Cell signaling in trophoblast-uterine communication.

A critical pathway that regulates embryo-uterine interactions and placentation is Wnt signaling. The term “Wnt” combines the Wg (Wingless) and Int1 genes, which were found to be homologous ( Nusse et al ., 1991 ). The Wnt pathway includes a large number of ligands and receptors (19 and 10 respectively), resulting in 190 different potential interactions, which can be enhanced or attenuated by additional co-receptors and co-activators with multiple signaling activities ( van Amerongen et al ., 2012 ). The Wnt signaling pathway is pleiotropic, as abnormalities in Wnt signaling are associated with cancers, osteoperosis, degenerative diseases and developmental disorders ( Sonderegger et al ., 2010 ). Wnt signaling operates through both canonical and non-canonical pathways. The canonical pathway functions through β-catenin, as shown in Fig. 2 . In its unstimulated state, β-catenin is complexed with adenomatous polyposis coli (APC), Axin, GSK3b and CK1a, which mediates constitutive degradation of β-catenin ( Liu et al ., 2002 ). Upon binding of Wnt ligands to transmembrane frizzled receptors (FZD) and low density lipoprotein receptor-related protein coreceptors−5/6 (LRP5/6), β-catenin is liberated from its complex ( Metcalfe and Bienz, 2011 ), accumulates intracytoplasmically and translocates into the nucleus where it directs transcription of target genes ( Behrens et al ., 1996 ). Several diverse non-canonical path ways operate independently of β-catenin ( Semenov et al ., 2007 ). Wnt pathways are crucial for blastocyst-uterine communication and subsequent implantation. Mouse morula and blastocyst stage embryos express Wnt genes in response to uterine factors induced by estrogen prior to implantation ( Mohamed et al ., 2004 ). In mice, the canonical β-catenin pathway is dynamically activated in the uterus, initially in the circular smooth muscle of the myometrium, followed by activation in the luminal epithelium directly opposed to the blastocyst at implantation sites, both of which require the presence of a blastocyst ( Mohamed et al ., 2005 ). Inhibition of the Wnt/b-catenin pathway prevents implantation ( Mohamed et al ., 2005 ). Interestingly, it has been shown that interruption of nuclear β-catenin signaling in the developing embryo does not adversely affect the development of the embryo to the blastocyst stage; however it does interfere with implantation ( Xie et al ., 2008 ). Through embryo transfer experiments, these investigators found that implantation failure is due to silencing of the canonical pathway in the blastocyst rather than its deficiency in the acquisition of uterine receptivity. Wnt signaling is also necessary in uterine decidualization. Uterine Wnt4 expression is absent prior to implantation, but increases during the period of receptivity for implantation and continues to be expressed in the decidua ( Paria et al ., 2001 ). Additionally, Wnt 5a, Wnt 7a, Wnt 11, and Wnt 16 are expressed in the mouse uterus during the window of implantation ( Hayashi et al ., 2009 ). Wnt signaling systems are critical for proper placentation and trophoblast function in mice. Wnt 2 deficient mice have a 50% prenatal mortality due to placental defects with histological findings of increased fibrinoid and decreased capillary formation within the placentas ( Monkley et al ., 1996 ). Furthermore, mice lacking R-spondin3, an activator of the WNT/β-catenin pathway, die prenatally due to improper placental development ( Aoki et al ., 2007 ), and FZD5 knockout mice die prenatally with placental abnormalities ( Ishikawa et al ., 2001 ). Wnt7b is required for fusion of the chorion and allantois and Wnt7b deficient mice die prenatally due to placental defects ( Parr et al ., 2001 ). In humans, numerous Wnt ligands and receptors have been identified in the endometrium, indicating their importance for endometrial function ( Tulac et al ., 2003 ). Wnt signaling in human endometrium during implantation differs from mice. In mice, proper uterine decidualization and implantation is estrogen dependent, and estrogen dependent Wnt signaling and β-catenin expression is crucial. In humans, however, decidualization and implantation are progesterone driven. Studies have shown that during the early luteal to mid luteal phase of human endometrium, Dickkopk1 (DKK1), an inhibitor of the canonical Wnt pathway, is significantly upregulated ( Carson et al ., 2002 ). This is likely progesterone driven, as progesterone can increase DKK1 expression in human endometrial stromal cells. Furthermore, after treatment with mifepristone, a progesterone antagonist, an increase in β-catenin, Wnt ligands and Wnt receptors have been observed in human endometrium ( Catalano et al ., 2007 ), suggesting that progesterone driven human decidualization down regulates the canonical Wnt pathway. Abnormalities in Wnt signaling could contribute to infertility. Significantly higher levels of β-catenin expression have been observed in the endometrium of the mid-secretory phase in infertile patients with endometriosis, compared to fertile counterparts ( Matsuzaki et al ., 2010 ). Impaired Wnt signaling could also contribute to the etiology of endometriosis, as DKK1 mRNA expression was significantly decreased in endometriotic fibroblast of endometriosis patients compared to women without disease ( Aghajanova et al ., 2010 ). This deficiency could play a role in the development of endometriosis and in maintaining a proliferative phase of the lesions. As found in mice, Wnt signaling is essential for human placentation. Expression of 14 out of 19 Wnt ligands and 8 out of 10 FZD receptors has been documented in human placentas ( Sonderegger et al ., 2007 ). Interestingly, certain Wnts, including Wnt1, Wnt7b, Wnt10a, and Wnt10b, were absent in term placentas, although they are expressed in first trimester placentas, suggesting a role in the critical early period of placental development and trophoblast differentiation ( Sonderegger et al ., 2007 ). Wnt3a can stimulate trophoblast invasion, which could be blocked by DKK1 ( Pollheimer et al ., 2006 ). Methylation of certain genes encoding inhibitors of Wnt signaling has been found in human placentas and trophoblast cells, further supporting a role of Wnt signaling in trophoblast invasion ( Novakovic et al ., 2008 ). Interestingly, hypermethylation of the APC gene promoter, an inhibitor of b-catenin, has been reported in choriocarcinoma trophoblast cell lines ( Wong et al ., 2008 ). Similarly, DKK1 is decreased in choriocarcinoma cell lines, and its over-expression induces mitotic arrest ( Peng et al ., 2006 ). Nuclear localization of β-catenin is significantly elevated in invasive trophoblasts of complete hydatidiform mole, compared to normal placentas ( Pollheimer et al ., 2006 ), suggesting that although activation of the Wnt system is critical for placentation, over activation could cause excessive invasiveness that may be associated with carcinomas. Conversely, repression of the canonical Wnt pathway could lead to pregnancy related complications associated with abnormal placentation. Wnt2 and β-catenin are both decreased in placentas of women with preeclampsia, whereas DKK1 is increased ( Zhang et al ., 2013b ). High levels of DKK1 were also found in patients with recurrent miscarriages ( Bao et al ., 2013 ). Wnt signaling plays crucial roles in many aspects of reproduction and in humans its inhibition and activation must be properly balanced for normal reproductive function. Inhibition of the Wnt canonical pathway is critical for proper endometrial decidualization and implantation, whereas excessive activation is linked with infertility. In placentation, activation of the canonical pathway is necessary for normal placental development and differentiation; however, overexpression may lead to rare placental carcinomas.

Another secreted factor shown to be necessary for implantation and placentation is Indian Hedgehog (IHH). In humans, blastocyst implantation is progesterone dependent ( Lee et al ., 2006b ). IHH is a key element in the progesterone priming of the uterus for implantation ( Zhang et al ., 2013a ). The highly conserved hedgehog family of genes was discovered in 1980 in a study of abnormalities in the Drosophila body plan ( Nusslein-Volhard and Wieschaus, 1980 ). The requirement of IHH for mouse development was first described in 1993 ( Echelard et al ., 1993 ). Since then IHH has been linked to the formation of many organs, including bone, intestines, and heart ( Ingham and McMahon, 2001 ). In mice, IHH increases in response to progesterone in the uterine glandular epithelium during the implantation window ( Takamoto et al ., 2002 ). The hedgehog receptor, PATCHED (PTC), and GLI-Kruppel family transcription factors, GLI1, 2 and 3, are simultaneously upregulated in the underlying stroma ( Matsumoto et al ., 2002 ). The epithelial-stromal interaction mediated by progesterone and IHH is regulated by chicken ovalbumin upstream promoter transcription factor II (Coup TF II), a transcription factor found in the uterine stroma of mice ( Kurihara et al ., 2007 ). Deletion of COUP TFII results in decidualization and implantation failure, similar to that of IHH deletion, indicating that IHH may work in a paracrine fashion within the uterus to orchestrate progesterone-induced inhibition of epithelial proliferation necessary for implantation ( Matsumoto et al ., 2002 ). The uterine effects of IHH knockout mice cannot be studied due to its lethality; however, targeted ablation of IHH in progesterone positive uterine cells shows that these mice are infertile due to impaired implantation, resembling mice with progesterone receptor knockout uteri ( Lee et al ., 2006b ). In the human endometrium, IHH is upregulated in conjunction with progesterone receptor during the secretory phase ( Talbi et al ., 2006 ). IHH, PTC, GLI1 and GLI2 all increase during the secretory phase ( Wei et al ., 2010 ). In contrast to the mouse, IHH mRNA has been found in both the stroma and epithelium in human endometrium. This provides evidence that in humans, similar to mice, IHH is likely upregulated by progesterone and plays an essential role in uterine decidualization and implantation. Interestingly, it has been shown that IHH is abnormally expressed in the endometrium of endometriosis patients as compared to healthy controls, implicating a role in the decline of fertility experienced by endometriosis patients and possibly others, as well ( Smith et al ., 2011 ). Pathology occurs throughout the reproductive tract, causing decreased fertility rates in these patients. Recently, impaired endometrial receptivity has been postulated ( Wei et al ., 2009 ). It would be instructive to know whether IHH signaling is disrupted in patients with progesterone resistance ( Al-Sabbagh et al ., 2012 ).

Leukemia inhibitory factor (LIF), a pleiotropic cytokine, is essential in the early stages of pregnancy. LIF is a member of the interleukin-6 family that is highly glycosylated and exerts its physiological effects by interacting with the LIF receptor (LIFR) and gp130 to activate the Jak/STAT downstream pathway ( Heinrich et al ., 1998 ). The first evidence linking LIF to implantation was obtained in experiments with transgenic mice. Blastocyst implantation does not occur in LIF gene deficient mice; however, their embryos are able to implant in non-LIF deficient mice ( Stewart et al ., 1992 ). Furthermore, delivering exogenous LIF to LIF deficient dams enables implantation ( Stewart et al ., 1992 ). It has also been shown that inhibition of the LIF-dependent STAT 3 pathway in mice significantly reduces embryo implantation ( Catalano et al ., 2005 ). In humans, evidence suggests that LIF plays a crucial role in implantation and is secreted maximally during the luteal phase. LIF mRNA is detected in the human endometrium, increasing during the secretory phase ( Arici et al ., 1995 ). More specifically, LIF and LIFR are maximally expressed during the “window of implantation” and correspond with the presence of pinopodes in endometrial biopsies ( Aghajanova, 2004 ). Immunohistochemical analysis of LIF in 89 biopsies of fertile women reveals its highest expression during the mid-luteal phase in the luminal epithelium, significantly greater than in glands or stroma ( Leach et al ., 2012 ). Thus, LIF is positioned for signaling directly to the peri-implantation blastocyst. LIF secretion by the glandular endometrium depends on progesterone, as treatment with mifepristone, a progesterone receptor antagonist, reduces LIF secretion from glandular endometrium ( Danielsson et al ., 1997 ). Endometrial LIF transcription and biosynthesis is complex and is regulated by various growth factors and cytokines, including interleukin-1, tumor necrosis factor (TNF), platelet-derived growth factor and epidermal growth factor (EGF), while it is inhibited by interferon gamma ( Lass et al ., 2001 ). LIF plays a critical role in uterine communication to the embryo, as it is secreted from the endometrium and human embryos contain LIFR ( Charnock-Jones et al ., 1994 ). Earlier in embryonic development, LIF is secreted from the ampullary portion of the fallopian tube ( Keltz et al ., 1996 ). Furthermore, LIF concentration in follicular fluid positively correlates with embryo quality ( Arici et al ., 1997 ). Interestingly, supplementation of serum-free culture medium with LIF increases human blastocyst formation, as compared to non-supplemented controls ( Sargent et al ., 1998 ). These findings suggest that LIF could regulate embryonic growth and development in preparation for implantation. Any aberration in LIF secretion is likely to have an adverse effect on fertility. In fertile patients, LIF increases two-fold between proliferative and secretory phases, whereas it fails to increase in infertile women ( Hambartsoumian, 1998 ). Even during the proliferative phase, overall endometrial LIF levels decrease in infertile women as compared to fertile women ( Wu et al ., 2013 ). It has also been suggested that heterozygous mutations of the LIF gene could lead to decreased availability or decreased biological activity of LIF in the uterus and contribute to infertility in women ( Giess et al ., 1999 ). Significantly better in vitro fertilization (IVF) outcomes are associated with elevated LIF expression in the endometrium of infertility patients ( Serafini et al ., 2009 ). The presence of LIF and LIFR in decidua and placenta implicates a role not only in implantation, but also in placentation ( Kojima et al ., 1994 , Sharkey et al ., 1999 ). Moreover, LIFR deficient mouse embryos display significant placental defects and growth abnormalities in other organs ( Ware et al ., 1995 ). LIF is required for expression of the EGF family of growth factors during implantation, illustrated by uteri of LIF knockout mice that are deficient in the subepithelial expression of EGF-like growth factors at implantation sites ( Song et al ., 2000 ). Conversely, the upregulation of LIF expression by human uterine epithelial cells cultured with heparin binding EGF-like growth factor (HBEGF) suggests that LIF expression is dependent upon the EGF signaling system ( Lessey et al ., 2002 ). Studies of human preimplantation embryos cultured in growth factors indicate that LIFR expression is dependent on HBEGF signaling, and that the HBEGF receptor ERBB4 is upregulated by LIF ( Kimber et al ., 2008 ). These data highlight the complexity of cross talk amongst growth factors during implantation and placentation.

The EGF family of growth factors and their receptors regulate a variety of biological processes that include proliferation, differentiation and survival. Growth factors in the EGF superfamily include EGF, HBEGF, transforming growth factor-alpha (TGFA), amphiregulin (AREG), betacellulin (BTC), epiregulin (EREG), epigen and the neuregulins (NRG) ( Holbro and Hynes, 2004 , Riese and Stern, 1998 ). Recent studies have examined the roles of each EGF family member in implantation and placentation. The EGF-like growth factors activate receptor tyrosine kinases in the plasma membrane that include epidermal growth factor receptor (EGFR/ERBB1) and related ERBB family proteins (ERBB2, ERBB3, ERBB4). Growth factor binding initiates an intrinsic protein-tyrosine kinase activity, autophosphorylation of the cytoplasmic domain and receptor dimerization to generate a downstream signaling cascade ( Holbro and Hynes, 2004 ). The EGF-like growth factors are synthesized as membrane bound proteins that signal to their receptors on adjacent cells, also called juxtacrine signaling, or they are secreted through proteolytic cleavage of their extracellular domain, carried out by metalloproteinases, for autocrine or paracrine signaling ( Riese and Stern, 1998 ). The importance of the EGF signaling system for blastocyst implantation is suggested by the EGFR knockout ( Miettinen et al ., 1995 , Sibilia and Wagner, 1995 , Threadgill et al ., 1995 ), which, depending on the genetic background of host mice, has a perinatal-lethal phenotype. However, the picture is less clear from studies of embryos with deletion of EGF family growth factors where several knockouts produce mice that are viable and fertile, while others have roles in early post-implantation embryonic development that make interpretation difficult. It remains to be determined whether the EGF signaling system has a specific role in the process of implantation or if it is more involved in embryonic survival and differentiation. Alternative approaches provide evidence to support the hypothesis that this complex signaling system directly influences the success of blastocyst implantation. Experiments in mice establish that HBEGF is present in the uterus shortly before embryo implantation, providing the earliest known signaling between the blastocyst and maternal cells ( Lim and Dey, 2009 ). In situ hybridization demonstrates that HBEGF is expressed in the luminal epithelium specifically at sites surrounding blastocysts 6–7 hours prior to the attachment reaction ( Das et al ., 1994 ). HBEGF is also expressed by blastocysts prior to implantation ( Liu and Armant, 2004 ), suggesting its role in a two-way dialogue. Experimentally induced delayed implantation in ovariectomized mice reveals no HBEGF expression at implantation sites; however, HBEGF becomes upregulated in both the uterus ( Das et al ., 1994 ) and blastocyst ( Hamatani et al ., 2004 ) with administration of estrogen to activate the blastocyst to implant. Its receptors ERBB1 and ERBB4 are concomitantly regulated by estrogen along with HBEGF in the blastocyst ( Paria et al ., 1993 , Paria et al ., 1999 ). HBEGF signaling between the trophoblast and uterine epithelial cells is most likely initiated by embryonic HBEGF shedding, as depicted in Fig. 1A . Crosstalk with other signaling pathways mobilizes HBEGF from intracellular stores in the trophectoderm and demonstrates the complexity of the maternal-embryo dialogue. Lysophosphatidic acid (LPA) produced in the maternal reproductive tract accelerates blastocoel formation at the morula stage and trophoblast differentiation at the early blastocyst stage through downstream intracellular Ca 2+ signaling ( Liu and Armant, 2004 , Stachecki and Armant, 1996 ). LPA activates cognate G protein-coupled receptors that signal through phospholipase C to generate inositol−1,4,5-trisphosphate and diacylglycerol that, in turn, mobilize cytoplasmic free Ca 2+ and activate protein kinase C (PKC), respectively ( Contos et al ., 2000 ). Signaling by both cytoplasmic Ca 2+ and PKC is known to stimulate HBEGF shedding ( Dethlefsen et al ., 1998 , Dong and Wiley, 2000 ) and the subsequent transactivation of ERBB signaling ( Umata et al ., 2001 ). Immunofluorescence microscopy demonstrates that HBEGF is sequestered from the plasma membrane during blastocyst development, but accumulates transiently at the embryo surface in response to LPA-induced intracellular Ca 2+ release ( Liu and Armant, 2004 ). Therefore, uterine LPA secretion could activate HBEGF shedding from blastocysts in apposition to the uterine wall, signaling its presence to the adjacent uterine epithelium ( Fig. 1A ). Several studies support the hypothesis that expression of HBEGF in the peri-implantation endometrial epithelium is induced by embryo derived HBEGF signaling. HBEGF is expressed in both mouse and hamster preimplantation blastocysts ( Hamatani et al ., 2004 , Liu and Armant, 2004 , Wang et al ., 2002 ). Interestingly, when growth factor-soaked beads the approximate size of blastocysts are transferred into the uteri of pseudopregnant mice only HBEGF and insulin-like growth factor-1 (IGF1) provoke discrete implantation-like responses ( Paria et al ., 2001 ). Furthermore, highly localized expression of HBEGF occurs in the uterine epithelium adjacent to apposition stage blastocysts ( Das et al ., 1994 ). These studies suggest a positive feed-forward regulation of uterine HBEGF expression initiated by the blastocyst during the apposition stage of implantation. While HBEGF secreted by the embryo is sufficient to induce HBEGF expression in the luminal epithelium, other factors secreted by embryos (e.g., IGF1) could have equivalent activity and perhaps compensate in HBEGF deficient embryos, which are able to implant normally when transferred to wild type pseudopregnant dams ( Xie et al ., 2007 ). HBEGF protein appears on the surface of the uterine epithelium early on gestation day 5 in mice as the attachment reaction commences ( Das et al ., 1994 ). The primary target of HBEGF in the luminal epithelium is the trophoblast ( Fig. 1B ). Late on day 4, HBEGF signaling in the blastocyst becomes robust as a result of ERBB4 trafficking to the trophoblast surface ( Wang et al ., 2000 ). ERBBs 1–3 are already localized at the apical surface as the blastocyst forms. However, ERBB4 is the preferred receptor for HBEGF in blastocysts ( Paria et al ., 1999 ). Thus, ERBB1 and ERBB4 both become available on the trophoblast at approximately the time when uterine HBEGF first appears. These events position HBEGF and its receptors to mediate blastocyst adhesion and trophoblast differentiation. HBEGF-induced ERBB autophosphorylation advances trophoblast adhesive differentiation ( Das et al ., 1994 , Paria et al ., 1999 , Wang et al ., 2000 ). In blastocysts cultured serum-free, HBEGF induces trafficking of integrin subunits, including ITGA5 ( Fig. 1B ), into the plasma membrane of trophoblast cells approximately 24 hours earlier than in control embryos, which promotes their adhesion to fibronectin in the maternal extracellular matrix ( Wang et al ., 2000 ). In vitro experiments suggest that after apoptosis of the apposing luminal epithelial cells, ligation of the integrin ITGA5/ITGB1 by the exposed basement membrane activates phospholipase Cγ ( Wang et al ., 2007 ), leading to trafficking of ITGA2B into the plasma membrane ( Rout et al ., 2004 ), which strengthens cell adhesion as trophoblast invasion commences ( Fig. 1C ). Experiments have been reported suggesting that HBEGF supports the attachment reaction during implantation in addition to its ability to advance hatching and trophoblast differentiation. Using engineered cells that express membrane bound HBEGF to simulate receptive uterine epithelial cells, it has been demonstrated that cell surface HBEGF binds to day 4 mouse blastocysts, but not to delayed blastocysts ( Raab et al ., 1996 ). HBEGF signaling activates Ca 2+ influx across the plasma membrane through N-type voltage-gated Ca 2+ channels ( Fig. 1B ) that stimulates PKC and calmodulin, consequently accelerating trophoblast invasion ( Wang et al ., 2000 ). It appears that there are complex autocrine, paracrine and juxtacrine pathways that integrate HBEGF signaling with input from other growth factors, as well as through cell-cell interactions between the trophectoderm and uterine epithelium during implantation ( Fig. 1 ). Maternal HBEGF deficiency targeted to the uterus defers the window of implantation and compromises pregnancy outcome, thus, implicating the importance of two-way communication between the embryo and the uterus ( Xie et al ., 2007 ). However, another heparin-binding member of the EGF family, AREG, partially compensates for HBEGF loss. Progesterone upregulates AREG throughout the uterine luminal epithelium of the mouse early on gestation day 4 independently of the blastocyst ( Das et al ., 1995 ), but it becomes restricted to the implantation site by day 5. Studies conducted to clarify the mechanism of AREG signaling at the embryonic-maternal interface suggest that the presence of fertilized embryos in the reproductive tract elevates expression of AREG on day 4 ( Lee et al ., 2006a ). Therefore, the presence of embryos in an HBEGF-deficient uterus could induce partially-compensating production of AREG, which can activate only ERBB1 ( Riese and Stern, 1998 ). Another EGF family member, TGFA, has extensive sequence homology to EGF and comparable affinity for ERBB1, its only receptor. Immunohistochemical, in situ hybridization, northern blot and RT-PCR analyses clearly demonstrate TGFA expression in the uterus and embryo ( Rappolee et al ., 1988 , Tamada et al ., 1991 ). Additionally, a reduction in the number of implantation sites in rats after administration of intraluminal anti-TGFA injections supports the involvement of TGFA in implantation ( Tamada et al ., 1997 ). Addition of TGFA to in vitro culture medium is favorable for preimplantation embryo development and trophoblast differentiation ( Haimovici and Anderson, 1993 , Machida et al ., 1995 ). However, this involvement is controversial, as TGFA knock out mice have normal implantation ( Luetteke et al ., 1993 , Mann et al ., 1993 ). Similarly, AREG and EGF are not essential for implantation, indicated by mice deficient in AREG and triple knockout mice lacking TGFA, AREG and EGF, which are all fertile ( Luetteke et al ., 1999 ). These studies support the concept of a compensatory response by other EGF family members, perhaps HBEGF, during implantation. Another group of EGF-like growth factors that mediate cell-cell interactions, the NRGs, are encoded by four distinct genes, NRG-1, NRG-2, NRG-3 and NRG-4 ( Burden and Yarden, 1997 , Falls, 2003 ). There is evidence for 15 splice variants of NRG-1 that are grouped into three subtypes ( Ben-Baruch and Yarden, 1994 , Meyer et al ., 1997 ). Of the three subtypes, Type 1 is comprised of neu differentiation factor (NDF), heregulin and acetylcholine receptor-inducing activity ( Ben-Baruch and Yarden, 1994 ). Type 2 consists of the glial growth factor (GGF) and type 3 comprises the sensory and motor neuron derived factor (SMDF). ( Brown et al ., 2004 ). NDF and its isoforms are expressed in the preimplantation mouse uterus ( Reese et al ., 1998 ). Mice with delayed implantation do not express NDF; however, upon administration of estrogen, the expression of NDF returns. Alternately there is evidence that SMDF has a similar cellular distribution to NDF and its expression becomes prominent at the time of implantation ( Brown et al ., 2004 ). However, no significant expression of GGF is observed in the peri-implantation uterus. In the blastocyst, both ERBB3 and ERBB4 are present on trophoblast cells, suggesting the capacity for signaling with NRG family members ( Wang et al ., 2000 ). A parallel study observed a similar expression pattern of the EGF family members BTC and EREG ( Das et al ., 1997 ), suggesting that optimal expression of NDF, BTC and EREG requires an activated blastocyst. Therefore, it remains possible that these EGF family members contribute to mouse implantation. In contrast to its regulation in mice that have a short estrus cycle, HBEGF is regulated independently of the presence of an embryo in the stromal and epithelial compartments during the menstrual cycle of the human endometrium ( Chobotova et al ., 2002a , Leach et al ., 1999 , Yoo et al ., 1997 ). There is evidence that HBEGF mediates maternal-blastocyst signaling and attachment in humans as its does in rodents owing to the switch in expression of HBEGF mRNA and protein from the subepithelial stroma during early secretory phase to glandular and luminal epithelium in the mid-secretory phase ( Leach et al ., 1999 ). ERBB4 is expressed by the trophoblast cells of human peri-implantation blastocysts, suggesting that HBEGF present in the luminal epithelium could stimulate the blastocyst through paracrine or juxtacrine signaling ( Chobotova et al ., 2002b ). HBEGF not only accelerates the development of human embryos to the blastocyst stage, but also aids in hatching from the zona pellucida ( Martin et al ., 1998 ). Evidence indicates that proliferation in human endometrial stromal cells is regulated by both the soluble and transmembrane forms of HBEGF, along with other EGF family members, EGF and BTC, suggesting key functions for the EGF family of growth factors in regeneration and maturation of the human endometrium and to prepare the embryo for implantation ( Chobotova et al ., 2002a ). However, this induction is under the cooperative effect of TNF. Another study shows that the endometrial stromal cells produce both the soluble and transmembrane forms of HBEGF to regulate TNF and TGFB ( Chobotova et al ., 2005 ). This provides a survival function that prevents apoptosis of endometrial cells exposed to apoptotic factors, which highlights the importance of HBEGF and its receptors in the human endometrium. HBEGF appears to participate in a broader dialogue that includes other molecules, exemplified by trophinin-bystin signaling. Both human trophoblast cells and the uterine epithelium express the transmembrane protein trophinin that mediates cell adhesion through homophilic binding ( Suzuki et al ., 1999 ). Trophinin forms a complex with its cytoplasmic partner bystin and ERBB4, which inhibits ERBB tyrosine phosphorylation required for invasive trophoblast differentiation ( Fig. 1B ). However, during the attachment reaction between the blastocyst and luminal epithelium, homophilic trophinin binding releases bystin, thus derepressing ERBB4, which can then become activated by HBEGF to promote trophoblast differentiation ( Sugihara et al ., 2007 , Tamura et al ., 2011 ). Meanwhile, in the luminal epithelium, homophilic trophinin binding disrupts its tethering of protein kinase Cδ to the plasma membrane, freeing the latter to enter the nucleus where it induces apoptosis ( Tamura et al ., 2011 ). Thus, trophinin signaling removes the cellular barrier to blastocyst implantation as it licenses activation of trophoblast adhesion competence. Most components of the EGF signaling system are present throughout pregnancy. Transcripts from placental tissues obtained in all three trimesters demonstrate expression of HBEGF, with the highest abundance in first trimester decidua ( Yoo et al ., 1997 ). Strong staining for HBEGF is observed in villous and extravillous trophoblast cells in both normal placentas and those delivered preterm ( Leach et al ., 1999 , Leach et al ., 2002 ). Other EGF family members and ERBB receptor tyrosine kinases are expressed in placental trophoblast populations ( Hofmann et al ., 1992 , Tanimura et al ., 2004 ), indicating that the EGF signaling system is highly active during placental development. There is evidence to support the hypothesis that HBEGF, and possibly other EGF family members, promote trophoblast invasion ( Jessmon et al ., 2009 ). The capacity for invasion weakens as gestation progresses ( Damsky et al ., 1994 , Librach et al ., 1991 ), indicating the importance of the EGF signaling system in first trimester trophoblast cells ( Bass et al ., 1994 ). EGF, TGFA and HBEGF can stimulate first trimester trophoblast cells to become invasive and differentiate to the extravillous phenotype ( Bass et al ., 1994 , Leach et al ., 2004 ). However, the effectiveness of EGF diminishes with trophoblast cells obtained from the second or third trimester ( Bass et al ., 1994 ), suggesting that their loss of invasion competence is intrinsic rather than a product of their environment. Invading extravillous trophoblast cells are necessary for adequate remodeling of the uterine spiral arteries that perfuse the developing placenta ( Norwitz et al ., 2001 ). The first 10 weeks of pregnancy are a period of low O 2 within the conceptus ( Burton and Jauniaux, 2004 , Burton et al ., 1999 ). Intrauterine measurements report low (~2%) O 2 concentrations in placental tissues compared to the surrounding decidua ( Jauniaux et al ., 2001 , Rodesch et al ., 1992 ). This environment favors the proliferation of the trophoblast cells, while repressing invasion ( Genbacev et al ., 1996 , Genbacev et al ., 1997 ). As a result, trophoblast cells accumulate and occlude the maternal blood vessels, which maintains the relatively hypoxic state ( Burton and Jauniaux, 2004 ). The most distal trophoblast cells contact maternal blood, exposing them to higher O 2 concentrations, which promotes the invasive phenotype. As gestation progresses beyond 10 weeks, the spiral arteries are remodeled to an extent that reperfuses and oxygenates the placenta by dislodging the occluding trophoblast cells. The widespread expression of HBEGF in placental trophoblast cells and surrounding decidua ( Leach et al ., 1999 ) could facilitate invasion and contribute to remodeling of the spiral arteries. It has been demonstrated that when the HTR-8/SVneo human first trimester cytotrophoblast cell line is cultured at 2% O 2 , HBEGF, but not other EGF family members, is upregulated and secreted ( Armant et al ., 2006 ). However, trophoblast cells in villous explants from term placentas do not elevate HBEGF in response to low oxygen and their survival is compromised by hypoxia ( Imudia et al ., 2008 ). Addition of HBEGF to term villous explants cultured at 2% O 2 inhibits apoptosis, demonstrating its capacity as a survival factor. The placenta is programmed to survive at a low O 2 tension during the first trimester, while trophoblast invasion progresses slowly at the vascular interface in preparation for the full onset of placental perfusion ( Norwitz et al ., 2001 ). However, if reoxygenation does not occur on time, placental insufficiency ensues, which can increase risk for obstetric disorders that include pregnancy loss, intrauterine growth restriction or preeclampsia ( Burton and Jauniaux, 2004 ). These disorders are associated with poor trophoblast invasion and increased apoptosis ( Brosens et al ., 1972 , DiFederico et al ., 1999 ), two functions that are strongly regulated by HBEGF signaling ( Jessmon et al ., 2009 ). Placental tissues from preeclamptic pregnancies have reduced expression of HBEGF compared to gestational age-matched normotensive placentas, as well as reduced HBEGF mRNA expression ( Leach et al ., 2002 ). These findings suggest the hypothesis that the reduction in expression of HBEGF could contribute to the shallow trophoblast invasion and poor survival observed in preeclampsia. It remains to be established whether the reduced expression of HBEGF precedes the onset of preeclampsia or if it is merely associated with late placental demise. There is also clinical evidence of HBEGF disruption associated with infertility in women ( Aghajanova et al ., 2008 ). The decreased HBEGF expression in infertile couples was not confirmed in a larger clinical study, but it was noted that a significant increase between the early and mid secretory phases was absent in the infertile patients ( Leach et al ., 2012 ). The clinical significance of HBEGF signaling during human pregnancy will not be fully appreciated without more sophisticated experimental approaches that take into account its interactions with other signaling pathways.

Significant advancements have been made in unraveling the molecular mechanisms governing growth factors involved in implantation and trophoblast invasion. The mouse has provided a powerful experimental model; however, its correlation with human implantation and trophoblast development is not entirely consistent. Infertility affects approximately 15% of reproductive age women and in the past four decades, IVF has become a viable and widely used treatment option for infertile couples, with over 1.5% of babies in the U.S currently conceived through IVF ( Sunderam et al ., 2013 ). Although relatively successful compared to natural fecundity, one of the major barriers to higher IVF implantation rates is a non-receptive endometrium ( Li and Jin, 2013 ). A better understanding of the molecular mechanisms governing the expression of growth factors in implantation and placentation could better define the window of implantation and contribute towards higher IVF pregnancy rates. Furthermore, this additional information could provide a foundation for development of new treatments for clinical pathologies such as recurrent pregnancy loss and placental insufficiencies that reduce fetal growth rates or cause preeclampsia. Therefore, continued investigation of growth factors and the molecular mechanisms that ensure successful implantation and placentation has significant clinical relevance.

Successful implantation and placentation in humans and other eutherian mammals depend upon complex interactions between the embryo and a receptive uterus. Steroid hormones, particularly estrogen and progesterone, are well established as having a critical role. However, an abundance of evidence shows that various cytokines and growth factors are crucial for implantation to succeed; in part, because they mediate the actions of the steroid hormones ( Zhang et al ., 2013a ). In this brief review, several key signaling systems will be discussed that coordinate the distinct developmental programs of the endometrium and embryonic trophoblast during the constitution of a viable conceptus. The mouse, which is amenable to genetic and experimental approaches, has been a useful model for understanding the molecular dialogue between embryo and maternal tissue. However, there are important physiological differences between murine and human reproductive biology that are beginning to surface through the increasing impact of translational studies with human tissues and cell lines. Where possible, signaling pathways critical to human reproduction will be highlighted.