Uterine sensing of the embryo
The uterus uses the Wnt pathway to sense the embryo during initial implantation.
In the Hans Christian Andersen tale, a young princess is identified by her ability to sense a pea buried deep within a thick layer of mattresses. In this issue of PNAS, Mohamed et al. (1) provide new insight into the molecular mechanism of embryo implantation by identifying the Wnt pathway as part of the system the uterus uses to sense and respond to the relatively tiny embryo during the initial stage of implantation in mice. Ovarian steroids coordinate embryonic development to an attachment-competent state with maturation of the uterus to a receptive state. The latter is a transient process that is followed by a refractory period requiring reinitiation of the uterine cycle before again becoming receptive. A small rise in estrogen levels, i.e., nidatory estrogen, is required to trigger the implantation process, including expression of the molecules mentioned above in rodents. It has long been recognized that local events occur in the uterus in response to the implanting embryo, and a number of early molecular markers of these events have been described (2).
Mohamed et al. (1) used a mouse model in which a lacZ reporter gene is activated by T cell factor/lymphoidenhancer factor, indicating that nuclear β-catenin signaling, a downstream event in canonical Wnt pathway activation, has occurred. Their studies demonstrate not only that this pathway is activated locally in response to the implanting embryo but also that embryo-derived Wnts mediate the process. Wnt7a, but not Wnt5a, appears to be particularly effective in triggering implantation-related responses. Consistent with this observation, previous studies by Mohamed et al. (3) demonstrated that mouse embryos increase expression of Wnt7a mRNA during development from the morula to the blastocyst stage. Furthermore, they show that the Wnt antagonist, secreted frizzled related protein 2 (sFRP-2), but not sFRP-1, partially inhibits formation of the regularly spaced swellings typical of mouse embryo implantation sites. They report that the embryos that do implant appear to be developmentally delayed, suggesting that all implantation sites, rather than a subset, are impacted. Closer examination of the sites to determine whether multiple embryos can implant at one site in the sFRP-2-treated animals is warranted because other observations by Mohamed et al. indicate that Wnts may be involved in aspects of embryo spacing. In addition, it is of interest to determine whether all sites containing blastocysts in the sFRP-2-treated animals display molecular markers of early implantation, e.g., heparin binding-epidermal growth factor-like factor and cyclooxygenase-2 (2).
Wnts and Uterine Estrogen Responses
Other reports indicate that the Wnt pathway is likely to play an important role in uterine physiology. A screen of estrogen-stimulated genes in an ovariectomized mouse model revealed that a select set of Wnts, namely Wnt4 and Wnt5a, as well as the Wnt receptor Frz-2, were up-regulated within 6 h of 17-β-estradiol or catechol estrogen treatment (4). In contrast, a variety of other Wnts, including Wnt7a, were not stimulated in this time frame; in fact, Wnt7a is barely detectable under either condition (S. Das, personal communication). Uterine Wnt7a expression may be a late response to estrogen but seems unlikely to account for rapid, local effects associated with embryo implantation. Thus, the notion that Wnt7a is derived from the embryo is consistent with previous studies. Nonetheless, canonical Wnt signaling is needed for embryo-independent, estrogen-stimulated uterine growth in the mouse (5). As is the case for implantation-associated events, these studies demonstrated that sFRP-2 blocks these responses. Estrogen-driven and embryo-stimulated uterine growth involve both epithelial and stromal compartments. In this regard, mRNA-encoding Wnt receptors are expressed in both compartments, indicating cells in both compartments can respond to these signals. In the case of embryo implantation, it is possible that all Wnt signals are derived from the embryo or that only one, an initiating Wnt signal, acts on the epithelium, which, in turn, produces additional Wnt signals to drive the subsequent events. It should be possible to discriminate between these possibilities in embryo transfer experiments in which Wnt expression selectively is blocked in either the embryo or recipient uterus.
Estrogen-driven uterine growth may have parallels with uterine growth associated with implantation, although the source of the Wnt signal may differ. Under normal conditions, formation of the primary and secondary decidual zones, at least in part, may be driven by Wnt pathway activity. Nonetheless, the data of Mohamed et al. (1) indicate that Wnt pathway activation is not a requisite for this response because decidual responses artificially induced by physical trauma or oil injection were not accompanied by activation of the T cell factor/lymphoidenhancer factor-driven reporter. Thus, pathways apart from canonical Wnt pathways must exist that also can trigger uterine differentiation.
Uterine Activation
If uterine differentiation can be activated in the absence of an embryo without activating a canonical Wnt pathway, does the presence of an embryo invariably result in activation of this pathway? Apparently not, because blastocysts fail to do so in the absence of nidatory estrogen. In addition, the finding that embryo-sized beads coated with BSA or Wnt5a fail to trigger similar responses destroys the notion that activation is the result of an ability of the uterus to physically sense the presence of the embryo via local perturbation of cellular architecture or baroreceptors. In these cases, the hormonal milieu is appropriate to support uterine responses, yet these do not occur. Nonetheless, it appears that the estrogen influence is critical to establish the conditions for the uterine response to the embryo. What does this mean in a molecular sense? Expression of a number of uterine gene products associated with early implantation require nidatory estrogen, and it is easy to imagine how estrogen receptors might be required in concert with other transcription factors activated locally in response to the implanting embryo to drive expression of these genes; however, the rapid (6 h) estrogen effects on the Wnt pathway in the mouse do not appear to require either nuclear estrogen receptor α or β (4). Thus, it is possible that estrogen's influence is membrane-initiated. The identity of membrane-associated estrogen receptor(s) remains enigmatic. Some suggestions indicate that these may be related or identical to the nuclear receptors that perhaps form complexes with membrane or cytoplasmic signal-transducing proteins (6, 7). A more recent study indicates that at least one member of the G protein-coupled receptor family, GPR30, is a signal-transducing 17-β-estradiol receptor (8). No information on any of these candidates in the uterus or in the context of implantation is available, but these possibilities should be examined. With regard to the nuclear receptor-related proteins, this demonstration would require high-resolution examination of early implantation sites because it is possible that these proteins redistribute among nuclear, cytoplasmic, and plasma membrane pools.
The Wnt pathway is activated in well defined bands at the side on which embryo attachment takes place.
The Wnt pathway also may be important in embryo implantation-related events in humans, although there may be some differences in how the response is controlled. High-density cDNA microarray screening of changes in endometrial gene expression occurring during the transition to the receptive stage in women did not indicate that either Wnts or their receptors change to a significant extent from either the proliferative (estrogen-dominated) or early secretory (progesterone-dominated, prereceptive) to the midsecretory (receptive) phase (9, 10); however, large changes occur in expression of Wnt antagonists with impressive increases in Dickkopf and equally impressive decreases in FrzHE (11, 12). It is not obvious why these two antagonists are regulated so differently in this context; however, it is possible that they display different abilities to neutralize specific Wnts and/or interact with other extracellular components. The transcripts encoding these proteins are localized to the uterine stroma. Thus, it appears unlikely that Dickkopf can function at the site of initial embryo attachment unless it is transported either through or between the uterine epithelium. Nonetheless, Dickkopf may restrict Wnt activity to the implantation site and retard uterine differentiation. For ethical reasons, it is not feasible to examine the pattern of Dickkopf expression in human implantation sites, but it is possible that a gradual reduction in expression may accompany the human decidual response to permit expansion of the implantation site. Wnt antagonists also may attenuate Wnt responses in the mouse because expression of the soluble form of Wnt receptor, sFRP2, is markedly inhibited in mouse endometrium by estrogen (4). Interestingly, Mohamed et al. (1) demonstrate that sFRP-2 effectively inhibited the embryo-dependent decidual reaction. Thus, one role of nidatory estrogen might be to suppress expression of this Wnt antagonist to permit uterine responsiveness to the embryonic Wnt signal.
Finally, Mohamed et al. (1) provide the remarkable, but puzzling, observation that the Wnt pathway is activated in well defined bands in circular myometrium at the antimesometrial aspect of the uterus, i.e., the side on which embryo attachment takes place, shortly before embryo implantation occurs. Although this observation appears to be related to the implantation process, it is very difficult to imagine how blastocyst products could trigger discrete expression at a distal site. If the initializing signal was a Wnt, then activation of intervening tissues would be expected. Therefore, this event appears not to be initialized at the uterine lumen by Wnts and is likely to be patterned by the uterus itself. This idea is supported by Mohamed et al.'s contention that the numbers of bands exceed the number of embryos found in the horns. Moreover, the response is transient. Could Wnts be involved in controlling uterine spacing of embryos? There are many reports of Wnt pathway and β-catenin actions on smooth muscle cells, but none that indicate a role in modulating contractility that might account for embryo spacing. Thus, these studies provoke thoughts about novel roles for the Wnt pathway in embryo implantation as well as in other developmental and physiological contexts. The capacity to sense the blastocyst “pea” continues to be a vital, remarkable, and revealing aspect of uterine physiology.
