Role of Fibroblast Growth Factor Signaling in Gonadotropin-Releasing Hormone Neuronal System Development

Transgenic Mouse Models

This review will analyze GnRH system defects in four lines of transgenic mice deficient in FGF signaling. Each transgenic mouse line provides a unique tool for disrupting FGF signaling, but also has its own limitations (table 1).

A transgenic mouse (dnFGFR mouse) was generated by targeting the expression of a dominant-negative FGFR (dnFGFR) to GnRH neurons using the rat GnRH promoter [20]. Although this mouse allows the tissue-specific disruption of FGFR function in GnRH neurons, it comes with two limitations. First, it is not suitable for studying the very early phase of GnRH neuronal development (including fate specification) because GnRH promoter activity at this early period is presumably too weak to drive the expression of dnFGFR. Second, since dnFGFR disrupts the function of FGFRs 1, 2, and 3 [20], the observed phenotypes cannot be attributed to the disruption of a specific FGFR subtype. To circumvent these limitations, we utilize mice that are hypomorphic for Fgfr1 (Fgfr1 hypomorphs) [21] and knockout for Fgfr3 (Fgfr3 KO) [22] to elucidate the contribution of each receptor and to ensure their deficiency/deletion at the very beginning of GnRH neuron genesis. The reasons for concentrating on these two FGFRs are described in the section below. Since FGF8 has been identified as an FGF ligand indispensable for the morphogenesis of the olfactory region [23], the birthplace of GnRH neurons, we also use mice hypomorphic for Fgf8 (Fgf8 hypomorphs) to examine the specific contribution of this ligand. It is important to note that Fgfr1 hypomorphs, Fgf8 hypomorphs, and Fgfr3 KO bear the global deletion/deficiency in FGF signaling components; thus, indirect effects upon GnRH neurons cannot be ruled out. However, compared to mice with tissue-specific disruption, the hemizygous state of global disruption in these mice may be more comparable to humans harboring Fgfr1 and/or Fgf8 mutations.

Regulation of Gonadotropin-Releasing Hormone Neuronal Fate Specification by Fibroblast Growth Factor Signaling

Landmark studies showed that the vast majority (∼80%) of mouse GnRH neurons undergo their last division in the olfactory placode between E9.5 and E10.5 [2]. Interestingly, these postmitotic olfactory placode progenitor cells do not express detectable levels of GnRH mRNA or peptide prior to E11.5 [1, 2]. In this sense, one can define the period of GnRH neuron fate specification to occur between E9.5–E11.5. The absence of measurable GnRH peptide prior to E11.5 is a major hurdle for studying the molecular processes that regulate GnRH neuronal fate specification, because at present the only marker for GnRH neurons is GnRH itself. Therefore, postmitotic olfactory placode progenitor cells that are destined to become GnRH neurons cannot be identified during a large portion of the fate specification, before they express measurable GnRH levels. Notwithstanding this major caveat, our laboratory has accumulated mounting evidence that the activation of FGF signaling is important for the fate specification of the mouse GnRH neuronal system.

In mice, FGF signaling is mediated by four specific types of FGFRs [24]. Localization studies from our laboratory showed that the E10.5 mouse olfactory placode contained abundant levels of FGFR1, FGFR2 and FGFR3, but not FGFR4 protein [25]. Furthermore, dual immunocytochemistry detected the presence of FGFR1 and FGFR3, but not FGFR2 protein in E15.5 and postnatal-day-3 GnRH neurons [25]. This is consistent with an earlier observation showing that Fgfr1 and Fgfr3 mRNA is expressed in immortalized mouse GnRH neuronal cell lines [6]. The strongest data indicating that FGF signaling plays a role in GnRH neuronal fate specification are derived from in vitro nasal explants isolated from E10.5 embryos. In this experimental paradigm, the number of GnRH neurons in the nasal explants increased after 2 or 3 days in vitro [25]. This increase in GnRH neurons was inhibited by the FGFR antagonist SU5402, indicating the activation of FGFR is required for GnRH neuron fate specification [25].

The aforementioned studies provided the impetus for identifying the specific FGF signaling component(s) required for GnRH neuronal fate specification. Based on the presence of FGFR1 and FGFR3 in GnRH neurons, we hypothesized that activation of these receptors was critical for mouse GnRH neuronal fate specification. Therefore, disruption of FGFR1 or FGFR3 function is expected to result in a defective GnRH neuronal system. We tested this hypothesis using mice globally deficient in FGFR1 (Fgfr1 hypomorphs) or FGFR3 (Fgfr3 KO), which have a major advantage. Unlikely tissue-specific KOs that require the GnRH promoter to be active before disruption can occur, the global disruption of genes guarantees FGFR1 or FGFR3 deficiency at the time of GnRH neuronal fate specification. We first examined the total number of GnRH neurons in the entire heads of newborn wild-type, heterozygous and homozygous Fgfr1 hypomorphs, and found that the number of GnRH neurons in homozygous Fgfr1 hypomorphs was about 88% lower than their wild-type littermates [26]. Furthermore, there was no significant difference between heterozygous Fgfr1 hypomorphs and their wild-type littermates [26]. On the other hand, the number of GnRH neurons was unaffected in newborn Fgfr3 KO mice [26]. These whole animal studies corroborated the nasal explant studies and supported the idea that activation of FGFR function may be a key pathway required for the emergence of GnRH neurons [25]. Furthermore, by comparing the GnRH neuronal system between Fgfr1 hypomorphic and Fgfr3 KO mice, we established that the activation of FGFR1, but not FGFR3, is required for the normal number of GnRH neurons in the newborn mouse head.

Of the 22 identified FGF ligands, there are several lines of experimentally derived evidence indicating that FGF8 is the most likely FGF candidate to be involved in the early development of GnRH neurons. First, Fgf8 mRNA is found in the ectodermal region ventral/lateral of the telencephalic commissural plate on E9.5, which will form the olfactory placode [23, 27]. This expression profile suggests that FGF8 signaling is critically important for the induction and differentiation of the mouse olfactory placode. Second, a partial loss of FGF8 function resulted in the absence of the olfactory bulbs and vomeronasal organs (VNO) [26, 28]. The absence of the VNO has a special relevance given that the VNO also originates in the olfactory placode [2]. Third, FGF8 has been shown to bind and readily activate FGFR signaling, including FGFR1, in cell lines [10]. Therefore, we hypothesized that disruption of FGF8 signaling will severely impair the genesis of the GnRH neuronal system. To test this hypothesis, we examined the GnRH neuronal system in newborn Fgf8 hypomorphic mice, which in the homozygous state shows a ∼54% downregulation of Fgf8 mRNA [28]. Much to our surprise, the GnRH neuronal system in homozygous Fgf8 hypomorphs was not partially affected, but eliminated in its entirety [10, 26]. An equally striking observation was that heterozygous Fgf8 hypomorphs had approximately 50% fewer GnRH neurons remaining as compared to wild-type littermates [10, 26]. These results led to our conclusion that the prenatal development of the mouse GnRH neuronal system is extremely dependent on the presence of FGFR1 and FGF8 signaling.

A major constraint of studies in newborn mice mentioned before is that they do not elucidate whether FGF8 signaling is required for GnRH neuronal fate specification or other developmental processes, such as neuronal survival after emergence. Therefore, we expanded our studies to include the analysis of the E11.5 GnRH neuronal system, when GnRH neurons are first detectable using immunocytochemistry. By examining the emergence of GnRH neurons from the E11.5 olfactory placode, we were able to directly answer the question: Is FGF8 signaling required for the fate specification process of GnRH neurons in the olfactory placode? In contrast to wild-type embryos, we did not detect any GnRH neurons in the olfactory placode of homozygous Fgf8 hypomorphs on E11.5 or in any subsequent time points that were examined [26]. Together with the previously described nasal explant studies, these data strongly suggest that FGF8 signaling acting through FGFR1 is critical for GnRH neuron fate specification in the mouse olfactory placode.

Interestingly, we found that the secondary recess of the E11.5 olfactory placode from where GnRH neurons emerge is virtually absent in homozygous Fgf8 hypomorphs. The dysgenesis of this secondary recess is likely the underlying cause for the failure of GnRH neurons to emerge (fig. 1). Our preliminary data showing that the VNO is missing in the homozygous Fgf8 hypomorphs are consistent with the notion that this recess fails to develop and give rise to differentiated derivatives. Our data also suggest the presence of discrete developmental zones in the olfactory placode [13], with FGF8 signaling deficiency more likely to impact the zone that gives rise to the epithelial cells of the VNO and GnRH neurons. Currently, we are studying how a partial loss of FGF8 function could severely disrupt the developmental organization of the secondary recess of the olfactory placode.

Regulation of Gonadotropin-Releasing Hormone Neuron Migration by Fibroblast Growth Factor Signaling

The organization and functionality of the GnRH neuronal system depends not only on the proper emergence of GnRH neurons from the olfactory placode, but also on the accurate migration from the olfactory placode to the preoptic/hypothalamus region. Indeed, the failure of GnRH neurons to migrate and transition from the nasal cavity into the anterior forebrain was considered to be the major cause of GnRH deficiency in KS [29]. This view was based on the observation that GnRH migration was arrested just prior to entering the forebrain in a 19-week-old fetus with the X-linked form of KS [30]. Indeed, no GnRH neurons were detected in the forebrain regions where they were expected during this stage of embryonic development, but were instead found beneath the forebrain within the meninges [29]. Moreover, the VNN fibers in this KS fetus also ended within the meninges and failed to contact the developing olfactory bulbs [29].

GnRH migratory defect is frequently extrapolated to be the key cause of IHH in all forms of KS, including KS patients who harbor loss-of-function mutations in the Fgfr1 or Fgf8 genes [7, 10]. It is possible that GnRH migration is affected in these KS patients, but there are no studies verifying this in human fetuses with mutations in Fgfr1 or Fgf8. In fact, mouse studies done in our laboratory found little evidence supporting the view that defects in FGF signaling cause impaired GnRH neuronal migration. Only a slight arrest (<10%) in GnRH neurons was detected in the intra-hemispheric area of newborn dnFGFR mice as compared to their wild-type counterparts [31]. This slight arrest in GnRH neuron migration is inconsistent with the significant reduction of GnRH neurons in the preoptic/hypothalamus of adult dnFGFR mice. The majority of GnRH neurons in dnFGFR mice still migrated normally into the brain [20], and the loss of GnRH neurons in dnFGFR mice is due primarily to postnatal disappearance instead of migratory defects. These data suggest that the disruption of FGFR function directly at the level of GnRH neurons plays only a minor role in their migration [31].

One might also question if global deficiency of FGF signaling could alter the path-finding of axons that guide GnRH neuronal migration, thereby indirectly disrupting this migratory process. It is interesting to note that the peripherin-positive fibers of VNN were in fact disrupted and failed to target the telencephalon of homozygous Fgf8 hypomorphs (fig. 1) [26]. However, it is important to emphasize that GnRH neurons were already missing in these transgenic mice prior to migration; thus, the cause of GnRH system defect in these mice is failure to emerge, not failure to migrate. In heterozygous Fgf8 hypomorphs with 50% of GnRH neurons still remaining, the reduction in GnRH neurons was also not due to defective GnRH migration. We base this conclusion on our analysis of the entire head of these newborn mice, in which we did not find any evidence of abnormal accumulation of GnRH neurons along the migrational pathway prior to entering the brain [26]. A similar observation was made for homozygous Fgfr1 hypomorphs. Consequently, data from mice suggest that IHH in KS patients harboring Fgfr1 or Fgf8 mutations is more likely caused by defects in GnRH neuronal fate specification than in migration.

Regulation of Gonadotropin-Releasing Hormone Axon Targeting by Fibroblast Growth Factor Signaling

During axon targeting, GnRH fibers extend ventrally in a highly directional manner to terminate upon the fenestrated capillaries within the external zone of the ME. GnRH is released into the primary capillary plexus to reach the hypothalamo-pituitary portal veins, and ultimately the anterior pituitary. Thus, this is the stage that confers final fertility. In order for GnRH axons to target the ME properly, two events must occur: (1) the extension of the axons, and (2) the directional growth of axons to the proper target. There is virtually nothing known about the mechanisms of GnRH axon targeting in humans. However, there is a wealth of information on how FGF signaling regulates GnRH axon targeting in mice.

Regarding the extension of axons, in vitro data from our laboratory showed that this process is regulated by FGF signaling [25]. The addition of FGF2, a prototypic FGF, significantly stimulated neurite outgrowth in cultured E15.5 GnRH neurons [25]. Importantly, GnRH neurons isolated from dnFGFR mouse embryos exhibited attenuated neurite outgrowth in culture [32], suggesting FGF signaling is an endogenous cue that maintains GnRH axon growth. Regarding the directional axonal growth to the proper target, Silverman et al. [33] showed that GnRH neurons grafted into the third ventricle targeted axons to the ME, and this directional axonal growth was due to the presence of a diffusible chemoattractive signal(s) in the target [34]. To date, the identity of the diffusible chemoattractant(s) remains elusive, but several lines of evidence suggest FGFs might be one of the components. FGF2, a molecule produced in abundance by both endothelial cells and ventricular tanycytes of the ME, served as a diffusible cue to attract the directional growth of GnRH axons in an organotypic culture system [35]. The caveat is that while the study showed an attractive response of GnRH axons to an exogenously added FGF, it did not verify the role of FGFs as endogenous regulators in this process. To satisfy the criterion of FGF signaling as an endogenous chemoattractive regulator, we have demonstrated that in cultured embryonic GnRH neurons, the directional outgrowth of neurites toward the cocultured ME explant is significantly reduced in GnRH neurons expressing dnFGFR [32]. Further indirect evidence arises from the recent observation that anosmin-1, the diffusible extracellular product of KAL1, in turn exerts a modulatory role of FGF2/FGFR1 signaling [36].

Overall, our in vitro results showed attenuated FGF signaling disrupted the ability of GnRH neurites to extend and to grow in a directional manner. Do these in vitro alterations translate into GnRH system defects in the whole animal? There is evidence that suggests they do, but these in vivo defects are somewhat less striking. GnRH neurons in dnFGFR mice showed a significant but small reduction (15%) in the uptake of peripherally injected fluorogold, a retrograde tracer, indicating compromised contact between GnRH axons and circumventricular organs such as the ME [32]. That GnRH axon targeting is only modestly disrupted in dnFGFR mice argues for the presence of other compensatory mechanisms guiding GnRH axons in vivo. Since GnRH axon targeting is a highly axophilic event in which GnRH axons travel with other growing axons [34], these guiding axons could drive the extension and directional growth of GnRH axons via physical or other FGF-independent means. Overall, despite significant defects observed in vitro, adequate GnRH fibers have targeted the ME of dnFGFR mice to support reproduction. Current evidence suggests that GnRH axon targeting in the whole animal does not appear to be a process that critically requires FGF signaling.

Regulation of Postnatal Gonadotropin-Releasing Hormone System Function by Fibroblast Growth Factor Signaling

The proper timing of pubertal onset and reproductive senescence is key to a normal reproductive lifespan. In this regard, a healthy GnRH system becomes activated in a timely fashion and remains functional for a period of time optimal for the propagation of offspring. At present, there is direct and circumstantial evidence for a role of FGF signaling in the onset and termination of the mouse GnRH system function, thereby determining the duration of the reproductive lifespan.

Mouse GnRH neurons undergo dynamic remodeling in their morphology after birth, and this remodeling could have a functional significance [37]. During the pubertal transition, mouse GnRH neurons transform from having a complex dendritic tree bearing a few spines to having a simple dendritic structure bearing a large number of spines [37]. Consequently, the remodeled GnRH neurons in adults possess high dendritic spine densities, indicating the acquisition of abundant excitatory synaptic inputs [38] that could drive puberty. These observations contradict the previous reports that GnRH neurons receive sparse synaptic inputs [39, 40] and demonstrate pubertal onset is accompanied by the morphological maturation of GnRH neurons. Could FGF signaling be involved in the morphological maturation of postnatal GnRH neurons? Results from our laboratory suggest this is likely. We have shown that FGF signaling stimulates in vitro neurite branching of postnatal, but not embryonic, GnRH neurons [25]. Although the significance of this finding is unclear, it implicates FGF signaling in the postnatal alterations of GnRH neuronal morphology. Of further importance, dnFGFR female mice exhibit delayed puberty [20]. It stands to reason that delayed puberty in dnFGFR mice may be correlated with the reduced ability of their GnRH neurons to undergo morphological remodeling conducive to the receipt of excitatory inputs.

In addition to the postnatal maturation of GnRH neuronal morphology, FGF signaling could be required to maintain the postnatal GnRH system, thus promoting a healthy reproductive lifespan. Wu et al. [41] reported that there were ca. 1,100 GnRH neurons in neonatal mice, and this number declined to ca. 800 in adult mice. From then on, the number of GnRH neurons remained either unchanged [42] or declined only modestly by 18% in aging mice [43]. Overall, there does not appear to be a marked decrease in GnRH neurons with aging, suggesting GnRH neurons are unlikely to undergo dedifferentiation or cell death as they age. Rather, the aging GnRH system is more likely to undergo functional declines in GnRH synthesis and secretion [44]. Contrary to these observations, we show that postnatal dnFGFR mice, as they age, suffer a 70% loss in GnRH neurons and 95% loss in hypothalamic GnRH peptide concentration, respectively [26]. Our data suggest neurotrophic support by FGF signaling is absolutely critical for the maintenance of the postnatal GnRH system. In this sense, GnRH neurons are not unlike other neuronal systems such as motor neurons that rely heavily upon neurotrophic support for postnatal survival [45, 46]. Future studies are needed to examine the time course of postnatal GnRH neuronal loss and the mechanism underlying this loss.

Clinical Implications

Defective GnRH neuron migration was thought to be one of the major causes of IHH in all forms of KS. The lack of gross defects in GnRH neuronal migration in our dnFGFR, Fgfr1 hypomorphic and Fgf8 hypomorphic mice suggests that this extrapolation may not hold true for KS patients harboring loss-of-function mutations in Fgfr1 and/or Fgf8. On the other hand, we have evidence indicating that disruption of FGFR1 and FGF8 signaling may have caused major defects during the fate specification phase of the GnRH system. Furthermore, our studies showed that FGF signaling is required for the postnatal maintenance and functionality of the GnRH neuronal system. Alltogether, a single mutated copy of the Fgfr1 and/or Fgf8 gene may lead to multiple insults on the GnRH neuronal system during prenatal and postnatal life.

Studies from mouse clearly showed that FGF signaling is involved, albeit not of equal importance, in all phases of GnRH neuronal development, maintenance, and functionality. For example, the global disruption of FGF signaling impacts primarily the fate specification phase of the GnRH neuronal system, whereas we did not observe gross abnormalities in GnRH neuronal migration. One might argue that if GnRH neurons were able to emerge, FGF signaling deficiency may still cause major migratory impairment later on. Our data from dnFGFR do not fully support this notion. In dnFGFR mice in which the tissue-specific disruption of FGFR occurs only after fate specification, GnRH neurons emerged normally, as expected (unpubl. data). After the emergence of these neurons, we observed only a slight arrest in GnRH neuronal migration [31]. The major GnRH system defect in these mice appears to be the precipitous postnatal loss of GnRH neurons that likely led to premature reproductive senescence. Based on these observations, we hypothesize that a heterozygous Fgfr1 and/or Fgf8 mutation in KS patients may lead to multiple insults on the GnRH neuronal system during fate specification and after birth. The postnatal loss of GnRH neurons may, in part, explain the manifestation of adult-onset IHH seen in a patient harboring Fgf8 mutation [10].

Clinical observations showed that carrying a single identical gene mutation within and across IHH families is not predictive of the clinical reproductive and non-reproductive phenotypes, suggesting incomplete penetrance of a specific gene defect [47]. Another interesting and novel explanation for the lack of predictive precision could be that the severity of IHH depends on the number of gene defects found in a patient [47]. Indeed, Pitteloud et al. [47] identified a KS patient harboring heterozygous defects in both Fgfr1 and Nelf. Recently, compound defects in Fgfr1 and Fgf8 have also been identified in two cases with normosmic IHH [10]. Although these invaluable clinical observations reveal the genetic complexity of IHH and KS, it is not possible to study directly in humans how these multiple mutations exacerbate IHH in KS patients. In theory, the mouse model could be extremely useful for testing the compound effects of such gene defects. For example, by crossbreeding Fgfr1 hypomorphs with Fgf8 hypomorphs, we can examine whether heterozygous defects on multiple IHH or KS-associated genes have compounding effects on the GnRH neuronal system.

KS patients represent a unique ‘human model’ that provides insights into signaling factors involved in the organization of a critical neuroendocrine system. Further elaboration on the molecular mechanisms induced by these factors is accomplished by studying transgenic mice. Using the mouse models, we have shown that FGF signaling is required not only for olfactory placode induction and differentiation, but also for fate specification and postnatal maintenance of the GnRH neuronal system. In sum, we speculate that IHH in KS is not merely a result of defective GnRH neuronal migration, but also insults accumulated in the GnRH system during pre- and postnatal development.