Abstract
Schwann cells are the main glial cell type in the PNS. They develop along nerves during embryogenesis and rely on the HMG domain containing Sox10 transcription factor for specification, lineage progression, and terminal differentiation. Sox10 deletion in immature Schwann cells caused peripheral nerve defects in mice that were not restricted to this glial cell type, although expression in the nerve and gene loss were. Formation of the perineurium as the protecting sheath was, for instance, heavily compromised. This resembled the defect observed after loss of Desert hedgehog (Dhh) in mice. Here we show that Sox10 activates Dhh expression in Schwann cells via an enhancer that is located in intron 1 of the Dhh gene. Sox10 binds this enhancer in monomeric form via several sites. Mutation of these sites abolishes both Schwann-cell-specific activity and Sox10 responsiveness in vitro and in transgenic mouse embryos. This argues that Sox10 activates Dhh expression by direct binding to the enhancer and by increasing Dhh levels promotes formation of the perineurial sheath. This represents the first mechanism for a non-cell-autonomous function of Sox10 during peripheral nerve development.
Introduction
Sox10 regulates several aspects of vertebrate nervous system development (Stolt and Wegner, 2010). In the CNS, it is selectively expressed in oligodendrocyte lineage cells in which it affects several stages of development and is indispensable for terminal differentiation and myelination (Stolt et al., 2002; Finzsch et al., 2008). In the forming PNS, Sox10 influences the generation of several cell types, including sympathetic and sensory neurons (Kim et al., 2003; Carney et al., 2006; Reiprich et al., 2008). Even more important is its requirement for the specification of all glia, including Schwann cell precursors along nerves (Britsch et al., 2001; Paratore et al., 2001).
In Schwann cells, Sox10 is expressed throughout development (Kuhlbrodt et al., 1998a). After its role in specification (Britsch et al., 2001), it continues to be functional as Schwann cell precursors develop first into immature Schwann cells and then via promyelinating to myelinating Schwann cells. This lasting requirement has been confirmed by the analyses of mice with hypomorphic Sox10 alleles and mice in which a loxP-flanked Sox10 allele was deleted during specific stages of Schwann cell development by Cre recombinase (Schreiner et al., 2007; Finzsch et al., 2010; Bremer et al., 2011; Fröb et al., 2012). Although the essential role of Sox10 for Schwann cell development is evident, much remains to be learned about its mode of action. One of the key issues is the identification of stage-specific Sox10 target genes that are themselves important for peripheral nerve development.
Sox10 activates the expression of peripheral myelin genes during terminal differentiation, including Mag, Mbp, Mpz, Pmp22, and Connexin 32 (Peirano et al., 2000; Bondurand et al., 2001; Jones et al., 2007, 2011). Sox10 is also directly involved in the activation of Krox20, which is a key activator of peripheral myelin genes (Ghislain and Charnay, 2006; Reiprich et al., 2010). Available evidence suggests that Sox10 first activates and then cooperates with Krox20 to activate the myelination program (Svaren and Meijer, 2008).
Two other Sox10 targets are the genes for the POU domain transcription factor Oct6 and the receptor tyrosine kinase subunit ErbB3 (Jagalur et al., 2011; Prasad et al., 2011). Oct6 is required for Schwann cells to progress to the pro-myelinating stage and to prepare for terminal differentiation (Bermingham et al., 1996; Jaegle et al., 1996), whereas ErbB3 acts as part of the ErbB2/ErbB3 heterodimer that allows Schwann cells to respond in various phases of their development to neuregulin 1 (Newbern and Birchmeier, 2010). Although these known targets help to explain several Sox10 functions, they are by no means sufficient to fully understand the mechanisms by which Sox10 influences peripheral nerve development.
We have discovered recently that Sox10 deletion in immature Schwann cells also led to a dramatic decrease in Dhh levels (Finzsch et al., 2010). Here we show that Dhh is also a direct target gene of Sox10 in Schwann cells and that, through its influence on Dhh, Sox10 exerts a non-cell-autonomous effect on perineurial cells and the formation of the perineurium.
Materials and Methods
Plasmids.
The Dhh genomic region analyzed in this study was located between positions 98720506 and 98739191 of mouse chromosome 15 (mouse genome version mm9). From this region, fragments were obtained by PCR or using restriction nucleases, as indicated in Figures 1A, 2A, and 3A, and inserted into reporter plasmids. For autofluorescence-based reporter gene studies, pTATA–tdTomato was used. It carried the red fluorescing GFP derivative tdTomato under control of the β-globin minimal promoter. For luminescence-based reporter gene assays, Dhh–luc was used. This plasmid is based on pGL2basic (Promega) and contained the firefly luciferase gene downstream of the Dhh minimal promoter (positions −277 to +31 relative to the transcriptional start site). In the context of the E4–Dhh–luc construct (see Fig. 3A), potential Sox10 binding sites were mutated using the QuickChange XL Site-Directed Mutagenesis kit (Stratagene). The eukaryotic pCMV5-based expression plasmids for full-length Sox10 (pCMV5–Sox10) and the short MIC version of Sox10 (pCMV5–Sox10 MIC) have been described as well as the pCAGGS–Sox10–IRES–nls–GFP plasmid in which Sox10 and GFP reading frames are under control of the chicken β-actin promoter (Kuhlbrodt et al., 1998b; Cossais et al., 2010). For knockdown experiments, pSuper–Neo–GFP plasmids (Oligoengine) were used that either expressed a Sox10-specific shRNA (targeted region, 5′-CTGCTGTTCC TTCTTGACCT TGCCC-3′) or a control shRNA. E4–lacZ transgene constructs contained the E4 fragment in wild-type or mutant version (see Fig. 7C) from the Dhh locus in front of the Dhh minimal promoter followed by a lacZ reporter gene and poly(A) cassette.
Cell culture, transient transfection, extract preparation, electrophoretic mobility shift assays, and reporter gene assays.
HEK293 cells were maintained in DMEM containing 10% fetal calf serum (FCS) and transfected by the polyethylenimine technique using 10 μg of pCMV5-based expression plasmid per 100 mm plate. At 48 h after transfection, cells were harvested for extract preparation (Schlierf et al., 2002). After verification of ectopic expression by Western blotting, electrophoretic mobility shift analyses (EMSAs) were performed (Küspert et al., 2011) using 32P-labeled 25 bp double-stranded oligonucleotides containing putative Sox10 binding sites.
For reporter gene assays, rat S16 Schwann cells were maintained in DMEM containing 10% FCS and transfected with SuperFect Transfection Reagent (Qiagen) on 24-well tissue culture plates. For autofluorescence-based assays, cells were transfected with 500 ng of pTATA–tdTomato-based reporter plasmids and scored for fluorescence 48 h after transfection using a Leica inverted microscope (DMIRB) equipped with a cooled MicroMax CCD camera (Princeton Instruments). For luminescence-based assays, 500 ng of Dhh–luc-based reporter plasmids were used in the presence of 100 ng of pCMV5- or pSuper–Neo–GFP-based expression vectors. Cells were generally harvested 48 h after transfection, except for knockdown experiments in which analysis took place 72 h after transfection. Luciferase activity was determined in the presence of luciferin substrate by detection of chemiluminescence.
Chromatin immunoprecipitation.
Chromatin immunoprecipitation (ChIP) assays were performed on S16 cells as described previously (Schlierf et al., 2006). After crosslinking proteins to DNA in the presence of 1% formaldehyde, chromatin was prepared and sheared to an average fragment length of 200–500 bp using a Sonoplus HD2070 homogenizer (Bandelin). Immunoprecipitations were performed overnight at 4°C using guinea pig antiserum against Sox10 (1:400 dilution) (Maka et al., 2005) as well as guinea pig preimmune serum and protein A Sepharose beads. Quantitative PCR was performed on input and precipitated chromatin after crosslink reversal and purification. The following primer pairs were used at an annealing temperature of 63°C: 5′-GCAGCCAAGATAACTGTGGC-3′ and 5′-GCAGTACAATGGCCATTCTC-3′ for E4a; 5′-CAATGCCCAGTGCCAGGGAAG-3′ and 5′-CTCCCAGCGTTTGGGAGTCG-3′ for E4b; 5′-GGCAGAGAGCTGGGATTGTC-3′ and 5′-CCAGGGTTGGCCTATACACG-3′ for Con1; and 5′-CAATTGACATATGCCAGCCC-3′ and 5′-GATCACACATCTAAGGCCTC-3′ for Con2.
Generation and analysis of transgenic animals.
Transgenic mouse embryos were generated by microinjecting the E4–lacZ transgene in wild-type or mutant form (E4m3–lacZ and E4mAll–lacZ; see Fig. 7C) as an NheI/KpnI fragment into the male pronucleus of fertilized F1 (C57BL/6 × CBA) oocytes according to standard techniques. Foster mothers were killed when embryos were at 14.5 d postcoitum. Transgenic embryos were identified and genotyped by PCR on embryonic tail DNA using 5′-GCCCAGGAAGATAGTTTGGTG-3′ and 5′-GATAGGTTACGTTG GTGTAGATGG-3′ as primers under standard PCR conditions.
The E4–lacZ transgene was injected as plasmid into the neural tube of live chicken embryos at Hamburger–Hamilton stage 10–11 in the presence of pCAGGS–IRES–nls–GFP-based expression plasmids and electroporated as described previously (Cossais et al., 2010). Successfully electroporated embryos were identified 48 h later by GFP expression and dissected.
After fixation, mouse and chicken embryos were cryoprotected in sucrose and frozen at −80°C in tissue freezing medium (Leica). After transverse sectioning on a cryotome, β-galactosidase activity was detected on 20 μm sections by incubation in 1% X-gal at 37°C. For immunohistochemistry, 10 μm cryotome sections were incubated with anti-β-galactosidase goat antiserum (1:500 dilution; Biotrend), anti-Sox10 guinea pig antiserum (1:1000 dilution) (Maka et al., 2005), or anti-GFP mouse monoclonal antibody (1:100 dilution; Roche). Secondary antibodies conjugated to Cy2, Alexa Fluor 488, and Cy3 immunofluorescent dyes (Dianova and Molecular Probes) were used for detection. Sections were analyzed and documented with either a Leica TCS SL confocal microscope or a Leica MZFLIII stereomicroscope equipped with an Axiocam (Carl Zeiss).
Results
The first intron of the Dhh gene has enhancer activity in S16 Schwann cells
Previous studies had shown that Dhh is not only strongly expressed in the male gonad but also in the developing peripheral nerve in which it is predominantly produced from Schwann cells (Bitgood and McMahon, 1995; Parmantier et al., 1999). Loss of this paracrine signaling molecule in mice leads to massive defects of the perineurium, the cellular sheath that surrounds and protects the endoneurium with its axon–Schwann cell units and blood vessels (Parmantier et al., 1999). It also prevents invasion of immune cells into the nerve. The perineurium consists of several layers of flattened fibroblasts that secret collagens and other extracellular matrix molecules as well as poorly characterized signaling factors. Many of the extracellular matrix molecules end up in the epineurium and thereby help to form the outer nerve cover. The signaling molecules from perineurial cells influence many aspects of nerve development. In the absence of Dhh, the perineurium no longer provides an effective barrier between surrounding tissue and nerve. It appears thin, less compacted, and disorganized, with patchy basal lamina and abnormal tight junctions (Parmantier et al., 1999). In Dhh-deficient mice, perineurial cells divide the nerve into mini-fascicles. Alterations in perineurium-derived signals cause defects in myelinating as well as non-myelinating Schwann cells. They lead to an abnormal permeability of the blood–nerve barrier and elevated levels of immune cells in the nerve (Sharghi-Namini et al., 2006). Similar consequences have also been observed in humans in which Dhh loss leads to a peripheral neuropathy with mini-fascicle formation and gonadal dysgenesis (Umehara et al., 2000).
Here we analyzed whether Dhh is a direct target of Sox10. Transcription of the Dhh gene has not been studied, and the regulatory region that mediates its expression in Schwann cells has not yet been identified. However, work by Jaegle et al. (2003) had shown that 19 kb of genomic sequence around the Dhh gene are sufficient to achieve expression of a Cre transgene in Schwann cells. This region encompasses all three Dhh exons as well as upstream and downstream sequences. To identify the part responsible for Schwann-cell-specific expression, we divided the Dhh locus in eight overlapping fragments (Fig. 1A) and combined each with the β-globin minimal promoter (Fig. 1B). All eight combinations were then tested for their ability to drive expression of a tdTomato reporter gene in the S16 Schwann cell line using autofluorescence of transfected cells as readout (Fig. 1C). Combinations of the same β-globin minimal promoter with the U3 enhancer of the Sox10 gene or with the enhancer of the Mpz gene served as controls and verified that the assay is sensitive enough to detect the activity of Schwann-cell-specific enhancers (LeBlanc et al., 2007; Werner et al., 2007).
Scanning the Dhh genomic region for an enhancer. A, Topology of the Dhh genomic region and localization of the fragments F1–F8 in this 19 kb. Restriction enzymes used for generation of the fragments are indicated. B, Scheme of the reporter construct. Fragments F1–F8 were inserted in front of the β-globin minimal promoter, followed by tdTomato and SV40 poly(A) signal. C, At 48 h after transient transfection of the reporter plasmids, S16 cells were scored for expression of the reporter by detection of tdTomato-specific autofluorescence. The experiment was repeated three times in duplicates. Reporter plasmids that carried the U3 enhancer or the Mpz enhancer in combination with the β-globin minimal promoter served as positive control. +, Autofluorescence detected; −, no autofluorescence.
In contrast to the two known Schwann cell enhancers, none of the four fragments from the upstream region of the Dhh gene (F1–F4; Fig. 1A) exhibited autofluorescence (Fig. 1C). Equally inactive as the upstream fragments was F5. This fragment spanned the immediate upstream region, exon 1, and part of intron 1 and contained the Dhh promoter. It follows that the promoter is not sufficient to secure more than basal expression in S16 cells. In contrast, autofluorescence was observed with fragment F6 (Fig. 1A,C). This contained all three exons of the Dhh gene, the intervening introns, and the immediate downstream region. A comparable autofluorescence was also obtained with a fragment that corresponded to intron 1 (F7; Fig. 1A,C), whereas a fragment representing the complete downstream region remained inactive (F8; Fig. 1A,C). These results point to enhancer activity in intron 1.
Autofluorescence was fairly weak and is inconvenient for quantification. We were furthermore concerned that enhancer activity may be suboptimal in combination with the heterologous β-globin promoter. Therefore, we switched to a luciferase reporter assay and exchanged the β-globin promoter for a 308-bp-long Dhh minimal promoter (−277 to +31) (Fig. 2B). This minimal promoter increased luciferase activity barely above levels observed with the promoterless construct. High activity was observed in S16 cells not before the Dhh minimal promoter was combined with the U3 enhancer. This construct served again as positive control (Fig. 2C). In contrast, combination of the Dhh minimal promoter with the Sox10 D7 enhancer remained inactive, in agreement with its lack of activity in Schwann cells (Werner et al., 2007).
Mapping enhancer activity to intron 1 of the Dhh gene. A, Schematic representation of fragments F9–F16 and their location relative to the Dhh gene. Restriction sites used for cloning are indicated at the ends of each fragment. F14 was generated by PCR. B, Reporter constructs consisted of fragments F9–F16 inserted in front of Dhh minimal promoter (positions −277 to +31, Dhh), luciferase gene, and SV40 poly(A) signal. C, Transient transfections of these reporter constructs were performed in S16 cells. Luciferase activities in extracts from transfected cells were determined 48 h after transfection in three experiments each performed in duplicates. The luciferase activity obtained for a reporter plasmid containing only the minimal Dhh promoter (Dhh) was arbitrarily set to 1. Activities in the presence of additional Dhh regions were calculated relative to minimal promoter activity and are presented as mean ± SD. A reporter in which the Dhh minimal promoter was combined with the U3 enhancer served as positive control, one in which it was combined with the D7 enhancer as negative control. −, Reporter plasmid without the minimal promoter.
As observed for combinations with the β-globin promoter, none of the fragments from the Dhh upstream region (F9–F12; Fig. 2A) nor any from the downstream region (F15 and F16; Fig. 2A) showed activity in S16 cells when combined with the Dhh promoter (Fig. 2C). Activity was instead confined to F13 and F14. These fragments strongly overlapped and contained most of the sequences from intron 1, confirming that this intron likely contains the enhancer element.
The Dhh enhancer is confined to the second half of intron 1
To determine more exactly the location of the enhancer, we generated subfragments of F14 (Fig. 3A) and tested them again in S16 cells for their ability to induce expression of a luciferase reporter (Fig. 3B). Division of F14 into a larger 2.8 kb fragment spanning the first two-thirds and a smaller 1.4 kb one containing the remaining one-third led to the two, equally inactive F17 and F18 fragments. In contrast, enhancer activity was preserved in a 2.8 kb fragment that corresponded to the hind two-thirds of F14 (Fig. 3B).
Fine mapping of the Dhh enhancer to an evolutionary conserved region in the hind part of intron 1. A, Location of fragments F14, F17–F19, and E1–E5 relative to exons I–III of the Dhh gene and the region with highest evolutionary conservation (ECR) in the gene. Restriction sites used for cloning are indicated. All other fragments were amplified by PCR. B, C, Luciferase reporter constructs containing one of the F (B) or E (C) fragments inserted in front of the Dhh minimal promoter were transiently transfected in S16 cells, and luciferase activities were determined 48 h after transfection in three experiments each performed in duplicate. The luciferase activity obtained for a reporter plasmid containing only the minimal Dhh promoter (Dhh) was arbitrarily set to 1. Activities in the presence of additional Dhh regions were calculated relative to minimal promoter activity and are presented as mean ± SD.
This F19 fragment furthermore strongly overlapped with a region that exhibits conservation among mammals and reaches from the middle of intron 1 into exon 3 (Fig. 3A, ECR). Conservation is, however, confined to mammals and not seen in other vertebrates. Restricting our search to the evolutionary conserved part of F19, we generated five more fragments (Fig. 3A, E1–E5) and combined them with the Dhh minimal promoter in reporter plasmids to assess their enhancer activity in S16 cells. Of these five fragments, all except E5 exhibited enhancer activity (Fig. 3C). Activity varied considerably, with highest levels being found in E4. Therefore, we regard the 927-bp-long E4 region in intron 1 as the core of the Dhh enhancer.
The Dhh enhancer is responsive to Sox10
Because Dhh expression may be under control of Sox10 in Schwann cells (Finzsch et al., 2010), we wanted to find out whether activity of the newly identified Dhh enhancer depends on Sox10. To interfere with the function of endogenous Sox10 in S16 cells (Reiprich et al., 2010) and see how this affects activity of the Dhh enhancer, we chose a dominant-negative approach. It has been shown that transfection of a Sox10 version truncated immediately behind the DNA-binding HMG domain such as Sox10 MIC (Kuhlbrodt et al., 1998b) suppresses the function of cotransfected wild-type Sox10 (Inoue et al., 2004; Prasad et al., 2011). As a consequence, reporter gene expression in S16 cells should be lowered by cotransfected Sox10 MIC for those constructs that contain Sox10-responsive sequences of the Dhh locus. A luciferase reporter containing U3 enhancer and Dhh minimal promoter helped to test the concept (Fig. 4A). Its activity in S16 cells was on average 16-fold lower in the presence of Sox10 MIC. Activity of the Dhh minimal promoter alone, in contrast, was refractory to Sox10 MIC.
Sox10 responsiveness of the Dhh enhancer. Luciferase reporter constructs containing a fragment from the Dhh locus (see Figs. 2A, 3A) inserted in front of Dhh minimal promoter and luciferase gene were transiently transfected in S16 cells in the presence of a Sox10 MIC expression plasmid or an empty expression plasmid. Luciferase activities were determined 72 h after transfection in three experiments each performed in duplicate. The luciferase activity obtained for a reporter plasmid in the presence of empty expression plasmid was arbitrarily set to 1. Activities in the presence of the Sox10 MIC expression plasmid were calculated for each luciferase reporter in relation to it and are presented as mean ± SD. A, Fragments F9–F16; −, reporter plasmid without the minimal promoter. B, Fragments F17–F19. C, Fragments E1–E5.
When the genomic Dhh fragments that were used to map the enhancer were tested for their responsiveness toward Sox10 MIC in transfected S16 cells by luciferase assays, many showed a slight reduction in activity. However, only those fragments that contained the enhancer were repressed more than threefold. These included fragments F13 and F14 from the original screen (Fig. 4A), as well as fragments F19 and E1–E4 from the fine mapping (Fig. 4B,C). These findings therefore support the assumption that Sox10 regulates Dhh expression in Schwann cells at least partly through this newly identified Dhh enhancer.
Intron 1 of the Dhh gene contains potential binding sites for Sox10
Activation of the Dhh enhancer may involve direct binding of Sox10. Intron 1 of the Dhh gene contains eight potential Sox binding sites, site 1 through site 8 (Fig. 5A). These were defined as sites that either conform to the Sox consensus 5′-(A/T)(A/T)CAA(A/T)G-3′ or deviate from this consensus at no more than one position outside the central CAA core. Only site 8 carried two mismatches at the first and last positions. Each of the sites was tested in the context of a larger oligonucleotide in EMSA with control and Sox10-containing extract. Site C/C′ from the Mpz promoter (Peirano et al., 2000) helped to identify DNA complexes with Sox10 monomers and dimers by their respective mobility (Fig. 5B). Of the eight potential binding sites present in intron 1, site 1, site 4, and site 7 bound Sox10 strongly. Weaker binding was additionally detected to site 2 and site 3. All of these sites bound Sox10 as monomers. Dimer binding was not detected.
Mapping Sox10 binding sites in intron 1 of the Dhh gene. A, Location of the eight potential Sox10 binding sites in intron 1 and relative to the E4 fragment. B, EMSA with radiolabeled double-stranded oligonucleotides encompassing one of the potential Sox10 binding sites as indicated below the gels. Oligonucleotides were incubated in the absence (−) or presence (E, S) of protein extracts before gel electrophoresis as indicated above the lanes. Extracts were from mock-transfected HEK293 cells (E) or HEK293 cells expressing full-length Sox10 (S). Oligonucleotides with site C/C′ from the Mpz promoter (Peirano et al., 2000) served as positive control for Sox10 binding and as marker for the mobility of complexes containing either Sox10 monomers (m) or dimers (d). C, EMSA was performed with wild-type and mutant versions (m) for binding sites 3, 4, and 7. Oligonucleotides were incubated in the presence of protein extracts from mock-transfected HEK293 cells (E) or HEK293 cells expressing full-length Sox10 (S). The position of the Sox10-specific complex is highlighted (m). The asterisk marks a nonspecific complex.
Because fragment E4 was the region from intron 1 with strongest enhancer activity, we concentrated our analysis on those sites that were present in this core region and bound by Sox10. These were site 3, site 4, and site 7. Mutations were introduced into the CAA core of each of these sites, and the consequence of these mutations on binding was then tested in EMSA. As evident from Figure 5C, all mutations effectively abrogated Sox10 binding to the respective sites.
Mutations were introduced in the context of E4 into each of these sites as single mutations or in combinations, and the consequences on enhancer activity were assessed in luciferase reporter gene assays in transfected S16 cells (Fig. 6A). Compared with wild type, all E4 mutants exhibited dramatically reduced enhancer activity. Wild-type E4 enhanced the activity of the Dhh minimal promoter 150-fold, but activation rates were down to sixfold for the site 3 mutant (Fig. 6A, E4m3). The site 4 mutation (E4m4) reduced activation rates to ninefold, and the site 7 mutation (E4m7) allowed a residual 14-fold activation. E4 mutants with simultaneous mutation of two sites exhibited even lower activation rates. These results argue that E4 enhancer activity in S16 cells depends on the three identified Sox10 binding sites.
Functional relevance of Sox10 binding sites in the Dhh enhancer. A, Luciferase reporter constructs containing the E4 fragment in wild-type or mutant versions were transiently transfected in S16 cells, and luciferase activities were determined 48 h after transfection. Mutant E4 versions carried inactivating mutations in Sox10 binding site 3 (m3), site 4 (m4), site 7 (m7), or combinations thereof (m3m4, m3m7, m4m7, mAll). The luciferase activity obtained for a reporter plasmid containing only the minimal Dhh promoter (Dhh) was arbitrarily set to 1. Activities in the presence of E4 were calculated relative to minimal promoter activity and are presented as mean ± SD. B, The same luciferase reporter constructs were also transfected in S16 cells in the presence of a Sox10 MIC expression plasmid or an empty expression plasmid. The luciferase activity obtained for a reporter plasmid in the presence of empty expression plasmid was arbitrarily set to 1. Activities in the presence of the Sox10 MIC expression plasmid were normalized for each luciferase reporter to this value and are presented as mean ± SD. C, Additionally, transfections of luciferase reporters were performed in the presence of Sox10-specific shRNA and scrambled shRNA. Luciferase activities were determined, and the ratio of activities in the presence of Sox10-specific shRNA versus scrambled shRNA was calculated. Values were then normalized to the relative activity of the reporter plasmid containing only the minimal promoter and are presented as mean ± SD. All experiments were performed three times in duplicates.
We also analyzed to what extent the three mutations altered E4 responsiveness toward Sox10 by cotransfecting all E4 mutants with the dominant-negative Sox10 (Fig. 6B). Sox10 MIC not only effectively reduces the activity of wild-type E4 but also the remaining activity of E4m4, E4m7, and the E4m4m7 double mutant. In contrast, the residual activity of the E4m3 mutant or all E4 double and triple mutants with mutant site 3 were refractory to Sox10 MIC.
Very similar results were also obtained when S16 cells were transfected with the E4 reporter plasmids in the presence of Sox10-specific shRNAs (Fig. 6C). The activity of the Dhh minimal promoter was only marginally affected by the presence of Sox10–shRNA or scrambled shRNA. For reporter plasmids that additionally contained the U3 or the E4 enhancer, activity was reduced to a residual 5–20% in the presence of Sox10–shRNA. Similar to the wild-type, several E4 mutants were responsive to the Sox10–shRNA, arguing that they still contain residual functional Sox10 binding sites. These included E4 versions with site 4, site 7, and combined site 4/7 mutations. In contrast, site 3 mutations abolished this responsiveness when present alone or in combination with other mutations (Fig. 6C). Both cotransfections with Sox10-specific shRNA and Sox10 MIC therefore lead us to conclude that site 3 is most important among the three Sox10 binding sites present in E4.
Considering that we detected Sox10 binding to three sites and found mutation of these sites to reduce activity and Sox10 responsiveness, it is reasonable to assume that Sox10 exerts its stimulatory effect on E4 by direct binding. To verify this assumption, we performed ChIP from S16 cells with Sox10-specific antibodies (Fig. 7A,B). These antibodies readily precipitated chromatin fragments from the E4 enhancer, leading to their enrichment in the precipitate relative to two control fragments from the upstream region of the Dhh gene (Fig. 7B). No such enrichment was observed with control antibodies. Interestingly, the enrichment was much higher for the E4 region that contained site 3 and site 4 than for the region with site 7 (Fig. 7B, compare E4a, E4b). ChIP thus confirms the presence of Sox10 on the E4 enhancer in S16 Schwann cells.
In vivo requirement of Sox10 binding for Dhh activity. A, Schematic representation of the location of regions from the Dhh locus probed by PCR during ChIP studies, including the front (E4a, positions +2363 to +2779) and hind (E4b, positions +2929 to +3293) part of E4 and two control fragments from the distal upstream region of the Dhh gene (Con1, positions −118661 to −119032; Con2, positions −78544 to −78886). B, ChIP was performed on S16 cells using antibodies directed against Sox10 (α-Sox10) and control preimmune serum (PI). Quantitative PCR was applied on the immunoprecipitate. Values for each fragment correspond to the percentage of material precipitated from the input. C, Schematic representation of the transgenic constructs consisting of E4 in wild-type (E4) or mutant (E4m3 and E4mAll) versions, the minimal Dhh promoter (Dhh), the lacZ marker gene (lacZ), and an SV40 poly(A) signal (pA). D–F, Analysis of transgenic mouse embryos. D, LacZ activity was detected colorimetrically using X-gal substrate on transverse sections of 14.5-d-old transgenic mouse embryos. Only spinal nerve (marked by dotted lines) is shown. The number of embryos that exhibited X-gal staining in the nerve relative to the total number of transgenic embryos obtained for E4–lacZ (Wt), E4mAll–lacZ (mAll), and E4m3–lacZ (m3) by pronucleus injection are given on the left. Scale bar, 50 μm. E, Coimmunohistochemistry was performed on transverse sections of 14.5-d-old E4–lacZ mouse embryos using antibodies directed against β-galactosidase (in red) in combination with antibodies directed against Sox10 (in green). The confocal image is confined to spinal nerve. The inlay shows a higher resolution. Scale bars, 10 μm. F, Summary of mouse embryos obtained for each transgene and analyzed during this study including the determined number of transgene copies present on average in each embryo per cell. G–J, Analysis of electroporated chicken embryos. Chicken embryos were electroporated with the E4–lacZ in the presence of pCAGGS–IRES–nls–GFP (G, H) or pCAGGS–Sox10–IRES–nls–GFP (I, J). Electroporated neural tube regions were identified 48 h later on transverse sections by unilateral GFP expression (green) using DAPI as a nuclear counterstain (blue) (G, I). LacZ activity was detected on immediately adjacent sections using X-gal substrate (H, J). Scale bar (in G): G–J, 200 μm.
Schwann-cell-specific activity of the Dhh enhancer depends on Sox10 in vivo
Although the ChIP results in S16 cells are suggestive, they do not prove that E4 is truly active as a Sox10-dependent Schwann cell enhancer in vivo. To confirm enhancer activity in vivo, we generated transgenic constructs in which E4 was placed in front of the Dhh minimal promoter and a lacZ reporter gene (Fig. 7C, E4-lacZ). Injection of the E4–lacZ transgene into the male pronucleus of fertilized oocytes allowed us to generate five transgenic embryos (Fig. 7F) that were analyzed at embryonic day 14.5 by X-gal staining and immunohistochemistry (Fig. 7D,E). At the time of analysis, the embryos contained on average one to nine copies of the transgene per cell as determined by quantitative PCR (Fig. 7F). Four of the five embryos exhibited staining along the peripheral nerve. There was some variation in staining intensity and in the number of stained cells. Nevertheless, staining along the nerve was clearly detectable. Immunohistochemistry confirmed that β-galactosidase was present in the cytoplasm of those cells along the nerve that contain Sox10 in their nuclei and are thereby defined as Schwann cells (Fig. 7E). We therefore conclude that E4 is able to target transgene expression to Schwann cells in vivo.
In addition to transgenes with wild-type E4, we also injected transgenes in which all three Sox10 binding sites (E4mAll–lacZ) or site 3 (E4m3–lacZ) were mutated (Fig. 7C). Of the four transgenic embryos obtained for E4mAll–lacZ, none exhibited significant activity in peripheral nerves (Fig. 7D,F). Similarly, none of the seven embryos obtained for E4m3–lacZ at embryonic day 14.5 exhibited expression of the lacZ reporter along the nerve (Fig. 7D,F). The activity observed for the lacZ transgene in vivo thus crucially depends on E4 and not the minimal promoter, and within E4 on the identified Sox10 binding sites, in particular on site 3.
Finally, we electroporated the E4–lacZ transgene into the neural tube of early chicken embryos in the presence of GFP as electroporation marker. Considering that the neural tube at this stage consists of highly proliferative, Sox10-negative neuroepithelial cells, the lack of activity of the lacZ transgene was not surprising (Fig. 7G,H). However, lacZ activity became readily visible in the electroporated half of the neural tube after coelectroporation of Sox10 (Fig. 7I,J), arguing that Dhh enhancer activity indeed depends on this transcription factor in vivo.
Discussion
In this study, we have mapped the regulatory region that is likely responsible for Dhh expression in Schwann cells by a combination of cell culture and transgenic studies. Schwann-cell-specific expression was not mediated by the promoter but rather by an enhancer within the first intron of the Dhh gene. The enhancer was localized in the hind part of intron 1, and the E4 region constituted its core.
It is unclear at present whether E4 is identical to the enhancer or whether surrounding sequences additionally contribute to enhancer activity. Two observations argue for a role of E4 flanking sequences in enhancer activity. Although E4 effectively directed reporter gene expression to Schwann cells in transgenic embryos, expression levels were quite variable. This is a consequence of different copy numbers and integration sites in the transgenic embryos. However, the fact that activity is fairly sensitive to these parameters may indicate that E4 lacks some of the elements that stabilize enhancer activity in situ. Additionally, E4 is embedded in a larger region that exhibits sequence conservation among mammalian species, and this conservation may reflect the exact dimensions of the enhancer.
It is also noteworthy that conservation was only observed among mammals. Already in monotherians, the conserved region is much smaller than in metatherians and confined to 190 bp of E4 surrounding site 3 and site 4. No conservation was detected in other vertebrates in agreement with the fact that Dhh is a hedgehog gene specific to mammals (O'Hara et al., 2011).
We have also shown that the Dhh enhancer and its E4 core contain binding sites for Sox10 and that they respond to its presence. As a consequence, deletion of the Sox10 binding sites led to a dramatic decrease of enhancer activity in cultured cells and abolished Schwann-cell-specific activity in vivo. Considering further that Sox10 was detected on this enhancer by ChIP and that the presence of Sox10 allows this enhancer to function in neuroepithelial cells after electroporation in vivo, there is strong evidence that Sox10 contributes to the Schwann-cell-specific activity of the Dhh enhancer by directly binding to these sites.
Among the three sites identified, site 3 is the most important for Sox10-dependent activity, despite a rather modest Sox10 binding in vitro when compared with site 4 and site 7. This shows that binding strength in vitro is not a completely reliable predictor for functional importance in vivo. Several reasons for the discrepancy can be envisaged. It is possible that the accessibility of the sites is different in chromatin in vivo compared with naked DNA in vitro. Some of the sites may be more effectively masked in nucleosomes than others. Binding of other transcription factors to the enhancer has to be taken into account as well. Competitive binding to overlapping sites may reduce accessibility, whereas transcription factor binding to adjacent sites may allow additional contacts between the two bound proteins and thereby facilitate binding. In this respect, it is intriguing that there is a potential binding site for Krox20 adjacent to site 3. Both sites are separated by 11 bp, and the distance between their centers corresponds exactly to two helical turns. This is closer than most previously identified composite binding elements for Sox10 and Krox20 (Jones et al., 2007). It also differs from previously identified sites in that it combines the Krox20 site with a monomeric rather than a dimeric Sox10 binding site. Additional sites for Klf, Nfat, and YY1 proteins are a bit farther away. Future experiments will have to probe their relevance.
It is also worth mentioning that all of the identified sites bind Sox10 as monomer. Similar to the other SoxE proteins Sox8 and Sox9, Sox10 can bind to DNA as monomer or dimer (Wegner, 2010), and it has been shown for at least two Sox10-responsive enhancers that a dimer site cannot simply be replaced by a monomer site without a significant loss in enhancer activity (Peirano and Wegner, 2000; Jagalur et al., 2011). It has even been postulated that dimeric binding is such a crucial feature that it can be used to predict Sox10-dependent neural-crest enhancers (Antonellis et al., 2008). For Schwann-cell-specific enhancers that are activated by Sox10, there seems to be no strict reliance on dimeric binding. Some Schwann-cell-specific enhancers only contain dimer sites and crucially depend on them (i.e., the Oct6 SCE) (Jagalur et al., 2011), whereas others contain only monomeric sites (i.e., the Dhh enhancer), yet others carry both monomeric and dimeric sites and need both for their activity (i.e., Krox20 myelinating Schwann cell enhancer, U3 enhancer of the Sox10 gene) (Reiprich et al., 2010; Wahlbuhl et al., 2012). Considering that Sox10 monomers affect the overall topology of the enhancer in a different way than dimers, the most likely explanation is that each enhancer contains the type of site that is best suited at its specific position to guarantee the multiprotein complex formation of the enhanceosome.
The identification of Dhh as a direct target gene also provides the first mechanistic explanation for a non-cell-autonomus function of Sox10. By regulating Dhh expression in Schwann cells, Sox10 influences the development of perineurial cells and is responsible for formation and proper function of the perineurium. Loss or inactivation of Sox10 should therefore be associated invariably with defects in the perineurial sheath. The compromised blood–nerve and tissue–nerve barriers should furthermore lead to elevated immune cell numbers in the nerve, which in turn may evoke a chronic inflammation. The fact that these changes are indeed observed in mouse models with Sox10-deficient Schwann cells (Finzsch et al., 2010) confirms the relevance of this effector–target gene relationship for normal nerve development. Disturbance of Dhh signaling to the perineurial cells may also occur in patients with Sox10 mutations and contribute to disease manifestation and pathology in humans.
Footnotes
This work was supported by Deutsche Forschungsgemeinschaft Grant We1326/9 (M.Weg.). Sarah Homann is acknowledged for help with transgenesis.
- Correspondence should be addressed to Dr. Michael Wegner, Institut für Biochemie, Emil-Fischer-Zentrum, Friedrich-Alexander Universität, Fahrstrasse 17, 91054 Erlangen, Germany. m.wegner{at}biochem.uni-erlangen.de
