Correct development of the cerebellum requires coordinated sonic hedgehog (Shh) signaling from Purkinje to granule cells. How Shh expression is regulated in Purkinje cells is poorly understood. Using a novel tyrosinase minigene-tagged Sleeping Beauty transposon-mediated mutagenesis, which allows for coat color-based genotyping, we created mice in which the Ski/Sno family transcriptional co-repressor 2 (Skor2) gene is deleted. Loss of Skor2 leads to defective Purkinje cell development, a severe reduction of granule cell proliferation and a malformed cerebellum. Skor2 is specifically expressed in Purkinje cells in the brain, where it is required for proper expression of Shh. Skor2 overexpression suppresses BMP signaling in an HDAC-dependent manner and stimulates Shh promoter activity, suggesting that Skor2 represses BMP signaling to activate Shh expression. Our study identifies an essential function for Skor2 as a novel transcriptional regulator in Purkinje cells that acts upstream of Shh during cerebellum development.
The cerebellum is composed of two principal classes of neurons, Purkinje cells and granule cells, which originate from the roof plate of the metencephalon during early embryogenesis (Wang and Zoghbi, 2001; Sillitoe and Joyner, 2007). During embryonic day 11-13 of mouse development, Purkinje cells generated from the ventricular zone of the fourth ventricle become postmitotic and migrate radially within the developing cerebellar anlage. After birth, Purkinje cells settle into a monolayer where they differentiate and develop extensively arborized dendritic processes. Granule cells, which arise from the rhombic lip, migrate rostrally over the dorsal surface of the cerebellar anlage to form the external granule cell layer (EGL). During the first 2 weeks after birth, cells in the EGL undergo extensive proliferation to produce a granule cell progenitor (GCP) pool that is required for generating a large number of granule cells. Developing GCPs then exit the cell cycle, and migrate internally past the Purkinje cells to form the inner granule cell layer (IGL). Signaling between the granule cells and Purkinje cells is required to orchestrate the proliferation and migration of the granule cells to form the final, highly organized, structure of the mature cerebellum (Jensen et al., 2002). Sonic hedgehog (Shh) secreted from Purkinje cells serves as a potent mitogenic signal that causes the expansion of the GCP population (Wallace, 1999; Wechsler-Reya and Scott, 1999; Dahmane and Altaba, 1999). Deletion of Shh in mouse Purkinje cells disrupts normal cerebellar development, principally by blocking the proliferation of GCPs in the EGL (Lewis et al., 2004). Thus, a tight regulation of Shh expression by Purkinje cells is required for appropriate GCP expansion during cerebellum development. However, mechanisms involved in the control of Shh expression within the Purkinje cells are poorly understood.
Skor2 was first identified by an in silico search for novel genes homologous to Ski/Sno family of transcriptional co-repressors (Arndt et al., 2005). It has also been referred to as FUSSEL18 (functional Smad suppressing element on chromosome 18) (Arndt et al., 2005), or Corl2 (Minaki et al., 2008), owing to its high degree of homology to co-repressor for LBX1 (Mizuhara et al., 2005), of which the human counterpart is termed FUSSEL15 (Arndt et al., 2007). Although Corl1 has been shown to be a co-repressor for LBX1 (Jagla et al., 1995), the role of Skor2 has not been determined. Similar to Ski/Sno, Skor2 possesses two structural domains in the N-terminal region, a DHD (Dachshund homology domain) (Wilson et al., 2004), and an adjacent SAND domain that is necessary for Ski/Sno to interact with Smad4 (Wu et al., 2002). Ski/Sno has been shown to negatively regulate transforming growth factor β (TGFβ)/bone morphogenetic protein (BMP) signaling pathways through binding to Smad proteins (Deheuninck and Luo, 2009). Given the sequence similarity to Ski/Sno, Skor2 may have similar repressive function on TGFβ/BMP signaling pathways (Arndt et al., 2005).
Transposons are mobile genetic elements that can be used as tools for the generation of insertional mutations, as exemplified by the use of the P element in Drosophila genetics (Cooley et al., 1988; Hummel and Klambt, 2008). Although endogenous mouse transposons have not been identified, the Sleeping Beauty (Ivics et al., 1997) and piggyBac transposons have been shown to be functional in mice (Ding et al., 2005; Dupuy et al., 2001; Ivics et al., 2009). In order to simplify the genetic monitoring of transposition in vivo, we tagged the Sleeping Beauty transposon with a tyrosinase minigene. Most albino strains of laboratory mice have a mutation in their endogenous tyrosinase gene (Yokoyama et al., 1990). Therefore, transgenic mice that carry the transposon with a functional copy of tyrosinase will lead to dose-dependent and integration site-sensitive pigmentation, thus allowing for coat-color based genotyping. We have created a series of stable insertional mutants in mouse using this strategy, and have recovered two loss-of-function alleles of Skor2 that exhibit severe defects in Shh expression and cerebellar development.
MATERIALS AND METHODS
RNA in situ hybridization
Sagittal serial sections of brains of wild-type control and Skor2–/– mice were cut with a cryostat, and placed with adjacent sections on separate slides. After paraformaldehyde fixation and acetylation, the slides were assembled into flow-through hybridization chambers and placed in a Tecan Genesis 200 liquid-handling robot (Mannedorf). Templates for synthesis of digoxigenin-labeled riboprobes for Skor2 and Shh are N-terminal 810 bp of Skor2 cDNA and full-length Shh cDNA, respectively. Antisense and sense probes were transcribed with T7 and Sp6 polymerase, respectively, from linearized vector. Hybridized probes were detected by catalyzed reporter deposition using biotinylated tyramide; this was followed by colorimetric detection of biotin with avidin coupled to alkaline phosphatase. Hybridization with sense control probes did not yield signals above background.
Histology and immunohistochemistry
Animals were perfused with 4% paraformaldehyde. Tissues were post-fixed in 4% paraformaldehyde, dehydrated in ethanol, embedded in paraffin, and sectioned at 10 μm. Alternatively, fixed tissues were incubated in 30% sucrose at 4°C overnight, frozen in OCT (TissueTek) and sectioned at 30 μm. For histological analysis, paraffin sections were stained with Hematoxylin and Eosin (Sigma) or with thionin for Nissl staining. Immunohistochemistry on frozen and paraffin sections was performed by incubation overnight at 4°C using the following primary antibodies: anti-calbindin D-28K (Chemicon), anti-phospho-Histone H3 (Chemicon), anti-Zic1 (Rockland). Secondary antibodies used were: goat anti-mouse Alexa Fluor 488 and goat anti-rabbit Alexa Fluor 555 (Invitrogen), which were added for 2 hours at room temperature. Sections were counterstained with Toto3 (Invitrogen) and mounted using Prolong Gold anti-fade reagent (Invitrogen) and images were acquired with a Zeiss LSM 510 confocal microscope. Apoptosis in the developing brain was assessed by the terminal deoxynucleotidyl transferase-mediated dUTP nick end-labeling (TUNEL) assay using DeadEnd Fluorometric TUNEL System (Promega) according to the manufacturer's protocol.
Plasmids, cell culture, reporter assays
Skor2 cDNA was amplified by RT-PCR from mouse cerebellum and cloned into pcDNA 3.1 expression vector (Clontech). Deletion and mutation constructs were generated by site-directed mutagenesis. Id1-Luc (Korchynskyi and ten Dijke, 2002) was kindly provided by Peter ten Dijke (LUMC, The Netherlands). p3TP-lux, pCMV5B-Flag-Smad3 and pCL-Neo HA-hSki were purchased from Addgene (Cambridge, MA). Shh-Luc reporter plasmids containing human Shh promoter (nucleotides 3347 to 1548 of a 1.9 kb human Shh promoter sequence) were kindly provided by Vladimir A. Botchkarev (University of Bradford, UK) (Sharov et al., 2009). HepG2 and HaCat cells were grown in MEM containing 10% FBS. HEK293 and N2a cells were grown in DMEM containing 10% FBS. Cells were transfected using Lipofectamine 2000 reagent (Invitrogen). Cells in 12-well plates were co-transfected with expression plasmids for Skor2 and reporter plasmids (p3TP-lux, Id1-luc or Shh-luc), together with Renilla luciferase vector. Twenty-four hours after transfection, cells were treated for 12 hours with or without 25 ng/ml BMP2 (R&D) and lysed with Passive Lysis Buffer (Promega). The Dual-Luciferase Reporter Assay System (Promega) was used to determine firefly and Renilla luciferase activities according to the manufacturer's instructions. Measurements were performed with a BD luminometer (BD), and firefly luciferase values were normalized to Renilla luciferase values. In all experiments, the internal control plasmid was used to compensate variable transfection efficiencies. All assays were repeated three times, data were pooled, mean ± s.e.m. was calculated, and statistical analysis was performed using unpaired Student's t-test.
Immunoprecipitation and western blotting
HEK293 cells were transfected with an appropriate combination of expression plasmids, and solubilized in a buffer containing 50 mM Tris HCl (pH 7.5), 150 mM NaCl, 1% Nonidet P-40 supplemented with Protease inhibitor cocktail (Roche). Lysates were cleared and incubated with anti-FLAG and anti-HA antibody (Sigma), followed by incubation with protein A/G-Sepharose beads (Amersham Pharmacia Biotech). The beads were washed with solubilization buffer, and the immunoprecipitates were eluted by boiling for 3 minutes in SDS sample buffer [100 mM Tris HCl (pH 8.8), 0.01% Bromphenol Blue, 36% glycerol, 4% SDS] containing 10 mM dithiothreitol and subjected to SDS-gel electrophoresis.
Sleeping Beauty-tyrosinase transposon-based mutagenesis
Transposable elements can be mobilized by the expression of a transposase, which causes transposons to excise and reinsert randomly in the genome. Thus, randomized mutagenesis can be performed by the generation of a parental animal that contains the transposon and transposase, and stable mutations screened in offspring which lack expression of the transposase.
Most albino strains of laboratory mice have a mutation in their endogenous tyrosinase gene. Expression of a tyrosinase reporter leads to gene therapy for albinism, with pigment production in the fur, skin and eyes (Overbeek et al., 1991). In order to simplify the genetic monitoring of transposition in vivo, we have engineered the Sleeping Beauty transposon, which expresses a tyrosinase minigene that confers four particularly useful features (Lu et al., 2007). First, the minigene is expressed at sufficient levels to give visible pigmentation for most sites of integration in the genome. Second, the level of tyrosinase expression is consistent for a given integration site, resulting in hemizygous siblings that exhibit coat colors that are nearly identical to each other and to their hemizygous parents. Third, the level of minigene expression is usually rate limiting for melanin synthesis, resulting in a gene dose effect for each integration site such that homozygous transgenic mice, if viable, are more darkly pigmented than their hemizygous littermates. Fourth, different integration sites give different levels of minigene expression, and yield mice with different coat colors for each different integration site. Therefore, transgenic mice that carry the transposon with a functional copy of tyrosinase will lead to dose-dependent and integration site-sensitive pigmentation, thus allowing for coat color-based genotyping.
The 4.1 kb tyrosinase minigene isolated from plasmid Ty811C (Yokoyama et al., 1990) was inserted into the pT2 version of the Sleeping Beauty transposon (Cui et al., 2002). We also inserted a Gal4-VP16-SV40 cassette in the opposite orientation to the tyrosinase minigene, but detected no expression in the hemizygous or homozygous mutants (see Fig. S1A in the supplementary material). The tyrosinase-tagged transposon DNA (Fig. 1A) was microinjected into one-cell stage inbred albino FVB/N embryos and two pigmented transgenic founder mice were obtained. Upon outbreeding to FVB males, both founder mice produced F1 offspring with two different coat colors, indicating two different integration sites for each founder. The F1 mice were used to establish four different transgenic families. Southern hybridization and fluorescent in situ hybridization to metaphase chromosomes were used to estimate the copy number for the transgenes in each family and the initial chromosomal integration sites (see Table S1 in the supplementary material).
In order to mobilize the transposons from their initial integration sites, we mated the transposon-carrying transgenic mice to transgenic mice that express the transposase for Sleeping Beauty (CAGGS-SB10) (Dupuy et al., 2001). Pigmented male offspring (transposon-positive) were screened for the presence of the transposase transgene to identify males that carried both transgenes. Excision and re-integration of the transposon is thus activated in the germline of these males and new integration events can be isolated from their offspring.
Males carrying either the Sleeping Beauty transposon only, or both the transposase and the Sleeping Beauty transposon were mated to albino FVB females, and the offspring were screened for new coat colors. As expected, all offspring from crosses of the single transgenic Sleeping Beauty transposon alone displayed the identical coat color to the parental transgenic male (Fig. 1B). By contrast, an average of 30% of the offspring from double transgenic males showed a change in coat color (Fig. 1C). Overall, 64 mice with coat color changes were recovered using the four initial integration sites. These mice were bred to FVB partners to remove the transposase and to establish new transgenic lines with stable transposon integration, followed by interbreeding to generate homozygous mice (Fig. 1D). Inverse PCRs [described previously (Yang et al., 2004)] were carried out to obtain molecular information about the transposition events for 52 of the families (see Table S2 in the supplementary material). Sequencing results were used for BLAST searches against the mouse genome database to determine the precise location of each integration site. Putative homozygous mice were mated to albino partners to test for fertility and to confirm homozygosity (Fig. 1E).
In 15 of the 52 transposition events that were molecularly characterized, the transposon was found to have integrated into Gal4 sequences or tyrosinase sequences (see Table S2 in the supplementary material), implying local hopping within the original transgenic arrays. For the other 37 sites, 17 were located within 15 different endogenous mouse genes. Fifteen of these sites were intronic. Two integrations occurred in exons. For four of the 15 genes (Ppp3ca, Ace, Axin2 and Rgs9), mice with targeted mutations have previously been described. However, 11 of the integration sites were located in `new' genes, i.e. genes for which live mice with targeted alleles have not previously been described. Nine of the integration sites produced homozygous mutant phenotypes. Here, we describe our studies of one of the mutated genes isolated from family 1799B. The other integration sites and mutations will be described in later manuscripts.
Transposon integration sites for family 1799B
Founder 1799B had 15 tandem copies of the transposon integrated in chromosome 18B3 (Fig. 1F). New genomic transposition sites were characterized from founder 1799B and are shown in Table 1. Two of the transposition sites (CA3 and CA7) were located in the same predicted gene: Skor2 (NCBI accession number XM_972803.2) (Fig. 1G). The CA3 integration site was located in exon 2. For CA7, the right junction between transgenic and genomic sequences was located in intron 8, right after the stop codon of Skor2; the left junction was located 237 kb upstream from the start codon of Skor2. Integration of the transgenic DNA was accompanied by a deletion of 273 kb in mouse chromosome 18, including the entire coding region of Skor2, but excluding any exons from the predicted adjacent genes (Fig. 1G). Other investigators have noted that concatemeric transposons can produce complicated genomic alterations during transposition (Geurts et al., 2006). Using primers located 5′ upstream of Skor2 gene, we showed that the predicted 196 bp band was present in DNA from wild-type (+/+) or heterozygous (+/–) mice, but absent from DNA taken from mice homozygous (–/–) for the transgene, indicating that the transposition resulted in a deletion (Fig. 1H). Reverse transcriptase-PCR (RT-PCR) analysis of RNA from P0 cerebellum using primers located in exon 2 of Skor2 showed a complete absence of the Skor2 transcript in homozygous mutant mice (Fig. 1I; see Fig. S1 in the supplementary material). These mutants are thus referred to as Skor2 null or Skor2–/–. The same analysis failed to amplify any transcripts containing the Gal4-VP16 sequences (see Fig. S1A in the supplementary material), indicating that Gal4-VP16 is not expressed in Skor2 mutants and cannot contribute to the phenotypes present in the mutant cerebellum. The expressions of known genes adjacent to Skor2, including the upstream Smad2 and the downstream Ler3ip1 and Hdhd2 were assessed by quantitative real-time PCR and the levels were found to be similar between the Skor2-null mutants and the littermate controls (Fig. 1J). Genomic PCR using primers immediately upstream and downstream from the Skor2 integration sites also failed to detect any aberrant bands, indicating that there were no unexpected genomic DNA rearrangements in Skor2-null mutants (see Fig. S1B in the supplementary material). These results, combined with the fact that the same phenotypes were seen in two independent alleles (see below) strongly support the notion that the phenotypes were specifically caused by the disruption of the Skor2 gene.
Homozygous mice in both families CA3 and CA7 were found to display very similar phenotypes. Most (90%) of the homozygotes die within 48 hours of birth, probably owing to defective neuromuscular junction formation and respiration failure (B.W. and H.Z., unpublished). Homozygotes that survive are smaller than their siblings and show an unstable gait throughout the life span. The subsequent studies were performed with mice from strain CA7.
Skor2 deficiency severely disrupts cerebellar development
Given phenotypes indicative of cerebellar anomalies, we performed morphological analysis of the Skor2-null cerebellum at various developmental stages. Nissl staining of sagittal sections from E15 wild-type and Skor2-null embryos revealed equivalent cerebellar development (Fig. 2A, E15). However, in P0 pups, the cerebellum from Skor2-null mice remained as a flat structure, lacking the four principal fissures observed in the control cerebellum. Additionally, the size of the cerebellum was much smaller when compared with the control littermates (Fig. 2A, P0). This phenotype persisted during postnatal development (Fig. 2A, P5) and to adulthood (Fig. 2B-D). Morphological examination of the Skor2-null mice that survived to adulthood revealed that, although the lateral cerebellar hemispheres were relatively unaffected, the central vermis region was severely reduced in size (Fig. 2B). In addition, analysis of serial sagittal sections through the vermis revealed an abnormal foliation pattern (Fig. 2C). Lobes 4-5 failed to develop entirely, whereas lobes 6-8 did not divide normally into the sublobules seen in wild-type littermates. Although the Skor2-null mutant maintained a normal laminar structure, the molecular layer of the cerebellar cortex was thinner compared with the wild-type cerebellum (Fig. 2D). Examination of histological sections from other brain regions did not reveal any gross morphological abnormalities (see Fig. S2 in the supplementary material).
Skor2 mediates Purkinje cell differentiation and granule cell proliferation
We analyzed Skor2 expression in both the developing and adult CNS by in situ hybridization. At E15, strong expression was detected in the central region of the cerebellum where primitive Purkinje cells are located (Fig. 3A,B). At P0, we observed predominant expression of Skor2 in developing Purkinje cell layers (PCL). Interestingly, higher levels of Skor2 expression were observed in the rostral lobes compared with caudal lobes (Fig. 3C). In situ hybridization of whole adult mouse brain revealed that Skor2 was exclusively expressed in the cerebellum (see Fig. S3 in the supplementary material). Examination of higher magnification images showed that Skor2 mRNA is restricted to Purkinje cell bodies (Fig. 3E). The specificity of the probe was confirmed by the absence of hybridization signal in Skor2-null brain slices (Fig. 5A). Overall, our results demonstrate that Skor2 expression is specific to developing and adult Purkinje neurons in the cerebellum.
The prominent expression of Skor2 in Purkinje cells prompted us to ask whether phenotypes seen in Skor2-null mutants are due to an intrinsic defect in Purkinje neurons. We performed immunofluorescence labeling of brain slices with the Purkinje cell marker calbindin D-28K at P0, P10 and 2 months of age (Fig. 3F). At P0, Purkinje cells were present in multiple layers in both the control and Skor2–/– mutant (Fig. 3F, P0), indicating that Skor2 is not required for the specification or migration of Purkinje cells. At P10, wild-type and Skor2-null Purkinje cells both settled into a monolayer and developed dendritic processes. However, the depth of the molecular layer in which the dendrites of the Purkinje cells project was significantly reduced when compared with the control (Fig. 3F, P10), suggesting that the maturation and differentiation of Purkinje cells is defective in the absence of Skor2. This phenotype persisted into adulthood (Fig. 3F, 2 months). Quantification of the molecular layer thickness showed that there was a 32.3% and 27.2% thinning in this layer in Skor2-null mice at P10 and 2 months, respectively.
Purkinje cells migrate along the Bergmann glia fibers and form a single layer postnatally. The absence of ectopic Purkinje cells in the Skor2–/– cerebellum indicates that Skor2 is not required for normal positioning of Purkinje cells. Further examination of Bergmann glia morphology by immunostaining for GFAP showed ordered and linear morphology of glial fibers extending to the pial surface in the Skor2 mutant, indistinguishable from the wild-type control (see Fig. S4 in the supplementary material), suggesting that Bergmann glia differentiation is not affected in the Skor2-null mutants.
Given that Purkinje cell signaling is known to be required for the proliferation and expansion of the GCPs, we performed phospho-Histone H3 staining to visualize the mitotic, immature granule cell precursors at P0 and P5. Indeed, compared with wild-type controls, there was a significant reduction in the number of proliferating cells in the EGL of Skor2–/– mice at both stages (Fig. 4A,B). Using the Zic1 transcription factor as a marker for differentiating and mature granule neurons, we observed abundant granule cells expressing Zic1 in both the EGL and IGL of the control and Skor2 mutant animals (Fig. 4C), indicating normal granule cell localization and maturation in the absence of Skor2. Likewise, the levels of apoptosis measured by TUNEL staining were similar in the control and Skor2–/– mutant (Fig. 4D,E). The data combined provide strong support for the notion that reduced GCP proliferation resulting from defective Purkinje cell signaling is the cause for the smaller cerebellum observed in the Skor2-null mutant.
Skor2 positively regulates sonic hedgehog expression
Shh is secreted from Purkinje cells and acts as a potent mitogenic signal to expand the GCP population. We performed in situ hybridization to determine the levels of Shh expression in the Skor2 mutant and control cerebellum. Consistent with earlier reports (Lewis et al., 2004; Corrales et al., 2004), in wild-type animals Shh is predominantly expressed in Purkinje cells with higher levels in rostral lobes than in caudal regions (Fig. 5A). In sharp contrast, Shh expression in Purkinje cells was greatly diminished in Skor2 mutant cerebellum (Fig. 5A), demonstrating that proper Shh expression is dependent on Skor2. To further confirm that this is associated with impaired Shh signaling, we performed real-time RT-PCR analysis using P0 cerebellum from control and Skor2 mutant mice, and found significant reduction in the expression levels of Shh and Shh target gene Gli1, but not of Gli2 or Gli3, in Skor2–/– cerebellum, suggesting that Skor2 plays a role in regulating specific Shh downstream targets (Fig. 5B).
We next performed in vitro luciferase assay using a Shh promoter-luciferase reporter construct (Sharov et al., 2009). Transfection of a Skor2 expression vector led to dose-dependent increase in Shh luciferase activity (Fig. 5C), demonstrating that Skor2 positively regulates Shh transcription. Together, these findings suggest that decreased Shh signaling from Purkinje to granule cells in the absence of Skor2 leads to cerebellar hypoplasia.
Skor2 transcriptional activity is correlated with nuclear localization and HDAC binding
Next, we investigated the biochemical and functional properties of Skor2 by transfecting a deletion series of Flag- or Gal4-tagged Skor2 expression vectors in cell culture (Fig. 6A,B), and monitoring their subcellular localizations (Fig. 6C; see Fig. S5 in the supplementary material) and transcriptional activities (Fig. 6D), respectively. Expression of the deletion mutants was validated by western blotting (Fig. 6B). Immunostaining revealed that full-length Skor2 is localized to the nucleus (Fig. 6C, Skor2), as is construct expressing the first 385 amino acids of the protein (Fig. 6C, 1-385). However, expressing the first 256 amino acids of Skor2 leads to its localization in both the nucleus and cytoplasm (Fig. 6C, 1-256), indicating that amino acids 256-385 of Skor2 contains a nuclear localization signal (NLS). Analysis of the primary sequences within this region using PSORTII (http://www.psort.org/) revealed that the KRPR residues at amino acid 306-309 could potentially serve as a NLS. We, accordingly, deleted the KRPR sequence from the full-length Skor2 (Skor2ΔKRPR) and found that, indeed, this mutant resulted in a predominantly cytoplasmic expression (Fig. 6C). We therefore conclude that the KRPR motif spanning residues 306-309 of Skor2 directs its nuclear targeting.
The sequence similarity of Skor2 to the Ski/Sno family of co-repressors prompted us to assess whether Skor2 has a repressive function on transcription. We co-transfected Gal4-fused full-length Skor2 or various deletion mutants (Fig. 6A, #1-#7) with a reporter construct containing the GAL4 DNA-binding site and the Elb promoter upstream of the luciferase sequence (Cao and Sudhof, 2001), and measured the luciferase activities of the transfected cells (Fig. 6D). We found that full-length Skor2 potently repressed basal transcription (Fig. 6A,D, #2), as was construct deleting the DHD and HAND domains (Fig. 6A,D, #7), demonstrating that the repression activity is conferred by the C-terminal sequences. Transfection of the C-terminal deletion constructs (Fig. 6A,D, #3-#6) showed that, interestingly, deletion mutant #4, which contains the NLS but lacks other C-terminal domains, failed to suppress luciferase activity (Fig. 6D, #4), suggesting that nuclear localization is not sufficient for the repressor function of Skor2. The full repression by expressing amino acids 1-592 of Skor2 (Fig. 6D, #5) indicates that sequences within amino acids 385-592 are also crucial for the repression activity.
Because Ski and Sno have been shown to complex with nuclear co-repressors (N-CoRs) and HDACs to repress transcription (Akiyoshi et al., 1999), we hypothesized that Skor2 may exert repression activity through association with HDACs. We co-transfected HEK293 cells with Flag-tagged Skor2 and HA-tagged HDAC1 expression plasmids and performed immunoprecipitation using anti-HA and anti-Flag antibodies. We found that Skor2 and HDAC1 strongly interact (Fig. 6E,F). By performing interaction assays using a panel of Skor2 deletion mutants, we narrowed down the HDAC1 interaction domain to residues 385 and 592 of Skor2 (Fig. 6G). As this region is also important for transcription repression (Fig. 6D, #4 versus #5), the results combined suggest that interaction with HDAC1 may be required for the repressive activity of Skor2.
Skor2 represses BMP-mediated transactivation
Having established the repressor function of Skor2 in a heterologous reporter system, we next sought to test whether it mediates TGFβ or BMP signaling similar to Ski. Previous studies have shown that the N terminus of Skor2 binds Smad2/3 and inhibits TGFβ signaling (Arndt et al., 2005). In order to ascertain the effect of full-length Skor2 on TGFβ signaling, we co-transfected Skor2 with a TGFβ-responsive promoter fused to luciferase (p3TP-Lux) in Hep3B cells (Wrana et al., 1992) (Fig. 7A, –Smad3). A subset of the cells was also co-transfected with Smad3 to activate the TGFβ promoter (Fig. 7A, +Smad3). Expression of Ski was used as a positive control (Ski), whereas transfection with the empty vector was used as the negative control (Ctrl). Both the basal and Smad3-induced TGFβ reporter activities were suppressed by co-expressing the TGFβ co-repressor Ski (Fig. 7A, Ski). However, Skor2 failed to suppress either the basal or Smad3-activated TGFβ signaling pathway (Fig. 7A, Skor2), suggesting that Skor2 cannot repress TGFβ-induced transcription. We then tested the ability of Skor2 to repress BMP-dependent transcriptional activity using Id1-Lux, a BMP responsive reporter (Korchynskyi and ten Dijke, 2002), in the absence or presence (+ BMP2) of exogenous BMP2 expression (Fig. 7B) (Rios et al., 2004). Similar to that of Ski, expression of Skor2 in Hep3B cells potently suppressed BMP2 signaling under both basal and BMP2-activated conditions (Fig. 7B). These data suggest that Skor2 selectively inhibits BMP- but not TGFβ-dependent transcriptional activation.
As BMP signaling activates the Smad family of proteins, we tested whether Skor2 interacts with Smads 1, 2, 3 and 4 by performing immunoprecipitations on lysates from co-transfected cells. We detected strong interactions between Skor2 and Smad 1 and 3, and weaker interactions with Smad 2 and 4 (Fig. 7C-F), suggesting that Skor2 may mediate transcriptional co-repression through physical association with Smad1 and Smad3.
Using Sleeping Beauty transposon tagged with a tyrosinase reporter, we demonstrate here that transposition events can be identified by simple visual inspection of the offspring from bigenic males, providing a valuable tool to uncover genes that play crucial, but heretofore uncharacterized, roles in normal vertebrate development. Transposition and generation of new insertional mutations requires only a two generation breeding strategy. There is no need for micromanipulation of mouse embryos to generate new integration sites. Identification of each transposition event is accomplished by visual inspection of the mice, making screening for insertional mutations straightforward and reliable. Additionally, owing to the dose-dependent expression of the tyrosinase minigene reporter, homozygous mice can be visually distinguished from hemizygous animals. Even though transposition usually results in reintegration near the original integration site (Keng et al., 2005), a significant fraction of the integration sites can result in new insertional mutations. Here, we describe one of the insertion mutants generated, the Skor2 loss-of-function animal that displays defects in cerebellar development. Loss of Skor2 leads to defective Purkinje cell development, a severe reduction of granule cell proliferation and a malformed cerebellum.
Skor2 contains DHD and SAND homology domains characteristic of the Ski/Sno family of transcription co-repressors. Although Skor2 shares structural similarities, it is unique in several aspects. First, in contrast to the widespread expression of Ski/Sno in multiple tissues including the CNS (Reed et al., 2001), Skor2 is specifically expressed in cerebellar Purkinje cells in the brain. Consequently, Skor2 deletion results in a cerebellum-specific phenotype, whereas the forebrain structure is preserved. Second, different from Ski/Sno, which are co-repressors of both TGFβ and BMP pathways, Skor2 represses only BMP but not TGFβ signaling, even though it binds to Smads associated with both BMP and TGFβ pathways. Third, Skor2 has weak binding to Smad4. Ski/Sno have been shown to bind to Smad4 through its SAND domain, which is an evolutionarily conserved sequence motif involved in chromatin-dependent transcriptional regulation (Bottomley et al., 2001). Crystal structure of the Ski SAND domain in complex with the MH4 domain of Smad4 established that the highly conserved structure motif (I loop) from Ski/Sno proteins is essential for binding to Smad4 (Wu et al., 2002). Comparison of the Skor2 SAND domain with that of Ski revealed that the essential hydrogen bonds in the I loop, which provides the specificity for the recognition of Smad4 by Ski, are divergent in Skor2 (see Fig. S6 in the supplementary material). The zinc-binding motif (CCHH) within the SAND domain of Ski, which contributes to its structure stability (Wu et al., 2002), is preserved in Skor2, indicating that Skor2 shares the conserved feature of zinc binding. It is worth pointing out that the biochemical assays performed here are based on in vitro overexpression of tagged proteins. The physiological formation of the Skor2 complexes in vivo remains to be confirmed.
Shh produced from Purkinje cells functions as a potent mitogenic molecule to promote cerebellar granule progenitor proliferation. Differential signaling by Shh across the cerebellar cortex, resulting from its uneven distribution along the rostral to caudal areas of the Purkinje cell layer, leads to the differential growth of the EGL and the complexity of the cerebellar foliation (Corrales et al., 2004; Corrales et al., 2006; Lewis et al., 2004). Our findings that Skor2 and Shh are co-expressed in Purkinje cell layers in almost identical patterns, that Shh expression is diminished in Skor2-deficient cerebellum, and that Skor2 potently activates the Shh promoter demonstrate that Skor2 is a crucial upstream regulator of Shh expression in Purkinje cells. The differential expression of Skor2 in rostral and caudal regions suggests that Skor2 might be one of the genetic cues responsible for establishing the Shh signaling gradient to determine the final size and shape of the cerebellum. The severely reduced size and compromised foliation of Skor2-null cerebellum support this conclusion.
In addition to Skor2, previous studies on staggerer mice have implicated retinoid-related orphan receptor α (RORα) as a positive regulator of Shh expression (Gold et al., 2003). Whether Skor2 regulated Shh transcription involves the RORα pathway is not clear. Our studies suggest that the transcriptional activation of Shh by Skor2 is mediated through a repression of BMP signaling involving Smad and HDAC complexes. The negative regulation of Shh expression by BMP has been previously suggested in a number of developmental and neoplastic contexts (Bastida et al., 2009; Zhang et al., 2000; Piedra and Ros, 2002; Sharov et al., 2009). We propose the existence of a similar antagonistic effect of BMP on Shh expression in Purkinje cells with Skor2 as an upstream regulator. The suppression of BMP signaling by Skor2 during normal cerebellum development ensures sufficient Shh expression and granule cell proliferation. This model is supported by the finding that, even though BMP ligands such as BMP2, BMP4 and BMP7 are abundantly expressed in all three layers of the developing cerebellar cortex (Rios et al., 2004), BMP signaling in Purkinje cells detected by phospho-Smad immunostaining is very low under physiological conditions (Qin et al., 2006), suggesting the presence of negative regulators. We performed phospho-Smad1/5/8 staining and found no detectable difference between Skor2-null and control cerebellum (see Fig. S7 in the supplementary material). Combining the in vivo result with our in vitro Skor2 localization studies, we propose that Skor2 modulates BMP signaling in the nucleus, downstream of Smad phosphorylation, and by actions involving binding of both Smads and HDAC. Although the exact mechanism is not clear, our data indicate that Skor2 act as a negative regulator to suppress BMP signaling, which in turn allows Shh expression in Purkinje cells. Nevertheless, it remains possible that Skor2 directly activates the Shh promoter independently of BMP signaling.
In conclusion, using a novel Sleeping Beauty transposon mutagenesis strategy coupled with coat color-based genotyping, we created mutant alleles with a disrupted Skor2 locus. We reveal here an essential physiological function for Skor2, a new member of the Ski/Sno family of transcriptional co-repressors, in cerebellar development. We demonstrate that Skor2, probably via a HDAC- and Smad-dependent co-repressor function on BMP signaling, is a regulator of Shh expression in Purkinje cells and thereby exerts a crucial influence over the proliferation of granule cells during cerebellum development.
We thank C. Thaller and A. Liang and R. Atkinson for assistance in in situ hybridization and confocal imaging, respectively, and the Eunice Kennedy Shriver Intellectual and Developmental Disabilities Research Center (IDDRC HD024064) at BCM for supporting the core facilities. We are grateful to D. Largaespada for the CAGGS-SB10 line, to P. ten Dijke for Id1-luc, to V. Botchkarev for Shh-luc, to A. Geurts for tyrosinase minigene constructs and to N. Justice for critical reading of the manuscript.
This work was supported by grants from NIH (AG020670 and AG032051). Deposited in PMC for release after 12 months.
Competing interests statement
The authors declare no competing financial interests.
Supplementary material for this article is available at http://dev.biologists.org/lookup/suppl/doi:10.1242/dev.067264/-/DC1
- Accepted August 10, 2011.
- © 2011.