Crkl Deficiency Disrupts Fgf8 Signaling in a Mouse Model of 22q11 Deletion Syndromes

Fgf8 and Crkl Genetically Interact during Mouse Morphogenesis

We used an Fgf8;Crkl allelic series in genetically engineered mice to determine if Crkl and Fgf8 cooperatively affect embryonic survival and morphogenesis of Fgf8-sensitive structures including the cardiovascular system, pharyngeal glands, and the craniofacial and appendicular skeleton.

The fourth pharyngeal arch artery (PAA4) ultimately forms the arch of the aorta; disrupted formation or remodeling of PAA4 results in interrupted aortic arch or right aortic arch as well as subclavian artery anomalies (vascular defects seen in human del22q11 syndrome). PAA4 formation occurs normally in Fgf8 null heterozygotes (Fgf8^+/-), but it is disrupted at lower levels of Fgf8 that occur in Fgf8 hypomorphic and conditional mutants, revealing the extreme sensitivity of PAA4 vasculogenesis to Fgf8 levels in the developing pharynx (Abu-Issa et al., 2002; Frank et al., 2002; Macatee et al., 2003). Although PAA4 forms normally in Crkl^-/- embryos, combined Crkl/Fgf8 deficiency profoundly disrupts PAA4 formation: 50% of Fgf8^+/-;Crkl^+/- and 90% of Fgf8^+/-;Crkl^-/- mutants have aplastic or hypoplastic fourth PAAs at embryonic day (E) 10.5 (p = 0.05 and p < 0.0001, respectively, relative to Crkl^-/-; Figures 1A-1D). Some Fgf8^+/-;Crkl^+/- and Fgf8^+/-;Crkl^-/- mutants die at mid-gestation and display aortic sac dilatation and no caudal PAAs (12% and 14%, respectively,Figure 1D); this phenotype was not observed in Crkl^-/- embryos. Recovery from PAA4 formation defects observed at E10.5 can occur during remodeling; however, decreased Fgf8;Crkl dosage frequently prevents recovery, resulting in aortic interruptions and subclavian artery defects in E15.5 Fgf8^+/-;Crkl^-/- mutants (Figure 1F;Table 1).

Ablation of a single copy of Fgf8 in Crkl null homozygotes increases the incidence and severity of intracardiac and outflow tract defects (Table 1; Figure 1). Atrioventricular valve dysplasias, which reflect abnormal formation and/or remodeling of the outflow and endocardial cushions, are more common and markedly more severe in Fgf8^+/-;Crkl^-/- mutants compared to Crkl^-/- (Figures 1H-1L). Survival of Fgf8^+/-;Crkl^+/- and Fgf8^+/-;Crkl^-/- mutants to E15.5 is also compromised relative to single heterozygotes and Crkl^-/- mutants (data not shown). Thus, Crkl deficiency reduces the effective Fgf8 signals required to support early cardiovascular development and embryonic survival.

The pharyngeal pouch endoderm gives rise to the primordia of the thymus and parathyroids; defects in these tissues contribute to the morbidity of DiGeorge syndrome patients in the form of immunodeficiency and hypocalcemia, respectively. Fgf8 hypomorphs have frequent aplasia of these glands and associated T cell deficiency (Frank et al., 2002). The size and location of the thymus and parathyroids are normal in Fgf8 or Crkl heterozygotes; however, Fgf8^+/-;Crkl^+/- mutants have the same incidence of glandular ectopy as Crkl^-/- embryos. Moreover, glandular phenotypes are exacerbated in Fgf8^+/-;Crkl^-/- mutants (Figure 1F; Table 1), indicating a dosage-sensitive interaction between Fgf8 and Crkl during pharyngeal pouch development.

Humans with del22q11 have variable appendicular and craniofacial skeletal malformations and short stature (Bird and Scambler, 2000; Emanuel et al., 2001; Motzkin et al., 1993). Patterning, outgrowth, and the ultimate length of limb bony elements are all sensitive to Fgf8 signaling from epithelial cells to underlying mesenchyme (Boulet et al., 2004; Lewandoski et al., 2000; Moon and Capecchi, 2000). Femurs of Fgf8^+/-;Crkl^+/- and Fgf8^+/-;Crkl^-/- mutants are an average of 35% shorter than Fgf8^+/- and Crkl^-/- littermates (Figures 1O-1Q). Cleft palate and mandibular hypoplasia are frequently observed in patients with del22q11 syndrome, and in Fgf8 hypomorphic or conditional mutant mice (Abu-Issa et al., 2002; Frank et al., 2002; Macatee et al., 2003; Trumpp et al., 1999). A synergistic interaction between Fgf8 and Crkl during formation of the bony palate is manifest by a dramatic increase in the incidence of cleft palate from 5% in Fgf8^+/-;Crkl^+/- or Crkl^-/- mutants to 55% in Fgf8^+/-;Crkl^-/- mutants (Figures 1O-1S). Mandibular hypoplasia is apparent in 50% of Fgf8^+/-;Crkl^+/-, 60% of Crkl^-/-, and 85% of Fgf8^+/-; Crkl^-/- mutants (Figure 1U).

Loss of Crkl Decreases Survival of Neural Crest and the Contribution of Neural Crest to the Outflow Tract and Pharyngeal Arches

Neural crest cells (NC) migrate from the dorsal neural tube and make crucial contributions of mesenchyme to the pharyngeal arches, OFT septum, and pharyngeal glands (Jiang et al., 2000). The OFT myocardium is composed of cells derived from the Fgf8-dependent anterior heart field in splanchnic mesoderm (SM) (Kelly and Buckingham, 2002; E. Park and A.M.M., submitted). Abnormal apoptosis of NC and SM in Fgf8 hypomorphs and domain-specific conditional mutants is associated with craniofacial, vascular, and OFT defects (Abu-Issa et al., 2002; Frank et al., 2002; Macatee et al., 2003; E. Park and A.M.M., submitted). We found a marked excess of apoptotic cells in the pharynx of Crkl^-/-, Fgf8^+/-;Crkl^+/-, and Fgf8^+/-;Crkl^-/- mutants, including cells streaming into the pharynx from the neural tube (Figures 2A-2E). Since Crkl is expressed in NC adjacent to Fgf8-expressing domains, we immunohistochemically labeled NC with the marker AP-2a and assayed for apoptosis. This revealed multiple regions of apoptotic NC en route to the pharyngeal arches in Crkl^-/-, Fgf8^+/-;Crkl^+/-, and Fgf8^+/-;Crkl^-/- mutants (Figures 2F-2J). Sections revealed that excess apoptosis occurred in the pharyngeal epithelia and SM, in addition to the NC (Figures 2K-2M′). Levels of apoptosis mirror decreases in Fgf8 and/or Crkl gene dosage.

To evaluate NC populations in Crkl^-/- embryos, we labeled this cell lineage by using the Rosa26-Cre reporter and the Wnt1Cre transgene (Jiang et al., 2000). All Crkl^+/+ and Crkl^+/- embryos had dense populations of lacZpositive cells throughout the PAs and OFT (Figure 2N, arrows; data not shown), whereas contribution of NC to these structures was obviously decreased in Crkl^-/- mutants (Figure 2O, arrows).

These phenotypes demonstrate that Fgf8 and Crkl genetically interact to regulate the development of structures affected in patients with del22q11 syndrome and suggest that Crkl is required for normal responses to Fgf8 in multiple morphogenetic pathways during embryogenesis. To investigate whether this interaction reflects the function of Crkl downstream of FgfRs (versus functions in parallel molecular pathways that influence development of these structures), we examined the effects of Crkl deficiency on expression of Fgf8 target genes, activation of MAP kinases, and intracellular biochemical signal transduction in vivo.

Crkl Regulates Expression of Fgf8 Target Genes

Fgf8 signaling stimulates transcription of the ETS domain DNA binding proteins Erm and Pea3 (Firnberg and Neubuser, 2002; Raible and Brand, 2001; Roehl and Nusslein-Volhard, 2001). Fgf/FgfR interactions trigger phosphorylation of Erm protein via activated MAPK/ERKs, causing it to bind its transcriptional targets (Janknecht et al., 1996; O’Hagan et al., 1996; Tsang and Dawid, 2004). Whole-mount in situ hybridization revealed decreased Erm and Pea3 expression in the pharyngeal arches of Crkl^-/- mutants, consistent with a role for Crkl in mediating Fgf signals (Figures 3A-3J). Decrements in Erm and Pea3 expression in the pharyngeal arches mirrored those in Fgf8 and Crkl gene dosage, and, in Fgf8^+/-;Crkl^-/- mutants, Erm expression was less than that detected in Fgf8 hypomorphs (data not shown). Altered expression of the Fgf8 target gene Barx1 was noted in the hypoplastic caudal arches of Fgf8^+/-;Crkl^-/- mutants (Figures 3K-3O). Decreased Erm and Pea3 expression correlated with levels of phosphorylated Erk1/2 in the pharyngeal arches and limb buds of Crkl^-/- and Fgf8;Crkl compound mutant embryos. In Crkl heterozygote controls (Figure 3Q), high levels of phosphorylated Erk are detected in the pharyngeal epithelium and mesenchyme. In compound mutants and Fgf8 hypomorphs (Figures 3R-3T, and data not shown), this pattern of Erk phosphorylation was lost. Thus, expression of known Fgf8 target genes and activation of MAP kinases in Fgf8-responsive tissues depend on Crkl function in vivo.

Crkl Binds Activated Fgf Receptors 1 and 2 and Is Required for Downstream Cellular Responses to Fgf8

We investigated whether Crkl interacts with FgfRs to mediate Fgf8 signaling. In primary cultures of mouse embryonic fibroblasts (MEFs), FgfR1 and FgfR2 are rapidly tyrosine phosphorylated in response to Fgf8b stimulation (Figures 4A and 4B). These activated receptors bind directly to the SH2 domain of Crkl in vitro (Figure 4C). Furthermore, Crkl forms stable complexes with endogenous FgfR1 and FgfR2 proteins in normal Fgf8b-stimulated MEFs in vivo, indicating that formation of these complexes is a physiologic event (Figure 4D). While association of Crkl with these receptors was readily detectable, only a minute amount of Crk was coimmunoprecipitated with the FgfRs compared to the level of Crk in the cell. Thus, Crkl interacts with activated FgfRs far more efficiently than Crk. It is noteworthy that, in contrast to results obtained with overexpressed or chimeric FgfRs in transformed cell lines (Chellaiah et al., 1999; Larsson et al., 1999; Ornitz et al., 1996), in this primary embryonic culture model, Fgf8 activates FgfR1 in addition to FgfR2.

We further interrogated the role of Crkl in mediating functional responses to Fgf8 in MEFs. Wild-type MEFs migrate robustly toward a source of Fgf8b in a modified Boyden chamber assay, whereas MEFs isolated from Crkl^-/- embryos do not (p = 0.0001, Figure 5A). Chemotaxis to Fgf8 can be rescued by expression of Crkl from a plasmid transfected into the mutant cells prior to assay (Figure 5A, Crkl^-/-+CRKL). The fact that Crkl^-/- MEFs migrate normally toward serum, epidermal growth factor (EGF), or platelet-derived growth factor (PDGF) (Figures 5A and 5B) demonstrates that the requisite machinery for cell motility is intact, and that migratory defects observed in Crkl^-/- cells reside in Crkl-dependent pathways specific to Fgf8 signal transduction.

The cellular response to Fgfs occurs after Fgf ligand binds to the high-affinity cell surface FgfRs, triggering receptor dimerization, autophosphorylation, and subsequent activation of the Ras/Raf/Mek pathway upstream of Erk1/2 MAP kinases (Hadari et al., 2001; Reuss and von Bohlen und Halbach, 2003). We found that Fgf8b induces robust and sustained activation of Erk1/2 MAP kinases in wild-type MEFs (Figure 5C). In contrast, in Crkl^-/- MEFs, Fgf8b induces only minor activation of Erk1/2. Furthermore, the effective minimal concentration of Fgf8b required to induce detectable activation of Erk1/2 is approximately 10-fold greater in Crkl^-/- cells (20 ng/ml) than in wild-type cells (2.0 ng/ml,Figure 5D).

Since FgfRs are RTKs, we examined the profile of tyrosine-phosphorylated proteins in MEFs after Fgf8b stimulation, with particular attention paid to the docking protein Frs2. In response to Fgf stimulation, Frs2 links FgfR activation to the Ras/MAPK signaling pathway by targeting signaling molecules to the plasma membrane. For example, activated Frs2 recruits the adaptor protein Grb-2/Sos, which sequentially recruits Ras to the FgfR complex (Hadari et al., 1998, 2001; Kouhara et al., 1997; Larsson et al., 1999). While the overall content of protein tyrosine phosphorylation is comparable in Crkl^-/- and wild-type MEFs, some species are relatively hypophosphorylated in Crkl^-/- cells after Fgf8b stimulation (Figure 5E). One species has a molecular weight of approximately 90 kDa and comigrates with tyrosinephosphorylated species of Frs2 when the same blot is probed with anti-Frs2 antibody (Figure 5F). Immunoprecipitation confirms that Fgf8b-induced tyrosine phosphorylation of Frs2 depends partially on the presence of Crkl (Figure 5G). Furthermore, the level of Frs2 tyrosine phosphorylation can be titrated in Crkl^-/- cells by varying the level of Crkl produced from a transgene (Figure 5H). These findings strongly support the role of Crkl in transducing intracellular biochemical responses to Fgf8 during morphogenesis.

Mice

The Crkl null (Crkl^-), Fgf8 hypomorphic (Fgf8^H), and Fgf8 null (Fgf8^-) alleles were described previously (Frank et al., 2002; Guris et al., 2001; Moon and Capecchi, 2000). These alleles were originally targeted in 129Sv-derived ES cells and then backcrossed 3-4 generations into C57BL/6J prior to interstrain crosses.

In Situ Detection of Apoptosis

Primary and secondary antibodies and the methods employed to detect AP2α in neural crest, TUNEL, and Hoechst staining of nuclei in whole-mount and on cryosectioned specimens were previously described (Macatee et al., 2003). In whole-mount specimens, apoptotic cells were detected with anti-cleaved caspase 3 (Cell Signaling Technology, 9664S).

Marking the Neural Crest

Neural crest cells were genetically labeled by using the R26R and Wnt1Cre strains (kind gifts from P. Soriano and A.P. McMahon, respectively). These strains were maintained by continual backcross with C57BL/6J for more than 11 generations. Embryos were stained with Salmon-gal (Biosynth) by standard protocols.

Whole-Mount In Situ Hybridization

Whole-mount in situ RNA hybridization was performed by using a standard protocol (Guris et al., 2001). Experiments were repeated a minimum of three times.

Whole-Mount Detection of MAP Kinase Activation

Phosphorylated Erk1 and Erk2 were detected as described (Corson et al., 2003), but with minor modifications: rabbit polyclonal antiphosphorylated p44/p42 was used (Cell Signaling Technology, 9101S). After dissection in cold PBS, the embryos were fixed overnight at 4°C, washed in 1× PBS and 0.5% NP40, then dehydrated/rehydrated through a methanol series. Embryos were blocked with 1× PBS/5% goat serum/0.1% Triton X-100 and were incubated overnight at 4°C with primary antibody diluted 1:350 in PBST. Washes were followed by repeated blocking and incubation overnight with secondary antibody (Alexa 594 goat anti-rabbit; diluted 1:200 in PBST). Fluorescence of 2 μm optical sections under a10× objective were recorded with an Olympus 1X70 Fluoview confocal microscope.

Cellular and Biochemical Analysis

Preparation of mouse embryonic fibroblasts and assay of cell motility with a modified Boyden chamber were described previously (Li et al., 2003). Western blots and immunoprecipitation were performed by using standard procedures. GST-CRKL SH2 generated in E. coli was isolated by glutathione-Sepharose beads (GE Biotechnology) prior to pull-down assays.

Rabbit anti-CRKL (C-20), anti-CRK-II (C-18), anti-FGFR1 (C-15), anti-FGFR2 (C-17), and anti-FRS2 (H-91) antibodies were purchased from Santa Cruz Biotechnology. In some experiments, mouse monoclonal anti-CRKL (Upstate Biotechnology), anti-CRK (BD Pharmingen), and anti-FGFR1 (Zymed) were used for compatible combinations of antibody host species for immunoprecipitation, followed by immunoblot analysis, and for confirmation. Mouse monoclonal anti-phosphotyrosine (PY99) antibody was obtained from Santa Cruz Biotechnology. Anti-phospho Erk p44/42 (pTEpY) monoclonal (E10) and anti-p44/42 polyclonal antibodies were obtained from Cell Signaling Technology.