Intercellular Genetic Interaction Between Irf6 and Twist1 during Craniofacial Development

Abstract

Interferon Regulatory Factor 6 (IRF6) and TWIST1 are transcription factors necessary for craniofacial development. Human genetic studies showed that mutations in IRF6 lead to cleft lip and palate and mandibular abnormalities. In the mouse, we found that loss of Irf6 causes craniosynostosis and mandibular hypoplasia. Similarly, mutations in TWIST1 cause craniosynostosis, mandibular hypoplasia and cleft palate. Based on this phenotypic overlap, we asked if Irf6 and Twist1 interact genetically during craniofacial formation. While single heterozygous mice are normal, double heterozygous embryos (Irf6 ^+/- ; Twist1 ^+/- ) can have severe mandibular hypoplasia that leads to agnathia and cleft palate at birth. Analysis of spatiotemporal expression showed that Irf6 and Twist1 are found in different cell types. Consistent with the intercellular interaction, we found reduced expression of Endothelin1 (EDN1) in mandible and transcription factors that are critical for mandibular patterning including DLX5, DLX6 and HAND2, were also reduced in mesenchymal cells. Treatment of mandibular explants with exogenous EDN1 peptides partially rescued abnormalities in Meckel's cartilage. In addition, partial rescue was observed when double heterozygous embryos also carried a null allele of p53. Considering that variants in IRF6 and TWIST1 contribute to human craniofacial defects, this gene-gene interaction may have implications on craniofacial disorders.

Figure 1

Craniofacial abnormalities in Irf6 and Twist1 double heterozygotes murine animals and morphometric measurements of embryonic mandible. Compared to wild type murine embryo (A), littermate mutant embryo of double het (B) has a short mandibular prominence start at E12.5. At P0, mutant pup has lower jaw abnormalities (C versus D), and complete cleft of the secondary palate (E versus F). Phase contrast images show mutant embryo with absent lower jaw and low-set ear (H) but grossly normal body (G versus H). Skeletal preparations show that the mutant pup is missing a mandible and has abnormalities in inner ear bones (I versus J). Hematoxylin and eosin staining of coronal sections of heads from wild type (K–M) and double heterozygous affected murine pups (N–P), from anterior (K,N), middle (L,O) to posterior (M,P). Relative to wild type littermate (WT), mutant (MT) has cleft palate, malformed tongue, detached retina and holoprosencephaly (O,P). Tongue (T), mandible (M), molar tooth germ (Mo), retina (R), brain (B). Scale bars = 1000 μm. Figures (Q,R and S) are average length, width and area of murine mandibles of different genotypes at E17.5. A total of 28 craniofacial skeletons of murine embryos were used to calculate the averages. The black bars represent standard error for each genotype. Average mandibular lengths of murine embryos of all different genotypes (Q). Average mandibular widths of murine embryos of all different genotypes (R). Average mandibular area of murine embryos of all different genotypes (Q). The average area of Irf6, Twist1 double hets vs. wild type is smaller relative to mean area of wild types, but the difference is not statistically significant using the multi-variants ANOVA test. Wild type (WT), Twist1 heterozygous (Tw.het), Irf6 heterozygous (Irf6.het) and double heterozygous (D.het).

Figure 2

TWIST1 suppresses IRF6 expression in vitro via its enhancer element. RTqPCR data shows that TWIST1 is not expressed in HaCaT keratinocytes, while IRF6 is highly expressed in these cells (A). Exogenous overexpression of TWIST1 using pGL3 expression vector remarkably reduced IRF6 expression by 12-fold 48 h post-transfection (A). In human embryonic kidney cells (HEK293), TWIST1 is moderately expressed compared to IRF6 that is expressed 2-fold more (B). Knockdown of TWIST1 expression with siRNA slightly increased the expression level of IRF6 in HEK293 cells (B). However, knockdown of TWIST1 expression significantly increased luciferase activity of a luciferase gene that is under direct control of IRF6 enhancer element (C). ChIP-seq data showed that TWIST1 binds to IRF6 enhancer element accompanied by enrichment of RNA PolII protein at the enhancer and promoter element of IRF6 gene, but the enrichment of PolII at the enhancer and promoter regions of IRF6 may indicate an active transcriptional state. However, the amount of PolII at the IRF6 enhancer and promoter compared to IgG was not significant and the enrichment was moderate compared to PolII at PCNA promoter (D).

Figure 3

Immunostaining for detection of IRF6 and TWIST1 expression and immunoblot for total protein analysis. IRF6 and TWIST1 are not expressed at E7.5 (A,B), while their expression is detected in neuroectoderm for IRF6 and in mesoderm for TWIST1 at E8.5 (C,D). IRF6 expression persists in neuronal- and non-neuronal ectoderm, while TWIST1 expression is detected in CNC cells at E9.0 (E,F). Cytosolic IRF6 and nuclear TWIST1 are co-localized to subset of CNC cells in first and second pharyngeal arches at E9.0 (White arrowheads, F,H). Subsequently, IRF6 is expressed in neural tube (I,J) and oral epithelium (K,L), while TWIST1 is expressed in migratory CNC cells (I,J) and mesenchymal cells (K,L) at E10.5. For assessment of cell proliferation and apoptosis, immunofluorescent images show the level of cell proliferation by highlighting the cells undergoing cell division in red in wild type embryos (M,N) compared to double heterozygous mutant for Irf6 and Twist1 at E12.5 (O,P). Relative to wild type (Q), immunostaining of activated caspase 3 indicates a higher level of cell apoptosis in the mutant at E12.5 (S). Stronger signal is observed in retina, tongue and mandible. Immunohistochemical staining show the level of P53 in wild type mandibular tissues (R) compared to mutant samples (T). For expression of EDN1 and TWIST1, EDN1 is secreted from epithelial and mesodermal cells of the mandible (m) but not in the maxilla (mx) (U), while TWIST1 is expressed in the adjacent mesenchymal cells in mandible and maxilla. Relative to wild type (U), EDN1 and TWIST1 are significantly reduced in epithelium and CNC cells in mutant samples, respectively, but the expression of TWIST1 in maxilla is not affected (V). Scale bars = 100 μm. Immunoblot was performed to validate the IF data of EDN1 at quantitative level. EDN1 proteins were remarkably reduced in mutant mandibular processes compared to wild type littermates at both embryonic stages E10.5 and E12.5 (W). Slight reduction in IRF6 expression and DLX5 was noticed at E10.5 and E12.5, respectively (W). GAPDH was used as internal loading control and full-length gels are included in supplementary.

Figure 4

Craniofacial skeletal abnormalities in Irf6 null pups and quantitative expression of selected genes. (A–H) Bone (red) and cartilage (blue) of head and mandible of P0 pups. Relative to wild type littermates (A,C,E,G), mutant pups (B,D,F,H) have less cartilage (B,F), and the mandible is missing a distal bone where the incisor is normally formed (E versus F). The mandibular processes are shorter (F). The coronal sutures of the skull of Irf6 null pups are fused (arrows) (D,H) compared to wild type littermate (C,G). Scale bars: 1000 μm. Expression of Irf6, Tw1 and Runx2 is reduced in Irf6 null embryos at E14.5 compared to wild type littermate (I–K). Similarly, the expression of Irf6 and Runx2 genes is reduced to a lesser extent in compound alleles for Irf6 null and p53 heterozygous embryos at E14.5, except for Tw1 where the expression is slightly increased (J). The relative gene expression is the average of two independent experiments with four technical replicates for each genotype. The asterisks represent the level of statistical significance of the average of gene expression compared to wild type embryos. The data are the average of two independent experiments. Similarly, the expression of Irf6 and Grhl3 is significantly reduced in Irf6 null embryos at both E10.5 and E12.5 (L,M).

Figure 5

Protein expression of DLX5, DLX6, TWIST1 and P53 in mandibular tissues. The counter staining color used in the immunohistochemistry is purple for nuclei and protein expression is in red. Staining without primary antibodies was used as negative control (A,B,G,H,M,N,S,T). Expression intensity of DLX5 (C,D), DLX6 (I,J) and TWIST1 (O,P) in wild type tissues is reduced compared to double heterozygous mutant embryos for DLX5 (E,F), DLX6 (K,L), and TWIST1 (Q,R) in mandibular processes at E12.5. In contrary, expression of P53 is low in wild type (U,V), but increased in mutant (W,X) at E12.5. Scale bars = 20 μm.

Figure 6

Ex-vivo mandibular organ for the rescue experiment with Endothelin1. Top row shows the Meckel’s cartilage of mandibular explants that were cultured in minimal medium without any treatment for all different genotypes (A–D). Schematic drawing for the Meckel’s cartilage is depicted in (A’–D’) images for illustrative purposes. The third row shows the Meckel’s cartilage of explants of all different genotypes treated with exogenous Edn1 peptides (E–H). Similarly, images (E’–H’) are schematic drawing of the Meckel’s cartilage. Scale bar = 1.5 mm.