1 INTRODUCTION

The tongue development requires a proper regulation of its molecular mechanisms to attain its stereotypical shape, while developmental defects including microglossia, macroglossia and aglossia1 can hamper the normal function of the tongue, for example taste sensing, speaking and food processing. It has been shown that molecular signalling pathways and their interactions play important roles in the proper formation of the tongue.2-8 However, our current understanding of tongue development in relation to molecular signalling pathways is far from complete. In this study, we report the important role of Neurofibromin 2 (Nf2) in the neural crest (NC)-derived tongue mesenchymal cells in regulating mouse tongue shape and size which is distinct from that in other organs.

In mice, the tongue forms from four branchial arches (BAs) 1–4, that is tongue primordia. Among these four BAs, BAs 1–2 give rise to the anterior two thirds of the tongue – the oral tongue, and BAs 3–4 give rise to the posterior third of the tongue – the pharyngeal tongue.9 The tongue emerges as three lingual swellings, two lateral and one posterior (tubercula impar) on the floor of the mandible at embryonic day (E) 11.5.10 These swellings fuse and grow into a spatulate tongue at E12.5. Thereafter, the tongue organ continues to grow and various types of cells are differentiated and highly organized, including formed appendages such as taste papillae and taste buds.

Nf2 is considered as a tumour suppressor gene that activates the Hippo signalling pathway through its gene product known as merlin.11 Under normal cellular conditions, merlin recruits mammalian sterile 20-like protein kinase (Mst1/2), a large tumour suppressor (Lats1/2) and adaptor protein Salvador (Salv) to activate the Hippo signalling pathway by phosphorylating the transcriptional activator Yes-associated protein (Yap).12, 13 Phosphorylation causes Yap to remain in the cytoplasm and prevents the transcription of proliferation-specific genes.14 On the other hand, the absence of Nf2 results in the lack of merlin, hence prevents Yap from phosphorylation and promotes the nuclear translocation of Yap to transcribe cell proliferation-specific genes.14 Deficiencies in Nf2 function and Hippo signalling activity lead to excess cell proliferation, organ overgrowth, tumourigenesis15, 16 and metastasis,17 which are often promoted by the nuclear translocation of Hippo signalling repressor form, Yap protein.18-20

However, our results indicated paradoxically distinct roles of Nf2 in regulating Hippo signalling and cell proliferation in different regions of tongue organ at different stages. In Wnt1-Cre driven conditional knockout of Nf2 (Wnt1-Cre/Nf2cKO) mice, Hippo signalling activities and cell proliferation were altered differently in anterior versus posterior tongue at early (E11.5–E13.5, macroglossia) versus late (E15.5–E18.5, microglossia) embryonic stages. In combination with previously reported data in the literature, our results indicate tissue context-specific roles of Nf2 in regulating Hippo signalling, cell proliferation and shape and size of organs.

2 RESULTS 2.1 The immunosignals of NF2 and Yap are primarily distributed in the tongue mesenchyme

Neurofibromin 2 (Nf2) is expressed in migrating neural crest cells21 that give rise to a large population, if not all, of the lingual mesenchyme.6, 22 To understand where Nf2 plays its regulatory role during tongue development, we examined the distribution of Nf2 and Yap proteins. At E18.5, Nf2 immunosignals were extensively distributed in the tongue mesenchyme (Figure 1A), but not apparent in the tongue epithelium (Figure 1A). Yap immunoproducts were broadly detected in the mesenchyme including the papilla core region subjacent to the tongue epithelium and bright signals were observed in the deeper layers of the mesenchyme of E18.5 tongue (Figure 1B). The concurrent distribution of Nf2 (bright) and Yap (faint) immunosignals was evident in the mesenchyme including the mesenchymal zone immediately under the epithelium (arrowheads in Figure 1A,B).

Single-plane laser scanning confocal images of E18.5 tongue sections immunoreacted with Nf2 (A, green), Hippo signalling transcription factor Yap (B, green) or epithelial cell marker E-cadherin (C, green). RFP+ cells in C were labelled by Wnt1-Cre. Sections were counterstained with DAPI (blue). Arrowheads in A and B point to the mesenchymal layer immediately under the epithelium. Dashed lines separate the epithelium from the underlying mesenchyme. Scale bars: 50 μm

Wnt1-Cre has been widely used to label NC cell lineage4 and labelled cells are largely distributed in the tongue mesenchyme,6, 23, 24 though rare labelled cells have also been found in the tongue epithelium.6 We have previously reported that Wnt1-Cre-labelled NC-derived cells populate the mesenchyme of tongue primordium – all four branches 1–4 at E10.5.2To confirm Wnt1-Cre-driven genetic alterations in neural crest-derived tongue mesenchyme at later stages of tongue development, E18.5 Wnt1-Cre/RFP tongues were analysed and we found that RFP+ cells were dominantly detected in the tongue mesenchyme and no RFP+ cells were seen in the tongue epithelium in the examined serial sections (Figure 1C).

2.2 Wnt1-Cre induces deletion of Nf2 in the tongue mesenchyme and reduced Nf2 transcripts in the epithelium

Nf2 in situ hybridization analyses revealed that mRNA transcripts were broadly distributed in both epithelium and mesenchyme of Cre−/Nf2fx/fx littermate control tongues at all three stages tested (E12.5, E15.5 and E18.5) (Figure 2A, B). In addition, our RNA sequencing data also showed that Nf2 mRNA transcripts were present in both epithelium and mesenchyme in E12.5, E14.5 and postnatal day 1 (P1) tongues with a significant difference at E12.5 when tongue mesenchyme had a higher level of Nf2 mRNA transcripts compared to the epithelium (p < 0.05 in Figure 2C). In Wnt1-Cre/Nf2cKO mutants, Nf2 transcripts were significantly reduced in both epithelium and mesenchyme at all three stages tested (Figure 2A,B). Quantitative RT-PCR analyses confirmed the significantly low Nf2 expression in both epithelium and mesenchyme of E12.5 Wnt1-Cre/Nf2cKO mutant tongues compared to that of Cre−/Nf2fx/fx littermate controls (p < 0.05 in Figure 2D).

Light microscopy images of E12.5, E15.5 and E18.5 tongue sections from Cre−/Nf2fx/fx littermate control and Wnt1-Cre/Nf2cKO mutants at a low (A) or high (B) magnification. In situ hybridization was performed using an antisense probe for Nf2. Dashed lines in B separate the epithelium from the mesenchyme. Scale bars: 100 µm in A; 25 µm in B. (C-D) Histograms (X±SD; n=3) to present the Nf2 transcripts level in tongue epithelium (Epi) and mesenchyme (Mes), that is fragments per kilobase of transcripts per million mapped reads (FPKM) in wild-type (C) or 2−ΔCT values (D) in Cre−/Nf2fx/fx and Wnt1-Cre/Nf2cKO mutants. *p ≤ 0.05 compared to the corresponding tissues of Cre−/Nf2fx/fx littermate control using two-way ANOVA followed by Fisher's least significant difference (LSD) analyses. (E) Histograms to present the number of differentially expressed genes (DEGs) associated with Epi, muscle (Mus), molecular signalling (Mol) and other (Oth) functions (left), and FPKM (X±SD; n=3) values of Hippo signalling target gene Yod1 (right) in E14.5 Cre−/Nf2fx/fx and Wnt1-Cre/Nf2cKO mutant tongues. *p ≤ 0.05 Student's t-test compared to Cre−/Nf2fx/fx littermate control group

To understand the changes of genetic profile that resulted from mesenchymal Nf2 deletion, RNA-Seq analyses were performed for a comparison between Wnt1-Cre/Nf2cKO mutant tongue versus Cre−/Nf2fx/fx littermates at E14.5 (n = 3). A total of 140 differentially expressed genes (DEGs) were detected by Cuffdiff in the Wnt1-Cre/Nf2cKO versus Cre−/Nf2fx/fx littermate controls (|FC| >1, p < 0.05, FDR q < 0.05). GO terms of the DEGs showed that 13 DEGs were relevant to epithelial development and cell behaviour (Figure 2E), 29 DEGs related to muscle. In addition, 12 DEGs that were associated with multiple molecular signalling pathways, including Shh (1 DEG), Wnt (4 DEGs), FGF (2 DEGs), Rho (1 DEG) and Jak (3 DEGs) (Figure 2E). Furthermore, down-regulation of Hippo signalling -regulator gene Yod1 was detected in the Wnt1-Cre/Nf2cKO mutant tongue compared to the Cre−/Nf2fx/fx littermate control tongue (Figure 2E).

2.3 Nf2 cKO in neural crest-derived tongue mesenchyme leads to macroglossia at early but microglossia at late stages of tongue development

To understand the roles of Nf2 in the tongue development, phenotypic analyses were performed in Wnt1-Cre/Nf2cKO mutant tongues and Cre−/Nf2fx/fx littermate controls at multiple embryonic stages. At E11.5 when tongue swellings emerged from the branchial arches, tongue swellings were clearly seen in Cre−/Nf2fx/fx littermate controls (Figure 3A). In Wnt1-Cre/Nf2cKO mutants, tongue swellings were less profound compared to the Cre−/Nf2fx/fx littermates (Figure 3A).

(A) Scanning electron microscopy (SEM) images of tongues on mandible in Cre−/Nf2fx/fx littermate control and Wnt1-Cre/Nf2cKO embryos at E11.5, E12.5, E13.5, E15.5, E16.5 and E18.5. Open arrowheads point to tongue swellings. Scale bars: 200 µm. (B) Histograms (X±SD; n=3) to present the width of the anterior and posterior oral tongues and length of the oral tongue in Cre−/Nf2fx/fx littermate controls and Wnt1-Cre/Nf2cKO mutants. *p ≤ 0.05, **p ≤ 0.01 Student's t-test compared to Cre−/Nf2fx/fx littermate control group

At E12.5 and E13.5 (Figure 3A), tongue shape and size in Wnt1-Cre/Nf2cKO mutants were altered significantly compared to Cre−/Nf2fx/fx littermate controls (Figure 3A). The Wnt1-Cre/Nf2cKO mutant tongues were significantly narrower in the anterior oral tongue region and significantly wider towards the posterior oral tongue region (Figure 3A–B; p < 0.05) compared to that of Cre−/Nf2fx/fx littermates (Figure 3). Even though the length of the oral tongue was not significantly altered in the Wnt1-Cre/Nf2cKO mutants (Figure 3A–B; p > 0.05), increase in the tongue width resulted in an overall enlarged tongue (i.e. macroglossia) in Wnt1-Cre/Nf2cKO mutants compared to Cre−/Nf2fx/fx littermate control (Figure 3).

At E15.5-E16.5, Wnt1-Cre/Nf2cKO mutants continued to develop a significantly wider posterior oral tongue (Figure 3A-B; p < 0.05) compared to that of Cre−/Nf2fx/fx littermates (Figure 3). In contrast to the early stages (i.e. E12.5 and E13.5), the E15.5–E16.5 oral tongues of Wnt1-Cre/Nf2cKO mutants were significantly shorter (Figure 3A-B; p < 0.05) than those of Cre−/Nf2fx/fx littermates (Figure 3). No significant differences in anterior oral tongue width were noticed between Wnt1-Cre/Nf2cKO mutants (Figure 3A, B; p > 0.05) and Cre−/Nf2fx/fx littermates (Figure 3A). As a result of the collective changes in the tongue length and width, E15.5–E16.5 Wnt1-Cre/Nf2cKO mutants had a relatively smaller tongue (i.e. microglossia, Figure 3A) compared to the Cre−/Nf2fx/fx littermates (Figure 3A).

At E18.5, microglossia phenotype in the Wnt1-Cre/Nf2cKO mutants was evident compared to Cre−/Nf2fx/fx littermates (Figure 3A). Wnt1-Cre/Nf2cKO mutant mice had a significantly shorter oral tongue compared to those of Cre−/Nf2fx/fx littermate controls (Figure 3A,B; p < 0.05). Unlike the earlier stages (E12.5–E16.5), no significant changes in the width of posterior oral tongue were observed in the Wnt1-Cre/Nf2cKO mutants compared to Cre−/Nf2fx/fx littermates (Figure 3A,B; p > 0.05).

2.4 Region- and stage-specific alterations of Hippo signalling in Wnt1-Cre/Nf2 cKO mouse tongues

Nf2 is a known cell proliferation suppressor and often acts via Hippo-YAP signalling to regulate organ size.13, 25 To understand the potential cause of the alteration of tongue shape and size, we examined the Hippo signalling activity in different tongue regions at different stages of Wnt1-Cre/Nf2cKO mutants and Cre−/Nf2fx/fx littermate controls.

Western blot analyses on downstream signalling components of Hippo pathway revealed that at E12.5 the deactivated form of transcriptional regulator p-Yap was at a significantly lower level in the anterior (Figure 4A,B; p < 0.05), but not posterior (Figure 4A,B; p > 0.05), oral tongue mesenchyme of Wnt1-Cre/Nf2cKO mutants compared to the corresponding regions of the Cre−/Nf2fx/fx littermate control. In contrast to p-Yap, the transcriptional activator of Hippo signalling pathway Yap that promotes cell proliferation was at a significantly higher level in the posterior, however lower in the anterior, oral tongue mesenchyme in Wnt1-Cre/Nf2cKO mutant compared to the corresponding tongue regions of the Cre−/Nf2fx/fx littermates (Figure 4A,B; p < 0.05).

(A) Western blot bands of p-Yap, Yap and GAPDH in the E12.5, E15.5 and E18.5 tongue mesenchyme from the anterior (Ant) and posterior (Post) oral tongues (E12.5 and E15.5) or the entire tongue (E18.5) of Cre−/Nf2fx/fx (Ctrl) and Wnt1-Cre/Nf2cKO (cKO) embryos. (B) Histograms (X ± SD; n = 3) to present the normalized band intensities of p-Yap and Yap relative to the GAPDH; and p-Yap/Yap ratios in the E12.5, E15.5 and E18.5 tongue mesenchyme. *p ≤ 0.05, **p ≤ 0.01 Student's t-test compared to Cre−/Nf2fx/fx littermate control group. (C) Histograms (X±SD; n=3) to present the 2−ΔCT values of Yap target genes in posterior and anterior tongue epithelium (Epi) and mesenchyme (Mes) of E12.5 Cre−/Nf2fx/fx and Wnt1-Cre/Nf2cKO mutants. *p ≤ 0.05, **p ≤ 0.01 compared to Cre−/Nf2fx/fx littermate control using two-way ANOVA followed by Fisher's least significant difference (LSD) analyses

At E15.5, the anterior oral tongue mesenchyme of Wnt1-Cre/Nf2cKO mutants had a significantly lower level of p-Yap than the corresponding region of the Cre−/Nf2fx/fx littermates (Figure 4A,B; p < 0.05). However, no significant changes of p-Yap level were detected in the posterior oral tongue mesenchyme of E15.5 Wnt1-Cre/Nf2cKO mutants compared to Cre−/Nf2fx/fx littermates (Figure 4A,B; p > 0.05). In contrast to p-Yap, Yap was detected at a significantly lower level in both posterior and anterior oral tongue mesenchyme of E15.5 Wnt1-Cre/Nf2cKO mutants compared to the respective regions of Cre−/Nf2fx/fx littermates (Figure 4A,B; p < 0.01). At E18.5, no significant changes of p-Yap and Yap level were detected in the tongue mesenchyme of Wnt1-Cre/Nf2cKO mutants compared to Cre−/Nf2fx/fx littermates (Figure 4A,B; p > 0.05).

The ratio of p-Yap/Yap was calculated to represent the Hippo pathway activity. The p-Yap/Yap ratio was altered distinctly in different tongue regions at different stages in Wnt1-Cre/Nf2cKO mutants compared to the Cre−/Nf2fx/fx littermates. A significantly lower ratio of p-Yap/Yap was detected in the posterior, but not anterior, E12.5 tongue mesenchyme (Figure 4B). At E15.5, a significantly higher p-Yap/Yap ratio was detected in both posterior and anterior tongue mesenchyme of the Wnt1-Cre/Nf2cKO mutants (Figure 4B). No significant changes in p-Yap/Yap ratio were found in E18.5 Wnt1-Cre/Nf2cKO mutant tongue mesenchyme (Figure 4B).

Quantitative RT-PCR analyses on Yap target genes that are known regulators of cell proliferation in E12.5 Wnt1-Cre/Nf2cKO and Cre−/Nf2fx/fx tongues revealed region- and tissue-specific changes in gene expression levels. Among the genes tested, distinct expression patterns with respect to tongue regions (anterior vs. posterior) and tissue types (epithelium vs. mesenchyme) were found (Figure 4C).

In the mesenchyme of E12.5 tongues of Wnt1-Cre/Nf2cKO mutants compared to the Cre−/Nf2fx/fx littermate controls: (1) Areg, Ctgf, Cyclin D1 and Amotl1 expression levels were significantly up-regulated in the posterior, and on the contrary, down-regulated in the anterior, oral tongue mesenchyme (p < 0.05); (2) Ankrd1 expression was significantly down-regulated in the posterior, but up-regulated in the anterior oral tongue mesenchyme (p < 0.05); (3) Cyr61 expression was significantly up-regulated, while Birc5 down-regulated in both posterior and anterior oral tongue mesenchyme (p < 0.05); (4) Axl gene expression was significantly down-regulated in the posterior (p  < 0.05), but not altered in the anterior (p > 0.05) tongue mesenchyme.

In the epithelium of E12.5 tongues of Wnt1-Cre/Nf2cKO mutants compared to the Cre−/Nf2fx/fx littermate controls, of all eight genes tested, Areg, Ctgf and Cyr61 had significant alterations in gene expression levels (Figure 4C), that is Areg expression levels were significantly lower in both posterior and anterior tongue epithelium, while Ctgf and Cyr61 expression were significantly up-regulated in the anterior oral tongue epithelium (p <0.05).

2.5 Region- and stage-specific alterations of cell proliferation in Wnt1-Cre/Nf2 cKO mouse tongues

To understand the effects of altered Hippo signalling activity in the Wnt1-Cre/Nf2cKO mutant mesenchyme and the cause of tongue shape/size alterations, cell proliferation was analysed in Wnt1-Cre/Nf2cKO mutants at multiple stages. We found distinct alterations in different tongue regions at different stages.

At E12.5 (Figure 5), more BrdU+ cells were observed in the posterior (p < 0.05), however fewer in the anterior (p < 0.05), oral tongue mesenchyme of Wnt1-Cre/Nf2cKO mutants (Figure 5B) compared to the Cre−/Nf2fx/fx littermate control (Figure 5A). No obvious changes (p > 0.05) in the number of BrdU+ cells were seen in the tongue epithelium of E12.5 Wnt1-Cre/Nf2cKO mutants (Figure 5B) compared to Cre−/Nf2fx/fx littermates (Figure 5A).

Single-plane laser scanning confocal images of tongue sections in E12.5 Cre−/Nf2fx/fx littermate control (A) and Wnt1-Cre/Nf2cKO (B) mutants. Sections were immunoreacted with S-phase cell proliferation marker BrdU (green) and epithelial cell marker E-cadherin (magenta). A1–2 and B1–2 are high magnification images of the posterior (A1, B1) and anterior (A2, B2) oral tongue regions. Scale bars: 200 μm in A, B; 50 μm in A1–2, B1–2. (C) Histograms (X±SD; n=3) to present the number of BrdU+ cells per mm2 in the anterior (Ant) and posterior (Post) oral tongues of Cre−/Nf2fx/fx littermate control and Wnt1-Cre/Nf2cKO mutants. *p ≤ 0.05 compared to the corresponding regions of Cre−/Nf2fx/fx littermate control using two-way ANOVA followed by Fisher's LSD analyses

At E15.5 (Figure 6), in Wnt1-Cre/Nf2cKO mutants as compared to the corresponding regions of the Cre−/Nf2fx/fx littermates, a significantly higher amount of BrdU+ cells were detected in the epithelium (p < 0.05) and mesenchyme just beneath the epithelium (p < 0.05) of both posterior (Figure 6B2) and anterior (Figure 6B4) oral tongue. In contrast, a significantly lower amount of BrdU+ cells were detected in the deeper mesenchyme layers of both posterior (Figure 6B1; p < 0.05) and anterior (Figure 6B3; p <  0.05) oral tongue (Mes2 in Figure 6B).

Single-plane laser scanning confocal images of tongue sections in E15.5 Cre−/Nf2fx/fx littermate control (Ctrl, A) and Wnt1-Cre/Nf2cKO (cKO, B) mutants. Sections were immunoreacted with S-phase cell proliferation marker BrdU (green) and epithelial cell marker Krt5 (magenta). A1–4 and B1–4 are high magnification images of the posterior (A1–2, B1–2) and anterior (A3–4, B3–4) oral tongue regions. Epi: epithelium (A2, B2 and A4, B4); Mes1: mesenchyme layer just beneath the epithelium (A2, B2 and A4, B4); Mes2: deeper mesenchymal layers (A1, B1 and A3, B3) of the posterior (A1, B1) and anterior (A3, B3) oral tongues. Dashed lines in A and B demarcate the borders between Epi, Mes1 and Mes2. Scale bars: 200 μm in A, B; 50 μm in A1–4, B1–4. (C) Histograms (X±SD; n=3) to present the number of BrdU+ cells per mm2 in the anterior and posterior oral tongue epithelium and mesenchyme of Cre−/Nf2fx/fx littermate control and Wnt1-Cre/Nf2cKO mutants. *p ≤ 0.05, **p ≤ 0.01 compared to Cre−/Nf2fx/fx littermate control using two-way ANOVA followed by Fisher's LSD analyses

At E18.5, in Wnt1-Cre/Nf2cKO mutants (Figure 7B) compared to the Cre−/Nf2fx/fx littermate controls (Figure 7A, A1, and A2), a significantly lower number of BrdU+ cells (p < 0.01) was detected in all tongue regions and tissue types, including epithelium and mesenchyme in both posterior (Figure 7B1) and anterior (Figure 7B2) oral tongue regions.

Single-plane laser scanning confocal images of tongue sections in E18.5 Cre−/Nf2fx/fx littermate control (A) and Wnt1-Cre/Nf2cKO mutants (B). Sections were immunoreacted with S-phase cell proliferation marker BrdU (green) and epithelial cell marker Krt5 (magenta). Insets A1–2 and B1–2 are high magnification images of the posterior (A1, B1) and anterior (A2, B2) oral tongue regions. Epi: epithelium; Mes1: mesenchyme layer just beneath the epithelium; Mes2: deeper mesenchymal layers. Dashed lines in A and B demarcate the borders between Epi, Mes1 and Mes2. Scale bars: 200 μm in A, B; 50 μm in A1–4, B1–4. (C) Histogram (X±SD; n=3) to present the number of BrdU+ cells per mm2 in the tongue epithelium and mesenchyme of Cre−/Nf2fx/fx littermate control and Wnt1-Cre/Nf2cKO mutants. **p ≤ 0.01 compared to the corresponding regions of Cre−/Nf2fx/fx littermate control using two-way ANOVA followed by Fisher's LSD analyses

2.6 Distinct alterations of cell proliferation in different regions of tongue primordium (branchial arches) in mesenchymal Nf2 cKO mice

To understand the potential cause of the unproportioned alterations in tongue shape and size at early stages, we examined the Nf2 expression, Hippo signalling activity and cell proliferation in E10.5 branchial arches (BAs) 1–4 of Wnt1-Cre/Nf2cKO mutants and Cre−/Nf2fx/fx littermate controls.

Nf2 in situ hybridization revealed that in Cre−/Nf2fx/fx littermate controls Nf2 mRNA transcripts were abundantly distributed in all four BAs that will give rise to the developing tongue (Figure 8A). In Wnt1-Cre/Nf2cKO mutants, Nf2 transcripts were absent in the corresponding regions of all four BAs (Figure 8B). Similar to Cre−/Nf2fx/fx littermate controls (Figure 8C), all four BAs (n = 3) were developed in the Wnt1-Cre/Nf2cKO mutants (Figure 8D). However, the distance between the lateral edges of each BA was significantly smaller in Wnt1-Cre/Nf2cKO mutants (Figure 8D,E, p < 0.05) compared to those of Cre−/Nf2fx/fx littermate controls (Figure 8C).

Light microscopy images of sections of E10.5 branchial arches (BAs) from a Cre−/Nf2fx/fx littermate control (Ctrl, A) and Wnt1-Cre/Nf2cKO mutant (cKO, B). In situ hybridization was performed using an antisense probe for Nf2. Bottom panel in A and B is high magnification images of the individual BAs. Scale bars: 25 µm. Whole-mount images (bright field in C1, D1, and SEM in C2, D2) of the Cre−/Nf2fx/fx littermate control (C) and Wnt1-Cre/Nf2cKO (D) BAs. Roman numerals I–IV in C1 and D1 represent BAs 1–4. Scale bars: 200 μm. (E) Histogram (X±SD; n=3) to present the width of each BA (between the lateral edges) in Cre−/Nf2fx/fx littermate controls and Wnt1-Cre/Nf2cKO mutants. *p ≤ 0.05, **p ≤ 0.01 compared to the corresponding BA of Cre−/Nf2fx/fx littermate control using two-way ANOVA followed by Fisher's LSD analyses. (F) Western blot bands of p-Yap, Yap and β-actin in the mesenchyme of individual BAs from Cre−/Nf2fx/fx littermate controls and Wnt1-Cre/Nf2cKO mutants. (G) Histograms (X±SD; n=3) to present p-Yap/Yap ratio in the mesenchyme of individual BAs from Cre−/Nf2fx/fx littermate controls and Wnt1-Cre/Nf2cKO mutants. No statistical significance was detected in Wnt1-Cre/Nf2cKO mutants compared to Cre−/Nf2fx/fx littermate control group. (H) Single-plane laser scanning confocal images of sagittal BA sections immunostained with S-phase cell proliferation marker BrdU (green) in Cre−/Nf2fx/fx littermate control (H1-4) and Wnt1-Cre/Nf2cKO mice (H5–8). Scale bars: 50 μm. (I) Histograms (X±SD; n=3) to present the number of BrdU+ cells per mm2 in individual BAs of Cre−/Nf2fx/fx littermate controls and Wnt1-Cre/Nf2cKO mutants. **p ≤ 0.01 compared to the corresponding BA of Cre−/Nf2fx/fx littermate control using two-way ANOVA followed by Fisher's LSD analyses

The downstream transcriptional regulators of Hippo signalling, that is Yap and p-Yap, were both detected in the Wnt1-Cre/Nf2cKO and Cre−/Nf2fx/fx BAs (Figure 8F). No significant differences in Western blot band intensities of p-Yap, Yap and p-Yap/Yap ratios were detected between the BAs of Cre−/Nf2fx/fx littermate control and those of Wnt1-Cre/Nf2cKO mutants (Figure 8F,G; p > 0.05).

Cell proliferation was altered distinctly in different BAs. BAs 1–2 of Wnt1-Cre/Nf2cKO mutants had a significantly lower, while BAs 3–4 had a significantly higher number of BrdU+ cells in Wnt1-Cre/Nf2cKO mutants compared to the Cre−/Nf2fx/fx littermate controls (Figure 8H,I; p < 0.05).

3 DISCUSSION

Our study demonstrated that the absence of Neurofibromin 2 (Nf2)/Merlin in the neural crest (NC) derived-mesenchyme leads to macroglossia at early (E12.5–E13.5) and microglossia at later (E15.5–E18.5) stages of embryonic tongue development. The tongue deformity (i.e. pointed anterior oral tongue and wider posterior oral tongue) along the anteroposterior axis and distinct changes in the Hippo signalling activity accompanied by changes in cell proliferation in different regions of the Wnt1-Cre/Nf2cKO mutant tongue indicate that NC-derived mesenchymal cells in developing tongue respond to the absence of Nf2 in a region- and stage-specific manner. Overall, our results indicated that Nf2 in NC-derived mesenchyme plays an essential role in regulating Hippo signalling activity and cell proliferation for the proper development of tongue shape and size.

3.1 Nf2/Hippo signalling activity in the NC-derived mesenchyme regulates tongue organogenesis in a stage-specific manner

NC cells migrate into the tongue primordium (i.e. branchial arches) at early embryonic stages2, 26, 27 and fully occupy the mesenchyme of the tongue bud.28 Structural and molecular integrity of the NC and NC-derived cells in the tongue primordium is critical for proper tongue organ development. As an example, the absence of primary cilia in NC cells causes aglossia.8 Deficient molecular signalling, such as BMP,2 TGF-β3, 29, 30 and Wnt,4, 5, 31, 32 results in microglossia due to defects in structural organization and cell migration in the developing tongue. In this study, we report the importance of continuous regulation of Nf2/Hippo signalling in NC and NC-derived tongue mesenchymal cells for proper tongue organogenesis.

Nf2, the product of the gene responsible for the disease neurofibromatosis type 2,33 is pivotal for embryo development, including the development of NC34, 35 and oral structures.36 Nf2 is well-known as a suppressor of cell proliferation and tumour.37-39 In many cases, Nf2 activates Hippo signalling, an evolutionarily conserved potent regulator of cell proliferation and organ size.13, 25 The Hippo pathway is driven by a core kinase cascade that includes Mst1/2 and Lats1/2, which in turn phosphorylate and inactivate Yes-associated protein 1 (YAP) and TAZ transcriptional co-activators.13

In most organs, the absence of Nf2/Hippo signalling causes an enhanced organ growth,40, 41 that is liver-specific genetic knockout of Mst1/2, Sav1, Nf2 or over-expression of Yap resulted in an enlarged liver,18, 42-44 deletion of Sav1, Mst1/2 or Lats1/2 caused a hyperproliferation and enlarged heart.45-50 However, our data indicate that Nf2/Hippo signalling regulates tongue organogenesis in a region- and stage-specific manner. The primordia (branchial arches/BAs) of the tongue organ were narrower at E10.5 and lingual swellings were less profound in Wnt1-Cre/Nf2cKO mutants. At E12.5–E13.5, deletions of Nf2 in NC lineage driven by Wnt1-Cre lead to a non-proportionally enlarged tongue, that is, macroglossia, with a widened posterior region and pointed tip. At E15.5-E18.5, in contrast to many other organs, we observed a smaller tongue (microglossia) suggesting that the effects of altered Nf2/Hippo signalling activity are distinct in different organs.

3.2 Stage- and tongue region-specific regulation of Hippo signalling and cell proliferation in response to mesenchymal Nf2 deletion

As aforementioned, Nf2/Hippo signalling serves as a cell proliferation suppressor.51 To understand the stage-specific alterations of tongue shape and size in Wnt1-Cre/Nf2cKO mice, we analysed the levels of Yap and p-Yap and cell proliferation in different tongue regions and tissue types at various stages. We detected regional and dynamic changes of Yap and p-Yap levels over the developmental course and found that cell proliferation was altered in general corresponding to the Yap level as the Yap typically promotes cell proliferation.18-20

At E10.5 BAs of Wnt1-Cre/Nf2cKO mutant and Cre−/Nf2fx/fx littermate control mice, even though individual BAs of Wnt1-Cre/Nf2cKO mutant mice had no significant changes of levels of Hippo signalling downstream transcriptional regulator Yap and deactivated form p-Yap, significantly different amounts of BrdU+ cells suggest that cell proliferation is very sensitive and responded to subtle yet undetectable changes of Yap and p-Yap levels in individual BAs of Wnt1-Cre/Nf2cKO mutant. Consequently, the initial formation of lingual swellings at E11.5 was deficient on the Wnt1-Cre/Nf2cKO mutant as a result of low amount proliferative cells in BAs1–2. It is reasonable to speculate that a low number of proliferative cells in BAs 1–2 and more proliferative cells in BAs 3–4 might be one of the reasons to have a pointed anterior tongue and widened posterior tongue since BAs 1–2 and BAs 3–4 give rise to the anterior and posterior tongue regions9 respectively.

Our thorough phenotypic analyses along with Hippo signalling activities in Wnt1-Cre/Nf2cKO mutants demonstrated that continuous expression of Nf2 in NC-derived mesenchymal cells is required for the proper tongue formation in shape and size. At E12.5, a high Yap level and Yap target genes (Areg, Ctgf, cyclin D1 and Amotl1) that are known regulators of cell proliferation were concurrent with more proliferative cells in the posterior oral tongue where the Nf2 cKO tongue is wider, and low Yap level with fewer proliferative cells in the anterior oral tongue where the tongue is smaller and more pointed. It is important to note that some Yap target genes (Cyr61, Birc5, Ankrd1 and Axl) had different expression levels in contrast to the Yap levels detected in the E12.5 tongue suggesting that Yap might regulate the expression of different target genes distinctly.

Notably, at later stages of tongue development (E15.5–E18.5), the loss of Nf2 in the NC lineage paradoxically resulted in decreased cell proliferation (reduced Yap level), thus resulting in a smaller tongue (i.e. microglossia) in Wnt1-Cre/Nf2cKO mutants compared to the Cre−/Nf2fx/fx littermates. Even though cell proliferation was reduced as a whole in the E15.5 Wnt1-Cre/Nf2cKO mutants, different tissue compartments had significantly different numbers of BrdU+ cells in the Wnt1-Cre/Nf2cKO mutants, for example the increased versus decreased cell proliferation in the shallow and deep layer of mesenchyme respectively. The increased cell proliferation in the mesenchyme layer subjacent to the epithelium represents a direct effect of Nf2 deletion on the NC-derived mesenchymal cells as the cells populated in this region are mostly, if not all, from NC, and as a result, Nf2 cKO in these cells leads to an inactivation of Hippo signalling, hence promotes the cell proliferation.

With regard to the alterations of Hippo signalling (Yap level) and cell proliferation in tongue regions at late embryonic stages (E15.5–E18.5), which are not consistent with conventional Nf2/Hippo/Yap-cell proliferation pathway, it is possible that Nf2 deletion causes changes in other signalling pathways that overwrite the effects of diminished Hippo signalling in the tongue, for example Wnt/β-catenin, TGF-β, Hedgehog.52 Moreover, Wnt5a/non-canonical Wnt signalling has been reported to be important for tongue outgrowth4 and is likely to be involved in the regulatory role of Nf2/Hippo signalling in tongue mesenchyme in promoting tongue formation. Moreover, tongue mesenchyme acts as a scaffold for the organization of migrating myoblasts into the myogenic cord and operates as a niche that releases molecular instructions (e.g. TGF-β2) to direct survival, proliferation and differentiation of myogenic progenitors, as well as patterning of muscle fibres.53, 54 Thus, disrupted myogenesis might be another cause of microglossia observed in Wnt1-Cre/Nf2cKO mutants at E15.5–E18.5. Indeed, our RNA-sequencing data reveal an altered gene expression profile in muscles and multiple molecular signalling that interact with Hippo signalling.55-57

3.3 Mesenchymal deletion of Nf2 alters the development of overlying tongue epithelium

It is intriguing that although the genetic deletion of Nf2 driven by Wnt1-Cre is largely tongue mesenchyme-specific,