RORalpha coordinates reciprocal signaling in cerebellar development through sonic hedgehog and calcium-dependent pathways

Abstract

The cerebellum provides an excellent system for understanding how afferent and target neurons coordinate sequential intercellular signals and cell-autonomous genetic programs in development. Mutations in the orphan nuclear receptor RORalpha block Purkinje cell differentiation with a secondary loss of afferent granule cells. We show that early transcriptional targets of RORalpha include both mitogenic signals for afferent progenitors and signal transduction genes required to process their subsequent synaptic input. RORalpha acts through recruitment of gene-specific sets of transcriptional cofactors, including beta-catenin, p300, and Tip60, but appears independent of CBP. One target promoter is Sonic hedgehog, and recombinant Sonic hedgehog restores granule precursor proliferation in RORalpha-deficient cerebellum. Our results suggest a link between RORalpha and beta-catenin pathways, confirm that a nuclear receptor employs distinct coactivator complexes at different target genes, and provide a logic for early RORalpha expression in coordinating expression of genes required for reciprocal signals in cerebellar development.

Figure 1. RORα has a systematic effect on developmental RNA expression profile in cerebellum

(A) In situ hybridization shows expression of Rora RNA in developing Purkinje cells as early as E12.5 and expressing cell populate the presumptive cortex by E15.5. Caudal is clockwise from top, position of the fourth ventricle is indicated in the first panel. (B) Distribution of F statistics generated by standard two-factor ANOVA indicate significant effects of genotype and time (but not gender) in our multidimensional data set compared to the distribution generated under the null hypothesis of no effect. (C) The number of standard deviations by which the data depart from expectation is plotted. The number of probe sets reaching given significance levels (p-values) for effect of time, Rora genotype, or gender is compared to the distribution of numbers expected by chance under the null hypothesis. (D) Clustergram of genes identified in our analysis indicates the strength of genotype dependence and that the majority of differentially expressed genes do not have a strong heterozygote effect. Color indicates relative expression level among experiments and brightness indicates absolute signal strength. Probe sets with a p-value less than 0.05 for genotype effect by ANOVA, a minimum average difference of 500 according to MAS and a minimum 1.25-fold change are shown. Genes are arranged in order by the product of F-statistic for genotype and approximate fold change. For upregulated genes, only those with consistent patterns across the full perinatal window are shown.

Figure 2. Progressive loss of proliferation markers in external granule layer

(A) Normalized expression data from Affymetrix Mu11K microarrays are plotted as line graphs. Staggerer samples are in red, littermate controls in blue. Average difference values for each probe set are normalized to make the average of all 24 hybridizations equal to 1. Horizontal lines indicate range of values among replicate samples. Multiple lines in a given plot represent overlay of normalized data from independent probe sets on the array. During this perinatal window, only the cells in the external granule layer should contribute in large number to pool of dividing cells in the cerebellum. (B) Paired serial sections from control and staggerer specimens were mounted together on single slides and processed for in situ hybridization. In situ hybridization shows Ccna2 and dUTPase RNA expression is restricted to the EGL within the cerebellum at P2. Note the thinning of the staggerer EGL by this time.

Figure 3. RORα regulates candidate mitogenic factors in Purkinje cells

Line plots of normalized array data from Kitl, Smst, and Gdf10. Q-PCR confirms reduced expression of Gdf10, but increased expression of Smst RNA in staggerer by P0. Difference by genotype has a p-value ≤0.02 for Gdf10 and p≤0.0023 for Smst by 2-sided t-test. In situ hybridization at postnatal day 2 (as well as E15.5 for Gdf10) indicates the relevant cell population and qualitatively confirms altered expression level for Kitl, Smst, and Gdf10, including dramatic increase in Smst-positive cells by P2. Although Gdf10 expression is roughly in the Purkinje cell layer at P2, expression at E15.5 is not consistent with Purkinje cells (compare with Pcp4 pattern in Figure 6), suggesting expression in Bergmann glia or inhibitory interneurons.

Figure 4. Shh signaling disrupted in staggerer

Q-PCR demonstrates reduced expression of Shh, Gli1, and Ptch RNA in staggerer cerebellum. Shh and Gli1 are significantly altered (p ≤0.05) at E15.5, while Ptch is not. All three genes are significantly different at both E17.5 and P0 (p ≤0.05).

Figure 5. Exogenous SHH stimulates staggerer granule cell precursor proliferation

(A) rSHH is sufficient to induce proliferation in sg granule precursors. 300 μm sections from a P4 staggerer were cultured in the absence (left) or presence (right) of rSHH. After 2 days, dividing cells were labeled by BrdU incorporation for 24 hours, then fixed and visualized with a FITC-conjugated antibody under epifluorescence. (B) Diminished proliferation of neonatal sg granule precursors is overcome by rSHH. 300 μm sections of wild-type and staggerer cerebellum at P0 were cultured immediately in media containing BrdU in the presence or absence of exogenous SHH. After 3 days in culture, sections were fixed, incubated with FITC-conjugated anti-BrdU antibody, and visualized by epifluorescence. (C) Fluorescence density was measured across each section in using MetaMorph software. Average values from multiple sections ± s.e.m. are shown. * p ≤0.05; ** p ≤0.01.

Figure 6. RORα regulates Purkinje cell genes involved in calcium signaling

(A) Line plots include normalized array data from Pcp4, Slc1A6, Itpr1, and Cals1. In situ hybridization at postnatal day 2 (E15.5 for Pcp4) indicates the relevant cell population and qualitatively confirms decreased expression. Quantitative real-time PCR at P0 demonstrates reduced expression for Cals1, Slc1A6 and Itpr1. Difference by genotype was significant for all three genes (p≤0.01). (B) SCA1 target genes that are relevant to calcium signaling are decreased in staggerer as well. Q-PCR showed 5-phosphatase levels decreased at P0, when Serca2/Atp2a2 is not reliably detected. Both genes are decreased in adult. Difference by genotype is significant for 5-phosphatase in both P0 (p ≤0.05) and adult (p ≤0.025) as is Serca2 (p≤0.03).

Figure 7. RORα and coactivator binding and activation of target genes

ChIP experiments on freshly dissected cerebella demonstrate in vivo binding of RORα to Pcp4, Pcp2, Slc1A6, Shh (at two sites), and Itpr1 promoters, but not to Baf53a and a second site in the Pcp2 promoter. RORα recruits specific sets of coactivators to each target promoter, including Tip60, β-catenin, and p300. None of the examined coactivators were bound to the Itpr1 promoter. Location of the PCR product relative to the known or presumed start of transcription is indicated for each gene; arrowhead indicates putative RORα binding site. (For the Shh distal site, a space in the image indicates different gels). (B) Co-immunoprecipitation assays were performed on freshly dissected cerebella from P0 mice. β-catenin was pulled down with an RORα antibody but not a control IgG from the same host species. β-catenin was detected with two separate antibodies. (C) Microinjection of blocking antibodies against β-catenin, SRC-1, and Tip60, but not CBP, p/CIP, and p/CAF blocked activation of the Pcp2/LacZ reporter in the presence of a CMV-RORα expression construct.

Figure 8. A model for reciprocal signaling in Purkinje and granule cell differentiation

(A) Speculative model for sequential intercellular signaling events surrounding RORα-stimulated gene expression. Wnt1 is expressed by migratory granule cell precursors (GCPs) at the rhombic lip (RL). WNT expression may influence the nuclear accumulation of β-catenin in Purkinje cells (PC). Nuclear β-catenin acts as a cofactor on RORα-regulated promoters, including Shh. Purkinje cell SHH stimulates proliferation of GCPs in the external granule cell layer (EGL). Granule cells in the internal granule cell layer (IGL) make glutamatergic synapses on Purkinje cells, where RORα also regulates expression of signal transduction molecules to receive and process this input. (B) Model for RORα-dependent gene expression in Purkinje cells. In Purkinje cell nuclei, RORα (red) recruits promoter-specific sets of coactivators (light red) to target genes in the context of additional, independent factors (gray). The products of known direct and potentially direct target genes are involved in receiving signals from afferent cells (blue), processing those signals via calcium release (orange), and the stimulation of proliferation of GCPs in the EGL (SHH).