Beyond the brain-Peripheral kisspeptin signaling is essential for promoting endometrial gland development and function

Uterine growth and endometrial gland formation (adenogenesis) and function, are essential for fertility and are controlled by estrogens and other regulators, whose nature and physiological relevance are yet to be elucidated. Kisspeptin, which signals via Kiss1r, is essential for fertility, primarily through its central control of the hypothalamic-pituitary-ovarian axis, but also likely through peripheral actions. Using genetically modified mice, we addressed the contributions of central and peripheral kisspeptin signaling in regulating uterine growth and adenogenesis. Global ablation of Kiss1 or Kiss1r dramatically suppressed uterine growth and almost fully prevented adenogenesis. However, while uterine growth was fully rescued by E2 treatment of Kiss1−/− mice and by genetic restoration of kisspeptin signaling in GnRH neurons in Kiss1r−/− mice, functional adenogenesis was only marginally restored. Thus, while uterine growth is largely dependent on ovarian E2-output via central kisspeptin signaling, peripheral kisspeptin signaling is indispensable for endometrial adenogenesis and function, essential aspects of reproductive competence.

Scientific RepoRts | 6:29073 | DOI: 10.1038/srep29073 and early antral stage; this results in infertility 10,23,24 . The observation that follicular development can advance to the antral stage likely reflects that FSH secretion is not completely abolished in these knockout (KO) mice 23,24 . While the follicles from these null mice maintain the capacity to produce and secrete 17β -estradiol (E2), Kiss1 −/− and Kiss1r −/− mice do not exhibit the pre-ovulatory E2 surge 10,24 , and consequently follicles do not undergo final maturation and ovulation. In addition, despite near normal circulating levels of E2 in adult KO mice 10,24 , the KO uterus is significantly smaller than the WT uterus 10,20,[24][25][26] , suggesting that other factors which stimulate uterine growth must be absent in mice congenitally devoid of kisspeptin signaling and/or that the uterus lacks ESR1 (estrogen receptor 1) expression in this model and hence is unresponsive to E2. However, an analysis of the pregnant uterus of Kiss1r −/− mice indicated there was normal Esr1 and ESR1 expression 20 . During the course of our investigations of the Kiss1 −/− and Kiss1r −/− mice, we also noted that the KO uterus is almost completely devoid of endometrial glands 20 ; an observation that is reminiscent of findings from a previous report by d' Anglemont de Tassigny et al. 24 , using an independently generated Kiss1 −/− mouse line (Kiss1 tm1Coll ).
Endometrial glands are found in all mammalian uteri where they produce and transport substances, such as leukemia inhibitory factor, that are required for the establishment of uterine receptivity and embryo implantation and survival [27][28][29] . In support of this, genetic inactivation of genes, such as Foxa2, Wnt4 and Wnt7a that positively regulate endometrial gland formation (adenogenesis) result in subfertility and infertility [30][31][32][33] . In humans, endometrial adenogenesis begins in the fetus, continues postnatally, and is completed during puberty. In contrast, in sheep, pigs and rodents, adenogenesis typically begins during the early postnatal period and involves differentiation and budding of glandular epithelia from the luminal epithelium 29,31 . In mice, this begins around postnatal day (PND) 5-6 and is followed by extensive cell proliferation within the nascent glands (evident by PND7) leading to their elongation and invasion of the surrounding stroma 29 . The adult uterine histoarchitecture is established between PND14-21 with adenogenesis persisting over the lifespan 34 .
Initial adenogenesis and uterine growth in the neonate occur independently of ovarian and adrenal hormones in many species, including rodents 35,36 and livestock [37][38][39] as well as independently of ESR1 in pre-weaning mice 40 . On the other hand, elevation of ovarian hormones at the onset of puberty alters the mechanisms regulating adenogenesis and uterine growth, shifting them into an ovarian-and ESR1-dependent phase that begins around PND29 and lasts throughout adult life 34,40 .
In this scenario, it is intriguing that despite the fact that circulating E2 levels are reported to be near WT levels in the adult Kiss1 −/− and Kiss1r −/− mice, these KO models exhibit markedly reduced uterine growth and adenogenesis. These observations would suggest that, in addition to E2, other peripheral factors (ovarian-derived or otherwise) driven by central and/or peripheral kisspeptin signaling and which are missing in these above kisspeptin null models, would physiologically contribute to uterine growth and endometrial adenogenesis. In this work, we aimed at elucidating this phenomenon using suitable genetically modified mouse models.

Results
Ablation of Kiss1 or Kiss1r results in loss of adenogenesis and reduction in uterine growth in the adult female mouse. The analysis of FOXA2 expression, a marker of endometrial glands, revealed that in transverse uterine sections from the Kiss1 −/− and Kiss1r −/− (global KO) adult mice (8-12 weeks old, non-pregnant, 129S1/SvImJ genetic background), gland formation was reduced by about 93% vs. WT littermate controls ( Fig. 1A-D,G). As an indication of uterine growth, the average area of transverse uterine sections was determined. In the Kiss1 −/− and Kiss1r −/− mice, this index was found to be reduced by about 70% relative to that of respective WT littermates ( Fig. 1A-D,H). The uterine phenotypes observed in Kiss1 −/− and Kiss1r −/− 129S1/SvImJ mice were fully recapitulated in age-matched Kiss1r −/− mice and WT littermates of the C57BL/6J genetic background ( Fig. 1E-H), where gland formation in the Kiss1r −/− mice was reduced by about 98% and the uterine area by about 78% compared to WT littermates (Fig. 1E,F,H). Our analyses also revealed that the endometrium of the C57BL/6J Kiss1r WT mouse contained about 5-fold more glands and the uterine area was about 3.3 times larger than age-matched 129S1/SvImJ WT Kiss1 and Kiss1r mice (Fig. 1). Despite these strain differences, our convergent findings of the impact of lack of kisspeptin signaling on adenogenesis in these two mouse strains unambiguously demonstrate that kisspeptin signaling regulates uterine growth and development. E2 therapy partially rescues adenogenesis but fully rescues uterine growth in the pregnant adult Kiss1 −/− mouse. Next, we sought to determine whether these striking uterine phenotypes in the Kiss1 −/− and Kiss1r −/− mice might be due solely to insufficient E2 drive (the end-point of the central failure of the hypothalamic-pituitary-ovarian axis). Of note, although circulating E2 concentrations have been reported to be nearly similar between adult Kiss1 −/− and Kiss1r −/− mice and their WT littermates 10,24 , the adult KO phenotypes are consistent with diminished ovarian function and E2 levels 34 . Furthermore, our recent results showed that chronic E2 therapy post-weaning coupled to gonadotropin treatment, rescued follicular development and triggered ovulation of fertilization-competent oocytes 20 . We therefore determined what effect E2 supplementation would have on the growth and development of the KO uterus. Since we were also interested in determining the impact of such treatment on early pregnancy, we examined uteri from E2-treated mice on D4 of pregnancy, the day on which embryo implantation occurs; a process dependent on glandular secretions 11,27,29,41 . Considering the commonalities in the phenotypes of the Kiss1 −/− and Kiss1r −/− mice, these studies were conducted only in the Kiss1 −/− mouse.
E2 administration for 5 weeks starting before puberty induced a significant increase in adenogenesis and uterine growth in both the adult KO and WT uterus (9-10 weeks old), as compared to closely age-matched (8-12 weeks old) adult untreated and non-pregnant KO and WT mice ( Fig. 2A,B,K,L vs. Fig. 1A,B,G,H). Admittedly, part of this response would reflect the pregnant state that the mice were in and the other part the E2 treatment. However, despite 5 weeks of E2 administration and 4 days of pregnancy, the KO uterus still exhibited significantly diminished adenogenesis ( Fig. 2A,B,K). Interestingly, however, while adenogenesis was only rescued by about Scientific RepoRts | 6:29073 | DOI: 10.1038/srep29073 24%, uterine growth was fully restored compared to WT littermates ( Fig. 2A,B,L). These results suggest that while central failure of the hypothalamic-pituitary-ovarian axis could be responsible for diminished uterine growth, it cannot solely account for severely reduced gland development, thereby implicating a role for peripheral kisspeptin signaling. E2-rescued glands in the Kiss1 −/− endometrium exhibit diminished FOXA2 expression, hyperplasia of the glandular epithelium (GE) and a lack of secretion. Initial analyses of the Kiss1 −/− endometrium revealed that the majority of glands displayed a consistent decrease in FOXA2 expression ( Fig. 2A,B). Additionally, in a smaller number of glands (about 20%), FOXA2 was detected only in a subset of cells comprising the GE (Fig. 2C-F) or not detected at all (Fig. 2C-F). In about 10% of the E2-rescued glands, the GE was comprised of a highly disorganized cellular layer that had undergone hyperplasia at one or more points (Fig. 2G,H); this was in striking contrast to the WT GE which was always comprised of a well-organized single layer of cells (Fig. 2I,J). While Stewart et al. 34 reported that E2 administration to the neonate triggered a hyperplastic glandular phenotype in adult mice, it does not appear that the post-weaning-administration of E2 is the underlying cause of this hyperplasia since WT littermates were also E2-treated but did not display this phenotype (Fig. 2I,J). We also determined that Ctnnb1 (β -catenin) expression was similar between the KO and WT littermates (data not shown) ruling out a deregulation of β -catenin in this phenotype 42 . Finally, it was observed that in all E2-rescued glands, the glandular lumen was completely devoid of any glandular secretions, while within a large number of WT glands secretions could be readily detected within the lumen ( E2-rescued glands in the Kiss1 −/− endometrium appear non-functional. Given the finding that E2-rescued glands in Kiss1 −/− mice lack luminal secretions, we hypothesized that kisspeptin signaling positively regulates gland function (that is, the expression, secretion and transport of important regulators of implantation and decidualization) and that E2-rescued glands would be deficient in factors that regulate gland function. To test this idea, we first examined the expression of SPP1 (secreted phosphoprotein 1 also known as osteopontin). SPP1 is expressed in both the endometrial stroma and glands on D4 of pregnancy in the mouse and is suggested to positively regulate implantation [43][44][45] .
Immunohistochemical analysis of the uteri from E2-treated Kiss1 −/− and WT littermates on D4 of pregnancy revealed that while SPP1 was expressed throughout the uterus in both KO and WT uteri, it was almost completely absent in the GE of all glands in the Kiss1 −/− endometrium ( Fig. 3A-D). While SPP1 is expressed throughout the uterus, spatial examination allowed us to confirm that SPP1 expression was diminished in the GE. To continue testing our hypothesis, in the absence of well-validated antibodies, we quantified the expression of genes reported to be highly expressed (relative to other uterine cells) in the GE 46 ; these were Prss28 (protease, serine, 28) 47 , Prss29 (protease, serine, 29) 47 , Spink3 (serine peptidase inhibitor, Kazal type 3) 48 and Ttr (transthyretin) 46 . As a positive control, Foxa2 was also included in this analysis. Our results show that all genes including Foxa2 exhibited a significant reduction in expression in E2-treated Kiss1 −/− mice (Fig. 3E) leading us to conclude that kisspeptin signaling directly regulates gland function. Ovarian activin A output might be reduced in the Kiss1r −/− Tg mouse. Collectively, the studies conducted on the Kiss1 −/− and Kiss1r −/− Tg mice clearly reveal that in addition to E2, other peripheral factors that regulate adenogenesis are missing in these mice. Studies from the Spencer laboratory 49,50 provided correlative data pointing out that components of the ovarian activin-follistatin system might regulate neonatal ovine uterine size and adenogenesis. These authors have also indicated the possibility that inhibins might be important regulators of these processes 50 . Therefore, we quantified the mRNA expression of the genes encoding follistatin (Fst) and the α subunit of inhibin (Inha) in the ovaries of Kiss1r −/− Tg, Kiss1r −/− (global KO) and WT mice. We also examined the genes encoding the other subunits of inhibin A and activin A (β A: Inhba) and inhibin B and activin B (β B: Inhbb); activin AB being comprised of β A and β B subunits.

Re-expression of
Global loss of Kiss1r (KO) resulted in the significant up-regulation of ovarian Fst expression relative to WT littermates (Fig. 5A), while the reactivation of the neuroendocrine axis (Kiss1r −/− Tg) reduced expression to WT levels (Fig. 5A). As for Inha, there was a significant down-regulation in the KOs relative to WT littermates (Fig. 5B), while in Kiss1r −/− Tg mice expression was fully restored, reaching significantly greater levels than in the WT mice (Fig. 5B). The pattern of responses for Inhba and Inhbb were identical to each other (Fig. 5C,D) and similar to Inha (Fig. 5B), except that levels were not significantly different between WT and Kiss1r −/− Tg mice (Fig. 5C,D). Finally, we calculated the Inhba/Inha and Inhbb/Inha ratios as an indirect measure of activin (A, B and AB) output by the ovaries. The average Inhba/Inha ratio in the WT mouse was significantly greater than in the global KO, while Kiss1r −/− Tg mice displayed a partial rescue, although this parameter was markedly lower than in the WT and there was no significant difference vs. the ratio detected in the KO mouse (Fig. 5E). In contrast, Inhbb/Inha ratio was similar between the three genotypes (Fig. 5F). Therefore, the possibility exists that reduced levels of activin A might in part account for the reduced adenogenesis observed in the Kiss1r −/− Tg mouse (Fig. 4). secretion from the hypothalamus 7 . However, unlike the Kiss1 −/− , Kiss1r −/− and Kiss1r −/− Tg mouse, peripheral kisspeptin signaling is preserved. Analysis of the Gnaq d/d ;Gna11 −/− uterus, relative to that of Gnaq fl/fl ;Gna11 −/− littermate controls, revealed that gland number and uterine growth were significantly reduced; findings consistent with the idea that the central axis regulates adenogenesis and uterine growth (Fig. 6). However, although gland number and uterine growth were significantly reduced by about 58% and 51%, respectively, the Gnaq d/d ;Gna11 −/− uterus still exhibited greater adenogenesis and uterine growth than the untreated Kiss1 −/− and Kiss1r −/− mouse lines, where adenogenesis was almost completely ablated and uterine growth reduced by about 81% (Fig. 1A-H). Although this smaller reduction might be partially due to the fact that these mice still exhibit central kisspeptin signaling, though greatly diminished 7 , it is highly probable that this milder phenotype is also caused by the fact that, in contrast to Kiss1 and Kiss1r null models, they retain peripheral kisspeptin signaling intact.

Reduced hypothalamic GnRH secretion in the adult
Finally, and unexpectedly, FOXA2 expression was visibly and consistently increased in the GE of all glands in the Gnaq d/d ;Gna11 −/− uterus as well as throughout the rest of the uterus, in particular the luminal epithelium (Fig. 6A,B). Elevated glandular and ectopic FOXA2 expression was never observed in the Kiss1 −/− , Kiss1r −/− or Kiss1r −/− Tg mouse of similar age (Fig. 1). Since the major difference between the Gnaq d/d ;Gna11 −/− mouse and  Adenogenesis and uterine growth are disrupted in the juvenile Kiss1 −/− mouse. While our data revealed that kisspeptin signaling regulates adenogenesis and uterine growth, at both central and peripheral levels, the question that remained was whether kisspeptin signaling regulates adenogenesis and uterine growth in the juvenile mouse, a developmental period in which these processes are thought to occur in an adrenaland ovarian-independent manner 34,36,40 . We therefore examined gland number and uterine growth in juvenile (PND21) Kiss1 −/− mice and their WT littermates. Results clearly showed that in juvenile Kiss1 −/− mice adenogenesis and uterine growth were significantly diminished (Fig. 7A,B,E,F), demonstrating that kisspeptin regulates these processes in the juvenile mouse; a phenomenon that, according to previous evidence, should occur independently of the ovaries. Of note, while loss of Kiss1 significantly reduced gland formation and uterine growth in the juvenile mouse, relative to WT littermates, these parameters were only reduced by about 37 and 49%, respectively, therefore suggesting the roles for other signaling pathways in regulating these ovarian-independent processes in the juvenile mouse. When FOXA2 expression was first inspected in the juvenile uteri from Kiss1 −/− and WT littermates, it was clear that FOXA2 expression was more intense on the glands in both the juvenile KO and WT uteri compared to that seen in the adult Kiss1 −/− , Kiss1r −/− , Kiss1r −/− Tg and Gnaq d/d ;Gna11 −/− mice and their respective WT controls (Fig. 7 vs. Figs 1,2,4 and 6). Importantly, in all studies FOXA2 was detected under identical conditions. Because the expression level was so intense in the juvenile period, we could not initially assess whether there was a difference in glandular expression between the KO and WT mice. The study was therefore repeated using the anti-FOXA2 antibody at a 10-fold dilution (0.18 μ g/ml IgG). Results show that, under these conditions, glandular FOXA2 levels did not appear strikingly different between Kiss1 −/− and WT littermates (Fig. 7C,D).

Discussion
Our study reveals that global inactivation of Kiss1 or Kiss1r results in an almost complete loss (about 97%) of total endometrial gland formation and a significant reduction (about 81%) in uterine growth in the adult mouse. Therefore, the kisspeptin signaling system must be a major regulator of adenogenesis and uterine growth in the adult mouse. Since E2 treatment of the Kiss1 −/− mouse and re-expression of Kiss1r in GnRH neurons in the Kiss1r −/− mouse restored uterine growth fully but adenogenesis only by about an average of 25%, our results document a striking dissociation in the kisspeptin-dependent pathways controlling these two related, but clearly distinct phenomena. Thus, while uterine growth in the adult is largely dependent on central kisspeptin signaling, endometrial adenogenesis is predominantly regulated by peripheral kisspeptin signaling, which accounts for about an average of 75% of kisspeptin-dependent gland development (Fig. 8). These findings are further reinforced by the data from the Gnaq d/d ;Gna11 −/− strain, a conditional KO that exhibits diminished central kisspeptin signaling but fully preserved peripheral kisspeptin actions, in which the reduction in adenogenesis and uterine growth was smaller than that seen in Kiss1 and Kiss1r null mice (Fig. 8).
Taken together, the studies conducted on the E2-treated Kiss1 −/− and Kiss1r −/− Tg (rescued) mice revealed that in addition to E2, other centrally stimulated ovarian factors are needed for endometrial adenogenesis. These could include progesterone, follistatin, inhibins and activins. It is also possible that, at the dose set, the exogenously administered E2 was insufficient for triggering maximal adenogenesis. Of note, follistatin, activins, and inhibins regulate growth and differentiation of many branched epithelia-mesenchymal organs 51,52 , and have been suggested to play a role in uterine growth and adenogenesis in ovine neonates 49,50 . Our analyses in the global It is important to note that while activin A might be one of the centrally-stimulated peripheral factors that induces kisspeptin-dependent adenogenesis, other factors might also exist. Additionally, although there is unambiguous evidence that kisspeptin-dependent adenogenesis is positively regulated by peripheral kisspeptin signaling, where this signaling is localized is currently unknown.
Kiss1 −/− and the Kiss1r −/− Tg mice strongly suggest that activin A, whose output, measured as Inhbb/Inhba ratio, is severely blunted in the global KO and significantly reduced in the rescued model, might co-operate with E2 in the control of kisspeptin-dependent uterine growth and development (Fig. 8). In addition, reduced activin B and inhibin, and/or increased follistatin production might contribute to the severe uterine phenotype of mice with global inactivation of kisspeptin signaling.
A number of genes are known to positively regulate adenogenesis in the mouse uterus. These include Wnt4 32 , Wnt7a 31,33 , Wnt5a 53 , Ctnnb1 (β -catenin) 42,54,55 , Foxa2 30 , Hoxa11 56 , Dicer1 57,58 , Lgr4 59 , Dlx5 and Dlx6 60 . In each case, conditional inactivation of these genes in the uterus resulted in an almost complete loss of gland formation. While the majority of these studies only examined adenogenesis in the adult mouse or shortly after weaning (PND21) and reported on a severe reduction in gland number compared to control mice, adenogenesis was also examined in the neonatal and juvenile periods in mice lacking Dicer in Müllerian duct mesenchyme-derived tissues of the reproductive tract and in the juvenile period in mice lacking Lgr4 in epithelial cells 59 . In the case of the conditional Dicer KO, initial adenogenesis at PND4 and 8 was similar between the conditional KO and control mice but by PND14 and 21 glands were almost absent 57,58 . However, from about 5 weeks to 4 months of age the number increased gradually but at all times was consistently reduced compared to control mice. In the conditional Lgr KO, glands were also almost completely absent at PND21 and this remained unchanged at 9 weeks of age 59 . Essentially, the same finding was made with the Kiss1 KO in the juvenile period (PND21), except that at PND21 a greater number of glands was detected in the Kiss1 −/− endometrium. Taken together, it appears that Kiss1, Dicer and Lgr4 are important regulators of adenogenesis in the juvenile period, which is characterized as being ovarian-and ESR1-independent.
Our study revealed that in the adult E2-treated Kiss1 −/− mouse and the Kiss1r −/− Tg mouse not only was gland number reduced but so was FOXA2 levels on the GE. This striking phenotype led us to conclude that kisspeptin signaling regulates adenogenesis in a FOXA2-dependent manner. Based on this, it was predicted that FOXA2 expression would have also been reduced in the Gnaq d/d ;Gna11 −/− mouse, but surprisingly the opposite was seen. Perhaps, this reflects a peripherally-stimulated compensatory response to diminished central signaling in this mouse. Why then was there not a similar central response to absent peripheral signaling in adult E2-treated Kiss1 −/− and Kiss1r −/− Tg mice? The answer to this interesting question is not known and might be linked to the observation that central signaling only accounts for about 25% of all adenogenesis while peripheral signaling accounts for the rest. More importantly, this observation further reinforces that mechanistically, central and peripheral kisspeptin pathways regulate adenogenesis differently.
This putative relationship between kisspeptin signaling and FOXA2 was only uncovered through our ability to rescue adenogenesis in suitable models, such as the E2-treated Kiss1 −/− and the Kiss1r −/− Tg mouse, before assessing FOXA2 expression. FOXA2 belongs to a family of three forkhead transcription factors encoded by different genes and is implicated in the development of organs such as the liver, pancreas, lung, prostate and uterus [61][62][63][64] . In the uterus, FOXA2 is uniquely localized to the GE in the WT endometrium and is essential for adenogenesis in the mouse 30,46 . Recently, Filant et al. 46 undertook a genome-wide investigation of in vivo FOXA2 binding target regions in the neonatal and adult uterus and found that in the neonatal uterus, FOXA2-bound genes in the GE were enriched for developmentally related processes including cell cycle, cell junction and focal adhesion while in the adult uterus there was an enrichment for functional processes including metabolic pathways, focal adhesion and WNT signaling. These important results further define how FOXA2 regulates endometrial gland development and function.
In our initial characterization of the infertility observed in Kiss1 −/− and Kiss1r −/− mice, we found that LIF was absent in all endometrial glands of E2-treated KO mice but if given exogenously could rescue the implantation defect 20 . This initially led us to conclude that LIF lies downstream of kisspeptin and that kisspeptin signaling is a positive regulator of glandular LIF expression and secretion. However, we now realize that diminished LIF expression is the indirect consequence of having non-functional glands and that it is less likely that kisspeptin signaling positively regulates its expression. This conclusion is based on the findings that (1) FOXA2 expression is diminished in glands from the Kiss1r −/− Tg mouse and sometimes even absent in the E2-rescued Kiss1 −/− mouse; (2) normal glandular morphology is disrupted in the E2-rescued Kiss1 −/− mouse; and (3) glands in the E2-rescued Kiss1 −/− mouse exhibit diminished expression of FOXA2 30 , LIF 27 , SPP1 [43][44][45] , Prss28 47 , Prss29 47 , Spink3 48 and Ttr 46 , molecules implicated in gland development and function 46 . We therefore suggest that kisspeptin signaling positively regulates both gland development and function and that in E2-treated Kiss1 −/− mice, while development is partially rescued, function is not. Interestingly, both development and function appeared to have been partially rescued in the Kiss1r −/− Tg mouse, again highlighting that other centrally-stimulated peripheral factors were missing in the E2-treated Kiss1 −/− mouse.
The data presented in this study reveal that kisspeptin-dependent adenogenesis is regulated by both central and peripheral pathways, but it is unknown whether both pathways contribute to the development of a single pool of glands or whether each contributes to a discrete pool. Additionally, while it is established that the central system resides in the hypothalamus, it remains unknown where the peripheral signaling system actually resides (Fig. 8). Based on a description of peripheral cells and tissues that express either kisspeptins and/or their receptors, possible sites are the ovary and uterus, although the contribution of other non-reproductive sites of action of kisspeptins, such as the liver and pancreas, cannot be excluded 1,13,16,18,20 . Although yet to be fully proven, we suggest that the uterus remains a strong candidate given our previously published data showing that on D4 of pregnancy, the uterus expresses a functional kisspeptin signaling system, on both the luminal and glandular epithelia 20,21 .
All mammalian uteri contain endometrial glands that secrete substances that positively regulate embryo implantation and subsequently support the survival and development of the conceptus (embryo and associated placental membranes) during pregnancy 28,29 . Human uterine secretions are enriched in cytokines, chemokines and growth factors and their levels appear to correlate positively with successful implantation and the establishment of a chemical pregnancy [65][66][67] . Despite these important findings, our understanding of gland function in human pregnancies lags behind our understanding in laboratory and domestic animals. Therefore, studies such as those described here with the Kiss1 −/− , Kiss1r −/− , Kiss1r −/− Tg the Gnaq d/d ;Gna11 −/− mouse will allow us to develop and test hypotheses designed to better understand gland function in human pregnancies. A better understanding could lead to higher implantation rates and successful pregnancy outcomes following assisted reproduction. Based on the current study, we conclude that while uterine growth in the adult is largely dependent on central kisspeptin signaling, endometrial adenogenesis is predominantly regulated by peripheral kisspeptin signaling, which accounts for about 75% of kisspeptin-dependent gland development (Fig. 8).
The Kiss1r −/− Tg (also referred to as Gpr54 −/− Tg) mouse is a GnRH neuronal-specific Kiss1r expressing (rescued) mouse line generated in the C57BL/6 background using BAC transgenesis, and is a generous gift to Dr. M. Tena-Sempere from the groups of Drs. G. Schuzt and M. Kirilov (German Cancer Research Center, Heidelberg, Germany) and A.E. Herbison (Centre of Neuroendocrinology, University of Otago, NZ) 26 . Kiss1r −/− Tg mice are fertile, and the line was maintained by crossing Kiss1r −/− Tg males to females.
The Gnaq d/d ;Gna11 −/− mouse, which was created in the Babwah laboratory 7 , is a global knockout for Gna11 but conditionally lacks Gnaq in its GnRH neurons. Consequently, Kiss1r-coupled Gα q/11 -signaling at the level of the GnRH neuron is abolished but Kiss1r continues to signal and mediate kisspeptin-dependent GnRH secretion, albeit weakly, via the β -arrestin-dependent pathway 7 . The mouse was generated in the C57BL/6J genetic background and is infertile. Therefore, the Gnaq d/d ;Gna11 −/− mouse and Gnaq fl/fl ;Gna11 −/− littermate controls were generated by crossing the Gnaq fl/fl ;Gna11 −/− line to a line bearing the GnRH-Cre transgene and the segregating genotypes were identified as previously described 7 . Hormonal treatments. Three to four week-old female mice (Kiss1 −/− and WT littermates) were administered E2 (100 μ g/100 μ l sesame oil) subcutaneously every 3-4 days over a 5-week period, then administered 7.5 IU pregnant mare serum gonadotropin (PMSG; Folligon; Intervet) intraperitoneally (i.p.) followed 48 hours later by 7.5 IU human chorionic gonadotropin (hCG; Chorulon; Intervet) i.p. Immediately after the hCG injection, mice were mated to WT males (D0 = day of mating) 20 . On D4 of pregnancy, uteri were collected and glands were characterized by analyzing FOXA2 and SPP1 immunoreactivity. The D4 uteri were used in quantifying the mRNA levels of Prss28, Prss29, Spink3, Ttr and Foxa2.

Animal husbandry. Animal studies involving the
Quantitative real-time RT-PCR studies. Gene expression studies were conducted independently in the Babwah and Tena-Sempere Laboratories. Protocols employed by each laboratory are described below.
Babwah Laboratory ( for the analysis of uterine Foxa2, Prss28, Prss29, Spink3 and Ttr). Gene expression was determined on total RNA prepared from the entire uterine horns of experimental and control mice. Freshly harvested tissues were collected in RNAlater (Life Technologies Inc., Burlington, ON, Canada) and RNA was isolated using the Qiagen RNeasy mini kit according to manufacturer's instructions (Qiagen, Missassauga, ON, Canada). One μ g of total RNA was reverse-transcribed using SuperScript II (Invitrogen, Burlington, ON, Canada). Reactions were performed according to the manufacturer's protocol using random hexamer primers (Amersham, Piscataway, NJ). Quantitative real-time PCR was performed in duplicate for each sample and done a total of three independent times using IQ SYBR Green Master Mix (Bio-Rad Laboratories, Mississauga, ON, Canada). To determine PCR efficiency, a 10-fold serial dilution of cDNA was performed as described previously 68 . Gene expression was normalized to Actb expression and presented as relative expression using the Pfaffl method 69 . Expression of the following genes was quantified using the following primers (presented 5′-3′ ). Tena-Sempere Laboratory ( for the analysis of ovarian Fst, Inha, Inhba and Inhbb). Total RNA was extracted using TRIsure isolation reagent (Bioline Reagents Ltd., UK) and treated with DNase Q1 (Promega corporation, USA). One μ g of total RNA was subjected to reverse transcription using IScript cDNA Synthesis kit (Bio-Rad Laboratories Inc., USA). For real-time PCR, we used Go Taq qPCR Master mix (Promega Corporation, USA) in a CFX96 Touch Real-Time PCR Detection System (Bio-Rad Laboratories Inc., USA). Primer-specific amplification and quantification cycles were run at 95 °C for 25 s, 62 °C for 25 s and 72 °C for 30 s and a final extension of 72 °C for 20 s. To normalize the quantification of inhibins subunits and follistatin mRNA level, we measured the amount of ribosomal 18S mRNA in each protocol. The corresponding standard curve for each gene was obtained by serial dilution of a reference ovarian cDNA sample. Expression of the following genes was quantified using the following primers (presented 5′-3′ ). Immunohistochemistry. Uteri were collected and processed for paraffin immunohistochemistry, as described previously 20 . Sections were then incubated in rabbit anti-FOXA2 IgG (1.8 μ g/ml, catalogue # AB108422, ABCAM, Cambridge, MA, USA) or rabbit anti-SPP1 IgG (1:10,000 dilution, catalogue number AB10910, Millipore, Etobicoke, ON, Canada). In experiments represented by Fig. 7C,D, anti-FOXA2 IgG was used at a final concentration of 0.18 μ g/ml. Antigen-bound primary antibodies were detected with the ImmunoCruz rabbit ABC Staining System (catalogue number sc-2018, Santa Cruz Biotechnology, Inc. Dallas, TX, USA). The secondary detection systems were used according to the manufacturers' guidelines without any adaptations. Experimental and control samples were processed in parallel and treated with the 3,3′ -diaminobenzidine substrate for an identical period of time. This allowed us to compare relative expression levels between experimental and control samples. Experimental conditions were carefully maintained between independent assays and analyses were conducted 5-20 independent times. We found it was visually easier to assess expression levels of FOXA2 and SPP1 in the absence of a counterstain; thus, tissue sections were not counterstained. Coverslips were affixed to slides with Permount mounting medium (Fisher Scientific, Ottawa, ON, Canada).
Slides were scanned using an Aperio ScanScope XT in conjunction with the ImageScope software and the area of transverse uterine sections determined using the annotation tool. Total uterine area (including the uterine lumen) and uterine luminal area were calculated and expressed as μ 2 . Uterine luminal area was then subtracted from the total area and the remaining area comprised of the myometrium and endometrium was used as an indication of uterine growth. The data in this study represent the average area (μ 2 ) of a transverse uterine section (excluding the uterine lumen) per uterine horn ± SEM. FOXA2 immunostaining was conducted to determine gland number and morphology. The data in this study represent the average number of FOXA2-positive glands/ transverse section of uterine horn ± SEM.
Statistics. The differences between groups were determined using unpaired Student's t-test or one-way ANOVA followed by post hoc Student-Newman-Keuls test (GraphPad Prism Software, Inc, La Jolla, CA). All values are expressed as mean ± SEM and a value of P < 0.05 was considered statistically significant.