Abstract
FGF signaling, an important component of intercellular communication, is required in many tissues throughout development to promote diverse cellular processes. Whether FGF receptors (FGFRs) accomplish such varied tasks in part by activating different intracellular transducers in different contexts remains unclear. Here, we used the developing mouse telencephalon as an example to study the role of the FRS adapters FRS2 and FRS3 in mediating the functions of FGFRs. Using tissue-specific and germline mutants, we examined the requirement of Frs genes in two FGFR-dependent processes. We found that Frs2 and Frs3 are together required for the differentiation of a subset of medial ganglionic eminence (MGE)-derived neurons, but are dispensable for the survival of early telencephalic precursor cells, in which any one of three FGFRs (FGFR1, FGFR2, or FGFR3) is sufficient for survival. Although FRS adapters are dispensable for ERK-1/2 activation, they are required for AKT activation within the subventricular zone of the developing MGE. Using an FRS2,3-binding site mutant of Fgfr1, we established that FRS adapters are necessary for mediating most or all FGFR1 signaling, not only in MGE differentiation, but also in cell survival, implying that other adapters mediate at least in part the signaling from FGFR2 and FGFR3. Our study provides an example of a contextual role for an intracellular transducer and contributes to our understanding of how FGF signaling plays diverse developmental roles.
SIGNIFICANCE STATEMENT FGFs promote a range of developmental processes in many developing tissues and at multiple developmental stages. The mechanisms underlying this multifunctionality remain poorly defined in vivo. Using telencephalon development as an example, we show here that FRS adapters exhibit some selectivity in their requirement for mediating FGF receptor (FGFR) signaling and activating downstream mediators that depend on the developmental process, with a requirement in neuronal differentiation but not cell survival. Differential engagement of FRS and non-FRS intracellular adapters downstream of FGFRs could therefore in principle explain how FGFs play several distinct roles in other developing tissues and developmental stages.
Introduction
A central question in biology that remains poorly addressed is how do so few families of extracellular signals such as FGFs mediate so many functions throughout development and in adulthood? One reason for this lack of understanding is that few studies have examined various intracellular signal transduction modules in specific contexts in vivo, relying instead on cell culture and biochemical analyses.
The FGF signaling system has undergone enormous expansion in mammals compared with invertebrates. Twenty-two FGF ligands and cell-adhesion molecules acting as ligands, can signal via four cell-surface FGF receptors (FGFRs) (Williams et al., 1994a; Mason, 2007). Tissue-specific alternative splicing of Fgfr genes confers differential ligand-binding properties (Johnson et al., 1991; Miki et al., 1992). This alone, however, cannot explain the pleiotropic roles played by FGFs in so many different contexts from development to physiology (Basson et al., 2008; Guillemot and Zimmer, 2011; Hébert, 2011). As an example, mutations in Fgfr1 result in diverse skeletal, hematopoietic, and nervous system diseases (Muenke et al., 1994; Xiao et al., 1998; Dode et al., 2003). Here, we address how specificity is governed in the FGF signaling pathway. Namely, can the choice of intracellular mediators by FGFRs govern functional specificity within a given context?
FGFRs can activate canonical intracellular transducers such as FRS and PLCγ directly, as well as other transducers such as Crk/Crkl and Grb14, thereby initiating multiple downstream signaling cascades (Williams et al., 1994b; Partanen et al., 1998; Turner and Grose, 2010; Goetz and Mohammadi, 2013; Brewer et al., 2015). The FRS family of docking proteins has two members, FRS2 (also called FRS2α) and FRS3 (also called FRS2β), which constitutively interact with the juxtamembrane region of FGFRs (Xu et al., 1998; Gotoh et al., 2004; Hoch and Soriano, 2006). After receptor activation induced by ligand binding, FRS proteins can become tyrosine phosphorylated and recruit Grb2, Gab1, and SHP2, leading to the activation of the MAPK and PI3K pathways (Hadari et al., 2001; Gotoh et al., 2005; Gotoh, 2008; Goetz and Mohammadi, 2013; Brewer et al., 2015).
Previously, we used telencephalon development as a model to study FGF signaling (Gutin et al., 2006; Storm et al., 2006; Tole et al., 2006; Hébert and Fishell, 2008; Paek et al., 2009; Paek et al., 2011). Fgfr1, Fgfr2, and Fgfr3 are expressed in precursor cells throughout telencenpalon development (Hébert et al., 2003; Tole et al., 2006) and deletion in mice of all three genes at once in early telencephalic precursors resulted in ablation of the telencephalon due to precursor cell death (Paek et al., 2009; Paek et al., 2011), whereas simultaneous deletion of two receptor genes revealed specific requirements of FGFRs in patterning the ventral telencephalon at later time points during development (Gutin et al., 2006). Although all FGFRs are likely capable of signaling through FRS proteins (Gotoh et al., 2005; Eswarakumar et al., 2006; Gotoh, 2008; Goetz and Mohammadi, 2013), in this study, we address using genetic approaches in mice to determine whether the dependence of FGFR function on FRS proteins varies in different processes of telencephalon development.
Materials and Methods
Mice.
The experiments described in this study were approved by the Institutional Animal Care and Use Committee of the Albert Einstein College of Medicine. Foxg1Cre (Hébert and McConnell, 2000), Frs2Flox (Zhang et al., 2008), Fgfr1Flox, Fgfr2Flox, Fgfr3Flox (Kang et al., 2014), Fgfr1ΔFRS (Hoch and Soriano, 2006), and Fgfr1Y776F (Partanen et al., 1998) alleles were maintained and genotyped as described previously. Frs3−/− mice were generated by replacing exon 3 with an IRES-LacZ/PGK-neo cassette in R1 embryonic stem cells (Fig. 1A). Homologous recombination was confirmed by Southern blot using a 3′ external probe (Fig. 1B,C). After blastocyst injections, chimeric males were mated to Swiss Webster mice and the offspring were identified by Southern blot analysis and PCR genotyped with: P1, 5′ATTAAGGGCCAGCTCATT; P2, 5′TTCAAGTGTTCGAGGGCA; and P3, 5′CAGGACTGATCCATCCTCAG, with P1 and P3 amplifying an ∼350 bp mutant product and P2 and P3 an ∼500 bp wild-type product (Fig. 1D and data not shown).
IHC analyses.
Whole embryonic day 9.0 (E9.0) to E13.5 embryos were fixed with 4% paraformaldehyde in PBS, incubated in 10% (4 h) and 20% (overnight) sucrose in PBS (4°C), and embedded in optimal cutting temperature medium. Cryosections (14 μm) were immunostained with the following antibodies: FRS3, goat polyclonal, 1:100 (Santa Cruz Biotechnology); DLX2, mouse monoclonal, 1:75 (Santa Cruz Biotechnology); NKX2.1, rabbit monoclonal, 1:300 (Abcam); PAX6, rabbit polyclonal, 1:250 (Invitrogen); LHX8, rabbit monoclonal, 1:300 (Abcam); LHX6, rabbit polyclonal, 1:200 (Abcam); p-HH3, rabbit monoclonal, 1:200 (Cell Signaling Technology); GSH2, rabbit polyclonal, 1:800 (Millipore); p-ERK-1/2, rabbit monoclonal, 1:150 (Cell Signaling Technology); and p-AKT, rabbit monoclonal, 1:100 (Cell Signaling Technology). Sections were mounted with Prolong Diamond Antifade Mounting Medium with DAPI (Invitrogen) and imaged using a Zeiss fluorescent microscope with Axiovision software.
RNA in situ hybridization.
35S in situ hybridizations were performed as described previously on fresh frozen sections (14 μm) with a cresyl violet counterstain (Frantz et al., 1994).
TUNEL assay.
TUNEL reactions were performed on 16 μm fresh-frozen sections following the manufacturer's protocol (Roche In situ Cell Death Kit).
Statistical analyses.
Quantitation was with ImageJ unless otherwise mentioned and data are presented as the average ± SEM. Significance was determined using Student's t test.
Results
FRS adapters are nonessential for FGF-dependent survival of early telencephalic precursor cells
Loss of three Fgfr genes from telencephalic precursor cells leads to a complete loss of the telencephalon due to precursor cell death at E8.75 (Paek et al., 2009). We therefore investigated whether FRS adapters were required first for telencephalon development and second for mediating FGFR1 signaling. The Frs gene family consists of Frs2 and Frs3 (Gotoh, 2008). Frs2 germline knock-out mice die by E7.5 due to extraembryonic deficits (Hadari et al., 2001; Gotoh et al., 2005). To assess the role of Frs2 in the telencephalon, a conditional Frs2Flox allele (Lin et al., 2007) was recombined using a Foxg1Cre allele, which recombines floxed alleles in telencephalic precursor cells by E9.0 (Hébert and McConnell, 2000; Paek et al., 2009). Because no knock-out or conditional allele of Frs3 had been reported, we generated an Frs3 knock-out allele by substituting exon 3, which encodes the essential phosphotyrosine-binding domain, with an IRES-LacZ/PGK-neo cassette (Fig. 1A–D). Adult Frs3−/− mice appeared normal and were fertile.
Frs2 mRNA was broadly detected in tissues including the telencephalon from E9.0 to E13.5 (Fig. 2A and data not shown; Gotoh et al., 2004), whereas FRS3 protein was restricted to ventral areas (Fig. 2B), consistent with its mRNA pattern (Gotoh et al., 2004). Despite mixed reports of compensatory and distinct functions for Frs2 and Frs3 (Gotoh et al., 2004; Huang et al., 2006; Minegishi et al., 2009; Hryciw et al., 2010), the overlap in their ventral telencephalic expression suggested a potential compensatory role for FRS2 and FRS3 in this region. Therefore, we examined several Frs2 and Frs3 compound mutant genotypes (Fig. 2C). Mutant embryos of all genotypes were obtained at the ages indicated in the expected Mendelian ratios with no signs of necrosis. Controls were littermates that did not carry Cre and/or were heterozygous for Frs2, Frs3, or both and, in none of these cases could a phenotype be detected. Whole E13.5 Frs3−/− and Frs2-cKO embryos appeared grossly normal (Fig. 2D). Although Frs3−/− mice with either one or two functional copies of Frs2 had a normal telencephalon at E13.5, Frs2-cKO mice had slightly smaller telencephalons along the A/P axis (by ∼13%) and slightly smaller eyes and frontonasal prominences (Fig. 2D). Consistent with compensatory roles for Frs2 and Frs3, Frs2,3-dKO mice, compared with either Frs2-cKO or Frs3−/− mutants, showed a more severe reduction in telencephalic size along the A/P (∼24%) and D/V axes (∼16%), along with smaller eyes and frontonasal prominences (Fig. 2D). These results suggest that the in vivo roles of FRS proteins are compensatory in nature.
Interestingly, in contrast to Fgfr triple-null mice, Frs2,3-dKO mice formed a telencephalon (Fig. 2E), suggesting that FRS proteins are not individually or combinatorially required for the early FGF-dependent survival of telencephalic precursor cells and that other FGFR adapters play roles in this process. Therefore, FRS proteins are not required for all FGF-dependent processes within the context of telencephalon development.
FRS2 and FRS3 are together required for phosphorylation of AKT and differentiation of a subset of neurons
General dorsoventral telencephalic patterning in Frs2 and Frs3 mutants was assessed with PAX6 (dorsal marker) and DLX2 (ventral marker). Although all mutants had normal dorsoventral boundaries marked by PAX6 and DLX2 immunoreactivity and normal sized PAX6+ domains, the ventral DLX2+ VZ was reduced in both Frs2-cKO and Frs2,3-dKO mice at E13.5 (Fig. 3A,B). Furthermore, Frs2-cKO and Frs2,3-dKO mice exhibited a reduction of the NKX2.1-labeled medial ganglionic eminence (MGE) and preoptic area and LHX8-labeled differentiating domain compared with control or Frs3−/− mice, even when these also lacked one copy of Frs2 (Fig. 3A,B). A greater reduction of the LHX8+ area relative to the NKX2.1+ area in both mutants suggests that FRS2 and FRS3 are together required for MGE differentiation, with FRS2 playing a dominant role.
Consistent with a role for FRS proteins in MGE differentiation, precursor cell division assessed by p-HH3 immunoreactivity within the SVZ, but not within the VZ, of the MGE was reduced in E13.5 Frs2,3-dKO mice (Fig. 4A,E). Precursor cell division at E10.5 before the onset of neurogenesis was unaffected (Fig. 5C,D), consistent with an initially normal NKX2.1+ field (Fig. 5A,B). Cell survival in the MGE measured by TUNEL staining was not detectably affected in Frs2,3-dKO mutants at either stage (Figs. 4B,F, 5C,D). Collectively, these results are consistent with FRS2 and FRS3 playing an important role in differentiation of NKX2.1+ precursor cells.
Because precursor cells in the ventral telencephalon give rise to several neuron populations, we investigated the contribution of FRS proteins in this process by examining subpopulations of neuron precursors in E13.5 control and mutant mice. Our results revealed that the domains of Shh and LHX6 expression, which mark predominantly the differentiating field of cortical GABAergic interneurons (Liodis et al., 2007; Xu et al., 2010), were comparable between control and mutant mice (Fig. 6B,D,F,H). This is in contrast to what we observed with reduced cells expressing LHX8 (Fig. 3), a marker of basal forebrain cholinergic neuron precursors induced by FGF8 (Zhao et al., 2003; Hoch et al., 2015). The lateral ganglionic eminence (LGE), although morphologically flatter, had approximately normal sized differentiating and progenitor fields marked by Ebf1 and GSH2 expression, respectively (Fig. 6A,C,E,G) (Corbin et al., 2000). However, PAX6+ cells in the dorsal LGE, which mark a subset of dopaminergic olfactory interneuron progenitors (Kohwi et al., 2005), were significantly reduced in Frs2,3-dKO mice when compared with control mice (Fig. 3C,D). Therefore, FRS adapters selectively regulate the differentiation of a subset of ventral telencephalic neurons. Their roles in the developing telencephalon are consistent with previously described roles for FGFRs (Hébert et al., 2003; Gutin et al., 2006; Hébert and Fishell, 2008).
To identify downstream targets of FRS proteins in the developing MGE, we stained control and double mutant sections with anti-p-ERK-1/2 and p-AKT antibodies (Fig. 4C,D). Although p-ERK-1/2 levels were comparable between control and mutant mice, p-AKT levels were reduced in SVZ, but not VZ, of mutant MGEs compared with control SVZ and VZ (Fig. 4G,H). These results show that FRS adapters are critical for AKT/PI3K activation, but are dispensable for the activation of ERK-1/2/MAPK pathway within MGE progenitors. Collectively, our results suggest that FRS adapters, via activation of the AKT/PI3K pathway, play a role in differentiation of a subset of MGE-derived neurons.
FRS adapters are required to transmit most or all FGFR1 activity
The Frs2-cKO and Frs2,3-dKO phenotypes are consistent with a role for FRS adapters in transmitting FGFR activity in the MGE. Although Foxg1Cre-driven Fgfr1 and Fgfr2 single mutants appeared largely normal (Fig. 7A,B) and Fgfr3 was previously suggested not to contribute significantly to MGE development (Gutin et al., 2006), Foxg1Cre-driven Fgfr1;Fgfr2 double mutants exhibit a smaller telencephalon and a severe ventral telencephalon deficit (Fig. 7A,B; Gutin et al., 2006). Notably, the Fgfr1;Fgfr2 double-mutant phenotype has a more severe loss of ventral tissue than the Frs2,3-dKO phenotype, suggesting that FRS transduces some but not all FGFR signaling.
Because a single wild-type allele of Fgfr1 in Foxg1Cre-driven mutant Fgfr2 and Fgfr2;Fgfr3-deficient backgrounds (Fgfr1+/−;Fgfr2−/− and Fgfr1+/−;Fgfr2−/−;Fgfr3−/−, respectively) is sufficient for grossly normal telencephalon and MGE development (Gutin et al., 2006; Paek et al., 2009; Fig. 8A), we tested directly the contribution of FRS proteins in transducing signal from FGFR1 by replacing the single wild-type Fgfr1 allele in these mice with an Fgfr1 allele that is deleted for the FRS-binding-site codons (Fgfr1ΔFRS; Hoch and Soriano, 2006). Fgfr1ΔFRS/ΔFRS mutants exhibit a specific deficit in FRS signaling with normal activation of FRS-independent signaling pathways downstream of FGFR1 (Xu et al., 1998; Hoch and Soriano, 2006). We found that the Fgfr1ΔFRS/−;Fgfr2−/− phenotype (with a loss of both LGE and MGE) and the Fgfr1ΔFRS/−;Fgfr2−/−;Fgfr3−/− phenotype (with early E8.75 telencephalic precursor cell death) were as severe as the Fgfr1−/−;Fgfr2−/− and Fgfr1−/−;Fgfr2−/−;Fgfr3−/− phenotypes, respectively (Figs. 7A–C, 8A,B), indicating that FRS proteins are required to mediate most or all of FGFR1 activity in the early telencephalon. Consistent with this finding, we did not detect telencephalic deficits in mutants in which the wild-type Fgfr1 allele was replaced with an allele that lacks the binding site for another classically defined FGFR signaling mediator, PLCγ (Fgfr1Y766F, Partanen et al., 1998), in an Fgfr2;Fgfr3 mutant background, Fgfr1Y766F/−;Fgfr2−/−;Fgfr3−/− (Fig. 8A and data not shown).
Discussion
With this work, we show a context-dependent requirement for the FRS2 and FRS3 adapters in the developing telencephalon. First, FRS2 and FRS3 are together required for neuronal differentiation of NKX2.1+ precursors, but not for early precursor cell survival (Figs. 2, 3, 4). Second, FRS adapters are required for differentiation of a subset of neurons in the NKX2.1+ domain (Figs. 3, 6). Third, within the differentiating cells, FRS is required for activation of the AKT pathway (Fig. 4). In contrast, FGFR1 signaling via FRS adapters is not required for AKT activation in the context of gross whole embryo development (Hoch and Soriano, 2006; Brewer et al., 2015). Finally, we find that in the telencephalon FRS adapters are required for mediating most or all of FGFR1's activity (Figs. 7A–C, 8A,B). The finding that FRS adapters are required to mediate FGFR1 signaling not only in MGE differentiation (Fig. 7), where FRS proteins are themselves required (Fig. 3), but also in early precursor cell survival (Fig. 8A,B), where, in contrast, FRS proteins are dispensable (Fig. 2E), suggests that mediators other than FRS proteins participate in transmitting FGFR2 and FGFR3 signaling in early precursors (Fig. 8C). Moreover, the less severe phenotype of Frs2;Frs3 double mutants compared with Fgfr1;Fgfr2 mutants (e.g., c.f. DLX2, NKX2.1, and LHX8 staining of Frs2;Frs3 mutants in Fig. 3 with staining of Fgfr1;Fgfr2 mutants in Fig. 7B and in Gutin et al., 2006) also suggests that, at least for FGFR2, adapters other than FRS proteins mediate some signaling in ventral patterning (Fig. 8C).
The finding that FRS adapters are required for mediating most or all of FGFR1's activity in the telencephalon was unexpected. First, FGFR1 has the ability to activate multiple mediators such as Crk/Crkl, PLCγ, and Grb14 in other contexts (Williams et al., 1994b; Partanen et al., 1998; Turner and Grose, 2010; Goetz and Mohammadi, 2013). Second, a previous study showed that, unlike in the telencephalon, in whole embryos, loss of the FRS2-binding site in FGFR1 does not recapitulate the severity of the complete loss of Fgfr1 function and that Fgfr1 mutants lacking the combined binding sites for FRS2, Crkl, PLCγ, and Grb14 exhibit a phenotype more severe than loss of the FRS2-binding site alone, indicating that FGFR1 uses transducers other than FRS in other tissues (Brewer et al., 2015).
Our findings are consistent with FRS2 playing a dominant role over FRS3 as a mediator of FGF signaling (Figs. 2D, 3). The phenotype of Frs2-cKO mice observed here is reminiscent of the eye and other CNS deficits of Frs22F/2F mutant mice in which the residues encoding the binding sites for Shp2 are mutated (Yamamoto et al., 2005), consistent with FRS2 playing a dominant role over FRS3 in these tissues as well, given the lack of phenotype in Frs3 mutants (Fig. 2D). Likewise, whereas germline deletion of Frs2 is embryonically lethal (Hadari et al., 2001; Gotoh et al., 2005), Frs3 germline knock-out mice develop normally into adulthood (Fig. 2 and data not shown). Despite similarities in the structures of FRS2 and FRS3 and their mechanism of activation, in vitro experiments had implicated FRS2 as the primary mediator of FGF signaling (Gotoh et al., 2005; Gotoh, 2008). Our results confirm a dominant role for FRS2, but also indicate a compensatory role for FRS2 and FRS3 in telencephalon development (Figs. 2, 3).
FRS adapters, in addition to transmitting FGFR signaling, can also transmit signaling from the Ntrk1-, Ntrk2-, and Ntrk3-encoded neurotrophin receptors (Ong et al., 2000), making it theoretically possible that part of the Frs2 and Frs3 mutant phenotypes observed here are the result of reduced neurotrophin signaling. However, the neurotrophin receptors are either not significantly or not detectably expressed until after neurogenesis has begun between E11.5 and E13.5 and their loss-of-function phenotypes, even when combined, do not include early patterning defects such as those observed here (Minichiello et al., 1995; Minichiello and Klein, 1996). Therefore, it is unlikely that the Frs2 and Frs3 phenotypes described here are the result of reduced signaling though neurotrophin receptors. Moreover, the phenotypes described using the Fgfr1ΔFRS allele confirm a role for FRS in mediating FGFR1 signaling in the early telencephalon.
In sum, our observations provide an example of how one intracellular transducer can mediate receptor function within given contexts and, in more general terms, provide a partial explanation for how FGF signaling can promote diverse cellular responses in development and disease.
Footnotes
This work was supported by the National Institutes of Health (Grants MH083804 and MH070596 to J.M.H. and Grant K12GM102779 to C.A.B.) and the Hirschl/Weill-Caulier Foundation (J.M.H.). We thank Drs. Philippe Soriano (Mount Sinai Hospital, NY), Juha Partanen (Helsinki), and Raj Ladher (RIKEN, Kobe) for the Fgfr1ΔFRS/+ and Fgfr1Y776F/+ mice.
The authors declare no competing financial interests.
- Correspondence should be addressed to either Jean M. Hébert or Sayan Nandi, Department of Neuroscience, Albert Einstein College of Medicine, 1399 Morris Park Avenue, Bronx, NY 10461. jean.hebert{at}einstein.yu.edu or sayan.nandi{at}einstein.yu.edu