FZD2 regulates limb development by mediating β-catenin-dependent and -independent Wnt signaling pathways

Abstract

Human Robinow syndrome (RS) and dominant omodysplasia type 2 (OMOD2), characterized by skeletal limb and craniofacial defects, are associated with heterozygous mutations in the Wnt receptor FZD2. However, as FZD2 can activate both canonical and non-canonical Wnt pathways, its precise functions and mechanisms of action in limb development are unclear. To address these questions, we generated mice harboring a single-nucleotide insertion in Fzd2 (Fzd2em1Smill), causing a frameshift mutation in the final Dishevelled-interacting domain. Fzd2em1Smill mutant mice had shortened limbs, resembling those of RS and OMOD2 patients, indicating that FZD2 mutations are causative. Fzd2em1Smill mutant embryos displayed decreased canonical Wnt signaling in developing limb mesenchyme and disruption of digit chondrocyte elongation and orientation, which is controlled by the β-catenin-independent WNT5A/planar cell polarity (PCP) pathway. In line with these observations, we found that disruption of FZD function in limb mesenchyme caused formation of shortened bone elements and defects in Wnt/β-catenin and WNT5A/PCP signaling. These findings indicate that FZD2 controls limb development by mediating both canonical and non-canonical Wnt pathways and reveal causality of pathogenic FZD2 mutations in RS and OMOD2 patients.

Keywords: Fzd2; Limb; Omodysplasia; Robinow syndrome; Wnt.

Fig. 1.

Fzd2 is broadly expressed in the developing limb bud. (A) RNAscope in situ hybridization (ISH) of sagittal sections shows that Fzd2 is expressed in forelimb bud at E9.5. (A′) Magnified view of the indicated region in A. Fzd2 expression is higher in the mesenchyme than in the ectoderm, including the apical ectodermal ridge (AER). (B) Fzd2 expression persists in the ectoderm and mesenchyme of the forelimb bud at E10.5 but is decreased in distal mesenchyme compared with its expression at E9.5. (B′) Magnified view of the indicated region in B. (C) Fzd2 expression in E10.5 hindlimb is similar to that in the forelimb. (C′) Magnified view of the indicated region in C. In all photomicrographs, dorsal limb is oriented at the top and distal to the right. Yellow arrows indicate the AER; white dashed lines mark the boundary between ectoderm and mesenchyme. n=3 samples analyzed per stage. Scale bars: 100 µm (A,B,C); 50 µm (A′,B′,C′).

Fig. 2.

Fzd2^INS/INS pups display craniofacial defects. (A) CRISPR/Cas9 editing introduced an extra guanine (G) directly after p.Ser552, mutating the DVL-binding motif (red font), removing the PDZ-interacting domain (green font) and causing a frameshift in the C-terminus of FZD2, which produces an additional 39 amino acids (orange font). The schematic indicates the structure of wild-type human and mouse FZD2 (h/m FZD2) with the FZD domain indicated in green, the FZD-like domains indicated in blue, and the three DVL-interacting domains indicated in red. The predicted effects of human pathogenic FZD2 mutations hFZD2(p.Phe130Cysfs*98), hFZD2(p.Trp377*), hFZD2(p.Trp547*) and hFZD2(p.Trp548*), and the mouse Fzd2^INS allele on protein structure are shown below. The aberrant C-terminus of the FZD2^INS protein is indicated in black. (B) Fzd2^INS/INS neonates had swollen abdomens lacking milk spots (black arrowhead), compared with those of wild-type controls. (C) qPCR and immunoblotting showed that Fzd2 mRNA and protein levels were similar in control and Fzd2^INS/INS skin samples. The black arrow indicates FZD2 protein; the red arrow indicates a nonspecific band. (D) Fzd2^INS/INS pups display severely clefted palates (pink arrows) and shortened skulls. Note the fusion of the bilateral palatal bones (pa) in control skull and separated palatal bones and visible presphenoid bone (ps) in Fzd2^INS/INS skull (yellow arrows in ventral view). Fzd2^INS/INS skulls did not display evidence of craniosynostosis. Quantitation of Fzd2^INS/INS versus control skull lengths showed that Fzd2^INS/INS skulls were statistically significantly shorter than control skulls (n=6 control and n=5 mutants analyzed). Littermate controls were wild type or Fzd2^INS/+. Unpaired two-tailed Student's t-test was used to calculate P-values. P<0.05 was considered significant. For the box and whisker plots in C and D, the box represents the 25-75th percentiles, and the median is indicated. The whiskers show the minimum and maximum measurements. Scale bars: 1 mm.

Fig. 3.

The FZD2^INS mutation causes defective limb development by altering both canonical Wnt and Wnt/PCP signaling. (A) Skeletal preparations show that the ulna and tibia of Fzd2^INS/INS pups are shorter than those of littermate controls at P0. (B) Quantitation of bone element lengths in Fzd2^INS/INS pups (n=4 controls and n=3 mutants). Unpaired two-tailed Student's t-test was used to calculate P-values. P<0.05 was considered significant. (C) Fzd2 deficiency causes reduced expression of Axin2 (yellow arrows) but has little effect on Wnt3 expression in forelimbs at E12.5 (left). qPCR shows that Axin2 mRNA levels are statistically significantly reduced in E12.5 Fzd2^INS/INS forelimb buds compared with those of controls (n=3 Fzd2^INS/INS mutants and n=3 controls assayed) (right). A Kolmogorov–Smirnov test was used to calculate the P-value. (D) Wheat germ agglutinin (WGA; green) and SOX9 (red) staining show that elongation and orientation of digit chondrocytes is affected in Fzd2^INS/INS mutants. (E) Quantitation of cell orientation in E12.5 forelimbs. One-hundred chondrocytes from each embryo were measured. Three pairs of control and mutant embryos were analyzed. The x-axis represents the angle of orientation; the y-axis represents the percentage of chondrocytes at angle X. Chondrocytes oriented horizontally are designated as 0°; chondrocytes oriented along the proximal–distal (P-D) axis are designated as ±90°. The Kolmogorov–Smirnov test was used to calculate the P-value. P<0.05 was considered significant. (F) Quantitation of the ratio of length to width of chondrocytes shows that this is significantly altered in E12.5 Fzd2^INS/INS mutant forelimbs. One-hundred chondrocytes from each embryo and three embryos of each genotype were analyzed. Littermate controls were wild type or Fzd2^INS/+. Unpaired two-tailed Student's t-test was used to calculate the P-value. P<0.05 was considered significant. For the box and whisker plots in B and C, the box represents the 25-75th percentiles, and the median is indicated. The whiskers show the minimum and maximum measurements. Scale bars: 1 mm (A); 100 µm (C); 25 µm (D).

Fig. 4.

Mesenchymal-specific recombination of Fzd2^fl causes defective limb development. (A) Prx1-Cre Fzd2^fl/+ and Prx1-Cre Fzd2^fl/fl mutants have hypoplastic forelimbs at P20. (B) Skeletal preparation of P20 forelimbs. Compared with control forelimbs, Prx1-Cre Fzd2^fl/+ pups have hypoplastic forelimbs; Prx1-Cre Fzd2^fl/fl pups only have a few residual bone elements in their forelimbs. (C-E) Whole-mount views of E17.5 control (C), Prx1-Cre Fzd2^fl/+ (D) and Prx1-Cre Fzd2^fl/fl (E) forelimbs show reduced forelimb size in the Prx1-Cre Fzd2^fl/+ mutant (D) and a small residual forelimb in the Prx1-Cre Fzd2^fl/fl mutant (E). (F-H) Skeletal preparations show that all the skeletal elements are hypomorphic in an E17.5 Prx1-Cre Fzd2^fl/+ mutant forelimb (G) and only a few skeletal elements develop in a Prx1-Cre Fzd2^fl/fl mutant (H) compared with control (F). Controls were littermates lacking Prx1-Cre and/or Fzd2^fl. Scale bars: 0.5 mm.

Fig. 5.

Reduced mesenchymal Fzd expression affects both canonical and non-canonical Wnt signaling. (A-C) ISH shows that mesenchymal Fzd2 expression in E10.5 forelimb (FL) mesenchyme is reduced in Prx1-Cre Fzd2^fl/+ mutants (B) and almost absent in Prx1-Cre Fzd2^fl/fl mutants (C) compared with control (A). (D-F) ISH shows that Axin2 expression is reduced in distal forelimb mesenchyme (yellow arrows) but not in the AER (pink arrows) in E10.5 Prx1-Cre Fzd2^fl/+ (E) and Prx1-Cre Fzd2^fl/fl (F) embryos compared with control embryos (D). (G-I) Ectodermal Wnt3 expression is unaffected in the forelimbs of E10.5 Prx1-Cre Fzd2^fl/+ (H) and Prx1-Cre Fzd2^fl/fl (I) embryos compared with control forelimbs (G). (J) qPCR quantification of Fzd2 (left) and Axin2 (right) mRNA levels in E10.5 forelimb mesenchyme shows statistically significantly decreased expression of Fzd2 and Axin2 expression in Prx1-Cre Fzd2^fl/+ and Prx1-Cre Fzd2^fl/fl samples compared with samples from controls lacking Prx1-Cre or Fzd2^fl. ‘−dCT’ refers to the delta between the cycle threshold (CT) for the target gene and the CT for Gapdh. n=3 per genotype. Unpaired two-tailed Student's t-test was used to calculate P-values. P<0.05 was considered significant. (K) WGA (green) stains cell membranes, revealing the shapes of SOX9 (red)-expressing chondrocytes. In E12.5 control forelimb digits, chondrocytes are elongated and predominantly lie perpendicular to the P-D axis; in Prx1-Cre Fzd2^fl/+ digits, the chondrocytes are more rounded and their orientation is more random. (L) Quantitation of chondrocyte orientation. At least 500 control and Prx1-Cre Fzd2^fl/+ chondrocytes from three embryos of each genotype were analyzed. The x-axis represents the angle of orientation; the y-axis represents the percentage of chondrocytes at each angle X. Chondrocytes oriented horizontally are designated as 0°; chondrocytes oriented along the P-D axis are designated as ±90°. A Kolmogorov–Smirnov test was used to calculate the P-value. P<0.05 was considered significant. (M) Quantitation of the length-to-width ratio of Prx1-Cre Fzd2^fl/+ chondrocytes compared with controls shows a statistically significant difference, demonstrating that chondrocyte cell shape is altered in the mutants. One-hundred chondrocytes from each embryo were analyzed from n=3 embryos per genotype. Unpaired two-tailed Student's t-test was used to calculate the P-value. P<0.05 was considered significant. For the box and whisker plots in J, the box represents the 25-75th percentiles, and the median is indicated. The whiskers show the minimum and maximum measurements. Controls were littermates lacking Prx1-Cre and/or Fzd2^fl. Scale bars: 100 µm (A-I); 20 µm (K).