Tfap2a-dependent changes in mouse facial morphology result in clefting that can be ameliorated by a reduction in Fgf8 gene dosage

Abstract

Failure of facial prominence fusion causes cleft lip and palate (CL/P), a common human birth defect. Several potential mechanisms can be envisioned that would result in CL/P, including failure of prominence growth and/or alignment as well as a failure of fusion of the juxtaposed epithelial seams. Here, using geometric morphometrics, we analyzed facial outgrowth and shape change over time in a novel mouse model exhibiting fully penetrant bilateral CL/P. This robust model is based upon mutations in Tfap2a, the gene encoding transcription factor AP-2α, which has been implicated in both syndromic and non-syndromic human CL/P. Our findings indicate that aberrant morphology and subsequent misalignment of the facial prominences underlies the inability of the mutant prominences to fuse. Exencephaly also occured in some of the Tfap2a mutants and we observed additional morphometric differences that indicate an influence of neural tube closure defects on facial shape. Molecular analysis of the CL/P model indicates that Fgf signaling is misregulated in the face, and that reducing Fgf8 gene dosage can attenuate the clefting pathology by generating compensatory changes. Furthermore, mutations in either Tfap2a or Fgf8 increase variance in facial shape, but the combination of these mutations restores variance to normal levels. The alterations in variance provide a potential mechanistic link between clefting and the evolution and diversity of facial morphology. Overall, our findings suggest that CL/P can result from small gene-expression changes that alter the shape of the facial prominences and uncouple their coordinated morphogenesis, which is necessary for normal fusion.

Keywords: AP-2α; BOFS; Branchio-oculofacial syndrome; Cleft lip/palate; Craniofacial; Fgf signaling pathway; Geometric morphometrics; TFAP2A.

Fig. 1.

The Neo/Null allele and the bilateral cleft phenotype. (A) Schematic diagram of the Tfap2a alleles. The Neo/Null model is a combination of the Neo (Brewer et al., 2004) and the Tfap2a null (Zhang et al., 1996) alleles. The positions of primers for RT-PCR and qRT-PCR are shown, along with the intron:exon structure of the gene, and the position of sequences introduced by gene targeting. See Materials and Methods for details of primers. Frt, flippase recombination target. (B–D) E18.5 embryos showing the control phenotype (B), and the bilateral cleft phenotype in Neo/Null with normal neural tube closure (C) and Neo/Null with exencephaly (D). (E–H) Bone and cartilage staining from P1 Neo/Wt (E,G) and Neo/Null (F,H) after mandible removal showing norma basilaris (E,F) and norma lateralis (G,H). An asterisk after the name of the bone represents an area where differences were observed between Neo/Wt control and Neo/Null embryos. (I,J) Images of the palate from Neo/Wt (I) and Neo/Null (J) E18.5 embryos. Note the cleft in J extending from the primary palate into the secondary. (K) RT-PCR analysis of the Tfap2a transcripts that are present in the E10.5 face using the primer pairs shown in A. ‘-RT’ is a no reverse transcriptase control, and ‘Wt’ is wild type. (L) qRT-PCR data showing the ratio of Tfap2a transcripts containing exons 6–7 to those containing exons 2–3. Transcripts lacking exon 6–7 would lack the DNA-binding domain. *P<0.05 to all other genotypes. The two genotypes marked by a bar are not statistically different from each other, but they are different from all other alleles.

Fig. 2.

Morphometric analysis of the orofacial clefting etiology. (A) CVA of shape change at E9.5 and E11.5. Neural tube was categorized ‘closed’, ‘delayed’ or ‘exencephalic’ as shown in supplementary material Fig. S2. (B) Left: PCA of E10.5 embryos (23–35 somites from forelimb to tail tip) showing separation along PC4. Right: wireframes for this component show changes in the nasal pit and maxilla (mxp). fnp, nasal prominences. The wireframes represent a ‘typical’ embryo for that genotype and are based on the number shown under the wireframe (blue: Neo/Wt; red: Neo/Null). (C) Left: PCA of E11.5 (40–50 somite) embryos with separation along PC1. Right: wireframe changes are seen primarily in the size and shape of the nasal pit and the maxilla. (D) Application of the wireframes in B to a control E10.5 embryo. (E) Application of the wireframes in C to a control E11.5 embryo. The arrow denotes a region of high deformation under the nasal pit where the cleft will develop to the ‘Neo/Null’ side of PC1. The insets show this region in greater detail. Blue dots/lines show the landmarks used in the analysis.

Fig. 3.

Reduced cell proliferation in the facial prominences of Neo/Null mice. Proliferation assessed in sections from the nasal pit of Neo/Wt (A,C) and Neo/Null (B,D) E10.5 embryos using EdU (A,B) or anti-phospho histone H3 (PH3) (C,D) detection. Draq5 was used to show nuclei; scale bars: 75 μm. Distal edge of the nasal pit is to the right. Quantitation of EdU (E,G)- or phospho-histone H3 (F,H)-labeled cells expressed as a percentage of total cells for the nasal (E,F) and maxillary (G,H) prominences. *P<0.05; **P<0.01. Note that any differences in cell density apparent in C and D result from slightly different planes of section rather than inherent differences between the Neo/Wt and Neo/Null models.

Fig. 4.

Alterations in gene expression in Neo/Null mice. (A,B) Volcano plots from the (A) maxillary and (B) nasal microarrays. Orange shows all values called ‘above absent’, blue all values called ‘present’, with a P-value of 0.05 (adjusted) and a fold change >1.25. Positive fold change is increased expression in Neo/Null compared with Neo/Wt. (C) Fold change of Fgf pathway members in Neo/Null E10.5 nasal prominences compared with control embryos. (D–K) In situ hybridization for Fgf8 (D–G) and Dusp6 (H–K) expression in Neo/Wt (D,F,H,J) and Neo/Null (E,G,I,K) E10.5 embryos showing frontal (D,E,H,I) and lateral (F,G,J,K) views. Arrow in G shows increased Fgf8 expression at the lateral edge of nasal pit. (L,M) Fgf8 staining visualized by OPT in E10.5 Neo/Wt (L) and Neo/Null (M) samples. (N) qRT-PCR examination of Fgf8 and Wnt9b transcript levels from E10.5 facial prominences. *P<0.05.

Fig. 5.

Reduced Fgf8 gene dosage leads to a rescue of bilateral cleft lip. (A) Numbers of various genotypes and phenotypes observed from the Tfap2a^(neo/neo) × Tfap2a^(+/−);Fgf8^(+/−) crosses (L and R are left- and right-sided cleft, respectively). (B,C) Frontal images of the heads of Neo/Null;Fgf8wt (B) and Neo/Null;Fgf8het (C) P0 pups. Black arrow shows the unilateral cleft line. (D–F) Ventral view of skulls using μCT for Neo/Wt;Fgf8wt (D), Neo/Null;Fgf8wt (E) and Neo/Null;Fgf8het (F) P0 pups. Red arrows show clefting at points where the premaxilla and the palatal process of the premaxilla should be fused.

Fig. 6.

Morphometric analysis of the genetic interactions between Tfap2a and Fgf8 in facial development. (A) Morphometric analysis of E10.5 embryos from Tfap2a^(neo/neo) × Tfap2a^(+/−);Fgf8^(+/−) crosses. CVA showing between-group differences. The wireframes visualize +10 of the axis in black and 0 in gray and the warp images show −6 and +6, respectively, along the axes. (B) PCA of E10.5 Neo/Null;Fgf8het compared with Neo/Wt;Fgf8wt. Wireframes and warp embryo images show −0.05 and +0.05 along PC1. (C) PCA of E10.5 Neo/Null;Fgf8het compared with Neo/Null;Fgf8wt. Wireframes and warp embryo images show −0.06 and +0.06 along PC1.

Fig. 7.

Analysis of variance and integration of facial morphology associated with Tfap2a and Fgf8 mutations. (A) Trace of the variance-covariance matrix to measure variation in shape at E10.5. P-value was determined based on overlap between the curves based on 1000 repetitions of the analysis. (B) Two-block PLS analysis to determine integration at E10.5 between the maxilla and the nasal prominences, the two regions with the largest differences in shape. Statistics can be found in supplementary material Table S5. The data concerning E10.5 C57Bl/6J is presented as an additional control for background or strain effects and was obtained by re-landmarking scans generated previously (Schmidt et al., 2010).