Wnt5a can both activate and repress Wnt/β-catenin signaling during mouse embryonic development

Abstract

Embryonic development is controlled by a small set of signal transduction pathways, with vastly different phenotypic outcomes depending on the time and place of their recruitment. How the same molecular machinery can elicit such specific and distinct responses, remains one of the outstanding questions in developmental biology. Part of the answer may lie in the high inherent genetic complexity of these signaling cascades, as observed for the Wnt-pathway. The mammalian genome encodes multiple Wnt proteins and receptors, each of which show dynamic and tightly controlled expression patterns in the embryo. Yet how these components interact in the context of the whole organism remains unknown. Here we report the generation of a novel, inducible transgenic mouse model that allows spatiotemporal control over the expression of Wnt5a, a protein implicated in many developmental processes and multiple Wnt-signaling responses. We show that ectopic Wnt5a expression from E10.5 onwards results in a variety of developmental defects, including loss of hair follicles and reduced bone formation in the skull. Moreover, we find that Wnt5a can have dual signaling activities during mouse embryonic development. Specifically, Wnt5a is capable of both inducing and repressing β-catenin/TCF signaling in vivo, depending on the time and site of expression and the receptors expressed by receiving cells. These experiments show for the first time that a single mammalian Wnt protein can have multiple signaling activities in vivo, thereby furthering our understanding of how signaling specificity is achieved in a complex developmental context.

Figure 1. A novel transgenic mouse model allowing inducible Wnt5a overexpression

(A) Schematic representation of the tetO-Wnt5A transgenic construct and the experimental strategy. Mouse Wnt5a was cloned downstream of an artificial signal sequence (SS) in frame with an N-terminal FLAG tag under the control of a doxycycline inducible promoter. Transgene expression is only switched on in the presence of both an rtTA driver and doxycycline (DOX). (B) Western blot analysis illustrating that Wnt5a protein can be detected with an anti-Flag antibody in tetO-Wnt5A transgenic mouse embryo fibroblasts infected with pBabe-rtTA3 in the presence (right) but not in the absence (left) of doxycycline. (C) External appearance of a newborn tetO-Wnt5a;R26rtTA double-transgenic animal (right) and a littermate control (left) following treatment with doxycycline from E11.5-P0. Wnt5a overexpressing mice are smaller, have shortened limbs and an easily identifiable craniofacial phenotype. Insert depicts treatment schedule.

Figure 2. Global overexpression of Wnt5a or Dkk1 results in loss of hair follicle formation in the dorsal skin

(A–D) H&E stained histological tissue sections from the dorsal skin of tetO-Dkk;R26rtTA (B) and tetO-Wnt5a;R26rtTA (D) double-transgenic mice and their respective littermate controls (A and C), showing that both Wnt5a and Dkk1 overexpression results in loss of hair follicle formation. Transgene induction was achieved by administering doxycycline from E10.5 to P0. (E–H) H&E stained histological tissue sections showing representative images of the dorsal skin from newborn tetO-Wnt5a;R26rtTA mice, illustrating that a reduction in hair follicle numbers occurs regardless of whether doxycycline treatment was initiated at E10.5 (E), E11.5 (F) or E13.5 (G–H). In addition to their overall number being lower, remaining hair follicles were less mature, although some did show signs of keratinization (white arrowhead in F). (I) Graph depicting the quantification of hair follicle numbers in the skin of newborn mice overexpressing Wnt5a from E10.5-P0, E11.5-P0, E12.5-P0 or E13.5-P0 relative to littermate controls. Sagittal sections from a total of 38 skins were counted. Control samples (n = 16) derived from the different doxycycline treated litters were pooled to calculate the average number of hair follicles per unit length in control skin. A comparable and statistically significant reduction in hair follicle numbers was observed in all Wnt5a overexpressing newborn skins (* p < 1×10⁻⁴ for onset at E10.5, ** p < 5×10⁻⁴ for onset at E11.5 and *** p < 1×10⁻⁵ for onset at E12.5 and E13.5, T-test). Error bars indicate standard deviation. Scale bar is 100 μm in A–H.

Figure 3. Wnt5a overexpression inhibits the second, but not the first wave of hair follicle formation

(A–B) H&E stained histological tissue sections of E14.5 dorsal skin, showing that hair placodes (white arrowheads) are formed in both tetO-Wnt5a;R26rtTA double-heterozygous embryos and littermate controls when doxycycline is administered from E10.5–E14.5. (C) Wholemount alkaline phosphatase staining of E14.5 embryos, revealing a normal pattern of primary hair follicle induction in Wnt5a-transgenic mice (right) following doxycycline administration from E10.5–E14.5. (D–E) H&E stained histological tissue sections of E16.5 dorsal skin, showing a reduction in the number of hair follicles (white arrowheads) in Wnt5a-transgenic mice following doxycycline administration from E10.5–E16.5. Scale bars are 100 μm in A–B and D–E.

Figure 4. Loss of Wnt/β-catenin signaling at sites of Ror2 expression in the dermis of Wnt5a-overexpressing mice

(A–F) Histological tissue sections of wholemount X-gal stained dorsal skin, demonstrating that expression of the Wnt/β-catenin reporter Axin2-lacZ is markedly reduced in Wnt5a-transgenic embryos at both E14.5 (A–B) and E16.5 (C–F). Mice depicted in A and B are littermates, as are the mice depicted in C and D. The activity of an independent Wnt/β-catenin reporter strain, TOPGAL, was also reduced at these timepoints (data not shown). (A–B) At E14.5 the overall activity of Axin2-lacZ is lower throughout the epidermis and dermis of Wnt5a overexpressing mice, but expression of the reporter is particularly reduced in the dermal condensates. (C–D) At E16.5 the Axin2-lacZ signal in the dermal papilla of existing hair follicles remains largely unaffected (white arrowheads in D). (E–F) Close-ups of the control and Wnt5a overexpressing skins shown in C and D, showing that the most dramatic reduction in Wnt/β-catenin reporter gene expression is seen in a thin layer of cells just underlying the basal layer of the epidermis in tetO-Wnt5a;R26rtTA double-heterozygous mice (F) but not in littermate controls (E). (G–H) Immunohistochemical detection of endogenous Ror2 protein expression in the dermis of control (G) but not Ror2-knockout (H) skin at E16.5. Dashed lines indicate the boundary between epidermis and dermis. Dotted lines in G and H indicate the outermost cell layer of the epidermis. Scale bars are 100 μm in A–B, 20 μm in C and D, and 50 μm in E–H.

Figure 5. Global overexpression of Wnt5a causes multiple skeletal defects

(A–L) Wholemount preparations of newborn Wnt5a-overexpressing and control skeletons, stained with Alizarin Red and Alcian Blue to visualize bone (red) and cartilage (blue) following transgene expression from E10.5-P0 (B,F,J), E11.5-P0 (C,G,K) or E12.5-P0 (D,H,L). TetO-Wnt5a;R26rtTA double-heterozygotes develop a split sternum (A–D) and have shortened bones in the fore– (E–H) and hindlimbs (I–L) at the time of birth. (M) Graph depicting the relative length of flat (scapula) and long bones (humerus, ulna and tibia) in control (n=3) and Wnt5a-overexpressing mice (n=4), revealing that limb outgrowth is more severely impaired than overall body size (Figure S3C in the supplementary data). Data were pooled for animals treated with doxycycline from E10.5-P0, E11.5-P0 and E12.5-P0. Error bars indicate standard deviation. (N–Q) Close-up images, demonstrating that bone formation in the extremities is delayed in Wnt5a-overexpressing mice compared to littermate controls. This effect is more pronounced for tarsal (compare Q to P) than for carpal bones (compare O to N). Scale bar is 2 mm in (A–L) and 1 mm in (N–Q).

Figure 6. Delayed calvarial ossification in Wnt5a-transgenic mice

Wholemount preparations of newborn Wnt5a-overexpressing and control skulls, stained with Alizarin Red and Alcian Blue to visualize bone (red) and cartilage (blue) following transgene expression from E10.5-P0 (B,G,L) to E11.5 (C,H,M) or E12.5 (D,I,N) or E13.5 (E,J,O), demonstrating that calvarial bone formation is reduced in the skull of Wnt5a-overexpressing mice compared to control littermates (A,F,K). (F–J) are top views of the skulls depicted in (A–E) and (K–O) are close-ups of the same samples. Scale bar is 2 mm.

Figure 7. Global Wnt5a overexpression causes an increase in Wnt/β-catenin signaling in the developing meninges

(A–D) Wholemount preparations of the skulls from newborn mice, stained with Alizarin Red and Alcian Blue to visualize bone (red) and cartilage (blue), showing that overexpression of Wnt5a (compare B to A), but not Dkk1 (compare D to C) from E10.5-P0 results in reduced calvarial ossification at birth. (E–H) Wholemount preparations of X-gal stained E14.5 embryos, demonstrating that expression of the Wnt/β-catenin reporters Axin2-lacZ (compare F to E) and TOPGAL (compare H to G) is increased in the developing skull of Wnt5a-overexpressing mice compared to control littermates. The skin was removed to better visualize the calvarial mesenchyme. (I–L) Histological tissue sections of the samples depicted in (E–F) following paraffin embedding, showing that the increase in Axin2-lacZ activity is restricted to a thin layer of cells immediately underlying the calvarial mesenchyme and overlying the brain, consistent with the location of the developing meninges. The sagittal (I–J) and coronal (K–L) planes of sectioning are indicated with the black lines in panel (E). (M–N) Wholemount preparations of X-gal stained E16.5 embryos, demonstrating that expression of the Wnt/β-catenin reporter Axin2-lacZ (compare N to M) is increased in the developing skull of Wnt5a-overexpressing mice compared to control littermates. The skin was removed to better visualize the developing skull. (O–P) Independent coronal tissue sections of wholemount X-gal stained E16.5 embryos, showing an increase in Axin2-lacZ activity in the meninges. Scale bars are 2 mm in A–D, 1 mm in E–H and M–N, 100 μm in I–J and 50 μm in K–L and O–P. (br) brain, (cm) calvarial mesenchyme, (dm) dura mater, (sk) skin.