The ciliary Frizzled-like receptor Tmem67 regulates canonical Wnt/β-catenin signalling in the developing cerebellum via Hoxb5

Abstract

Primary cilia defects result in a group of related pleiotropic malformation syndromes known as ciliopathies, often characterised by cerebellar developmental and foliation defects. Here, we describe the cerebellar anatomical and signalling defects in the Tmem67^tm1(Dgen)/H knockout mouse. At mid-gestation, Tmem67 mutant cerebella were hypoplastic and had aberrantly high canonical Wnt/β-catenin signalling, proliferation and apoptosis. Later in development, mutant cerebellar hemispheres had severe foliation defects and inferior lobe malformation, characterized by immature Purkinje cells (PCs). Early postnatal Tmem67 mutant cerebellum had disrupted ciliogenesis and reduced responsiveness to Shh signalling. Transcriptome profiling of Tmem67 mutant cerebella identified ectopic increased expression of homeobox-type transcription factors (Hoxa5, Hoxa4, Hoxb5 and Hoxd3), normally required for early rostral hindbrain patterning. HOXB5 protein levels were increased in the inferior lobe, and increased canonical Wnt signalling, following loss of TMEM67, was dependent on HOXB5. HOXB5 occupancy at the β-catenin promoter was significantly increased by activation of canonical Wnt signalling in Tmem67^-/- mutant cerebellar neurones, suggesting that increased canonical Wnt signalling following mutation or loss of TMEM67 was directly dependent on HOXB5. Our results link dysregulated expression of Hox group genes with ciliary Wnt signalling defects in the developing cerebellum, providing new mechanistic insights into ciliopathy cerebellar hypoplasia phenotypes.

Figure 1

Tmem67^−/− cerebellar hypoplasia and foliation defects at perinatal and early postnatal stages of development. (a) Whole-mount E15.5 mid-hindbrain images showing dorsal views of the normal cerebellar vermis in the Tmem67^+/+ embryo (white arrowhead) compared to the vermian hypoplasia/aplasia in the Tmem67^−/− mutant (red arrowhead). Scale bar = 1 mm. (b) Upper panel: dorsal view of whole mount P0 midbrain and hindbrain dorsal view of the whole brain and cerebellum at P0, with hemispheric hypoplasia indicated (bracket). Lower panels show magnified insets of the indicated boxes in the upper panel, showing vermian hypoplasia (red arrowhead). Scale bar = 1 mm. (c) H&E stained median sagittal brain sections showing vermian hypoplasia at the indicated developmental stages. The red arrowhead in the Tmem67^−/− E15.5 vermis denotes the thickened elongated superior cerebellar peduncle. At E18.5 note the foliation defects with delayed development (red asterisk) of the preculminate fissure (pc) in Tmem67^−/− animals. At P0 the non-principal fissure (arrowhead) in the anterobasal lobe developed, with formation of lobules I and II in Tmem67^+/+, whereas the Tmem67^−/− mutant did not develop a principal fissure (red asterisk). At P1, the vermis was hypoplastic with delayed foliation, especially in the central lobe (Cen; red asterisk). The axial diameter between the antero-dorsal and postererior lobes of the vermis (dashed lines) is indicated for P1 sections. Scale bar = 100 μm in all panels. (d) Bar graphs quantitate the axial diameter between the antero-dorsal and postererior lobes of the vermis at the indicated ages. Statistical significance of pair-wise comparisons between age-matched animals is indicated by: n.s. not significant; *p < 0.05; ***p < 0.001; for Student’s t-test (paired, two-tailed). Error bars indicate s.e.m. (e) H&E stained parasagittal P1 cerebellar section, showing lack of preculminate (pc) fissure development (red arrowhead), absence (red asterisk) of the inferior lobe (Inl) and posterolateral fissure (Pl) in the Tmem67^−/− mutant. Scale bar = 500μm. Abbreviations: Ab, antero-basal lobe; Ad, antero-dorsal lobe; Cen, central lobe; CH, cerebellar hemisphere; CV, cerebellar vermis; Inl, inferior lobe; MB, midbrain; pc, preculminate fissure; pl, posterolateral fissure; pr, primary fissure; Pol, posterior lobe; sec, secondary fissure; V4, fourth ventricle.

Figure 2

Over-proliferation and increased apoptosis at in the Tmem67^−/− cerebellum at E12.5. Mid-sagittal sections of the E12.5 cerebellar vermis stained for (a) active β-catenin; (b) proliferation (Ki67); (c) mitotic cells (P-H3, phospho-Ser10 histone H3; green, indicated by arrowheads), with bar graph on right to quantitate the number of P-H3⁺ cells per field of view for Tmem67^+/+ (+/+) and Tmem67^−/− (^−/−) cerebella; and (d) activated cleaved caspase-3, with bar graph on right to quantitate the intensity of IHC staining per field of view (AU; arbitrary units). In the wild-type embryonic cerebella, active β-catenin and Ki67 were largely restricted to the rhombic lip (arrowheads in a and b). Note the increase in canonical Wnt signalling (marked by active β-catenin), proliferation (Ki67) and apoptosis (active caspase-3), and the decrease in the number of mitotic cells (P-H3), in Tmem67^−/− sections. Asterisks indicate non-specific staining of red blood cells in (c). Scale bars = 50μm (a–c) and 20μm (d). Abbreviation: cp, choroid plexus; cv, cerebellar vermis; mb, mid-brain; rl, rhombic lip. Statistical significance of pairwise comparisons are indicated by: *p < 0.05; **p < 0.01; for Student’s t-test (paired, two-tailed). Error bars = s.e.m. for a minimum of 4 fields of view in n = 3 biological replicates.

Figure 3

Defects in Purkinje cell morphology, ciliogenesis and Shh signalling in the Tmem67^−/− cerebellum. (a) E18.5 cerebellar tissue with GPC cells stained for Pax6 (red) and PCs stained for calbindin (green). Boxed areas are magnified in the bottom panels, showing details of the cellular architecture of the GPC and PCs. Scale bar = 30μm. (b) E18.5 cerebellar tissue sections for the indicated genotypes stained for calbindin (green), acetylated α-tubulin (red) and nuclei (DAPI; blue). White line indicates the thickness of the external granular layer (EGL). Arrowheads indicate smaller, irregular PCs in the Tmem67^−/− cerebellum. Scale bar = 50 μm. (c) Ciliogenesis defects in perinatal Tmem67^−/− cerebellum in tissue sections stained for acetylated α-tubulin (red; indicated by red arrowheads), γ-tubulin (green) and nuclei (DAPI; blue). Note the less frequent cilia in Tmem67^−/− and the many basal bodies with no apparent ciliary axoneme. Scale bar = 20 μm. (d) Relative expression levels of Shh, Gli1 and Ptch1 in Tmem67^−/− P0 cerebellum compared to Tmem67^+/+ control littermate tissue. Values are the mean of at least n = 3 independent experiments, with qRT-PCR runs performed in duplicate. Statistical significance of pairwise comparisons are indicated by: n.s. not significant; *p < 0.05; **p < 0.01; for Student’s t-test (paired, two-tailed). Error bars = s.e.m.

Figure 4

Differential expression analysis of the Tmem67^−/− post-natal cerebellum identifies a cluster of co-regulated homeobox transcription factors. (a) Heat map representing relative gene expression levels in early postnatal (P0) cerebella from matched littermate Tmem67^+/+ and Tmem67^−/− animals (n = 3 biological replicates) for the indicated genes. All statisitically significant differentially expressed genes (total n = 41; Benjamini-Hochberg p-value adjusted for false discovery rate p_adj < 0.01) in all three data-sets are represented. Up-regulated genes in Tmem67^−/− mutants are coloured in dark-red/brown, with the co-regulated cluster of Hox genes indicated by orange arrowheads. Down-regulated genes are coloured in orange/yellow, with Tmem67 indicated by the brown arrowhead. The DESeq2 output table for significant differentially expressed genes is shown in Supplementary Table S1. (b) Volcano plot displaying the 41 genes differentially-expressed in early postnatal Tmem67^+/+ and Tmem67^−/− cerebella that pass statistical cut-offs of false discovery rate (FDR) p_adj < 0.05 and absolute log₂fold change (FC) > = 1. The vertical axis (y-axis) corresponds to the statsticial significance of mean expression log₁₀(p_adj value) and the horizontal axis (x-axis) displays the log₂ fold-change values. Positive log₂fold-changes (red points) indicate up-regulated expression in the Tmem67^−/− mutant compared to the Tmem67^+/+ wild-type data-sets. Negative log₂fold-changes (cyan points) indicate down-regulated expression in Tmem67^−/−. Genes discussed in the main text are highlighted with coloured text and larger data points.

Figure 5

Homeobox transcription factor Hoxb5 is up-regulated in the Tmem67^−/− post-natal cerebellum. (a) Box plots of gene expression for Tmem67 and Hoxb5 gene expression in wild-type Tmem67^+/+ (+/+) and mutant Tmem67^−/− (−/−) post-natal cerebellum RNA-seq datasets displayed as normalized counts of mapped reads. Whiskers show the minimum and maximum values, horizontal lines show the median and boxes indicate the 1st and 3rd quartiles. (b) Immunohistochemical staining for Hoxb5 in early postnatal (P0) cerebellum, showing increased protein expression in nuclei of mutant compared to wild-type lobes, especially for the inferior lobe (arrowheads). Scale bars are 50 or 100 μm, as indicated. (c) Western blotting of protein lysates from embryonic (E12.5) and perinatal (E19.5) Tmem67^+/+ (+/+) and Tmem67^−/− (−/−) cerebellum for Hoxb5, with β-actin as a loading control. Relative protein level ratios for Hoxb5 and β-actin (loading control) are indicated. Full scans of blots are shown in the Supplementary Information File.

Figure 6

Wnt3a induces aberrant Hoxb5 expression and GCP loss in the early post-natal Tmem67^−/− cerebellum. (a) Representative images of in vitro cerebellar neuronal cultures from Tmem67^+/+ and Tmem67^−/− early postnatal (P0) cerebellae stimulated with either exogenous, purified N-terminal Shh protein or negative control protein (BSA). Cells were stained for Hoxb5 (red), Pax6 to visualize GCPs (green), counterstained with DAPI (blue), and assayed for incorporation of BrdU (grey) to measure cell proliferation. Scale bars = 100μm. (b) Bar graphs to quantitate the percentages of Pax6⁺ nuclei (GCPs; top left panel), Hoxb5⁺ nuclei (top right panel) and BrdU⁺ nuclei undergoing proliferation (bottom panel) in cerebellar neuronal cultures for 10 fields of view in n = 3 independent experiments. (c) Representative images of in vitro cerebellar neuronal cultures from Tmem67^+/+ and Tmem67^−/− early postnatal (P0) cerebellae stimulated with either L cell control or Wnt3a conditioned media, as for (a) above. Scale bars = 50μm. (d) Bar graphs to quantitate the percentages of Pax6⁺ nuclei (GCPs; left panel), and Hoxb5⁺ nuclei (right panel) in cerebellar neuronal cultures for 10 fields of view in n = 3 independent experiments. For all panels, statistical significance of pairwise comparisons are indicated by: n.s. not significant; *p < 0.05; **p < 0.01; for Student’s t-test (paired, two-tailed). Error bars = s.e.m.

Figure 7

Increased canonical Wnt signalling following loss of TMEM67 is dependent on HOXB5. (a) Relative transcript expression levels of TMEM67 (left panel), IFT88 (middle panel) and HOXB5 (right panel), following knockdown of either TMEM67 (siTMEM67) or IFT88 (siIFT88) compared to scrambled negative control (siScr). Statistical significance of pairwise comparisons are indicated by: n.s. not significant; **p < 0.01; for Student’s t-test (paired, two-tailed). Error bars = s.e.m. Values are the mean of n = 3 independent experiments, with qRT-PCR runs performed in duplicate. (b) Increased HOXB5 protein expression following siRNA knock-down of TMEM67 but not IFT88. Relative protein level ratios for HOXB5 and β-actin (loading control) are indicated. (c) HOXB5 nuclear expression (green) is increased following TMEM67 knockdown (green arrowheads). Scale bar = 20 μm. (d) Exogenous over-expression of FLAG-tagged HOXB5 (predicted size 33 kDa), visualized as a double band by immunoblotting (IB) with anti-FLAG (upper panel), increased expression of active β-catenin and total β-catenin. Treatment with Wnt3a conditioned media further increased active β-catenin and total β-catenin levels, compared to L cell control. Relative protein level ratios for HOXB5 and β-actin (loading control) are indicated. Full scans of blots are shown in the Supplementary Information File. (e) TOPFlash reporter assays showing increased canonical Wnt/β-catenin signalling following FLAG-HOXB5 over-expression and treatment with Wnt3a conditioned media, as indicated. (f) Activation of Wnt target genes AXIN2 and DKK1 was dependent on loss of TMEM67 and over-expression of HOXB5. For all panels, statistical significance of pairwise comparisons are indicated by: n.s. not significant; *p < 0.05; **p < 0.01; for Student’s t-test (paired, two-tailed). Error bars = s.e.m. Values are the mean of n = 3 independent experiments.

Figure 8

Hoxb5 occupancy at the β-catenin promoter is increased by activation of canonical Wnt signalling in Tmem67^−/− cerebellar neurones, informing the model of disrupted ciliary signalling in cerebellar hypoplasia. (a) Chromatin immunopreciptations (ChIP) using anti-Hoxb5 and control IgG for in vitro cerebellar neuronal cultures from Tmem67^+/+ and Tmem67^−/− early postnatal (P0) cerebellae, following treatment with either control or Wnt3a conditioned media as indicated. Input and ChIP material was analysed by semi-quantative PCR for the β-catenin (Ctnnb1) promoter (top panel), and an intragenic region (intron 13) of the Ctnnb1 gene as a negative control (bottom panel). Full scans of gels are shown in the Supplementary Information File. (b) Relative levels of the Ctnnb1 promoter quantitated by qRT-PCR, for conditions as in (a). Values are the mean of at least n = 3 independent experiments, with qRT-PCR runs performed in duplicate. Statistical significance of pairwise comparisons are indicated by: *p < 0.05; **p < 0.01; for Student’s t-test (paired, two-tailed). Error bars = s.e.m. (c) Proposed model of disrupted signalling and gene regulatory pathways in the Tmem67^−/− mutant cerebellum (arrows indicate positive modulatory effects; blunt-ended arrows indicate repressive effects). Early in normal embryonic development (up to E14.5), active Wnt/β-catenin signalling is transient and restricted to either the cerebellar rhombic lip (time line indicated at the bottom). At the later perinatal period (E17.5 to P1) Shh stimulates granule cell proliferation, forming the EGL, in the absence of Wnt/β-catenin signalling. TMEM67 normally mediates non-canonical Wnt signalling to repress the expression of homeobox-type transcription factors, including the HOX5 gene group. Loss or mutation of TMEM67 (indicated by red text and symbols) releases the repression of HOX5 genes, causing increased β-catenin expression, dysregulated canonical Wnt/β-catenin signalling in early cerebellar development and aberrant proliferation of granule cell progenitors (upward-pointing red broad arrows). Tmem67 mutation also decreases cilia incidence (downward-pointing red broad arrow) leading to loss of responsiveness to Shh signalling and further loss of correct modulation of neuronal proliferation. The timings of possible feedback loops in Wnt signalling (dashed arrow) remain unknown.