FGF signaling regulates otic placode induction and refinement by controlling both ectodermal target genes and hindbrain Wnt8a

Abstract

The inner ear epithelium, with its complex array of sensory, non-sensory, and neuronal cell types necessary for hearing and balance, is derived from a thickened patch of head ectoderm called the otic placode. Mouse embryos lacking both Fgf3 and Fgf10 fail to initiate inner ear development because appropriate patterns of gene expression fail to be specified within the pre-otic field. To understand the transcriptional "blueprint" initiating inner ear development, we used microarray analysis to identify prospective placode genes that were differentially expressed in control and Fgf3(-)(/)(-);Fgf10(-)(/)(-) embryos. Several genes in the down-regulated class, including Hmx3, Hmx2, Foxg1, Sox9, Has2, and Slc26a9 were validated by in situ hybridization. We also assayed candidate target genes suggested by other studies of otic induction. Two placode markers, Fgf4 and Foxi3, were down-regulated in Fgf3(-)(/)(-);Fgf10(-)(/)(-) embryos, whereas Foxi2, a cranial epidermis marker, was expanded in double mutants, similar to its behavior when WNT responses are blocked in the otic placode. Assays of hindbrain Wnt genes revealed that only Wnt8a was reduced or absent in FGF-deficient embryos, and that even some Fgf3(-)(/)(-);Fgf10(-)(/+) and Fgf3(-)(/)(-) embryos failed to express Wnt8a, suggesting a key role for Fgf3, and a secondary role for Fgf10, in Wnt8a expression. Chick explant assays showed that FGF3 or FGF4, but not FGF10, were sufficient to induce Wnt8a. Collectively, our results suggest that Wnt8a provides the link between FGF-induced formation of the pre-otic field and restriction of the otic placode to ectoderm adjacent to the hindbrain.

Fig. 1. Use of Fgf3 and Fgf10 conditional alleles to produce double null mutants that lack otic vesicles at E10.5 and FGF/MAPK signaling markers at E8.5

(A) Each targeted conditional allele has loxP sites (red arrows) flanking exon 2. An FRT site (black arrow) remains from FLP-mediated removal of a Neo expression cassette. A tetracycline response element (T) is located upstream of exon 2. CRE-mediated deletion of the 104 bp exon 2 from the conditional (c) alleles generates the null alleles (designated Δ2). Thick boxes indicate the sequences included within the targeting vectors; exons are indicated in green; 5’ and 3’ UTRs in red; introns in black; recombinase recognition sequences in pink; and the T in blue. The cross generating double mutants at a frequency of 25% is indicated below the allele diagrams. (B) E10.5 Fgf3^Δ2/+;Fgf10^Δ2/+ embryo shows a normal phenotype. (C) E10.5 Fgf3^Δ2/Δ2;Fgf10^Δ2/Δ2 embryo lacks limbs and otic vesicles, and has a shortened tail. Dashed lines demark limbs (fl, forelimb; hl, hindlimb), tail (t), and otic vesicle (ov) where present. Whole-mount E8.5 (7-somite) embryos were probed with FGF signaling indicators, Erm (D) and Spry1 (E). Genotype is indicated below each embryo. (D₁, E₁) Transverse sections taken through the placodal region (planes numbered and indicated with dashed lines in D and E) show otic placode (op) expression (arrows) of each gene in double heterozygotes and loss of gene expression in the corresponding (thin) ectoderm in double mutants (D₂, E₂, arrows).

Fig. 2. Isolation of tissue and validation by in situ hybridization of microarray candidate genes down-regulated in placodal ectoderm of Fgf3/Fgf10-deficient embryos

(A) Schematic depiction of microdissection of the placodal region of an E8.5 5-8-somite embryo. Embryos were first bisected along the midline (vertical dashed line). The dorsal, presumptive otic region, was isolated from the ventral aspect with a second cut (horizontal dashed line), generating two fragments containing neural tube (nt), placodal ectoderm (ec), and mesendoderm (m, en). (B) Protease treatment of each hemisected otic region released the mesendoderm from the neural tube and attached placodal ectoderm, which was then severed from the nt and recovered for RNA extraction. (C-G) 7-8-somite Fgf3^Δ2/+;Fgf10^Δ2/+ (3^-/+;10^-/+) and Fgf3^Δ2/Δ2;Fgf10^Δ2/Δ2 (3^-/-;10^-/-) embryos hybridized with riboprobes for Hmx3 (C), Foxg1 (D), Sox9 (E), Has2 (F), and Slc26a9 (G). Anterior is to the top. Transverse sections taken through the placodal region (planes numbered and indicated with dashed lines in panels C-G) show otic placode (op) expression of each gene in double heterozygotes (C₁-G₁, black arrows) and loss of gene expression in the corresponding (thin) ectoderm in double mutants (C₂-G₂, red arrows). Carets and arrowheads indicate Fgf-independent expression in pharyngeal endoderm, neural tube (nt), and migrating neural crest (mnc), respectively.

Fig. 3. Other placode-expressed genes are differentially affected in Fgf3/Fgf10-deficient ectoderm

Somite-matched E8.5 Fgf3^Δ2/+;Fgf10^Δ2/+ (3^-/+;10^-/+) and Fgf3^Δ2/Δ2;Fgf10^Δ2/Δ2 (3^-/-;10^-/-) embryos were hybridized with probes for Fgf4 (A,B), Foxi3 (C,D), and Foxi2 (E-H). Anterior is to the left in A-D and to the right in E-H. Transverse sections taken through the otic region (planes numbered and indicated with dashed lines in panels A-H) show otic placode (op) expression of each gene in double heterozygotes (A₁,C₁,E₂,G₂, black arrows; red caret indicates lack of expression in pharyngeal endoderm). Fgf4 was absent from dorsal double mutant ectoderm (B₁, red arrow), but up-regulated in pharyngeal endoderm (B₁, caret) and Foxi3 was absent from dorsal ectoderm in double mutants (D₁, red arrow). Foxi2 expression was expanded both anteriorly (F₁) and posteriorly (F₃) in 7-8 somite double mutants, and overall expression in the placodal region was more intense (F₂). In 11-12 somite embryos, Foxi2 expression was restricted from the otic cup (oc) in control embryos (G, G₁,G₂,G₃), whereas Foxi2 was present throughout the dorsal ectoderm (ec) in double mutants (H,H₁,H₂,H₃). The ventrally localized cup-like structure (“oc”), which may represent the precursor to one of the microvesicles occasionally seen in double mutants, showed incomplete clearing of Foxi2 expression (H₂).

Fig. 4. Wnt8a is the only hindbrain Wnt that is dependent on Fgf3 and Fgf10 expression, and the otic field expresses at least two WNT receptor (Fzd) genes

Somite-matched E8.5 control Fgf3^Δ2/+;Fgf10^Δ2/+ (3^-/+;10^-/+; A,C,E) and Fgf3^Δ2/Δ2;Fgf10^Δ2/Δ2 (3^-/-;10^-/-; B,D,F) embryos were hybridized with probes for Wnt6 (A,B), Wnt3a (C,D), and Wnt8a (E,F). Wnt6 and Wnt3a were expressed in control neural plates (A,C), and expression was unchanged in double mutant embryos (B,D). r4 expression of Wnt8a (E) was reduced or absent (F) in all 4 double mutants. Fgf3^+/Δ2;Fgf10^Δ2/Δ2 (3^+/-;10^-/-; G) and Fgf10^Δ2/Δ2 (10^-/-; I) embryos showed normal levels of Wnt8a in 4/5 and 4/4 embryos respectively. In contrast, Fgf3^Δ2/Δ2;Fgf10^+/Δ2 (3^-/-;10^+/-; H) and Fgf3^Δ2/Δ2 (3^-/-; J) showed reduced Wnt8a expression in 3/6 and 2/4 embryos respectively. WNT receptor genes Fzd1 (K) and Fzd8 (L) were expressed in the otic placode (op).

Fig. 5. FGF4 or FGF3, but not FGF10, induces Wnt8a in chick ectodermal explants

HH stage 4-5 ectodermal explants were cultured with control or FGF protein-coated beads, and Wnt8a expression was assessed by in situ hybridization. Control beads (A) were unable to induce Wnt8a, whereas FGF4 induced robust expression of Wnt8a (B, arrows) and FGF3 induced weak Wnt8a expression in approximately 1/3 of the explant cultures (C, arrow). In contrast, FGF10 did not induce Wnt8a (D). See Table 2 for quantification of the results.

Fig. 6. Model for induction and resolution of the otic placode from pre-otic ectoderm

(A) Oblique view of a pre-placodal-stage embryo illustrating the initiation of placode specification by hindbrain-expressed FGF3 (light blue) and mesenchyme-expressed FGF10 (grey) acting on the pre-otic field (pink) to induce gene expression (data from this and previous studies). Both FGFs are also required for induction and/or maintenance of hindbrain Wnt8a (dark blue); FGF3 being the more potent activator. (B) Oblique view of a placodal-stage embryo showing the proposed role of WNT8A interacting with FZD receptors, repressing Foxi2 and limiting otic placode fate to hindbrain-proximal ectoderm, with remaining ectoderm (yellow) assuming a non-otic fate (epibranchial placode or epidermis).