and empty decidua suggestive of early degeneration of a postimplantation embryo. Furthermore, the size of the decidual swellings is reduced compared with wild-type uteri (22). These implantation defects are implicit to Hoxa10 (–/–) females irrespective of the genotype of the embryo. Similar implantation defects with deficient endometrial stromal, glandular, and decidual cell development in early gestation are also seen in Hoxa11 (–/–) female mice (32).
Hox genes interact with other key developmental signaling molecules during development of reproductive competence. Recently, cross talk has been demonstrated between Wnt signaling and Hoxa10 in the adult endometrium during the peri-implantation period (33). Wnt7a has a role in anteroposterior delineation of the female reproductive axis during development (34). It is likely that combined signaling of a uterine domain-specific morphogen such as Hoxa10 with signaling molecules necessary for boundary determination such as the Wnt proteins may be necessary to modulate embryo-decidual interactions during implantation. In murine reproduction, implantation commences at the antimesometrial pole and proceeds toward the mesometrial end. Hox-Wnt signaling has been proposed to be necessary in establishing an antimesometrial-mesometrial orientation of the implantation chamber (33). In turn, these interactive pathways recruit other genes that may be necessary for implantation, such as Msx-1 (33).
As discussed previously, expression of Hox genes is necessary for normal embryonic development. Mice with targeted disruptions in these genes have embryonic developmental defects and manifest reproductive anomalies. However, the role of Hox genes in adult tissue differentiation is difficult to discern in these mutant animals because the preexisting embryonic developmental defect would override and prohibit analysis of the role of these genes in adult terminal differentiation. To determine the necessity of Hoxa10 expression in adult uterine function, uteri of wild-type mice were transfected in the peri-implantation period with constructs that altered Hoxa10 expression levels (35). Sequence homology between murine Hoxa10 and human HOXA10 permitted the use of constructs containing the human ortholog in murine transfections. In vivo uterine transfection with a HOXA10 antisense construct was used to decrease endometrial Hoxa10 expression levels, whereas a full-length HOXA10 cDNA construct was used to increase Hoxa10 expression. A significant reduction in litter size in antisense transfected mice (mean number of pups = 6.5) compared with controls (mean = 13.3) was observed. In contrast, mice transfected with HOXA10 cDNA consistently delivered large litters (11–14 pups) compared with controls (4, 5, 6, 7, 8, 9, 10, 11). A clear role for Hox genes in both embryonic development and adult tissue plasticity was established.
Although the functional implications of deficiency or overexpression of endometrial HOXA10 are clear, the histological appearance of the endometrium in both cases is relatively unchanged. A difference in ultrastructural morphology is, however, evident in Hoxa10-sufficient compared with Hoxa10-deficient mice (36). Pinopods are endometrial epithelial surface projections that are markers of the window of endometrial receptivity (37, 38, 39, 40, 41, 42, 43, 44). HOXA10 antisense treatment diminishes pinopod number, whereas an increase is observed when uterine HOXA10 expression is up-regulated. Similar changes are noted in the epithelial microvilli. The direction of alteration in pinopod and microvilli number therefore parallels that of HOXA10 expression levels and correlates with implantation efficiency. Pinopod and microvilli development may therefore represent a subcellular morphological feature of HOXA10-induced endometrial functional differentiation. Although striking functional phenotypes may be observed after targeted disruption of Hox genes, the mechanism may be apparent only at the subcellular level.
Several human diseases affecting the female reproductive tract lead to increased infertility, implantation defects, and reproductive wastage (45). Deficient endometrial HOXA10 expression has been demonstrated in these conditions and may be a common mechanism resulting in the implantation defect phenotype (46, 47, 48). The correlation between tissue HOXA10 expression levels and implantation is further evident in ectopic pregnancies where aberrant high levels of HOXA10 mRNA are evident in the tubal mucosa of ectopic gestations, specifically at the implantation site (49). There is a common molecular mechanism involving Hox genes that results in embryo implantation.
Human infertility in some instances may be due to defective endometrial receptivity. Most cases of infertility suspected to be due to defective endometrial development do not show histological abnormalities; it is likely that subcellular and molecular defects are responsible for these conditions. Few genes are known to be necessary for endometrial receptivity. A better understanding of these genes may elucidate the mechanisms of some forms of human infertility.
Hoxa10/HOXA10 and Hoxa11/HOXA11 expression is regulated by endogenous estrogen and progesterone in the reproductive tract. As discussed later, these Hox genes act downstream of estrogen and progesterone to regulate endometrial function; consequently, aberrant expression of either sex steroids or HOX genes, as observed in several diseases affecting the reproductive tract, is likely associated with endometrial functional deficiency manifest clinically as implantation defects.
The paralogs HOXC10, HOXC11, HOXD10, and HOXD11 are also expressed in the endometrium, overlapping the spatial domain of expression of the genes HOXA10 and HOXA11. However, the expression is limited to stromal cells, and is not responsive to sex steroids (50). Furthermore, as described in other cell lines, the paralog HOXC10 is likely more closely associated with proliferation than differentiation in the endometrium (51, 52, 53). In keeping with its mitogenic role, endometrial expression levels are higher in the proliferative phase than in the secretory phase of the reproductive cycle (50).
2. Pregnancy.
In addition to their well-characterized role in functional differentiation of the adult endometrium and embryo implantation, endocrine–Hox gene signaling has recently been implicated in pregnancy as well. We have previously described adult endometrial HOXA10 expression and its regulation by estradiol as well as progesterone. It is the combined action of both estradiol and progesterone that results in high HOXA10 mRNA levels in the late secretory phase of the menstrual cycle (28). In the event of a successful pregnancy, persistent high HOXA10 mRNA expression levels are observed in the decidua (28).
Pregnancy is characterized by high systemic progesterone levels, which likely mediates uterine quiescence and prevents the premature onset of labor. Not surprisingly, the progesterone-HOXA10 relationship endures beyond the phase of endometrial receptivity, embryo implantation, and early pregnancy associated with prevalent high systemic sex steroid levels. HOXA10 and HOXA11 are expressed in term human decidua (54). It is therefore likely that the HOX genes subserve a role of progesterone for the duration of gestation.
A well-characterized role of progesterone is its ability to prevent premature onset uterine contractions and thereby allow pregnancies to reach term. The initiation of labor in many vertebrate animal species occurs as a result of progesterone withdrawal (55, 56, 57). Overt progesterone withdrawal, however, has not been demonstrated to signal the onset of labor in humans.
Intraamniotic infection (chorioamnionitis) and placental abruption frequently underlie preterm uterine contractions and often result in preterm delivery. Chorioamnionitis is accompanied by inflammation at the maternal-fetal interface, resulting in increased decidual and fetal membrane expression of proinflammatory cytokines such as IL-1ß (58). Placental abruption is associated with enhanced decidual thrombin generation (59). As at term, systemic progesterone levels remain unchanged in the setting of early onset uterine contractions. In the absence of decreased systemic progesterone levels, IL-1ß and thrombin directly decrease the endometrial expression of HOXA10. Treatment with thrombin and IL-1ß resulted in 86 and 72% decreases, respectively, in HOXA10 expression levels.
HOXA10 is associated with the development and maintenance of the decidua, which in turn supports the developing embryo. Until fetal development is completed at term, the progesterone-HOXA10 axis is necessary for the prolongation of pregnancy. Although in other species progesterone withdrawal signals the onset of labor, in humans HOXA10 down-regulation may transform the uterine environment into one characterized by progesterone resistance, in effect mimicking the progesterone withdrawal seen in other species and thereby leading to the onset of labor.
In keeping with this hypothesis, Hoxa10 (–/–) mice demonstrate progesterone resistance, demonstrating inappropriate regulation of such progesterone-regulated genes as the prostaglandin E receptors EP3 and EP4, as discussed above (60). The dramatic decrease of HOXA10 in response to IL-1ß or thrombin, with consequent activation of downstream endocrine signal transduction pathways such as those involving prostaglandins, has implications for the onset of labor.
3. Hematopoiesis.
As in the female reproductive tract, Hox genes function in endocrine-driven developmental pathways in the adult hematopoietic system as well. The different cell types constituting the hematopoietic system also undergo constant turnover in the adult organism analogous to that observed in the endometrium. Bone marrow hematopoietic stem cells undergo proliferation and differentiation along several distinct lines to constitute the mixed population of cells found in human blood. HOX genes have well-characterized expression patterns in this system as well (26, 27).
In the hematopoietic system, which lacks a spatial orientation, it is likely that spatial domain-specific expression of HOX genes is instead manifest as lineage-specific HOX expression. Hematopoietic lines each express a distinct set of HOX genes. For example, whereas genes of the HOXA and HOXB cluster are expressed in CD34+ cells, CD34– cells lack expression of genes from these clusters (61). Similarly, HOXC4 demonstrates a lymphoid-specific pattern of expression and has been observed in both T and B lymphocyte cell lines in the bone marrow as well as in peripheral blood. It is, however, not seen in myeloid cells such as granulocytes and monocytes. Likewise, HOXA10 is expressed in megakaryocytes and myeloid precursor cells such as promyelocytes (26, 62).
Specific HOX gene expression correlates not only with specific lineages within the hematopoietic system but also with specific stages of differentiation. The temporal expression pattern of the clustered HOX genes also depends upon their genomic topography. In the hematopoietic system, 3' Hox genes are activated earlier and in more primitive hematopoietic precursor cells than genes located progressively toward the 5' end. Differential patterns of HOX gene expression therefore correspond to cells representing various stages of hematopoietic differentiation. For example, the 3' gene HOXB2 is expressed in the pluripotent progenitor cell, whereas the more 5' gene HOXB7 is not. The opposite expression pattern is however, observed in more committed cells (63, 64). Likewise, HOXB4 is also expressed by the most primitive CD34+ hematopoietic stem cells, being absent from more mature committed progenitor cells (61). HOXB4 induces proliferation and self-renewal capacity of bone marrow stem cells and also the differentiation of downstream progenitor cells along the erythroid pathway (65, 66). Likewise, HOXA10, which is positioned more toward the 5' end of the HOXA cluster, is expressed in later stages of myeloid differentiation such as in promyelocytes, but is absent in more primitive precursors as well as in mature neutrophils and monocytes (67). Interestingly, in acute myeloid leukemia, increased numbers of HOXA10-expressing cells correlate with the increased number of blasts in circulation in this disease.
Undifferentiated cells must maintain a pluri- or totipotent phenotype by suppressing the expression of genes that modulate differentiation. The Polycombs genes regulate formation of chromatin complexes that repress gene transcription. Specifically, members of the Drosophila Polycombs complex and their mammalian orthologs have been found to transcriptionally silence the expression of multiple homeotic genes (68). Transcriptional repression of Hox gene expression is one mechanism by which embryonic stem cells do not initiate homeotic gene-regulated segmental differentiation, but instead retain pluripotency (69). Polycomb gene-mediated repression may also underlie mammalian female X chromosome inactivation (70).
Hox genes are cell fate determinants; differentiation stage-specific expression of each Hox gene programs the cell to appropriate cell development before advancing to the next stage of differentiation. In the myeloid cell line, one mechanism by which differentiation occurs is through sequential expression of previously unexpressed differentiation stage-specific genes and their cognate proteins. Two components of the phagocyte respiratory burst oxidase enzyme, CYBB and NCF2, are transcriptionally repressed by HOXA10 during the earlier stages of myelopoiesis (71). Expression of these two genes is up-regulated later in the development of these cells when HOXA10 expression diminishes. It is likely that the expression of these genes is necessary for functional capability in the differentiated cell.
We have previously described the role of the homeobox gene HOXA9, which is necessary for oviduct development and is also expressed in the adult human fallopian tube. In the hematopoietic system, like the other 5'-located HOX genes, it is also expressed in intermediate and later stages of myeloid differentiation such as in promyelocytes. Up-regulating its expression in these blast cells results in proliferation and leukemogenesis. Interestingly, HOXA9 is also expressed in endothelial cell progenitors as well as in mature endothelial cells and is necessary for their migration as well in tubular structure formation as a precursor to angiogenesis. It does so via regulation of a novel target gene EphB4 receptor (72). Consistent with its role in endothelial cell differentiation and function, Hoxa9 (–/–) mice demonstrate fewer endothelial cells and defects in neovascularization (73). It is likely that HOXA9 plays an analogous role in the female reproductive tract during normal development of the oviduct or in teratogenesis when uterine fundal overexpression of HOXA9 in diethylstilbestrol (DES)-exposed uteri transforms the region into a tubular structure. Like HOXA9, HOXA10 is also expressed in the uterus (with roles in uterine development and embryo implantation) as well as in myeloid progenitor cells (where it regulates the expression of respiratory burst enzymes). In the case of both HOXA9 and HOXA10, their apparently divergent roles in different tissues may be reconciled by the fact that Hox genes are a common mechanism for transmitting both spatial and temporal instructions to developing or maturing cells.
In summary, the adult expression of homeobox genes in certain organ systems permits the retention of developmental plasticity. These systems display ongoing cycles of differentiation that are more complex than simple cell turnover and replacement that is commonly seen in many adult cell types. The complex adult developmental pattern in such tissues as the hematopoietic system and female reproductive tract is necessary for their function. As selectors, HOX genes likely regulate differentiation in adult tissues through molecular mechanisms analogous to those used in embryogenesis. This includes imparting developmental spatial and temporal cues to specific cell types. The clustered genomic organization of HOX genes is designed to allow serial, nested expression that then relays graded spatial and temporal information. As in embryonic development, adult HOX expression is translated into functional signals that direct the ordered progression of differentiation along complex pathways as seen in the hematological and reproductive cells. As discussed further below, the endocrine regulation of homeobox genes also contributes to this control of embryonic as well as adult development.
|
III. Regulation of HOX Gene Expression in Development |
|---|
|
TopAbstractI. IntroductionII. HOX Gene Structure...III. Regulation of HOX...IV. Endocrine Regulation of...V. Endocrine Disruption of...VI. Targets of HOX...VII. HOXA10 Transcriptional...VIII. ConclusionReferences |
|---|
A. Retinoic acid
Retinoic acid has a well-characterized role in the development of the central nervous system during early embryogenesis, wherein it regulates the expression of 3' Hox paralogs. Although Hox genes themselves have no ascribed role in early zygotic cleavage, after differentiation of the three germ layers (gastrulation), anteroposterior axis determination is initiated, with concurrent Hox gene expression. According to the proposed activation-transformation model for early embryogenesis, the embryo is initially programmed for development of anterior structures (79). Initial anterior differentiation of early embryonic cells is followed by transformation of certain cells toward more posterior fates, likely mediated by a signal originating in the posterior mesoderm.
This process can be followed in the murine embryo, by sequential expression of various developmental genes that initiate the formation of the early anteroposterior body axis. Accordingly, in the early murine embryo, the nonclustered homeobox gene Otx2 (murine ortholog of Drosophila orthodenticle), is ubiquitously expressed. With further development, Otx2 expression becomes restricted to the most anterior segment that is associated with future forebrain development. The paralogous Hox genes, Hoxa1 and Hoxb1, are expressed more posteriorly in the region associated with midbrain and hindbrain development (80, 81, 82, 83). Each embryonic segment thus differentiates into unique adult structures; it is likely that discrepant, but domain-restricted expression of specific developmental genes is necessary for normal segmental differentiation. Not only is the segmental expression of specific developmental genes necessary for the development of various organ systems; in addition, the boundaries between each segment need to be accurately defined so as to allow coordinated axial development. One mechanism by which this occurs is through endocrine regulation of developmental gene expression. In the example discussed here, retinoic acid suppresses the expression of Otx2 (84). Additionally, retinoic acid treatment also up-regulates expression of Hoxa1 and Hoxb1 (85). The net effect of retinoic acid treatment therefore recapitulates the differential Hox gene expression pattern seen in early embryogenesis and likely represents the posterior transformation signal in early central nervous system development. This phenomenon is not just limited to murine species, because Xenopus embryos treated with retinoic acid also preferentially initiate the development of more posterior structures manifested by a reduction in forebrain volume, accompanied by a compensatory increase in hindbrain volume (84).
Retinoic acid acts via its cognate receptors to directly activate target genes. The retinoic acid receptors comprise two families, RAR and RXR, both of which are members of the ligand-activated nuclear receptor superfamily. Retinoic acid binds both types of receptors. Each receptor family is further divided into 
-, ß-, and 
-isoforms. The receptors interact to form RAR-RAR or RAR-RXR homo- or heterodimers, respectively. The RAR complex regulates target gene binding to enhancer RARE or RXRE (retinoic acid response element). RAREs have been found in the 5' regulatory regions of three Hox genes (Hoxa1, Hoxb1, and Hoxd4), suggesting that their expression may be responsive to retinoic acid signaling (86, 87, 88, 89, 90). In fact, as discussed above, retinoic acid directly regulates Hoxb1 expression in the posterior region of the embryo (88, 89, 91). Although the expression domain of retinoic acid is relatively posterior in early embryonic development, with further caudal growth of the embryo, these expression domains subsequently localize to a predominantly rostral region in the developed organism. Therefore, during embryonic development retinoic acid predominantly regulates axial expression of 3' Hox genes in rostral embryonic domains where they are necessary for central nervous system development.
Further evidence of a common pathway involving retinoic acid and Hox genes in early anteroposterior axis determination comes from colocalization of expression of Hox genes and RARs, as well as phenotypic similarities observed in animals with targeted mutations in either. retinoic acid-responsive transgene constructs colocalize within the domains of expression of the paralogs Hoxa1 and Hoxb1 in the early developing embryo, which is consistent with their regulatory relationship (92, 93, 94). Because retinoic acid regulates the 3' Hox paralogs via RAR-mediated mechanisms, there are similarities in the observed phenotypes of homozygous retinoic acid receptor mutants as well as loss-of-function Hox gene mutants (2, 95). In both instances, anterior transformations of the cervical vertebrae are seen. The specificity of the retinoic acid-Hox regulatory relationship is further established when either retinoic acid administration or constitutive Hox gene expression also results in a similar phenotype. However, as opposed to the anterior transformation associated with loss of retinoic acid-Hox signaling, augmented signaling of either results in a posterior transformation phenotype.
Retinoic acid regulation of Hox genes is limited to chromosomal 3'-located Hox paralogs that are expressed early in embryogenesis. As discussed previously, the 3' Hox paralogs are expressed early in embryogenesis in relatively anterior regions of the body, such as the head and cervical region. It is therefore likely that retinoic acid is necessary for determination of early anterior-posterior boundaries that are subsequently further subdivided with progressive growth of the embryo (96, 97, 98, 99). With further caudal embryonic growth, the early posterior boundaries defined by retinoic acid-mediated Hox gene regulation become confined to the rostral regions of the embryo. Retinoic acid, however, does not regulate the expression of more 5'-located, posteriorly, and later expressed Hox genes at subsequent phases of embryonic development.
The regulatory relationship between retinoic acid-RAR and Hox genes is dynamic and involves feedback regulation. Retinoic acid induces RARß expression in the presegmented embryo, which in turn induces the expression of specific Hox genes. Interestingly, the homeobox genes Hoxb4 and Hoxd4 that are induced via such retinoic acid-RARß interactions further modulate RARß expression by restricting it to a more posterior location (100). By continuous modulation of the temporal and spatial expression of the nuclear receptor, it is likely that a more graded expression of target Hox genes is achieved. Depending on the expression levels of RAR and its target Hox genes, the location of a definitive segmental boundary may therefore be temporally and spatially coordinated to the overall concomitant development in the region. Coordinated, continuous feedback between Hox genes and RARs is likely essential for axial development.
In addition to its role in central nervous system development, through a process that involves the up-regulation of multiple homeobox genes, retinoic acid also induces differentiation of murine embryonic F9 teratocarcinoma cells into extraembryonic endoderm. In F9 cells mutant for RAR, Hoxb1 expression is not up-regulated, whereas in RAR (–/–) F9 cells Hoxa1 and Hoxa3 are transcriptionally repressed (101, 102). This suggests that multiple Hox genes are differentially regulated by specific RARs in early embryogenesis.
B. Estrogen
Estrogens are necessary for embryonic development of the female reproductive tract. ER-deficient mice have morphologically normal appearing, although hypoplastic, uteri that are unresponsive to estradiol (103, 104). In these ER (–/–) mice, a lack of response to estradiol results in inability to functionally differentiate adult endometrium, causing sterility. The reproductive tracts of ERß mice appear morphologically normal and also have normal responses to estradiol, although females are subfertile (31).
As discussed previously, the paramesonephric ducts in the female embryo differentiate into adult reproductive tract structures. Terminal differentiation of the paramesonephric duct is accompanied by spatially restricted expression of the Hox genes orthologous to Drosophila Abd-B. Consequently, Hoxa9 is expressed in the developing oviduct, Hoxa10 in the uterine anlage, Hoxa11 in the uterine and cervical anlagen, and Hoxa13 in the developing vagina (24). Altered estrogen signaling resulting from exposure of animals to the synthetic nonsteroidal estrogen DES results in developmental anomalies of the reproductive tract. Estrogens may have a role analogous to that of retinoic acid in mediating segmental Hox gene expression.
The reproductive tracts of adult female offspring exposed in utero to DES on d 9 through 16 of pregnancy exhibit abnormal Hox gene spatial expression profiles. A posterior shift in the spatial domains of expression of Hoxa9, Hoxa10, and Hoxa11 is seen throughout the developing female reproductive tract (Fig. 2). Hoxa9 is expressed in the uterus, but not in the oviduct where it is normally expressed. Likewise, a caudal shift in the Hoxa10 expression domain is also observed. The cranial part of the uterus, normally exhibiting robust Hoxa10 expression, shows greatly diminished Hoxa10 expression; it is, however, normally expressed in the caudal part of the uterus. Additionally, Hoxa11 expression is diminished throughout the uterus compared with that observed in untreated wild-type animals. The expression domain and mRNA levels of the most caudal Hox gene, Hoxa13 is apparently unaffected, consistent with its lack of regulation by the sex steroids in vivo (25, 105). Given that the estrogens regulate expression of the 5' Hox paralogs such as Hoxa9, Hoza10, and Hoxa11, which are expressed in posterior and distal domains of the body axis, it may be postulated that by mediating homeobox gene expression in posterior structures, the estrogens function in determining caudal differentiation (105).
|
Estrogen-Hox gene signaling operative in reproductive tract development is ligand-specific. For example, DES, but not 17ß estradiol, causes a perturbation in the domain of expression of each Hox gene, resulting in reproductive tract developmental anomalies. The crystallographic structure of the DES-ER ligand-receptor complex is different from that of 17ß estradiol bound to ER. It is likely that this difference in structure causes differential cofactor recruitment to the complex, thereby resulting in distinct DES-related effects on target tissues (105, 106, 107). Interestingly, in contrast to the decreased uterine Hoxa10 expression seen in DES-treated wild-type mice, uterine Hoxa10 expression is unaffected in DES-treated ER (–/–) mice (108). Furthermore, when exposed to DES, these ER (–/–) mice do not manifest reproductive tract anomalies, indicating that during embryonic development, DES signals through the ER receptor (108). The endocrine regulation of the Abd-B orthologous Hox genes by the estrogens during embryogenesis is necessary for posterior patterning; however, as discussed later, homeobox gene regulation by estrogens likely mediates not only organogenic but also subsequent functional differentiation in the adult reproductive tract.
C. Vitamin D
Vitamin D has a well-characterized role in calcium homeostasis and bone metabolism. It also has a role in reproductive tract and hematopoietic development. Vitamin D is converted in vivo into its active metabolite 1,25-dihydroxycholecalciferol, which binds the vitamin D receptor (VDR) and regulates target genes via ligand-receptor binding to specific vitamin D response elements (VDREs) located in the enhancers of various target gene.
Familial VDR deficiency results in type II rickets. When homozygous VDR (–/–) mice were generated by targeted disruption of the murine VDR locus, the mice not only displayed phenotypic features characteristic of VDR deficiency rickets but also manifested uterine hypoplasia as well as infertility (109). In addition, vitamin D deficiency in the adult results in impaired functional endometrial differentiation with lack of decidualization and consequent implantation defects. As expected, these defects are reversible when the vitamin D signaling pathway is activated through 1,25-dihydroxycholecalciferol treatment, resulting in increased uterine weight and endometrial decidualization (110).
Uterine hypoplasia during embryogenesis generated by VDR mutation and implantation defects observed in adults with deficient vitamin D signaling are similar to the phenotypes observed in loss of function mutations of the homeobox genes Hoxa10 and Hoxa11, raising the possibility of a regulatory relationship. The regulatory region of HOXA10 contains a VDRE and has been shown to be regulated by 1,25-dihydroxycholecalciferol in both hematopoietic and endometrial cells (111).
Hormone-entrained cyclic functional differentiation is implicit to female reproductive physiology. Adult functional differentiation recapitulates embryonic developmental processes. As discussed earlier, expression of developmental genes such as the Hox genes confers a developmental plasticity, enabling reactivation of programmed differentiation in select adult organ systems where it is physiologically necessary. Endocrine regulation of Hox gene expression is operative during both fetal and adult life; hormonal signaling in both of these developmental periods is essential for reproductive capability. It is likely that endocrine-Hox signaling represents a conserved molecular mechanism by which developmental plasticity is retained in the adult. The significance of hormone-regulated homeobox gene expression in the adult is reviewed below.
|
IV. Endocrine Regulation of HOX Genes in the Adult Reproductive Tract |
|---|
|
TopAbstractI. IntroductionII. HOX Gene Structure...III. Regulation of HOX...IV. Endocrine Regulation of...V. Endocrine Disruption of...VI. Targets of HOX...VII. HOXA10 Transcriptional...VIII. ConclusionReferences |
|---|
A. Estrogen
In the previous section, we described the necessity of estradiol for functional differentiation of adult endometrium. ER (–/–) mice are sterile and do not undergo the endometrial differentiation that is essential for embryo implantation (103, 104). We have previously also seen that Hoxa10 (–/–) and Hoxa11 (–/–) mice exhibit aberrant endometrial functional differentiation resulting in infertility (20, 21).
Evidence for estradiol-mediated endometrial HOXA10 expression initially came from observations in adult endometrial cells (28). Primary endometrial cell lines obtained from human subjects were established and treated with varying doses of 17ß estradiol ranging from subphysiological to supraphysiological concentrations. Compared with untreated controls, a significant increase in HOXA10 mRNA expression levels is observed in response to estradiol (28). A dose-dependent increase in HOXA10 mRNA levels is seen across the physiological range of 17ß estradiol concentrations (10–6 to 10–10 M) with no further increase at supraphysiological concentrations. The effect of estrogen on endometrial epithelial cells was evaluated by treating Ishikawa cells with 17ß estradiol in an analogous manner (28). Ishikawa cells are a well-differentiated endometrial adenocarcinoma cell line, which is known to express the estrogen and progesterone receptors and serves as a model of endometrial epithelium (112, 113, 114). Up-regulation of HOXA10 mRNA levels is also observed in this epithelial cell line, which is consistent with that seen in estrogen-treated primary endometrial stromal cells. 17ß estradiol is therefore associated with up-regulation of HOXA10 expression in both adult endometrial epithelium and stroma. Consistent with this regulatory relationship, these findings observed in vitro in endometrial cells were also corroborated in vivo, where the endometrial expression of Hoxa10 parallels rising estradiol concentrations. High endometrial levels of expression of HOXA10 in the latter half of the menstrual cycle coincide with coincident high systemic estradiol concentrations (28).
Pretreatment with cycloheximide did not abrogate the increase in HOXA10 mRNA levels in response to treatment with 17ß estradiol either in vitro in endometrial cells or in vivo within the murine uterus (25, 28). Cycloheximide inhibits new protein synthesis; 17ß estradiol therefore likely directly regulates endometrial HOXA10 expression without requiring an intermediate protein. In fact, estradiol regulates HOXA10 through ER binding. Sequence analysis of the 5' regulatory region of HOXA10 revealed the presence of two putative estrogen response elements (EREs) (115). Both elements show partial sequence homology to the consensus ERE and bind the estrogen receptors ER and ERß. Although both of these elements demonstrate estrogen-responsive up-regulation of reporter gene expression, the response mediated by one of the elements (ERE1), heretofore referred to as HOXA10-ERE, is more robust (115). Regulation of HOXA10 expression by estradiol in the endometrium is therefore ER dependent.
Another interesting aspect of the molecular mechanism involved in estrogen-Hox signaling is that it is ligand specific. As described above, during embryonic development DES, but not 17ß estradiol causes a posterior shift in expression of the genes Hoxa9, Hoxa10, and Hoxa11 (25, 105). DES not only alters the spatial domains of expression of Hox genes in the reproductive tract, but additionally elicits significantly lower reporter gene expression mediated by the HOXA10-ERE compared with treatment with 17ß estradiol (115). Reporter gene expression is differential in magnitude in response to treatment with the cognate ligand 17ß estradiol compared with DES. In addition to posterior displacement of the spatial domains of expression of the Abd-B orthologs, decreased Hox gene expression levels may be another mechanism by which DES causes reproductive tract developmental anomalies (105, 115).
Estrogen-Hox signaling is likely to have clinical implications in assisted reproductive technology (ART), where controlled ovarian hyperstimulation is frequently used to induce oocyte growth in infertile patients. Treatment protocols are commonly based on the use of recombinant FSH. As opposed to natural reproductive cycles, treatment cycles typically are associated with multiple folliculogenesis and consequent high serum estradiol levels. ART cycles are also associated with a high rate of multiple gestation (116, 117). The cause of multiple pregnancies is likely related to spontaneous multiple ovulation and/or multiple embryo transfer (118, 119). It is also possible that elevated 17ß estradiol in such cycles could potentially up-regulate HOXA10 expression. As seen previously, high endometrial HOXA10 expression levels in mice resulted in consistently large litter sizes possibly due to increased implantation efficiency (35). However, no difference in secretory phase endometrial HOXA10 mRNA levels is seen in ART cycles compared with those seen in endometrium obtained from untreated subjects (120). It is therefore unlikely that multiple pregnancies associated with ART are due to increased endometrial receptivity from elevated HOXA10 expression; instead they are simply due to the generation of multiple embryos.
To synchronize follicular growth and prevent a premature LH surge in ART cycles, treatment regimens frequently involve the use of GnRH agonists or antagonists. Because GnRH receptors are found in endometrium, these agents may have direct and likely adverse effects on endometrial receptivity (121). When Ishikawa cells are treated with either a GnRH agonist or antagonist at comparable doses to those used for ART, no difference in HOXA10 mRNA levels is observed. It is therefore unlikely that GnRH agonist or antagonist therapy, as is currently used for ART, can iatrogenically decrease endometrial receptivity (120). Rather, diminished pregnancy rates seen with the use of GnRH antagonists likely are the result of diminished LH production. In contrast, the use of depot GnRH analog, accompanied by profound suppression of endogenous estrogens and pseudomenopausal amenorrhea, is associated with suppressed HOXA10 mRNA levels (28).
Another area where endocrine-Hox signaling is likely to be operative in the adult is in oncology. One molecular mechanism underlying the development of cancer likely involves dysregulation of developmental processes resulting in aberrant growth and differentiation of affected tissues. As described earlier, HOX genes have been associated with leukemogenesis (67). Both 17ß estradiol and Hox genes have been implicated in multiple cancers. Estrogen-Hox signaling is therefore likely to have implications in cancer biology and therapeutics. A recent large randomized clinical trial has associated conjugated equine estrogens and a progestin with increased occurrence of breast cancer (122, 123). The expression of certain HOX genes is associated with antiproliferative effects in human breast cancer. For example, HOXA5 directly up-regulates expression of the tumor suppressor gene p53, leading to apoptosis; loss of expression of HOXA5 as well as p53 is associated with cancer progression (124). Finally, in the ER+ human breast adenocarcinoma cell lines (MCF-7), estradiol up-regulates HOXA10 expression. HOXA10 in turn up-regulates p53 expression. In contrast, the ER(–) breast cancer cell line (BT-20), derived from highly invasive cancers, does not endogenously express HOXA10. When these cells are transfected with HOXA10, increased p53 expression is observed as well as decreased invasion (125). Estradiol treatment of BT20 cells, in the absence of HOXA10, does not increase p53. In sex steroid-responsive tissues such as the breast, a complex interplay between estradiol and progesterone and the expression levels of various HOX genes may be involved in tumorigenesis. Overexpression of such HOX genes as HOXA5 and HOXA10 has been associated with increased tumor suppressor expression and possibly decreased invasion. It is likely that, as in the endometrium, HOXA10 mediates terminal differentiation in other tissues as well. The terminally differentiated state cannot coexist with neoplasia, which may explain the antineoplastic effects of HOXA10 in breast tissue.
As discussed earlier, estrogen-Hox signaling is effected by multiple estrogenic ligands. In ER+ breast cancer cells, the selective ER modulator (SERM) tamoxifen also up-regulates HOXA10 and thereby increases p53. Tamoxifen may therefore mediate HOX gene-induced differentiation in breast cancer. Such a molecular mechanism may contribute to the successful clinical use of tamoxifen (125, 126, 127) .
It is well known, that the SERMs have tissue-specific effects; whereas tamoxifen has antineoplastic effects in breast cancer, it is proneoplastic in the endometrium, and it has been associated with endometrial hyperplasia and carcinoma (128). Another SERM, raloxifene, exhibits antiproliferative effects in the breast without increasing the incidence of endometrial cancer (129, 130, 131). Because HOX genes are regulated by estrogens, it is likely that they are differentially regulated by individual SERMs as well. Understanding the molecular interactions between the SERMs and the Hox genes is likely to have significant therapeutic implications.
B. Progesterone
Progesterone secreted by the corpus luteum programs the endometrium to undergo terminal differentiation, necessary for embryo receptivity (132). Progesterone receptor A knockout mice are infertile and demonstrate failed implantation of artificially transferred blastocysts (132). Additionally in these mice, administration of progesterone results in inflammation and proliferation rather than differentiation during the secretory phase (132). Endogenous progesterone, like estradiol, also regulates adult endometrial Hox gene expression during the reproductive cycle, and this signaling pathway is likely to modulate endometrial functional differentiation (25, 28).
As with 17ß estradiol, the levels of HOXA10 mRNA in primary endometrial stromal cells are also increased in response to treatment with the progestin medroxyprogesterone acetate (MPA). The increase in HOXA10 mRNA levels however, is greater in response to MPA than to 17ß estradiol, increasing 25-fold when the concentration of MPA is increased from 10–9 to 10–6 M (28). Furthermore, in cells treated with combined 17ß estradiol and MPA, HOXA10 expression levels are higher compared with those obtained after treatment with either hormone alone. This latter experimental condition recapitulates the hormonal milieu prevalent in the mid- and late secretory phase of the menstrual cycle (Fig. 3). As with 17 ß estradiol, the linear increase in HOXA10 mRNA levels with rising progestin concentrations plateaus at high physiological and supraphysiological dosage levels (28).
|
Some of the physiological events that accompany terminal differentiation in secretory phase endometrium include calcitonin secretion by endometrial epithelial cells and prolactin expression by stromal cells. Terminal differentiation in the endometrium is also accompanied by leukocyte infiltration. Cytokines elaborated by the leukocytes are necessary for critical mitogenic and immunological processes that occur at the time of embryo implantation (45, 133, 134, 135, 136). In murine uteri, whereas progesterone and HOXA10 result in concordant epithelial cell calcitonin and stromal cell prolactin expression, discordant effects are seen with respect to leukocyte infiltration. HOXA10, but not progesterone induces uterine eosinophil infiltration and degranulation (137). Endometrial eosinophil infiltration is a well-characterized estrogenic response. Typically, Hoxa10 mediates progesterone action; however, in some cases it may limit the effects of progesterone, allowing continued estrogenic effects. Because estradiol, progesterone, and their target HOXA10 are each necessary for embryo implantation, it is likely that differential spatiotemporal activation of this gene by estradiol and progesterone in the endometrium may be necessary to achieve distinct stages of functional development to support nidation.
HOXA10 expression is not confined to the endometrium; it has been demonstrated in the uterine myometrium as well. As in the endometrium, myometrial HOXA10 expression is dynamic and subject to progestational regulation. In contrast to the endometrium, however, HOXA10 is expressed at low levels in the human myometrium in the midsecretory phase of the human reproductive cycle (138). In vitro MPA down-regulates HOXA10 mRNA expression levels in myometrial cells. It is likely that the down-regulation of this HOX gene in myometrium by progesterone is necessary to prevent terminal differentiation, allowing the myocytes to retain proliferative capability necessary to accommodate fetoplacental growth during pregnancy (138). Although HOXA10 expression levels in myocytes are low during pregnancy, in labor the expression is up-regulated (139). At term gestation, increased HOXA10 likely enables functional development of myocytes necessary for myometrial contractility during labor.
Although progesterone-mediated up-regulation of HOXA10 in endometrium during the secretory phase of the menstrual cycle is necessary for embryo implantation and hence fertility, the opposite response is seen with the use of contraceptive progestin formulations. The progestin depot MPA is a commonly used contraceptive agent that acts primarily via suppression of the hypothalamo-pituitary-ovarian axis, resulting in anovulation. Depot MPA also has additional effects on the endometrium, attenuating endometrial thickness and possibly functional differentiation, thereby making it unfavorable for embryo implantation. Long-term treatment of endometrial cells with this contraceptive also results in suppression of HOXA10 mRNA expression levels, which may be one of the molecular mechanisms that contribute to the efficacy of this agent (28). The effects of sex steroids on endometrial HOX gene expression are both dose and time dependent.
Progesterone regulates not only uterine HOXA10 expression but also HOXA11. We have seen earlier that HOXA11 regulation by estradiol and progesterone is similar to that of HOXA10 throughout the estrus/menstrual cycle. However, in contrast to HOXA10, under physiological conditions both sense and antisense transcripts of Hoxa11/HOXA11 are synthesized (21, 140). Endometrial HOXA11 sense mRNA levels vary inversely with antisense mRNA levels. It is therefore likely that the translation of HOXA11 is regulated by the balance in levels of these two types of transcripts.
In primary endometrial stromal cells, progesterone is found to alter this balance, resulting in increased sense mRNA and decreased antisense mRNA transcript levels (140). Although it seems intuitive that the sense and antisense transcripts combine to form a duplex and thereby repress HOXA11 bioavailability, this mechanism is not observed when murine uteri are transfected with HOXA11 antisense. The litter size in antisense transfected uteri is unchanged from controls (140). It is therefore likely that the mechanism of action of HOXA11 antisense involves transcriptional interference. At any given time, transcription proceeds in only one direction on a single gene, resulting in either sense or antisense transcription; the relative amount of sense transcript determines HOXA11 bioavailability. By promoting sense over antisense transcription, progesterone augments endometrial HOXA11 expression levels. In addition to nuclear receptor-mediated transactivation, transcriptional interference is a novel molecular mechanism of endocrine regulation of HOX gene expression.
Progesterone is the major stimulus for cyclic endometrial terminal differentiation and is necessary for embryo implantation and fertility. Not unexpectedly, this functional differentiation is accompanied by high endometrial expression levels of the homeotic genes Hoxa10 and Hoxa11, driven by progesterone. Both progesterone and Hoxa10 induce similar phenotypic effects characteristic of terminal differentiation in endometrial epithelial as well as stromal cells. It is therefore likely that these homeotic genes function downstream of progesterone to mediate cyclic endometrial functional differentiation. Differing HOX gene expression in response to changing endocrine signals is a mechanism by which hormones can alter developmental identity.
C. Testosterone
Testosterone is necessary for mesonephric duct differentiation in male embryos; lack of androgen signaling results in anomalous development of the male reproductive tract (141, 142, 143, 144). Analogous to their role in paramesonephric duct differentiation, the Abd-B orthologous Hox genes are likely necessary for male genital tract development as well. The phenotype of Hoxa10 (–/–) male mice is characterized by anterior transformation of mesonephric derivatives; additionally Hoxa10 mutations are associated with cryptorchidism secondary to a lack of shortening of the gubernaculum in both mice and humans (20, 22, 145). The anterior transformation seen in Hoxa10 (–/–) male mice suggests that, as in the female, HOX genes are also expressed axially in male reproductive anlagen during embryogenesis; it is likely that testosterone regulates spatial expression of the individual Hox genes in a role analogous to that of estrogen in female reproductive organogenesis.
In the adult female, coordinated expression of estradiol and progesterone is sufficient for ovulation and menstrual cyclicity, and both hormones differentially regulate endometrial HOX gene expression throughout the reproductive cycle. Although androgens have physiological roles in adult females, abnormally increased androgen secretion is seen in women with polycystic ovary syndrome (PCOS) (146). Hyperandrogenemia is likely driven by insulin resistance and concomitant hyperinsulinemia in many of these patients (147, 148). Aberrant androgen secretion results in anovulation, menstrual anomalies, and hirsutism. Patients with PCOS often experience infertility and, despite successful ovulation induction, demonstrate high rates of implantation failure and early reproductive wastage. Human endometrium expresses the androgen receptor; it is likely that hyperandrogenemia interferes with cyclic endometrial functional differentiation (149, 150). In contrast to normal women, significantly lower levels of HOXA10 mRNA are seen in secretory phase endometrium from patients with PCOS. Low levels of expression of HOXA10 in peri-implantation endometrium may result in defective implantation and reproductive wastage that are commonly seen in these women.
In vitro, in Ishikawa cells dose-responsive decreases in HOXA10 mRNA are seen in response to physiological or supraphysiological concentrations (10–4 to 10–7 M) of testosterone. As discussed previously, Ishikawa cells are an endometrial adenocarcinoma cell line that expresses the estrogen and progesterone receptors. This cell line also expresses the androgen receptor (46). As with estradiol and progesterone, the testosterone regulates HOXA10 through an androgen receptor-mediated mechanism that is readily blocked by the androgen receptor antagonist, flutamide.
PCOS is a heterogeneous clinical entity that is associated with elevated circulating levels of multiple androgens. Not all of the circulating androgens, however, repress endometrial HOXA10 expression. In fact only, testosterone and dihydrotestosterone, but not dehydroepiandrosterone and dehydroepiandrosterone sulfate, decrease HOXA10 mRNA levels. Interestingly, testosterone blocks estradiol, progesterone, or combined estradiol and progesterone-mediated up-regulation of HOXA10 mRNA in vitro in Ishikawa cells. During the mid- and late secretory phase, high systemic estradiol and progesterone levels are associated with peak endometrial functional differentiation, high HOXA10 expression levels, and embryo receptivity. As discussed above, in patients with PCOS, hyperandrogenemia suppresses endometrial HOXA10 expression in the secretory phase of the menstrual cycle, overriding the stimulatory effects of estradiol and progesterone in the luteal phase. Such aberrant endocrine regulation of HOXA10 by testosterone may interfere with endometrial functional differentiation, resulting in inefficient reproductive capability seen clinically in patients with hyperandrogenemia.
When treated with the biguanide metformin, women with PCOS demonstrate
