This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Dey, S. K. Articles by Wang, H. Search for Related Content PubMed PubMed Citation Articles by Dey, S. K. Articles by Wang, H.

Molecular Cues to Implantation

Departments of Pediatrics (S.K.De., S.K.Da., J.R., B.C.P., T.D., H.W.), Cell and Developmental Biology, and Pharmacology (S.K.De.), Vanderbilt University Medical Center, Nashville, Tennessee 37232; and Departments of Obstetrics and Gynecology, and Cell Biology and Physiology (H.L.), Washington University School of Medicine, St. Louis, Missouri 63110

Correspondence: Address all correspondence and requests for reprints to: S. K. Dey, Ph.D., Department of Pediatrics, Vanderbilt University Medical Center, MCN D4100, Nashville, Tennessee 37232. E-mail: sk.dey{at}vanderbilt.edu.

	Abstract

Top Abstract I. Introduction II. Preimplantation Embryo... III. Species-Specific... IV. Delayed Implantation V. Window of Implantation:... VI. Embryo-Uterine Signaling... VII. Emerging Concepts VIII. Perspectives and Future... References

 
Successful implantation is the result of reciprocal interactions between the implantation-competent blastocyst and receptive uterus. Although various cellular aspects and molecular pathways of this dialogue have been identified, a comprehensive understanding of the implantation process is still missing. The receptive state of the uterus, which lasts for a limited period, is defined as the time when the uterine environment is conducive to blastocyst acceptance and implantation. A better understanding of the molecular signals that regulate uterine receptivity and implantation competency of the blastocyst is of clinical relevance because unraveling the nature of these signals may lead to strategies to correct implantation failure and improve pregnancy rates. Gene expression studies and genetically engineered mouse models have provided valuable clues to the implantation process with respect to specific growth factors, cytokines, lipid mediators, adhesion molecules, and transcription factors. However, a staggering amount of information from microarray experiments is also being generated at a rapid pace. If properly annotated and explored, this information will expand our knowledge regarding yet-to-be-identified unique, complementary, and/or redundant molecular pathways in implantation. It is hoped that the forthcoming information will generate new ideas and concepts for a process that is essential for maintaining procreation and solving major reproductive health issues in women.

I. Introduction

II. Preimplantation Embryo Development and Genomic Activation

III. Species-Specific Morphological Blueprint and Timing of Implantation

IV. Delayed Implantation

V. Window of Implantation: A Transient and Unique Moment

VI. Embryo-Uterine Signaling Pathways in Implantation

A. Steroid hormone signaling

B. Signaling via adhesion molecules: cell-cell interactions

C. Signaling by vasoactive factors

D. Signaling by growth factors

E. Signaling by cytokines

F. Homeobox genes in implantation

G. Ligand-dependent nuclear receptors and coactivators in implantation

H. Cell cycle regulation and signaling in implantation and decidualization

I. Matrix remodeling and angiogenesis during implantation and decidualization

VII. Emerging Concepts

A. Endocannabinoid signaling in implantation

B. Developmental genes in implantation

C. Discovery of novel implantation-related genes

VIII. Perspectives and Future Directions

	I. Introduction

Top Abstract I. Introduction II. Preimplantation Embryo... III. Species-Specific... IV. Delayed Implantation V. Window of Implantation:... VI. Embryo-Uterine Signaling... VII. Emerging Concepts VIII. Perspectives and Future... References

 
PROCREATION, BOTH SEXUAL and asexual, is a fundamental evolutionary process necessary to sustain life. Viviparity is a landmark in the process of evolution. Sexual procreation in higher eukaryotes, especially in mammals, is often inferior to asexual procreation in prokaryotes and in some eukaryotes with respect to shear number of progeny. Thus, mammalian reproduction is more complex and highly regulated for the propagation of superior offspring to carry on the task of procreation. The nurturing of an offspring within the body and producing a live birth is an enduring task. This process demands safeguard regulatory systems at various critical steps. The assembly of a new life first depends on the union between a sperm and an egg (ovum) culminating in fertilization; failure to achieve such a union leads to their demise. The one-cell fertilized egg, termed embryo, undergoes several mitotic cell divisions, eventually forming a differentiated tissue called the blastocyst with two distinct cell populations, the inner cell mass (ICM) and a layer of trophectoderm cells surrounding the ICM (1). The embryo proper is derived exclusively from the ICM, whereas the placenta and extraembryonic membranes are produced from cells contributed mainly by the trophectoderm. A two-way interaction between the blastocyst and maternal uterine luminal epithelium initiates the process of implantation, a process by which blood vessels of the embryo are brought into functional communication with the maternal circulation leading to the establishment of a functional placenta and pregnancy. Maternal resources filtered across the selective barrier of the placenta protect and nourish the conceptus. Placental types have been classified into three categories: hemochorial (rodents, humans, and nonhuman primates), epitheliochorial (horses, cows, sheep, and pigs), and endotheliochorial (most carnivores) (reviewed in Ref. 2).

A significant pregnancy loss resulting from preimplantation embryonic death is common to many mammals and is considered to be a selection process leading to the survival of superior embryos for implantation. However, dysregulation of the events before, during, or immediately after implantation also may often be a cause for poor pregnancy rates in eutherian mammals. Understanding the mechanism of preimplantation embryonic development and implantation in the uterus has been a challenge to reproductive and developmental biologists with the goal of alleviating the problems of human infertility and ensuring the birth of quality offspring. Such knowledge is also necessary for developing novel contraceptive approaches to restrict world population growth.

The current state of our knowledge of preimplantation and implantation physiology is the result of the accumulation of scientific observations gathered over many years. Implantation is a complex process involving spatiotemporally regulated endocrine, paracrine, autocrine, and juxtacrine modulators that span cell-cell and cell-matrix interactions. However, the precise sequence and details of the molecular interactions involved have not yet been defined. Furthermore, the implantation process varies among species, thus precluding the formulation of a unified theme. In addition, ethical restrictions and experimental difficulties prevent direct analysis of embryo-uterine interactions during human implantation. Thus, it is an onerous task to write a review on the molecular basis of embryo-uterine interactions during implantation that could be relevant to mammals in general. This review focuses primarily on the physiological and molecular basis of implantation in mice because more mechanistic information is now available for this species. However, an attempt has been made to indicate comparative analyses based on limited work in other species.

Despite experimental success in initiating embryonic development outside the womb and in identifying numerous molecules involved in the embryo-uterine dialogue (3, 4, 5, 6, 7), there is a yet-to-be-filled significant knowledge gap in understanding the in vivo events of implantation. The successful implantation of an embryo is contingent upon cellular and molecular cross-talk between the uterus and the embryo. The coordination of the endocrine, cellular, and molecular events via paracrine, autocrine, and/or juxtacrine factors in a dynamic manner produces within the uterus a favorable environment, the receptive state, to support implantation. The embryo also functions as an active unit with its own molecular program of cell growth and differentiation. Thus, deficiencies in uterine receptivity, embryo development, or the embryo-uterine dialogue will compromise fertility. This review of the implantation process focuses on the molecular basis of embryo homing and attachment, on elucidating the reciprocal signaling networks between the embryo and uterus, and on determining genetic causes of implantation failure.

	II. Preimplantation Embryo Development and Genomic Activation

Top Abstract I. Introduction II. Preimplantation Embryo... III. Species-Specific... IV. Delayed Implantation V. Window of Implantation:... VI. Embryo-Uterine Signaling... VII. Emerging Concepts VIII. Perspectives and Future... References

 
Preimplantation embryo development and differentiation, which culminate in the formation of a blastocyst, require the activation of the embryonic genome, a process that is essential to implantation. The maternal-zygotic transition occurs at the two-cell stage in mice and other rodents, between the eight- and 16-cell stages in cows and sheep, and between the four- and eight-cell stages in humans (reviewed in Ref. 8). Upon activation of the embryonic genome, the embryo grows rapidly to form a blastocyst. At the blastocyst stage, embryos mature and escape from their zona pellucidae to gain implantation competency. The differentiated and expanded blastocyst is composed of three cell types: the outer polarized epithelial trophectoderm, the primitive endoderm, and the pluripotent ICM. The ICM provides the future cell lineages for the embryo proper (9, 10), and the trophectoderm, the very first epithelial cell type in the developmental process, makes the initial physical and physiological connection with the uterine luminal epithelium. The ICM is not identifiable in marsupial blastocysts that appear as a hollow ball of cells with similar morphological characteristics. It is also not known which cells are programmed to form the embryo proper (reviewed in Ref. 10). The formation of the trophectoderm and its subsequent development into trophoblast tissue are crucial steps for the initiation of implantation and the establishment of pregnancy. Trophoblast cells produce a variety of growth factors, cytokines, and hormones that influence the conceptus and maternal physiology in an autocrine, paracrine, and/or juxtacrine manner (11, 12).

Preimplantation embryo development normally occurs within the zona pellucida. However, zona removal by various experimental manipulations does not deter embryonic development in vitro (13), suggesting that this glycoprotein barrier is not essential for development to progress. The nonadhesive nature of the zona pellucida is thought to facilitate the journey of embryos through the oviduct and from the oviduct to the uterus. In mice and rats, normal embryonic development to the blastocyst stage within the reproductive tract requires the presence of ovarian estrogen and progesterone (14). There is a reduced number of embryos and a reduced number of cells per embryo in the absence of these hormones (15), but treatment with estrogen and progesterone reverses these defects (16). Because there is no convincing evidence that estrogen and/or progesterone act directly on the preimplantation embryo (17), embryonic development is considered to depend on growth-promoting factors originating from the reproductive tract under the influence of these hormones. However, apparently normal development in simple defined media in culture suggests that preimplantation embryos are capable of producing their own growth-promoting factors (reviewed in Ref. 18). In fact, several growth factors, cytokines, and their receptors are expressed in the embryo, and the proliferative and differentiating effects of these factors on embryonic development and functions have been observed (reviewed in Refs. 3 and 18, 19, 20).

	III. Species-Specific Morphological Blueprint and Timing of Implantation

Top Abstract I. Introduction II. Preimplantation Embryo... III. Species-Specific... IV. Delayed Implantation V. Window of Implantation:... VI. Embryo-Uterine Signaling... VII. Emerging Concepts VIII. Perspectives and Future... References

 
Implantation is the process by which the blastocyst comes into intimate physical and physiological contact with the uterine endometrium. Enders and Schlafke (21, 22) have classified the process of implantation into three stages: apposition, adhesion, and penetration. Apposition is the stage when embryonic trophectoderm cells become closely apposed to the uterine luminal epithelium. This is followed by the adhesion stage in which the association of the trophectoderm and the luminal epithelium is sufficiently intimate as to resist dislocation of the blastocyst by flushing the uterine lumen. The stage of penetration involves the invasion of the luminal epithelium by the trophectoderm. Stromal cell differentiation into decidual cells (decidualization) is more extensive, and the loss of the luminal epithelium is evident at this stage. These three stages of implantation form a continuum.

In mammals, especially rodents, a generalized stromal edema occurs before the beginning of apposition. This event leads to the closure of the uterine lumen, which results in interdigitation of the microvilli of the trophectoderm and the luminal epithelia (apposition), followed by closer contact between them (the adhesion or attachment reaction). Luminal closure occurs throughout the entire uterus during pregnancy or pseudopregnancy and thus does not require the presence of blastocysts. Priming of the uterus with progesterone alone appears to be sufficient for this event to occur; luminal closure does not occur in the absence of progesterone. Although luminal closure and apposition occur in progesterone-treated delayed implanting mice, the attachment reaction does not occur. Estrogen treatment is required for attachment to occur.

Bonnet (23), on the basis of different types of blastocyst-uterine cell-cell interactions, classified implantation into three categories: central, eccentric, and interstitial. Central implantation occurs in mammals such as rabbits, ferrets, and some marsupials. In these animals, blastocysts grow and expand extensively before implantation. In contrast, the blastocysts of mice, rats, and hamsters are small and show modest expansion. In these species, an implantation chamber is formed by the invagination of the uterine epithelium, which is a characteristic of eccentric implantation. In guinea pigs, chimpanzees, and humans, the implantation process is of the interstitial type, i.e., blastocysts are embedded within the subepithelial stroma. The results of ultrastructural studies led Schlafke and Enders (24) to classify implantation into intrusive, displacement, and fusion types. In the intrusive type of implantation, which occurs in humans and guinea pigs, the trophoblasts penetrate through the luminal epithelium, reaching and extending through the basal lamina. The displacement type of implantation occurs in rodents; the luminal epithelium is freed from the underlying basal lamina, facilitating the spread of trophoblasts through the epithelium. The fusion type of implantation, in which trophoblasts make a connection with the luminal epithelium by forming symplasma, occurs in the rabbit. In many rodents, including mice and rats, implantation always occurs at the antimesometrial side of the uterus, whereas in some bats, implantation is mesometrial. In other animals, the embryos elongate and either attach over the entire endometrium (horse, pig, and wallaby) or only at specialized areas known as caruncles (cow and sheep) (2). Schematic diagrams for different types of implantation have been illustrated previously (2, 3).

The attachment reaction coincides with a localized increase in stromal vascular permeability at the site of the blastocyst, as can be demonstrated by iv injection of a macromolecular blue dye (uterine blue reaction) (25). The first sign of the attachment reaction (apposition stage) in the process of implantation occurs in the mouse and rat on the evenings of d 4 and d 5, respectively, and on d 6.5 in the rabbit (25, 26, 27). In the primates, the attachment reaction occurs approximately on d 8 in humans and baboons, on d 9 in macaques, and on d 11 in marmoset monkeys (28, 29). In large domestic animals, the first signs of attachment occur on d 13 in pigs, on d 20 in cows, on d 16 in sheep, and on d 19 in goats (30).

In both mice and humans, stromal cells surrounding the implanting blastocyst undergo decidualization, eventually embedding the embryo into the antimesometrial stromal bed. In mice, blastocysts are oriented with their ICMs directed mesometrially, whereas in humans the ICM is directed antimesometrially. The mechanism by which the blastocyst is directed to the antimesometrial luminal epithelium or by which the orientation of the blastocyst is achieved at the time of implantation remains elusive. There is evidence that in progesterone-treated delayed implanting mice, blastocysts are placed antimesometrially, and interdigitation (apposition) of luminal epithelial cell microvilli occurs with those of the abembryonic or lateral trophectoderm cells of the blastocyst with its ICM oriented toward the uterine lumen. This observation led to the suggestion that upon initiation of the attachment reaction and subsequently the implantation process by estrogen, blastocysts retain the orientation they adopted during delay. During normal implantation in mice with the onset of luminal closure, blastocysts are placed at the antimesometrial side of the lumen along the uterine axis. Shortly after the luminal closure, zona-encased blastocysts are located in implantation chambers with random orientation of the ICMs. However, by the beginning of the attachment reaction, blastocysts are correctly oriented with their ICMs directed at the mesometrial pole. This observation suggested that the trophectoderm of the entire blastocyst surface has the potential for attachment to the luminal epithelium, and that attachment occurs randomly immediately after the loss of the zona pellucida. Evidence was presented to suggest that the correct orientation of the blastocyst is achieved by free movement of the ICM. However, further investigation is necessary to resolve this issue. All of these events, from the luminal closure to the attachment reaction, occur between about 86 and 92 h after coitum in mice (reviewed in Refs. 31 and 32).

	IV. Delayed Implantation

Top Abstract I. Introduction II. Preimplantation Embryo... III. Species-Specific... IV. Delayed Implantation V. Window of Implantation:... VI. Embryo-Uterine Signaling... VII. Emerging Concepts VIII. Perspectives and Future... References

 
Delayed implantation is a process by which implantation is postponed for a certain period. The uterus remains in a quiescent state, and embryos at the blastocyst stage become dormant. Delayed implantation occurs in many vertebrate species, but the underlying mechanisms that direct this process are different in various species that have adapted to this reproductive strategy (33). In mice and rats, ovariectomy before the presumed estrogen surge in the morning of d 4 of pregnancy results in the failure of implantation and initiates a state of dormancy of the blastocyst within the uterine lumen (25, 34). This condition is referred to as delayed implantation and can be maintained for many days by continued treatment with progesterone. The process of implantation with blastocyst activation can be rapidly initiated by a single injection of estrogen in the progesterone-primed uterus (25, 34). The mechanisms by which estrogen mediates the processes of blastocyst activation and implantation are poorly understood. Delayed implantation also occurs naturally (facultative) during lactation after postpartum ovulation and fertilization of the egg in mice and rats (35, 36). However, implantation ensues rapidly after termination of the suckling stimulus. The occurrence of lactational delay in these species is due to the secretion of an insufficient amount of ovarian estrogen. Mustelids, marsupials, and many other species also exhibit obligatory seasonal delayed implantation (37, 38, 39, 40). Delayed implantation does not occur in some species such as the hamster, guinea pig, rabbit, and pig. Whether this phenomenon occurs in humans is not known. The delayed implantation models in mice and in other species could be exploited more extensively to better understand the molecular signaling that emanates from the embryo and influences uterine biology and vice versa.

	V. Window of Implantation: A Transient and Unique Moment

Top Abstract I. Introduction II. Preimplantation Embryo... III. Species-Specific... IV. Delayed Implantation V. Window of Implantation:... VI. Embryo-Uterine Signaling... VII. Emerging Concepts VIII. Perspectives and Future... References

 
In all eutherian mammals thus far studied, the uterus differentiates into an altered state when blastocysts are capable of effective two-way communication to initiate the process of implantation. This state is termed uterine receptivity for implantation and lasts for a limited period. At this stage, the uterine environment is able to support blastocyst growth, attachment, and the subsequent events of implantation (30, 41, 42, 43). The major factors that specify uterine receptivity are the ovarian steroids, progesterone and/or estrogens. Ovarian progesterone and estrogen are crucial for implantation in mice and rats, but ovarian estrogen is not essential for implantation in pigs, guinea pigs, rabbits, and hamsters (41, 44, 45, 46, 47, 48). Estrogen-synthesizing capacity has been demonstrated in rabbit and pig embryos, but whether embryonic estrogen plays a role in implantation in these species is still debatable. Recent evidence suggests that hamster blastocysts express the aromatase gene (Paria, B. C., unpublished results). The mouse embryo lacks aromatase activity necessary for estrogen synthesis (reviewed in Ref. 49). Whether preimplantation estrogen secretion by the ovary or embryo plays a role in human implantation is unknown.

In mice and rats, the coordinated actions of progesterone and estrogen regulating proliferation and/or differentiation of uterine cells in a spatiotemporal manner establish the window of implantation (50). For example, on the first day of pregnancy (vaginal plug) in mice, uterine epithelial cells undergo proliferation under the influence of the preovulatory estrogen secretion. Rising levels of progesterone secreted from freshly formed corpora lutea initiate stromal cell proliferation from d 3 onward. The stromal cell proliferation is further stimulated by a small amount of ovarian estrogen secreted on the morning of d 4 of pregnancy. These coordinated effects of progesterone and estrogen result in the cessation of uterine epithelial cell proliferation, initiating differentiation (50). During normal pregnancy, the presence of an active blastocyst in the uterus is the stimulus for the implantation reaction. After the attachment reaction is initiated on d 4 at 2400 h, stromal cells surrounding the implanting blastocyst begin to proliferate extensively and differentiate into decidual cells (decidualization) (30). In pseudopregnant mice, the steroid hormonal milieu within the uterus is similarly maintained due to the presence of newly formed corpora lutea. Thus, the sensitivity of the pseudopregnant uterus to implantation on d 1–4 is quite similar to normal pregnancy, and blastocyst transfer into the uterine lumen during the receptive phase provokes normal implantation reactions and subsequent decidualization. Although blastocysts are the normal inducers of these events, various nonspecific stimuli, such as intraluminal infusion of oil, air, and mechanical stimuli can also initiate certain aspects of the decidual cell reaction (deciduoma) in pseudopregnant or steroid hormonally prepared uteri (30). However, there is evidence that the initial uterine reactions induced by nonspecific stimuli are different from those induced by blastocysts (51, 52).

Uterine sensitivity with respect to steroid hormonal requirements and implantation has been classified as prereceptive, receptive, and nonreceptive (refractory) phases (30, 41). These phases have been defined by employing embryo transfer experiments in pseudopregnant mice. In the mouse, whereas the uterus is fully receptive on d 4, it is considered prereceptive on d 1–3 of pregnancy or pseudopregnancy. The mouse uterus can be rendered receptive to blastocyst implantation only if exposed to a small amount of estrogen after 24–48 h of progesterone priming (53). Evidence suggests that the uterus is most receptive to implantation on d 4 (43), and the efficiency of implantation decreases with time (54). By d 6, the uterus becomes completely refractory to blastocyst implantation. Recent evidence suggests that concentration of estrogen within a very narrow range determines the duration of the window of uterine receptivity in mice; uterine receptivity remains open for an extended period at lower estrogen levels but rapidly closes at higher levels. Uterine nonreceptivity induced at high estrogen levels is accompanied by aberrant uterine expression of implantation-related genes. These results suggest that careful regulation of estrogen levels could improve female fertility in in vitro fertilization programs (55). Another critical factor determining the window of implantation is the state of activity of the blastocyst, as described below.

In mice and rats, ovariectomy before the preimplantation estrogen secretion on the morning of d 4 of pregnancy induces delayed implantation (34, 43). This status can be maintained for many days if progesterone treatment is continuously provided. Under this condition, blastocysts undergo zona hatching, albeit at a slower pace, but they become dormant without initiating the attachment reaction, and the progesterone-primed uterus remains in the neutral stage. However, a single injection of estrogen promptly induces blastocyst activation with the initiation of implantation in the progesterone-primed uterus. Active and dormant blastocysts are molecularly and physiologically distinguishable. Epidermal growth factor (EGF) receptor (EGF-R), cyclooxygenase-2 (COX-2), and histamine type 2 receptor (H₂), the factors that are associated with blastocyst attachment reaction, are expressed in normal or active blastocysts but are down-regulated in dormant blastocysts (43, 56, 57, 58, 59). In contrast, the G protein-coupled cannabinoid receptor CB1, which is activated by natural and endocannabinoids, is down-regulated in active blastocysts but remains up-regulated in dormant blastocysts (60). Collectively, these findings suggest that a complex array of molecular networks regulates blastocyst activation and dormancy.

Although estrogen is essential for blastocyst activation and implantation in the progesterone-primed mouse uterus, the mechanisms by which estrogen initiates these responses remain elusive. We speculated that estrogen actions in uterine preparation and blastocyst activation for implantation are two distinct events. Embryo transfer experiments in delayed implanting recipient mice provide evidence that whereas the primary estrogen, 17ß-estradiol, initiates uterine events for implantation, its catechol metabolite, 4-hydroxy-17ß-estradiol (4-OH-E₂), participates in activation of dormant blastocysts (17). Blastocyst activation by 4-OH-E₂ involves COX-2-derived prostaglandins (PGs) and cAMP (17). The use of an estrogen receptor (ER) antagonist ICI-182,780 showed that, whereas estradiol via its interaction with the nuclear ERs participates in the preparation of the progesterone-primed uterus to the receptive state in an endocrine manner, 4-OH-E₂ produced from estradiol in the uterus mediates blastocyst activation in a paracrine manner that does not involve nuclear ERs. These results provide evidence that both primary and catecholestrogens are required for embryo-uterine interactions to ensure successful implantation and that implantation occurs only when uterine receptivity coincides with the activated state of the blastocyst. Molecular pathways that are potentially involved in uterine receptivity and blastocyst activation are discussed below in more detail.

	VI. Embryo-Uterine Signaling Pathways in Implantation

Top Abstract I. Introduction II. Preimplantation Embryo... III. Species-Specific... IV. Delayed Implantation V. Window of Implantation:... VI. Embryo-Uterine Signaling... VII. Emerging Concepts VIII. Perspectives and Future... References

 
Recent advances in molecular and genetic approaches have led to the discovery of numerous molecules involved in embryo-uterine interactions; however, the precise sequence and details of the signaling cascades for many of these molecules have not yet been defined. This review attempts to focus on a select number of signaling molecules and pathways that are implicated in embryo-uterine interactions in relation to implantation (Fig. 1). A large number of other growth factors, cytokines, lipid mediators, and vasoactive agents that could well be involved in implantation are not addressed here due to space limitations.

View larger version (70K): [in this window] [in a new window]   FIG. 1. A scheme of signaling networks in embryo-uterine communication during implantation in mice. Implantation in mammals absolutely depends upon synchronized development of the blastocyst to the stage when it is competent to implant and the uterus to the stage when it is receptive to blastocyst growth and implantation. Ovarian estrogen (E2) and progesterone (P4) are the primary effectors that direct the prereceptive uterus to a receptive state via a number of locally expressed growth factors, cytokines, transcription factors, and vasoactive mediators in the uterus, whereas uterine-derived catecholestrogen and regulated levels of endocannabinoids activate the blastocyst to an implantation-competent state. During the attachment phase, signaling and adhesive events embracing the uterus and the blastocyst lead to implantation. Le, Luminal epithelium; Ge, glandular epithelium; S, stroma; CE, catechol estrogen.

A large variety of natural and synthetic compounds that mimic natural estrogens and bind to the nuclear ER are present in the environment (79, 80). A few of these xenobiotics possess an even higher affinity for ERß than for ER (81, 82). The interaction of xenoestrogens with ER is considered as the basis for their reproductive toxicity, although their affinity for the receptor is quite low (83, 84). However, there is evidence that responses to an estrogenic compound in target tissues are not necessarily related to its affinity for the receptor (85, 86, 87, 88), suggesting the presence of other signaling pathways. Thus, xenoestrogens may interact with ER or other binding proteins that may not result in a similar kind of transactivation that normally occurs with natural ligands. Their estrogen-like effects in the uterus include an increase in water imbibition and ornithine decarboxylase activity, as well as increased DNA and protein synthesis. Xenoestrogens can also induce implantation in progesterone-primed rodents (89). Although the mechanisms of action of these xenoestrogens are not clear, these compounds have received a great deal of attention as a possible cause of certain cancers and impaired reproductive functions (reviewed in Refs. 80 and 90).

Emerging evidence indicates that many of the rapid uterine effects of estrogenic compounds do not involve the classical genomic effects (91, 92, 93, 94, 95). We and others have observed that ER mutant female mice, which show negligible classical responses to estrogen (96), nevertheless display estrogenic responses to catecholestrogens or xenoestrogens (91, 92, 93, 94, 95). For example, early uterine estrogenic responses, including gene expression and macromolecular uptake, are elicited by catecholestrogen or kepone independent of ER and/or ERß (91). Furthermore, genetic evidence suggests that several other early responsive target genes are induced in the mouse uterus by primary and catechol estrogens independent of nuclear ERs, suggesting nongenomic downstream events. These downstream signaling pathways represent Wnt signaling, protein processing, and calcium homeostasis (92). In contrast, estrogen, which induces the expression of many growth factors and their receptors and other secretory proteins, may facilitate their transport to the membrane and/or secretion by activating the membrane-trafficking pathway via nuclear ER (97). These results suggest that estrogen regulates diverse but interdependent signaling pathways via ER-dependent and -independent processes.

Despite the evidence for nongenomic steroid hormone actions, differential uterine cell-specific expression of nuclear ER and ERß and the progesterone receptor (PR) during the periimplantation period in mice suggests that the coordinated effects of estrogen and progesterone in uterine events for implantation are primarily mediated via these nuclear receptors (98). Furthermore, gene-targeting experiments in mice lacking ER or PR provide valuable information regarding their roles in uterine biology that were not previously recognized. ER(–/–) mice exhibit infertility due to hyperstimulated ovaries and hypoplastic uteri (99). Furthermore, blastocyst transfers in estrogen- and progesterone-treated uteri of ER(–/–) mice have established that functional ER is required for implantation (100). In contrast, ER(–/–) mice can induce and support decidualization (deciduoma) in response to artificial stimuli if primed appropriately with progesterone alone (101, 102). These results suggest that ER(–/–) mice fail to initiate and support implantation, perhaps due to the failure of the attachment reaction rather than the failure of decidualization events (91). Indeed, the genes and signaling pathways involved in decidualization are induced in ER(–/–) after application of an artificial stimulus (100). PR(–/–) mice show pleiotropic reproductive defects including impaired ovulation, uterine hyperplasia, and failure in decidualization (103). Selective deletion of the PR-A isoform also showed infertility but with a milder phenotype, suggesting that PR-A and PR-B serve as functionally distinct mediators of progesterone signaling in vivo (104). The results of experiments that used both PR(–/–) and PR-A(–/–) mice further reinforce a requirement of progesterone in decidualization (103, 104).

Gene-targeting experiments have established the importance of both ER and PR in uterine preparation for implantation in mice (99). However (98, 103), whether the preimplantation embryo is a direct target for steroid hormones remains unclear. Recent evidence suggests that preimplantation estrogen secretion on d 4 of pregnancy in mice has a dual role as primary estrogen and as a catecholestrogen with distinct targets (17). Whereas primary estrogen acts via the uterine ER to prepare the uterus for implantation, catecholestrogens formed locally in the uterus from the primary estrogen participate in blastocyst activation. However, it is still a mystery how catecholestrogens mediate activation of blastocysts (17). Although nuclear ER is present in both active and dormant blastocysts (105), dormant blastocysts do not respond to estradiol and fail to attain implantation competency in vitro. In contrast, dormant blastocysts do respond to a catecholestrogen 4-OH-E₂ and become implantation competent in vitro. The ER antagonist ICI-182,780 fails to reverse this response, suggesting that nuclear ER signaling is not critical to blastocyst activation (17). These observations are surprising in the light of other recent findings that ER, ERß, and efp mRNAs are expressed in the preimplantation embryos (106, 107). Examining the direct roles of estrogens and/or progesterone in preimplantation embryo function and how steroid hormone signaling in the embryo and uterus are coordinated for implantation will require further investigation.

Estradiol undergoes hydroxylation to 4-OH-E₂ by a P450-linked enzyme CYP1B1 (108). This enzyme is present throughout the mouse uterus on d 4, but disappears from the implantation site on d 5 (17). Activation of dormant blastocysts appears to involve an early response to 4-OH-E₂, because dormant blastocysts transferred into delayed implanting recipient uteri within 1 h of estradiol administration of the recipients show implantation, whereas similar blastocysts transferred beyond this 1-h period fail to implant (17). These results suggest that a rapid response that is critical to implantation occurs in utero. In contrast, dormant blastocysts cultured in the presence of 4-OH-E₂, but not estradiol, gain implantation competency and, upon transfer, implant in pseudopregnant recipients well beyond the 1-h window of estradiol treatment. Similar results were also obtained by culturing dormant blastocysts in the presence of PGE₂ or a permeable analog of cAMP. This effect apparently involves the COX-2 signaling pathway (17). For example, coincubation of dormant blastocysts with a selective COX-2 inhibitor and 4-OH-E₂ efficiently blocks their activation and implantation upon transfer to suitable recipients. This effect of the COX-2 inhibitor was partially reversed by addition of PGE₂ to the culture media. The results strongly suggest that the action of 4-OH-E₂ on dormant blastocysts is mediated via the COX-2 signaling pathway, leading to an increase in intracellular cAMP levels. Further investigation will reveal the types of PGs and receptors involved in this event.

B. Signaling via adhesion molecules: cell-cell interactions
Many glycoproteins and carbohydrate ligands and their receptors are expressed in the uterine luminal epithelium and blastocyst cell surfaces (reviewed in Refs. 109 and 110). Primary adhesion molecules that are implicated in implantation are selectins, galectins, heparan sulfate proteoglycans, Muc-1, integrins, cadherins, and the trophinin-tastin-bystin complex. Muc1 acts as an antiadhesive masking molecule (111). Muc1, a stretch of long carbohydrate moieties, is expressed in the mouse uterine epithelium before implantation. The physical hindrance created by these branches is thought to prevent interaction between the embryo and the luminal epithelium of the uterus before the attachment reaction. This is consistent with timely down-regulation of Muc1 from the luminal epithelium throughout the uterus before the attachment reaction in mice (reviewed in Ref. 3). In contrast, overall Muc1 expression increases in the rabbit and human uterus during the receptive period. However, careful examination revealed that there is indeed a decrease in Muc1 levels at the site of implantation in rabbits (112). In humans, the situation appears to be more complicated. During the apposition phase, the presence of an embryo increases the levels of Muc1 in the epithelium, but at the adhesion phase, the embryo induces a cleavage of Muc1 at the implantation site (113). Collectively, these findings suggest that Muc1 acts as an antiadhesive molecule that must be removed from the site of implantation.

Among the adhesion molecules, integrins have been studied more extensively in the human endometrium because of their cycle-dependent changes and the potential role in uterine receptivity. The members of the integrin family serve as receptors for various extracellular matrix (ECM) ligands and modulate cell-cell adhesion and signal transduction events (114). Each integrin is comprised of two subunits, and ß, and each ß-combination has its own binding specificity and signaling properties. As membrane-associated receptors, integrins possess short cytoplasmic tails with no enzymatic activity. Signaling by integrins is mediated by associating adaptor proteins that bridge them to the cytoskeleton, cytoplasmic kinases, and transmembrane growth factor receptors (114). Several members of the integrin family, including vß3, are known to interact with the RGD (Arg-Gly-Asp) peptide sequence present in many ECM proteins, such as fibronectin, laminin, and entactin. Although many integrin heterodimers show constitutive expression in the uterine epithelium or stroma, 1ß1, 3ß1, 6ß1, vß3, and vß1 heterodimers exhibit cycle-dependent changes (115, 116, 117, 118, 119). Of special interest is the expression of 1ß1, which shows implantation-related changes. For example, the expression of 1ß1 is restricted to early- and midsecretory phases both in the epithelium and stroma and primarily restricted to the stroma during the predecidual phase (117). Furthermore, unexplained infertility in women is associated with the deficiency of vß3 or 4ß1 in the uterus during the window of implantation (120). However, the functional definition of these markers in uterine receptivity or under pathological conditions still awaits further investigation. In mice, vß3 is expressed in both the uterine luminal epithelium and the blastocyst during implantation. It has been shown that an intrauterine injection of RGD peptide or neutralizing antibody against vß3 reduces the number of implantation sites in mice and rabbits (121).

Among many subunits, 5ß1, 6ß1, and vß3 are expressed in the mouse embryo throughout the periimplantation period, whereas several others exhibit stage-specific expression (122). Integrins are also expressed in the differentiating trophoblasts at later stages (122), suggesting their roles in trophoblast differentiation and adhesion. A role for fibronectin via integrin binding in blastocyst outgrowth was further confirmed in vitro using antibodies against the v, 5, ß1, or ß3, which inhibited adhesiveness on the outer surface of the trophoblast inducible by fibronectin (123). In addition, a gene-targeting experiment revealed that deletion of the ß1 gene results in ICM defects and embryonic lethality (124). However, the mutant embryos form morphologically normal blastocysts and initiate implantation, but trophoblast invasion becomes defective (124). Adhesion-competent, late-blastocyst-stage trophoblasts undergo intracellular signaling initiated upon ligation of 5ß1 and vß3 by fibronectin (125). Integrin signaling mobilizes cytoplasmic Ca²⁺ and induces the trafficking of intracellular vesicles, resulting in stronger adhesion to fibronectin at the apical surface. Therefore, blastocyst adhesion to the endometrium during implantation is considered to be regulated by the endogenous developmental program, as well as through interactions with ECM components in the local environment (126, 127). Although there is evidence that the embryo is a site of action for integrin signaling, it is not yet clear whether the uterus is a site of action. Results of gene-targeting experiments of integrin subunits are not very informative in relation to their roles in implantation because of the complex phenotypes and apparent compensation by other subunits (128, 129, 130, 131).

Invasive mouse trophoblasts adhere, spread, and migrate on ECM substrates (132, 133, 134, 135) and penetrate three-dimensional ECM structures (136, 137). Several ECM components that are up-regulated in the periimplantation endometrium, including fibronectin, laminin, and collagen type IV (138, 139, 140), support trophoblast outgrowth in vitro (132, 133). Trophoblast interactions with the ECM are mediated primarily by integrins (122, 123, 133, 134, 135, 141). Hexapeptides containing the RGD sequence recognized by integrins (142) block trophoblast outgrowth on fibronectin, collagen type II and IV, entactin, and vitronectin (133, 135, 141). However, trophoblast adhesion to type I laminin is independent of its RGD sequence and is primarily mediated through interaction of 7ß1 with the E8 integrin-recognition domain of laminin (143, 144).

Trophinin was identified by cDNA library screening of a human trophoblastic cell line (145). This transmembrane protein can mediate homophilic interactions between two different cell types. For example, it mediates an interaction between a human endometrial cell line and a trophoblastic cell line, but this interaction is complex (145). Trophinin requires the presence of a cytoplasmic protein tastin to sustain adhesion between these two cell types. In addition, the presence of bystin, another cytoplasmic protein, is required for effective interaction between trophinin and tastin. This adhesion complex, which is present in both trophoblastic teratocarcinomas and endometrial adenocarcinomas, mediates adhesion between them. In humans and monkeys, trophinin is specifically expressed in cells involved in implantation. Furthermore, the trophinin complex was detected in both trophoblast and decidual cells at the human fetalmaternal interface as early as the sixth week of pregnancy (146). Although trophinin expression in mice coincides with the timing of implantation, it is expressed in the luminal and glandular epithelium throughout the uterus irrespective of the presence or absence of blastocysts, raising doubts regarding its role specific to implantation (147).

E-cadherin, a calcium-dependent cell-cell adhesion molecule, participates in the formation of the epithelial adherens junctions in cooperation with - and ß-catenins (148, 149). E-cadherin is a critical factor for blastocyst formation, because its targeted deletion leads to defective embryonic development resulting in failure to form the trophectoderm (150, 151). E-cadherin is implicated in uterine-embryo interactions because of its homotypic adhesive activity (152). Whereas the trophectoderm highly expresses E-cadherin, the components of the adherens junctional complex are also expressed in the uterine luminal epithelium at the time of the attachment reaction. The expression subsequently becomes evident in the subepithelial stroma surrounding the implanting blastocysts with apoptosis occurring in the luminal epithelium (152, 153). Therefore, it is speculated that temporal and cell-specific expression of the adherence junction proteins in the uterus results in molecular guidance, which is important for blastocyst attachment and subsequent invasion.

A recent study has highlighted the critical role of the selectin adhesion system in human implantation (154). This adhesion system, which is involved in leukocyte capture from the bloodstream, is also operative during implantation and placentation. On the maternal side, selectin oligosaccharide ligands are expressed in the receptive uterine epithelium, and on the embryonic side, trophoblast cells expressed L-selectin receptors. It was observed that beads coated with specific selectin ligands adhere to the trophoblast, suggesting that the trophoblast cell surface receptors are functional. This investigation suggests that trophoblast L-selectin mediates interactions with the uterus to establish an adhesion mechanism for implantation. This is an exciting finding in several respects. First, this study affirms that specific ligand-receptor signaling pathways between the embryo and uterus are critical for implantation and subsequent pregnancy establishment. Second, it shows that the same mechanism operative during implantation is also operative during the later phases of pregnancy. Third, it shows that trophoblast cells share a system known to be active in the blood-vascular system. Similar adhesion signaling between the transmembrane form of heparin-binding EFG-like growth factor (HB-EFG) expressed on the luminal epithelial surface and ErbB receptors present on the trophectoderm cell surface for implantation in mice and humans has been reported previously (57, 155).

C. Signaling by vasoactive factors
It has long been speculated that vasoactive agents, such as histamine and PGs, are involved in many aspects of reproduction including ovulation, fertilization, implantation, and decidualization. Histamine functions as a ubiquitous mediator of cell-cell signaling and is synthesized from L-histidine by histidine decarboxylase (HDC) both in peripheral tissues and in the nervous system (156). Histamine is a well-known neurotransmitter in the brain (157), but it is also involved in other physiological responses including gastric acid secretion, regulation of allergic reactions, and vascular permeability (reviewed in Ref. 158). Because the process of implantation is considered a proinflammatory reaction and because increased vascular permeability at the site of blastocyst implantation is common to many species, it was suggested that histamine plays a role in implantation and decidualization (158). Earlier observations suggested that uterine mast cells are a possible source of histamine, and its release from mast cells by estrogen is important for implantation (159, 160). This suggestion was based on the observations that local histamine application stimulates uterine hyperemic and edematous responses (161) and that there is a reduction in uterine mast cell numbers and histamine content after estrogen treatment and during implantation (162). Furthermore, a histamine antagonist pyrathiazine or an inhibitor of HDC was shown to interfere with implantation when instilled into the uterine lumens of rats and rabbits (159, 163). Histamine works via at least four histamine receptor subtypes (H₁, H₂, H₃, and H₄) (164, 165, 166) and blocking both H₁ and H₂ receptors was shown to interfere with implantation (167). Subsequent studies also showed that histamine induces implantation in delayed implanting rats when injected with a suboptimal dose of estrogen (168). However, successful implantation and birth of live offspring in mast cell-deficient mice and other evidence suggest that uterine mast cell histamine is not essential for implantation (136, 169, 170). Thus, if histamine is involved in implantation, it should be provided either by major uterine cell types or by embryonic cells. A recent study showed that mouse blastocysts do not have the capacity for histamine synthesis (59). However, HDC is expressed in uterine epithelial cells on d 4 of pregnancy in mice before implantation, but not in decidual cells (171). Thus, although histamine may have a role in implantation in mice, its role in decidualization is unlikely. Whereas H₁, H₂, and H₃ receptor subtypes are not detectable in the uterus, H₂ receptors are expressed in preimplantation mouse blastocysts. These observations, as well as the inhibition of blastocyst zona-hatching and implantation by H₂ antagonists and an HDC inhibitor, suggest that uterine histamine targets the blastocyst for implantation (59). However, apparently normal implantation occurs in mice lacking HDC or H₂ type histamine receptor genes, suggesting the possible involvement of other vasoactive agents with overlapping functions in this process (165, 172).

PGs possess vasoactive, mitogenic, and differentiating properties (173) and are implicated in various female reproductive functions. COX, which exists in two isoforms, COX-1 and COX-2, is the rate-limiting enzyme in the biosynthesis of PGs. COX mediates the conversion of arachidonic acid into PGH₂, which is then converted to various PGs by specific synthases (173). The COX isoforms are encoded by two separate genes and exhibit distinct cell-specific expression, regulation, and subcellular localization, yet share similar structural and kinetic properties. COX-1 is considered to be a constitutive enzyme that mediates housekeeping functions. In contrast, COX-2 is an inducible enzyme and is induced in a variety of cell types by growth factors, cytokines, and inflammatory stimuli (173). PGs normally exert their function by interacting with cell surface G protein-coupled receptors, but they can also function as ligands for nuclear peroxisome proliferator-activated receptors (PPARs) (174, 175, 176, 177). Because COX-2 is primarily responsible for increased PG production during inflammation, this isoform is the target for development of selective antiinflammatory drugs (178, 179). COX-2 overexpression is also associated with tumorigenesis (180, 181).

The processes of ovulation and implantation are considered analogous to proinflammatory responses, and thus participation of PGs in these events has been speculated (182, 183). For example, PGs are considered to participate in follicular rupture during ovulation (reviewed in Ref. 184). This is consistent with gonadotropin-mediated induction of COX-2 in ovarian follicles preceding ovulation (85, 184). PGs are also implicated as important mediators of increased endometrial vascular permeability during implantation and decidualization (185). A unique pattern of expression of Cox-1 and Cox-2 genes in the periimplantation mouse uterus further suggests that PGs play important roles in these processes (185). Cox-1 is expressed in uterine luminal and glandular epithelial cells on the morning of d 4 of pregnancy, but its expression becomes undetectable in the luminal epithelial cells by the time of the attachment reaction. In contrast, Cox-2 is expressed in the luminal epithelium and underlying stromal cells solely at the site of blastocyst attachment. Using the delayed implantation model, this study also established that the expression of Cox-2 in the receptive uterus requires the presence of active blastocysts. The results suggested that Cox-2 expression during the attachment reaction is critical to implantation (185). Indeed, gene-targeting experiments have demonstrated that COX-2-derived PGs are essential for implantation and decidualization (186, 187, 188). Experiments with Cox-1(–/–) mice suggest that the loss of COX-1 is compensated by the expression of Cox-2 for implantation (189). Among various PGs, the levels of prostacyclin (PGI₂) are highest at the implantation sites of wild-type mice, and implantation defects are partially restored in Cox-2(–/–) mice by administration of a more stable PGI₂ agonist, carbaprostacyclin (190).

The role of PGs is further illustrated by the reduced fertility of female mice lacking cytoplasmic phospholipase A₂, which is involved in the liberation of arachidonic acid from membrane phospholipids for PG synthesis by the COX system (191, 192, 193). The reduced fertility in these females is due to deferral of on-time implantation, leading to subsequent retarded fetoplacental development and reduced litter size (54). Collectively, these results indicate that the cytoplasmic phospholipase A₂-COX-2 axis is crucial to implantation. However, one recent study described that wild-type (B6C3H) blastocysts transferred into COX-2-deficient female mice on a mixed (C57Bl6/JX129S7/SvEvBrd) background on d 3 of pseudopregnancy implanted and produced live offspring, although decidual growth was retarded (194). Because this study did not follow experimental protocols similar to other studies, the discrepancy between this study and others cannot be compared scientifically. However, some of the results reported by Cheng and Stewart (194) remain uninterpretable. For example, dissected decidual weights reported by these investigators seem abnormally high on various days of pregnancy, and the sizes of d 7 implantation sites shown appeared larger than d 8 implantation sites and at a more advanced stage of pregnancy. Nonetheless, recent evidence demonstrates that Cox-2 is expressed either in the uterus, blastocyst, or both during implantation in a variety of species with different modes of implantation, including sheep, mink, skunk, baboon, and pig (195, 196, 197, 198). COX-2 expression in human endometrium has also been reported (199, 200). These results suggest a conserved function of COX-2 in implantation in various species. However, it has been discovered recently that, depending on the genetic background, COX-1 can rescue female infertility in COX-2-deficient mice (201).

PGs exert diverse functions using both cell surface PG receptors and PPARs. Receptors for PGE₂, PGF₂, PGD₂, PGI₂, and thromboxanes have been named as EP₁–EP₄, FP, DP, IP, and TP, respectively; they belong to the G protein-coupled family of cell surface receptors (reviewed in Refs. 202 and 203). Although PGE₂ synthase is expressed at the implantation sites with the presence of PGE₂ and EP receptors (203, 204, 205) and although PGE₂ has been shown to be associated with implantation and decidualization (206), gene-targeting experiments show that three of the four EP receptor subtypes (EP₁–EP₃) are not critical for implantation. EP₄ deficiency results most frequently in perinatal lethality, and thus its role in implantation has not yet been determined (reviewed in Ref. 203). Furthermore, mice deficient in FP or IP show normal implantation. PGs can also exert their effects by utilizing PPARs that belong to a nuclear hormone receptor superfamily. The evidence that PG-mediated PPAR signaling is involved in implantation is discussed below in more detail. PGs also appear to be important for embryonic functions relevant to preimplantation embryo development and implantation. Preimplantation embryos produce PGs, and inhibitors of PG synthesis have been shown to inhibit embryonic growth, functions, and zona hatching in vitro (207 , 208). Dormant mouse blastocysts can achieve implantation competence if cultured in the presence of PGE₂ or a permeable analog of cAMP. This effect apparently involves the COX-2 signaling pathway (17). However, normal development of Cox-1(–/–) /Cox-2(–/–) double-mutant embryos in the uterus suggests that PGs of embryonic origin are not essential for embryo development (208). However, compensation by maternal PGs in embryonic development cannot be ruled out. Vascular endothelial growth factor (VEGF), also known as vascular permeability factor, is highly vasoactive in nature. It is a potent inducer of vasodilation and angiogenesis. Its role in implantation is discussed below.

D. Signaling by growth factors
The expression of various growth factors and their receptors in the uterus in a temporal and cell-specific manner during the periimplantation period suggests that these factors are important for implantation (3, 4, 5, 6, 209, 210). The present review highlights primarily the importance of the EGF family of growth factors; however, the roles of other growth factors, such as TGFßs, fibroblast growth factors (FGFs), IGFs, platelet-derived growth factors, and many others should not be ignored. The EGF family of growth factors includes EGF itself, TGF, HB-EGF, amphiregulin, betacellulin, epiregulin, and neuregulins (58, 211). HB-EGF is the earliest molecular marker found in the uterus exclusively at the sites of active blastocysts appearing several hours before the attachment reaction in mice (26). This induction is followed by the expression of betacellulin, epiregulin, neuregulin-1, and Cox-2 around the time of the attachment reaction (58, 185, 211). In contrast, amphiregulin is expressed throughout the uterine epithelium on the morning of d 4 of pregnancy and is well characterized as a progesterone-responsive gene in the uterus (212). Around the time of the attachment reaction, strong expression of amphiregulin in the luminal epithelium is found only around the implanting blastocysts, and this expression is absent by the morning of d 5. Although these results suggested that amphiregulin has a role in implantation, amphiregulin-deficient mice or compound knockout mice for EGF/TGF/amphiregulin do not exhibit implantation defects (213, 214). Because HB-EGF, betacellulin, epiregulin, neuregulin, and amphiregulin all show overlapping uterine expression patterns around the implanting blastocyst at the time of attachment reaction (reviewed in Refs. 3 and 211), it is assumed that a compensatory mechanism rescues implantation in the absence of one or more members of the EGF family.

The EGF-like growth factors interact with the receptor subtypes of the erbB gene family, which is comprised of four receptor tyrosine kinases: ErbB1 (EGF-R), ErbB2, ErbB3, and ErbB4. They share common structural features but differ in their ligand specificity and kinase activity (215). The initial dimerization between coexpressed receptors upon ligand binding constitutes the classical mechanism of action of EGF-like ligands. Spatiotemporal expression patterns of EGF gene family members and ErbBs in the uterus during the periimplantation period suggest compartmentalized functions of EGF-like growth factors in implantation (58).

A number of growth factors and their receptors are expressed in preimplantation embryos of several species, suggesting their roles in preimplantation mammalian development (216, 217). In this review, we focus on potential roles of the EGF family of ligands with respect to preimplantation embryo development and implantation. ErbB1 (EGF-R), ErbB2, and ErbB4, the receptor subtypes for the EGF family of growth factors, are expressed in the mouse blastocyst (Refs. 56 and 218 and our unpublished results), and EGF or TGF has beneficial effects on embryonic development in vitro (18). Using genetic and biochemical approaches, the roles of embryonic ErbB1 and/or ErbB4 in interacting with uterine HB-EGF in blastocyst implantation have recently been highlighted in mice (57, 218). HB-EGF is expressed as soluble and transmembrane forms in the uterine luminal epithelium at the site of the blastocyst before the attachment reaction, suggesting paracrine and/or juxtacrine interactions with embryonic ErbBs, as well as autocrine, paracrine, and/or juxtacrine interactions with uterine ErbBs that are expressed in a spatiotemporal manner during the periimplantation period (26, 58, 219). For example, the expression of both ErbB1 and ErbB4 is down-regulated in dormant blastocysts during delayed implantation but is readily up-regulated with blastocyst activation and initiation of implantation (56, 218). Furthermore, whereas a recombinant soluble HB-EGF can promote blastocyst growth and differentiation (26), cells that express the transmembrane form of HB-EGF can adhere to active, but not dormant, blastocysts in vitro (57), suggesting paracrine and juxtacrine functions of HB-EGF. In addition, by directing an HB-EGF-toxin conjugate toward wild-type and erbB1(–/–) blastocysts, it was shown that HB-EGF could also interact with embryonic ErbB4 and heparan sulfate proteoglycan molecules (218). Collectively, these results suggest that an interaction between uterine HB-EGF and blastocyst ErbBs is important for the attachment reaction. However, the absolute necessity of HB-EGF in implantation requires genetic evidence. A recent report shows that most HB-EGF mutant mice die early in postnatal life due to cardiac defects, precluding critical examination of the implantation phenotype (220). It is also to be noted that early events of implantation do not appear to be affected by blastocysts deficient in either ErbB1 or ErbB4 (221, 222), although the implantation-initiating efficiency of blastocysts deficient in more than one receptor type needs to be tested to delineate the functional redundancy among the receptor family. In conclusion, detailed expression and gene-targeting experiments with all of the ligands and receptors are required to define paracrine, autocrine, and/or juxtacrine roles of specific ligand or its receptors in implantation.

Among many growth factors that have been studied in humans, HB-EGF appears to play a role in implantation and embryonic development. Its expression is maximal during the late secretory phase (cycle d 20–24) when the endometrium becomes receptive for implantation (223, 224) and cells expressing the transmembrane form of HB-EGF adhere to human blastocysts displaying cell surface ErbB4 (155). Furthermore, HB-EGF was shown to be one of the most potent growth factors for enhancing the development of human in vitro fertilization-derived embryos to blastocysts and subsequent zona hatching (225). Thus, cumulative evidence suggests that HB-EGF has a significant role in preimplantation embryo development and implantation as a paracrine and/or juxtacrine factor in various species.

E. Signaling by cytokines
The expression of various cytokines and their receptors in the uterus and embryo during early pregnancy suggests their roles in various aspects of implantation (reviewed in Refs. 3 , 20 , 225 , and 226). However, gene-targeting studies show that mice lacking TNF, IL-1ß, IL-1 receptor antagonist, IL-1 receptor type 1, IL-6, and granulocyte/macrophage-colony stimulating factor apparently do not manifest overt reproductive defects (reviewed in Ref. 20). These observations suggest that either these molecules have minor roles in implantation or the loss of one cytokine is compensated by other cytokines with overlapping functions. In contrast, some cytokines are important for normal female fertility (227, 228, 229). For example, female op/op mice with a naturally occurring mutation of the M-CSF gene have markedly impaired fertility (227), and mice with a null mutation of the Lif gene encoding leukemia inhibitory factor (LIF) show complete failure of implantation, and blastocysts in these mutant mice undergo dormancy (228, 230). Studies using IL-11R mutant mice have also shown that IL-11 is crucial to decidualization, but not for the attachment reaction (229). Interestingly, both LIF and IL-11 are members of the IL-6 family, which includes IL-6 itself, oncostatin M, ciliary neurotrophic factor, and cardiotrophin (231). LIF and IL-11 bind to ligand-specific receptors, LIFR and IL-11R, respectively, and share gp130 as a signal transduction partner (231), suggesting that gp130 signaling is critically involved in implantation. Although the mechanism underlying implantation and decidualization failures in the absence of LIF still remains to be elucidated, recent evidence shows that there is a loss or an aberrant expression of certain implantation-related genes in pregnant Lif mutant mice (232). For example, uterine expression of HB-EGF and epiregulin is absent, and Cox-2 expression is aberrant at the sites of blastocysts in Lif mutant mice during the anticipated time of implantation.

LIF and its receptors, LIFR and gp130, exist in both soluble and membrane-bound forms, and soluble forms of these two receptors antagonize the actions of their ligands, implying the complexity of the LIF signaling pathway (233, 234, 235). Lif is transiently expressed in uterine glands on d 4 of pregnancy in mice, suggesting its role in implantation (236). However, our recent studies show that uterine Lif expression is biphasic on d 4. Not only is Lif expressed in glands, but it is also expressed in stromal cells surrounding the blastocyst at the time of the attachment reaction (232). This suggests that LIF has dual roles: first in the preparation of the uterus, and later in the attachment reaction. However, the molecular mechanism by which LIF executes its effects on implantation is not yet known. In this regard, it would be useful to establish the complicated ligand-receptor interactions and detailed expression patterns of LIF receptors that occur during the periimplantation period. However, a recent report shows that inactivation of gp130 by deleting all signal transducers and activators of transcription binding (STAT) sites results in implantation failure (237), reinforcing the importance of LIF signaling in implantation.

The uterine milieu in Lif mutant mice fails to induce implantation irrespective of the blastocyst genotypes, since Lif(–/–) blastocysts can implant after transfer to wild-type pseudopregnant recipients (228, 230). These reciprocal embryo transfer experiments suggest that maternal LIF is essential for blastocyst implantation. However, a role for this cytokine in embryonic functions cannot be ignored, because LifR and gp130 are expressed at the blastocyst stage, and administration of exogenous LIF improves embryo viability and hatching in several species (238, 239, 240). Taken together, these data suggest that both the preimplantation embryo and the uterus are sites of LIF action. However, embryos lacking either LIFR or gp130 develop to the blastocyst stage and implant normally but die during the perinatal period (241, 242). These results raise questions about the role of LIF signaling in preimplantation embryo development.

Lif expression in the uterus is maximal around the time of implantation in most species examined, although the steroid hormonal requirements for the preparation of uterine receptivity and implantation differ depending on the species. Whereas uterine Lif expression in several species appears to be regulated by P₄ (reviewed in Ref. 242), estrogen regulates Lif expression in the mouse uterus. This is evident from Lif expression on d 1 of pregnancy and during the estrous stage of the cycle when the uterus is under the influence of estrogen stimulation (236, 243, 244). In addition, Lif is not expressed in the uterus during experimentally induced delayed implantation but is rapidly induced by an injection of estrogen (232, 236). However, it has yet to be learned how estrogen induces Lif expression in the mouse uterus and the mechanism by which it is regulated by progesterone in other species. In humans, Lif is expressed in the endometrium and at higher levels in the glandular epithelium of the secretory endometrium (245). Furthermore, LIF deficiency has been associated with unexplained recurrent abortions and infertility in women (246).

F. Homeobox genes in implantation
Hox genes are transcription factors that belong to a multigene family. They are developmentally regulated and share a common highly conserved sequence element called the homeobox that encodes a 61-amino acid helix-turn-helix DNA-binding domain (247). Hox genes are organized in four clusters (A, B, C, and D) on four different chromosomes in mice and humans and follow a stringent pattern of spatial and temporal colinearity during embryogenesis (247). Several Hox genes at the 5'-end of each cluster are classified as AbdB-like Hox genes, because of their homology with the Drosophila AbdB gene. In vertebrates, AbdB-like Hox genes, similar to their Drosophila ortholog, are expressed in developing genitourinary systems (248). For example, Hoxa-10 and Hoxa-11 are highly expressed in developing genitourinary tracts and the adult female reproductive tract, suggesting roles in reproductive events (248, 249, 250). Hoxa-10 mutant mice exhibit oviductal transformation of the proximal one third of the uterus. Furthermore, adult female mice deficient in Hoxa-10 show failures in blastocyst implantation and decidualization unrelated to the oviductal transformation (248). Subsequent studies revealed that uterine stromal cells in Hoxa-10-deficient female mice show reduced proliferation in response to progesterone, leading to decidualization defects (248, 251). The uterus in Hoxa-11-deficient mice is hypoplastic and devoid of uterine glands due to developmental defects. Defective proliferation of stromal cells in Hoxa-10(–/–) female mice suggests that Hoxa-10 is involved in the local events of cellular proliferation by regulating cell cycle molecules. Indeed, cyclin D3 is aberrantly expressed in Hoxa-10 mutant uteri in response to a decidualizing stimulus (252). Furthermore, because several progesterone-responsive genes are dysregulated in the uterine stroma of Hoxa-10 mutant mice (251), Hoxa-10, as a transcriptional factor, may convey progesterone responsiveness in the uterine stroma by regulating gene expression. A similar, but more severe, phenotype was also noted in Hoxa-11-deficient female mice (250).

A recent study using microarray analysis further revealed that the absence of Hoxa-10 in the uterus in response to progesterone is associated with two cellular disturbances (253). First, among many genes that were up-regulated in the Hoxa-10-deficient uteri, two cell cycle molecules, p15 and p57, were notable. These two genes are both cyclin-dependent kinase inhibitors (CKIs), suggesting that the previously observed defect in stromal cell