Expression of Connexin47 in Oligodendrocytes is Regulated by the Sox10 Transcription Factor

J. Mol. Biol. (2006) 361, 11–21

doi:10.1016/j.jmb.2006.05.072

Expression of Connexin47 in Oligodendrocytes is Regulated by the Sox10 Transcription Factor Beate Schlierf, Torsten Werner, Gabi Glaser and Michael Wegner⁎ Institut für Biochemie, Universität Erlangen-Nürnberg, D-91054 Erlangen, Germany

In the central nervous system, Connexin32 and Connexin47 are confined to oligodendrocytes where they contribute to myelin formation and maintenance, and are essential for establishing a functional glial syncytium that ensures ionic homeostasis. Despite their importance, not much is known about the regulation of connexin gene expression in oligodendrocytes. Here, we identify group E Sox proteins, in particular Sox10, as essential transcriptional regulators of both connexins. Not only was expression of Connexin32 and Connexin47 severely compromised in spinal cords of mouse mutants with reduced amounts of group E Sox proteins. Sox10 also stimulated in transient transfections the Connexin32 promoter as well as Connexin47 promoter 1b which is the main Connexin47 promoter active in the postnatal spinal cord. Detailed characterization of Connexin47 promoter 1b identified a single monomer binding site that mediated Sox10-dependent promoter activation. The region containing this binding site was also occupied by endogenous Sox10 in 33B oligodendroglioma cells. These results add Connexin47 and Connexin32 to the list of Sox10 target genes and argue that Sox10 may influence transcription of many terminal differentiation and myelination genes in oligodendrocytes as an essential regulatory component of the myelination program. © 2006 Elsevier Ltd. All rights reserved.

*Corresponding author

Keywords: HMG; Sox; Sry; Krox-20; oligodendrocyte

Introduction Gap junctions represent clustered channels that allow ions and small molecules to pass across adjacent plasma membranes from one cytoplasmic compartment into another.1 They are formed by connexins of which there are approximately 20 different types. A particular gap junction channel can contain more than a single connexin, and the connexins present in each channel determine its permeability, conductance and mode of regulation. Glial cells of the central nervous system are linked into a functional syncytium by gap junctions. They express at least six different connexins: three in astrocytes and three in the myelin-forming oligodendrocytes. 2 Of the three connexins expressed in oligodendrocytes, Connexin29 is found in the innermost layer of myelin. 3 , 4 Connexin47, on the other hand, is Abbreviations used: HMG, high-mobility-group; siRNA, small interfering RNA. E-mail address of the corresponding author: [email protected]

predominantly present in oligodendrocyte-toastrocyte gap junctions and participates in linking the two types of glial cells. 5 Although oligodendrocyte-to-astrocyte gap junctions also contain Connexin32, this third connexin is primarily involved in forming reflexive gap junctions between myelin layers in Schmidt–Lanterman incisures and between paranodal loops of one and the same oligodendrocyte.5 It has been proposed that Connexin32 is needed to take up excess extracellular potassium ions released during neuronal activity at nodes of Ranvier along myelinated nerves, and that Connexin47 is consecutively involved in the redistribution of these ions and associated osmotic water throughout the glial syncytium to guarantee ionic homeostasis in the central nervous system. Supporting their importance in oligodendrocytes, mice deficient for Connexin47 and Connexin32 exhibit profound myelin abnormalities and oligodendrocyte cell death leading to action tremors, tonic seizures and finally death by postnatal week six.6,7 In spinal cord and brain, Connexin47 and Connexin32 are first detected in significant amounts at the end of the first postnatal week. 6 , 8 Their

0022-2836/$ - see front matter © 2006 Elsevier Ltd. All rights reserved.

12 expression in oligodendrocytes thus parallels the expression of many myelin genes. Co-expression of connexins and myelin genes has also been observed in the peripheral nervous system. Here, myelinating Schwann cells have similar functions as oligodendrocytes in the central nervous system. Their major connexin is Connexin32.9,10 In human patients, its mutation leads to demyelination of the peripheral nervous system in the X-chromosome linked form of Charcot-Marie-Tooth disease.9 Similar disease phenotypes have also been reported for mutations in the two transcription factors Krox-20 and Sox10.11,12 These transcription factors do not only appear to be general regulators of myelin gene expression in the peripheral nervous system,13,14 they also regulate the activity of the Connexin32 promoter.15 In contrast to Krox-20, Sox10 is also significantly expressed in myelinating oligodendrocytes.16,17 In the absence of Sox10, spinal cord oligodendrocytes fail to express myelin genes and fail to undergo terminal differentiation. 18 Sox10 belongs to the Sox family of transcription factors which all contain a high-mobility-group (HMG) box as their DNAbinding domain.19,20 Among the 20 mammalian Sox proteins, Sox10 is most closely related to the other group E Sox proteins Sox8 and Sox9, which are co-expressed with Sox10 during certain stages of oligodendrocyte development.21 Terminally differentiating oligodendrocytes, for example, express Sox8 in addition to Sox10.22 Sox8 appears to have similar functions as Sox10 in these cells21 but terminal differentiation of oligodendrocytes is less disturbed in the absence of both Sox8 alleles than in the absence of a single Sox10 allele. Accordingly, mice without Sox8 and with a single Sox10 gene copy have an oligodendrocyte defect in their spinal cord intermediate between mice with a single and no functional Sox10 allele.22 So far, very little is known about the regulation of connexin gene expression in oligodendrocytes. Here we show that Sox10 regulates expression of Connexin32 and Connexin47 in oligodendrocytes on the transcriptional level. This further extends our knowledge on the function of Sox10 in glia and at the same time identifies for the first time a transcriptional regulator of Connexin47.

Results Spinal cord oligodendrocytes require group E Sox proteins for expression of Connexin47 and Connexin32 Terminally differentiated oligodendrocytes in the mouse spinal cord express Connexin32 and Connexin47. Expression of both connexins is difficult to detect at the time of birth (data not shown). At postnatal day seven, however, both connexins were readily visualized by in situ hybridization in wildtype spinal cords in the forming white matter region (Figure 1(a) and (b)), although levels were lower than those of MBP and PLP (Figure 1(c) and (d)).

Sox10 Regulates Connexin47 Expression

Figure 1. In situ hybridization studies of Connexin32 and Connexin47 expression in early postnatal spinal cords of mice with combined Sox10 and Sox8 deletions. Transverse spinal cord sections from the forelimb region of wild-type mice ((a)–(d)) and Sox8−/−, Sox10+/− littermates ((e)–(h)) were hybridized at postnatal day seven with specific antisense riboprobes to detect expression of Connexin47 ((a) and (e)), Connexin32 ((b) and (f)), PLP ((c) and (g)) and MBP ((d) and (h)). All transcripts analyzed were severely reduced or absent in the mutant spinal cord. (i) Detection of Connexin47 transcripts in spinal cords of wild-type mice and Sox8−/−, Sox10+/− littermates by reverse transcription and 29 consecutive cycles of PCR amplification. Connexin47 transcripts were detected in toto (2) or according to the first exon (1a, 1b, 1c, 1d) present in the transcript. cDNAs were normalized to RPL8 levels.

As Sox10-deficient mice are not viable and die during birth,23 we could not analyze postnatal connexin gene expression in this genotype. However, mice with only a single functional Sox10 allele and additional loss of both Sox8 alleles live through the

Sox10 Regulates Connexin47 Expression

13

first postnatal week and exhibit a similar, but slightly less severe defect in spinal cord oligodendrocytes.22 When spinal cords of one week old Sox10+/− , Sox8−/− pups were analyzed by in situ hybridization, they not only exhibited severely reduced transcript amounts for the myelin genes MBP and PLP (Figure 1(g) and (h)), but also lacked expression of Connexin32 and Connexin47 (Figure 1(e) and (f)). Quantitative RT-PCR confirmed that expression of both connexins was reduced in the mutant spinal cord by at least 50-fold (data not shown). After 29 amplification cycles of standard PCR, Connexin47-specific transcripts were clearly visible only in the wild-type spinal cord (Figure 1(i)). The absence of connexin expression cannot be attributed to a loss of oligodendrocytes, as oligodendrocyte numbers are normal in spinal cords of Sox10+/− , Sox8−/− mice.22 Thus, we conclude that oligodendrocytes fail to properly express Connexin32 and Connexin47 in the absence of wildtype amounts of group E Sox proteins, in particular Sox10. Sox10 activates Connexin47 promoter 1b in transient transfections In situ hybridization studies do not allow us to distinguish between direct and indirect effects of group E Sox proteins on connexin gene expression. For Connexin32 at least, the effect appears to be direct as Sox10 activates the Connexin32 promoter in transiently transfected HeLa cells.15 We also observed such a Sox10-dependent stimulation of the Connexin32 promoter in luciferase reporter gene assays in transiently transfected HEK 293 cells (Figure 2(a)). As published for HeLa cells,15 the Sox10-dependent activation of the Connexin32 promoter was further enhanced by the presence of Krox-20. Whereas the Connexin32 promoter was activated approximately 20-fold by Sox10 and on average threefold by Krox-20, the presence of both transcription factors resulted in a synergistic 38-fold stimulation (Figure 2(a)). Next, we wanted to analyze whether Connexin47 is also under direct transcriptional control of Sox10. Of several promoters from which the Connexin47 gene is transcribed, promoter 1b is predominantly active in the central nervous system. 24 In the postnatal spinal cord, promoter 1b was the one with the highest activity, as promoter 1b-specific transcripts were present at significantly higher levels than transcripts originating from the 1a, 1c or 1d promoters (Figure 1(i)). Therefore, we concentrated on the Connexin47 promoter 1b and used a genomic fragment spanning positions −611 to +116 relative to the 1b transcriptional start site to drive expression of a luciferase reporter in HEK 293 cells (Figure 2(a)). It deserves to be noted, that this genomic region also contains the 1a promoter (Figure 3(a)). However, Connexin47 promoter 1a did not contribute significantly to overall transcriptional activity in the context of this reporter construct (data not shown). Upon co-transfection

Figure 2. Connexin32 and Connexin47 promoters are activated by Sox10. (a) Transient transfections were performed in HEK 293 cells with luciferase reporters carrying either the Connexin32 promoter (Cx32-416-luc), Connexin47 promoter 1b (Cx47-727-luc) or no promoter (pBKS-luc). Sox10 and Krox-20 were co-transfected either alone or in combination as indicated. (b) In addition to Connexin47 promoter 1b and Sox10 expression plasmid, plasmids were co-transfected in HEK 293 cells that code for a Sox10-specific siRNA (si Sox10) or a scrambled version thereof (si scrambled). Activation rates for each promoter are presented as fold inductions ± SEM. Luciferase activities were determined in three experiments each performed in duplicate.

of Sox10, reporter gene expression was activated approximately 15-fold compared to a promoter-less control. This Sox10-dependent activation of Connexin47 promoter 1b was no longer observed when Sox10-specific small interfering RNA (siRNA) was added during the transfection (Figure 2(b)). In contrast, a scrambled version of this siRNA did not significantly interfere with activation of Connexin47 promoter 1b. We conclude, that Connexin47 promoter 1b can also be activated by Sox10. Krox-20, on the other hand, did not activate Connexin47 promoter 1b either alone or in combination with Sox10 (Figure 2(a)). The presence of Krox-20 even counteracted Sox10 on Connexin47 promoter 1b, as it reduced Sox10-dependent activation rates from 15-fold to sixfold. Thus, Connexin32

14

Figure 3. The Sox10-responsive region of Connexin47 promoter 1b is located between positions −255 and −28. (a) Schematic diagram of the luciferase reporter constructs carrying successively shortened versions of Connexin47 promoter 1b. Exon 1a and exon 1b are indicated as well as the first and last position for each fragment relative to the 1b transcription start site (+1, marked by arrow). (b) Transient transfections were performed in HEK 293 cells with luciferase reporters depicted in (a) in the absence or presence of Sox10. A promoterless luciferase reporter served as a control. Sox10-dependent activations are presented for each construct ± SEM. Luciferase activities were determined in three experiments each performed in duplicate.

and Connexin47 promoters respond differently to Krox-20. Mapping the Sox10-responsive region within Connexin47 promoter 1b To map the region within Connexin47 promoter 1b that mediates the response to Sox10, we generated a deletion series by successively shortening the promoter from the end distal to the 1b transcription start site (Figure 3(a)). Truncation from position −611 to position −255 reduced Sox10dependent activation only marginally, yielding a ninefold instead of a 11-fold induction in this set of experiments (Figure 3(b)). Responsiveness toward Sox10 was, however, no longer observed with the minimal Connexin47 promoter spanning positions −28 to +116. Sox10 thus mediated its effect on Connexin47 promoter 1b through a region between positions −255 and −28. Connexin47 promoter 1b contains two monomeric Sox10 binding sites Inspection of this Sox10-responsive region within Connexin47 promoter 1b revealed two potential

Sox10 Regulates Connexin47 Expression

Sox10 binding sites, designated Cx47-D and Cx47-E, which are both highly conserved in the Connexin47 upstream region of various mammalian species (Figure 4(a)). To analyze whether Sox10 is indeed able to bind to these sites, we performed electrophoretic mobility shift assays using extracts of transiently transfected HEK 293 cells (Figure 4(b)). Both Cx47-D and Cx47-E formed a complex with Sox10. These complexes were absent when control extracts without Sox10 were used instead. Addition of a Sox10-specific antibody reduced the mobility of the complex on Cx47-D and Cx47-E, further proving that the complex was formed by Sox10. When a fragment with both Cx47-D and Cx47-E was used in electrophoretic mobility shift assays (Cx47-DE in Figure 4(b)), two complexes with different mobilities were observed which could both be super-shifted by antibodies directed against Sox10 (Figure 4(b)). The presence of two complexes for the larger fragment from Connexin47 promoter 1b is consistent with the assumption that no additional high-affinity Sox binding site is present in the region analyzed. Order of appearance and relative intensities of both complexes in tiration experiments furthermore did not yield any hint that binding of Sox10 to both sites was cooperative or that higher-order structures were formed by Sox10 on Connexin47 promoter 1b (data not shown). In most experiments, equal amounts of Sox10 protein yielded more intense complexes with Cx47-D than with Cx47-E indicating that Cx47-D might have a slightly higher affinity for Sox10 in vitro. We also compared the mobilities of the Sox10 containing complexes on Cx47-D and Cx47-E with the mobility of the Sox10 complex on the well characterized CC′ binding site from the MPZ promoter, which is known to bind two Sox10 molecules in a cooperative manner. 14 Despite comparable lengths of the oligonucleotide probes, complexes with Cx47-D and Cx47-E exhibited faster mobilities than the dimer-containing complex on CC′ (Figure 4(b)). Both sites from Connexin47 promoter 1b are thus more likely to bind a single molecule of Sox10. To analyze the mode of binding in greater detail, we performed electrophoretic mobility shift experiments in the presence of two Sox10 proteins with different lengths (Q377X corresponding to residues 1–376 of Sox10 and MIC corresponding to residues 1–189; Figure 4(c)). Each of these Sox10 proteins formed a complex on Cx47D, Cx47-E or CC′ that roughly corresponded to its size such that the longer version yielded a complex of lower mobility. When both the long and the short Sox10 version were present simultaneously in the binding reaction with Cx47-D and Cx47-E, the same two characteristic complexes were formed. Only with CC′ did we observe a third complex of intermediate mobility, which is indicative of both a long and a short Sox10 molecule binding simultaneously to the site (Figure 4(c)). These mixing experiments thus confirm that both sites from Connexin47 promoter 1b bind a single Sox10 molecule each.

Sox10 Regulates Connexin47 Expression

15 One of the two binding sites predominantly mediates the Sox10 response in transient transfections

Figure 4. Two Sox10 binding sites are present within Connexin47 promoter 1b. (a) Sequence comparison of the region (position −76 to −22) with Sox10 binding sites in Connexin47 promoter 1b. The positions conserved among mammalian species are marked by asterisks. (b) Electrophoretic mobility shift assays were performed with extracts from mock-transfected HEK 293 cells (Sox10, −) or HEK 293 cells expressing a truncated Sox10 protein (Sox10, +) containing residues 1–203.30 Double-stranded oligonucleotides representing Cx47-D and Cx47-E from Connexin47 promoter 1b in their natural context (see Figure 5(a)) and C/C′ from the MPZ promoter14 were used as probes. A larger DNA fragment spanning positions −82 to −21 of Connexin47 promoter 1b (Cx47DE) was additionally employed. Antibodies against Sox10 (Ab) were added as indicated. (c) Oligonucleotides for Cx47-D, Cx47-E and CC′ were incubated with extracts expressing the MIC variant of Sox10 (amino acid residues 1–189), the Q377X nonsense mutant (amino acid residues 1–366)29 or both as indicated. The mixture of both Sox10 proteins generated a novel complex with intermediate mobility only on CC′, indicating that CC′ but not the sites from Connexin47 promoter 1b is capable of binding a Sox10 dimer.

To analyze whether the identified Cx47-D and Cx47-E are indeed important for activation of Connexin47 promoter 1b by Sox10, we introduced base changes into both sites that should abolish Sox10 binding (Figure 5(a)). Electrophoretic mobility shift assays with the mutant sites (Cx47-mutD and Cx47-mutE; Figure 5) indeed confirmed both for a short and a long version that Sox10 failed to bind to the mutated sites (Figure 5(b)). We therefore introduced both mutations alone or in combination into Connexin47 promoter 1b and performed luciferase reporter gene assays with the mutant promoters (Figure 5(c)). The mutations reduced Sox10 responsiveness to different degrees. Whereas Sox10 activated the wild-type Connexin47 promoter 14-fold on average in this set of experiments, there was only an 11-fold induction after mutation of Cx47-E. Mutation of Cx47-D reduced Sox10-dependent reporter gene activation more drastically so that only a residual two- to threefold induction remained. Thus, Cx47-D appears to be much more important for Sox10-dependent activation of Connexin47 promoter 1b than Cx47-E. This conclusion was also corroborated by the finding that there was no further significant reduction of promoter activity in the presence of both Cx47-D and Cx47-E mutations as compared to the Cx47-D mutation alone (Figure 5(c)). To address the question whether the Sox10 binding sites also influence activity of the adjacent Connexin47 promoter 1a, we performed luciferase assays with reporter gene constructs in which the 1b transcription start site was deleted (Figure 6(a)). Following truncation from position +116 to position –28, the residual transcriptional activity was still 2.6-fold induced by Sox10 (Figure 6(b)). Further deletion from position –28 to position –345, however, led to a loss of Sox10 responsiveness, as did selective mutation of Cx47-D. Cx47-D may therefore also be important for conferring Sox10 responsiveness to Connexin47 promoter 1a. Activation of Connexin47 promoter 1b requires defined Sox10 domains In addition to wild-type Sox10, we also analyzed several mutants for their ability to activate Connexin47 promoter 1b. First, we investigated two Sox10 mutants in which amino acid substitutions had led to a selective loss of their ability to form dimers on DNA and to activate through dimer sites.25 These aa1 and aa2* mutants stimulated Connexin47 promoter 1b at least as efficiently as the wild-type Sox10 protein (Figure 6(c)) and therefore independently confirmed that interaction of Sox10 with Connexin47 promoter 1b is exclusively through sites that bind Sox10 monomers. In contrast, Sox10 mutants that are unable to bind to DNA such as the WS095 mutation that carries an

16

Sox10 Regulates Connexin47 Expression

insertion of two amino acids in the HMG-domain, completely failed to activate the Connexin47 promoter. Carboxy-terminally shortened versions of

Figure 5. Sox10 activates transcription through its binding sites within Connexin47 promoter 1b. (a) Sequence of oligonucleotides containing Cx47-D or Cx47-E from Connexin47 promoter 1b in wild-type version or after mutation (Cx47-mutD and Cx47-mutE). (b) Electrophoretic mobility shift assays were performed with these double-stranded oligonucleotides and extracts from mock-transfected HEK 293 cells (−) or HEK 293 cells expressing the isolated HMG-box or a truncated Sox10 protein (MIC). (c) Transient transfections were performed in HEK 293 cells in the absence or presence of Sox10 with luciferase reporters containing Connexin47 promoter 1b in its wild-type form or in versions in which Cx47-D and/or Cx47-E were converted into the sequences that no longer bind Sox10. A promoterless luciferase reporter served as a control. Sox10-dependent activations for each construct are presented ± SEM. Luciferase activities were determined in three experiments each performed in duplicate.

Figure 6. Connexin47 activation requires CX47-D as well as the DNA-binding and transactivation capacity of Sox10, but not its ability to dimerize. (a) Schematic diagram of the luciferase reporter constructs used for transient transfections. Exon 1a is indicated as well as the first and last position for each fragment relative to the 1b transcription start site (see Figure 3(a)). The 1a transcription start site is marked by an arrow. (b) Transient transfections were performed in HEK 293 cells with luciferase reporters depicted in (a) in the absence or presence of Sox10. A promoterless luciferase reporter served as control. Sox10-dependent activations are presented for each construct ± SEM. Luciferase activities were determined in three experiments each performed in duplicate. (c) Transient transfections were performed with the Cx47-727-luc reporter in the presence of wildtype Sox10 and several mutants.25,29 Mutants aa1 and aa2* do not dimerize, WS095 is unable to bind to DNA and the carboxy-terminally truncated Q377X, Y207X and MIC all lack the transactivation domain. Luciferase activities were determined in three experiments each performed in duplicate. Induction rates of Connexin47 promoter 1b were determined for all Sox proteins. The activity of wildtype Sox10 was then arbitrarily set to 100%, those of the Sox10 mutants are presented in relation to the wild-type.

Sox10 Regulates Connexin47 Expression

17

Sox10 also led to a strong reduction in Connexin47 activation (Figure 6(c)). Already a deletion of the last 90 amino acid residues by truncation to amino acid 376 (Q377X) reduced transactivation rates to 17%, indicating that the carboxy-terminal region of Sox10 indeed functions as the protein's major transactivation domain. Deletion of 170 further amino acid residues from the carboxy-terminal end (Y207X) led to a complete loss of Connexin47 promoter 1b activation arguing that in addition to the carboxyterminal transactivation domain, there might be a second region with at least some transactivation potential between amino acid residues 207 and 377. Connexin47 is also regulated by Sox10 in cells of oligodendroglial origin There are several cell lines of oligodendroglial origin that express Sox10 26 including the 33B oligodendroglioma (Figure 7(b)). To analyze whether endogenous levels of Sox10 could activate Connexin47 promoter 1b, we transfected a luciferase reporter containing Connexin47 promoter 1b into 33B cells. Its activity was approximately 6.5-fold higher than the promoter-less control (Figure 7(a)). Addition of an expression plasmid for a Sox10specific siRNA during the transfection lowered the activity of Connexin47 promoter 1b to 1.5-fold the activity of the promoterless version, whereas an expression plasmid for a scrambled version of this siRNA had no effect. This result indicates that endogenous Sox10 contributes strongly to the overall activity of a transfected Connexin47 promoter 1b in 33B cells. The importance of Sox10 for Connexin47 promoter 1b activity was also confirmed by mutational analysis. The combined mutation of both Cx47-D and Cx47-E in Connexin47 promoter 1b strongly reduced its activity in 33B cells (Figure 7(a)). Furthermore, the residual promoter activity was no longer Sox10-dependent as indicated by the fact that a Sox10-specific siRNA had as little effect on this mutant promoter as its scrambled version. RT-PCR analysis of 33B cells revealed that these cells also expressed Connexin47. Transcript levels are low as judged from the similar intensity of the Connexin47 signal from 33B cells and spinal cord of one day old rat pups, which at this time express only little Connexin47 (Figure 7(b)). The presence of both Connexin47 and Sox10 in 33B cells prompted us to analyze by chromatin immunoprecipitation whether Sox10 could be detected on endogenous Connexin47 promoter 1b in 33B cells. Whereas preimmune serum failed to precipitate Connexin47 promoter 1b fragments with the Sox10 binding sites, these DNA fragments were clearly present when immunoprecipitations were instead performed with a Sox10-specific antiserum (Figure 7(c)). Comparison of amplification products from input material and immunoprecipitate indicated that only a fraction of Connexin47 promoter 1b was precipitated. This correlates with the low expression levels of Connexin47 in 33B cells compared to

Figure 7. Sox10 is bound in vivo to Connexin47 promoter 1b. (a) Transient transfections were performed in 33B oligodendroglioma with luciferase reporters containing wild-type Connexin47 promoter 1b or a version with mutant sites Cx47-E and Cx47-D in the presence of expression plasmids for Sox10-specific siRNA (si Sox10) or a scrambled version thereof (si scrambled). A promoterless luciferase reporter served as a control. All luciferase activities were determined in three experiments each performed in duplicate. The values obtained were expressed relative to the activity of the promoterless reporter which was arbitrarily set to 1. (b) PCR experiments were performed on reverse transcribed RNAs from rat spinal cord and 33B cells to detect transcripts for Connexin32, Connexin47 and Sox10. cDNA amounts in each sample were normalized to comparable transcript levels for β-actin. Template-free control reactions (control) were performed in parallel. (c) Chromatin immunoprecipitation was performed on formaldehyde-fixed 33B cells with an antiserum specifically directed against Sox10 and the corresponding pre-immune antiserum. PCR was applied on the immunoprecipitate to detect the Connexin47 promoter 1b region, which contains the Sox10 binding sites identified in vitro. This fragment was also amplified from 1/20 of the input material used for immunoprecipitation. (–) Water control.

mature oligodendrocytes in vivo. Comparable experiments with the Connexin32 promoter were unsuccessful (data not shown), probably because of

18 the even lower expression levels of this connexin in 33B cells (Figure 7(b)).

Discussion Three lines of evidence suggest that expression of Connexin32 and Connexin47 is regulated by the HMG-box transcription factor Sox10 in oligodendrocytes. First, neither Connexin32 nor Connexin47 was expressed in vivo in spinal cord oligodendrocytes with significantly reduced levels of group E Sox proteins. Second, connexin gene promoters were activated by ectopic as well as by endogenous Sox10 in reporter gene assays in transiently transfected cells in a manner that completely depended on the ability of Sox10 to bind the promoters. And finally, Sox10 was found on Connexin47 promoter 1b in 33B oligodendroglioma cells by chromatin immunoprecipitation assays. The two latter results also argue that expression of the connexin genes is under direct transcriptional control of Sox10. Such an assumption is also supported by the fact that both DNA-binding and the transactivation domain had to be present and functional in the Sox10 protein to obtain connexin gene activation. This direct transcriptional control by Sox10 had already been shown for Connexin32 in Schwann cells,15 but is a novel finding for Connexin47. Although a part of our results were obtained by transient transfections in a non-oligodendroglial cell line and thus need to be interpreted cautiously, they are supported by findings in vivo and are therefore most likely relevant. Nevertheless, they do not exclude the possibility of additional Sox10-responsive regulatory elements outside the analyzed region. Similar to the Connexin32 promoter,15 Connexin47 promoter 1b contains one major binding site through which Sox10 exerts its function. The occurrence of a single major functional site in both connexin gene promoters is unusual because promoters of other myelin genes such as MPZ and MBP contain multiple cis-acting elements for Sox10.14,18 In the Connexin32 promoter, the essential binding site is a dimeric site that allows cooperative binding of two molecules of Sox10, whereas it is a monomeric binding site in the context of the Connexin47 promoter. Connexin47 promoter 1b is therefore unique in its ability to respond normally to mutant versions of Sox10 in which the ability for dimeric binding is selectively lost. How these differences in promoter organization between connexins and other myelin genes or between the two connexins affect the transcriptional response to Sox10 is currently a matter of speculation. Expression of the two connexin genes in maturing oligodendrocytes may be lower than other myelin genes because of the fewer binding sites. Furthermore, comparison of the two connexin gene promoters reveals that several functional binding sites are present for Krox-20 in the Connexin32 promoter,15 but missing in the Connexin47 promoter. As a consequence, only Connexin32 is

Sox10 Regulates Connexin47 Expression

activated by Krox-20 alone or in combination with Sox10. Krox-20-dependent activation is likely to be crucial for Connexin32 expression in Schwann cells, where this transcription factor is a major regulator of the myelination program.13 As Connexin47 is not expressed in myelinating Schwann cells and as Krox-20 does not occur in differentiating oligodendrocytes,16 there is no need for a Krox-20dependent stimulation of Connexin47 expression. Krox-20 even counteracted Sox10-dependent activation of Connexin47 promoter 1b in our transient transfections. A similar mechanism may actively suppress Connexin47 expression in Schwann cells. Sox proteins are commonly believed to require other transcription factors as partner proteins to activate gene expression in vivo.27 It is likely that Krox-20 is such a partner protein for Sox10 during Connexin32 expression in Schwann cells. Currently, we do not know the partners for Sox10 in oligodendrocytes that allow full expression of Connexin47 and also of Connexin32. We believe, however, that these partner transcription factors are missing in oligodendroglioma cell lines such as 33B and C6. This would explain why connexin gene expression and probably promoter occupancy by Sox10 in these oligodendrocyte-derived cell lines are comparatively low. Not only connexin levels, but also myelin protein levels are low in these oligodendrogliomas. It is therefore conceivable that the oligodendrocytic partner of Sox10 is involved in regulating many terminal differentiation genes as is the case for Krox-20 in myelinating Schwann cells. At present, the identified target genes in oligodendrocytes comprise MBP, PLP, Connexin32 and Connexin47. Sox10 thus regulates several terminal differentiation genes in oligodendrocytes. In addition to myelin sheaths, gap junctions between myelin layers and with astrocytes should be disturbed in Sox10-deficient oligodendrocytes or in oligodendrocytes expressing dominant-negative Sox10 mutants. Altered connexin expression and lost ionic homeostasis may thus contribute as much to the central dysmyelinating leukodystrophy observed in patients carrying such dominant negative Sox10 mutations28 as the reduced myelin gene expression. Sox10 may be a global regulatory component of the myelination program in oligodendrocytes.

Materials and Methods Plasmid constructs The reporter plasmid pBKS-luc contained luciferase coding sequences and simian virus 40 (SV40) poly(A) signal between the HindIII and BamHI restriction sites of pBluescript KS II (Stratagene). The mouse Connexin47 promoter 1b (corresponding to positions −611 to +116 as defined by Genbank accession number NM_175452) was inserted into pBKS-luc in front of the luciferase cDNA using EcoRV and HindIII sites, yielding Cx47-727-luc.

Sox10 Regulates Connexin47 Expression

Shorter fragments were generated by PCR and inserted between KpnI and HindIII to produce Cx47-583-luc (containing positions −611 to −28), Cx47-266-luc (containing positions −611 to −345), Cx47-371-luc (containing positions −255 to +116) and Cx47-min-luc (containing positions −28 to +116). A fragment corresponding to positions −269 to +145 of the mouse Connexin32 promoter (as defined by homology to the human Connexin32 promoter; Genbank accession number L47127) was similarly placed in front of the luciferase gene in pBKSluc using XhoI and HindIII restriction sites to generate Cx32-416-luc. Site-directed mutagenesis was used to mutate Sox10 binding sites in the context of the Cx47727-luc and the Cx47-583-luc vectors as indicated in Figures 5 and 6. pCMV5-based eukaryotic expression plasmids for full length human Sox10 and rat Krox-20 have been described as well as additional expression plasmids for carboxyterminally truncated or mutant Sox10 proteins.17,25,29,30 In analogy to the previously described siRNA expression plasmid for rat Sox10,30 a region corresponding to positions 704–724 of human Sox10 cDNA (Genbank accession number NM_006941) was inserted as small interfering RNA (siRNA) between the BglII and HindIII sites of the pSUPER.neo+gfp vector (Oligoengine™). Scrambled versions of both rat and human-specific Sox10 siRNAs were produced as controls. Fragments of the mouse Connexin32 cDNA (corresponding to positions 405–887; Genbank accession number NM _ 008124) and the mouse Connexin47 cDNA (corresponding to positions 1381–1692; Genbank accession number NM_175452) were retrieved by PCR and inserted into pGEM-T easy (Promega) to generate antisense riboprobes for in situ hybridization. Cell culture, transfections and luciferase assays HEK 293 cells were kept in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% (v/v) fetal calf serum (FCS). Transient transfection for preparation of protein extracts was performed in 100 mm plates with 15 μg of pCMV-based expression plasmid using the calcium phosphate technique. HEK 293 cells were seeded in 24 well plates for luciferase assays and transfected using Superfect reagent (Qiagen) with the following amounts of DNA: 50–100 ng of pCMV-Sox10 (wild-type and mutants), 200 ng of pCMV-Krox-20, 300 ng of pSuper.siSox10 (rat or human, wild-type or scrambled) and 0.5 μg of luciferase reporter plasmid. The total amount of plasmid DNA was kept constant using empty pCMV5. For luciferase reporter assays in 33B oligodendroglioma, cells were cultivated in DMEM containing 10% FCS and transfected in 35 mm plates with 1.5 μg of pSuper.siSox10 (rat-sepcific or scrambled) and 0.5 μg of luciferase reporter plasmid using Superfect reagent (Qiagen). The 48 h post-transfection, cells were harvested for extract preparation or luciferase assays.25 Protein preparations and electrophoretic mobility shift assays Transfected HEK 293 cells from one 100 mm plate were lysed in the presence of 2 μg of leupeptin and aprotinin each in ice-cold 20 mM Hepes (pH 7.9), 10 mM KCl, 5 mM MgCl2, 0.5 mM EDTA, 10% (v/v) glycerol, 2 mM dithiothreitol, 0.1% (v/v) Triton-X 100, and 300 mM NaCl. After extraction for 15 min under constant rotation, cell debris was removed from the extract by centrifugation.

19 For electrophoretic mobility shift assays, protein extracts were incubated with 0.5 ng of 32P-labeled oligonucleotide probes for 20 min on ice in the presence of 1 μg of poly(dG-dC) as unspecific competitor.25 The following double-stranded oligonucleotide probes were used: Cx47-D (positions −82 to −57 of the Connexin47 promoter, wild-type and mutant version; see Figure 5(a)), Cx47-E (positions −47 to −21 of the Connexin47 promoter, wild-type and mutant version; see Figure 5(a)), Cx47-DE (positions −82 to −21 of the Connexin47 promoter) and the prototypic dimeric binding site C/C′ from the myelin protein zero (MPZ) promoter.14 For supershift experiments, 0.1 μl of Sox10-specific antiserum was additionally added after 10 min, and incubation was continued for further 10 min. Samples were loaded onto native 4% (w/v) polyacrylamide gels and subjected to electrophoresis before gels were dried and exposed for autoradiography. RNA preparation, reverse transcription and PCR Total RNA was extracted from 33B cells and spinal cords from one week old mice or one day old rats using TRIZOL reagent (Invitrogen). A total of 2 μg of each RNA sample were reverse transcribed into cDNA.30 From each cDNA, 1 μl was amplified with primer pairs specific for rat Connexin47 (5′-GAGGAGCGAACGGAAGATGT-3′ and 5′GTTCGCCAGGTTCTGATCGT-3′, yielding a product of 545 bp), mouse Connexin47 (5′-GCAGTACAGGCAGCAGAGACG-3′ and 5′-CAGAGGGGTCATCCCTGCTC-3′, yielding a product of 314 bp), rat Connexin32 (5′-AGAAAATGCTACGGCTTGAG-3′ and 5′-CAGGCTGAGCATCGGTCGCTCTT-3′, yielding a product of 540 bp), rat Sox10 (5′-CCATGTCAGATGGGAACCCA-3′ and 5′GCTCAGCCCGTAGCCAGC-3′, yielding a product of 298 bp), β-actin (5′-CCTGGGCATGGAGTCCTG-3′ and 5′-GGAGCAATGATCTTGATCTTC-3′, yielding a product of 179 bp) and mouse ribosomal protein L8 (RPL8, 5′GTTCGTGTACTGCGGCAAGA-3′ and 5′-ACAGGATTCATGGCCACACC-3′, yielding a product of 363 bp). Transcripts from the various mouse Connexin47 promoters were amplified separately by combining primers specific for exon 1a, 1b, 1c and 1d with a primer located in exon 2 as described.24 In addition to the primers (20 pmol each), polymerase chain reactions contained 0.2 mM dNTP and one unit of Taq polymerase in 20 μl of 10 mM Tris–HCl (pH 8.8), 0.08% (v/v) NP40, 50 mM KCl and 2 mM MgCl2. Dimethyl sulfoxide (DMSO) was additionally added in case of Connexin47, Connexin32 (both 5%, v/v) and β-actin (10% DMSO, v/v). Chromatin immunoprecipitation assays Chromatin immunoprecipitation assays on 33B cells were essentially performed as described.31 Briefly, cellular protein and genomic DNA were cross-linked in living cells by incubation in 1% (v/v) formaldehyde before cells were lysed in 10 mM Tris–HCl (pH 8.0), 1 mM EDTA, 5 mM MgCl2, 0.5% NP40. After centrifugation, supernatants were discarded and pellets were resuspended in 1% (w/v) SDS, 10 mM EDTA, 50 mM Tris–HCl (pH 8.0), before the extracted chromatin was sonicated to an average fragment length of 300 bp to 600 bp, cleared by centrifugation and diluted into 16.7 mM Tris–HCl (pH 8.1), 167 mM NaCl, 0.01% SDS, 1.1% Triton X100, 1.2 mM EDTA. Immunoprecipitations were performed overnight at 4 °C with a rabbit antiserum directed against Sox1032 and the corresponding pre-immune serum. Precipitates

Sox10 Regulates Connexin47 Expression

20 were washed extensively, before cross-links were reversed during 4 h of incubation at 65 °C in 1% SDS and 0.05 M NaHCO3. DNA fragments were purified and subjected to PCR. For detection of the region from rat Connexin-47 promoter 1b containing the Sox10 binding sites, 5′-GGCAGGCAGGTATTCAGTCA-3′ and 5′-AACTCCACCTTAGCAGTCAG-3′ were used as primers. After an initial denaturation step for 2 min, 30 cycles were performed with each cycle consisting of a denaturation (30 s at 95 °C), an annealing (30 s at 63 °C) and an elongation step (30 s at 72 °C). Animal husbandry, tissue preparation and in situ hybridization Mice with a Sox8 null allele33 were kept as heterozygotes in the presence or absence of a Sox10 null allele on a mixed C3H/C57Bl6J background.23 For this study, Sox8−/+ mice were crossed with Sox8−/+, Sox10−/+ mice. Genotyping was performed by PCR as described.23,33 Spinal cords were dissected from pups at postnatal day seven. After fixation in paraformaldehyde and genotyping, spinal cords from Sox8−/−, Sox10−/+ pups and wildtype littermates were sectioned on a cryotome. The 20 μm sections were used for in situ hybridization according to standard protocols with DIG-labeled antisense riboprobes for Connexin47 and Connexin32 as well as MBP and PLP as described.18,32 Samples were analyzed and documented using a Leica MZFLIII stereomicroscope equipped with an Axiocam (Zeiss, Oberkochem).

6.

7.

8.

9.

10.

11.

12.

13.

Acknowledgements Petra Lommes is acknowledged for expert technical assistance. This work was supported by grant We1326/ 7-3 from the Deutsche Forschungsgemeinschaft.

References 1. Willecke, K., Eiberger, J., Degen, J., Eckardt, D., Romualdi, A., Gueldenagel, M. et al. (2002). Structural and functional diversity of connexin genes in the mouse and human genome. Biol. Chem. 383, 725–737. 2. Nagy, J. I., Dudek, F. E. & Rash, J. E. (2004). Update on connexins and gap junctions in neurons and glia in the mammalian central nervous system. Brain Res. Rev. 47, 191–215. 3. Li, X., Lynn, B. D., Olson, C., Meier, C., Davidson, K. G. V., Yasumura, T. et al. (2002). Connexin29 expression, immunocytochemistry and freeze-fracture replica immungold labeling (FRIL) in sciatic nerve. Eur. J. Neurosci. 16, 795–806. 4. Altevogt, B. M., Kleopa, K. A., Postma, F. R., Scherer, S. S. & Paul, D. L. (2002). Connexin29 is uniquely distributed within myelinating glial cells of the central and peripheral nervous systems. J. Neurosci. 22, 6458–6470. 5. Kamasawa, N., Sik, A., Morita, M., Yasumara, T., Davidson, K. G. V., J.I., N. & Rash, J. E. (2005). Connexin-47 and connexin-32 in gap junctions of oligodendrocyte somata, myelin sheaths, paranodal loops and Schmidt-Lanterman incisures: implications

14.

15.

16.

17.

18.

19.

20. 21.

for ionic homeostasis and potassium siphoning. Neuroscience, 136, 65–86. Menichella, D. M., Goodenough, D. A., Sirkowski, E., Scherer, S. S. & Paul, D. L. (2003). Connexins are critical for normal myelination in the CNS. J. Neurosci. 23, 5963–5973. Odermatt, B., Wellershaus, K., Wallraff, A., Seifert, G., Degen, J., Euwens, C. et al. (2003). Connexin47 (Cx47)deficient mice with enhanced green flujorescent protein reporter gene reveal predominant oligodendrocytic expression of Cx47 and display vacuolized myelin in the CNS. J. Neurosci. 23, 4549–4559. Teubner, B., Odermatt, B., Guldenagel, M., Sohl, G., Degen, J., Bukauskas, F. et al. (2001). Functional expression of the new gap junction gene connexin47 transcribed in mouse brain and spinal cord neurons. J. Neurosci. 21, 1117–1126. Ressot, C. & Bruzzone, R. (2000). Connexin channels in Schwann cells and the development of the X-linked form of Charcot-Marie-Tooth disease. Brain Res. Rev. 32, 192–202. Scherer, S. S., Deschenes, S. M., Xu, Y. T., Grinspan, J. B., Fischbeck, K. H. & Paul, D. L. (1995). Connexin32 is a myelin-related protein in the PNS and CNS. J. Neurosci. 15, 8281–8294. Pingault, V., Bondurand, N., Kuhlbrodt, K., Goerich, D. E., Prehu, M.-O., Puliti, A. et al. (1998). Sox10 mutations in patients with Waardenburg-Hirschsprung disease. Nature Genet. 18, 171–173. Warner, L. E., Mancias, P., Butler, I. J., McDonald, C. M., Keppen, L., Koob, K. G. & Lupski, J. R. (1998). Mutations in the early growth response 2 (EGR2) gene are associated with hereditary myelinopathies. Nature Genet. 18, 382–384. Nagarajan, R., Svaren, J., Le, N., Araki, T., Watson, M. & Milbrandt, J. (2001). EGR2 mutations in inherited neuropathies dominant-negatively inhibit myelin gene expression. Neuron, 30, 355–368. Peirano, R. I., Goerich, D. E., Riethmacher, D. & Wegner, M. (2000). Protein zero expression is regulated by the glial transcription factor Sox10. Mol. Cell. Biol. 20, 3198–3209. Bondurand, N., Girard, M., Pingault, V., Lemort, N., Dubourg, O. & Goossens, M. (2001). Human Connexin 32, a gap junction protein altered in the X-linked form of Charcot-Marie-Tooth disease, is directly regulated by the transcription factor SOX10. Hum. Mol. Genet. 10, 2783–2795. Sock, E., Leger, H., Kuhlbrodt, K., Schreiber, J., Enderich, J., Richter-Landsberg, C. & Wegner, M. (1997). Expression of Krox proteins during differentiation of the O2-A progenitor cell line CG-4. J. Neurochem. 68, 1911–1919. Kuhlbrodt, K., Herbarth, B., Sock, E., HermansBorgmeyer, I. & Wegner, M. (1998). Sox10, a novel transcriptional modulator in glial cells. J. Neurosci. 18, 237–250. Stolt, C. C., Rehberg, S., Ader, M., Lommes, P., Riethmacher, D., Schachner, M. et al. (2002). Terminal differentiation of myelin-forming oligodendrocytes depends on the transcription factor Sox10. Genes Dev. 16, 165–170. Bowles, J., Schepers, G. & Koopman, P. (2000). Phylogeny of the SOX family of developmental transcription factors based on sequence and structural indicators. Dev. Biol. 227, 239–255. Wegner, M. (1999). From head to toes: the multiple facets of Sox proteins. Nucl. Acids Res. 27, 1409–1420. Wegner, M. & Stolt, C. C. (2005). From stem cells to

Sox10 Regulates Connexin47 Expression

22.

23.

24. 25.

26.

27. 28.

neurons and glia: a soxist's view of neural development. Trends Neurosci. 28, 583–588. Stolt, C. C., Lommes, P., Friedrich, R. P. & Wegner, M. (2004). Transcription factors Sox8 and Sox10 perform non-equivalent roles during oligodendrocyte development despite functional redundancy. Development, 131, 2349–2358. Britsch, S., Goerich, D. E., Riethmacher, D., Peirano, R. I., Rossner, M., Nave, K. A. et al. (2001). The transcription factor Sox10 is a key regulator of peripheral glial development. Genes Dev. 15, 66–78. Anderson, C. L., Zundel, M. A. & Werner, R. (2005). Variable promoter usage and alternative splicing in five mouse connexin genes. Genomics, 85, 238–244. Schlierf, B., Ludwig, A., Klenovsek, K. & Wegner, M. (2002). Cooperative binding of Sox10 to DNA: requirements and consequences. Nucl. Acids Res. 30, 5509–5516. Kuhlbrodt, K., Herbarth, B., Sock, E., Enderich, J., Hermans-Borgmeyer, I. & Wegner, M. (1998). Cooperative function of POU proteins and Sox proteins in glial cells. J. Biol. Chem. 273, 16050–16057. Kamachi, Y., Uchikawa, M. & Kondoh, H. (2000). Pairing SOX off: with partners in the regulation of embryonic development. Trends Genet. 16, 182–187. Inoue, K., Khajavi, M., Ohyama, T., Hirabayashi, S.-i.,

21

29.

30.

31.

32.

33.

Wilson, J., Reggin, J. D. et al. (2004). Molecular mechanism for distinct neurological phenotypes conveyed by allelic truncating mutations. Nature Genet. 36, 361–369. Kuhlbrodt, K., Schmidt, C., Sock, E., Pingault, V., Bondurand, N., Goossens, M. & Wegner, M. (1998). Functional analysis of Sox10 Mutations found in human Waardenburg-Hirschsprung patients. J. Biol. Chem. 273, 23033–23038. Schlierf, B., Lang, S., Kosian, T., Werner, T. & Wegner, M. (2005). The high-mobility group transcription factor Sox10 interacts with the N-myc interacting protein Nmi. J. Mol. Biol. 353, 1033–1042. Weinmann, A. & Farnham, P. (2002). Identification of unknown target genes of human transcription factors using chromatin immunoprecipitation. Methods, 26, 37–47. Stolt, C. C., Lommes, P., Sock, E., Chaboissier, M.-C., Schedl, A. & Wegner, M. (2003). The Sox9 transcription factor determines glial fate choice in the developing spinal cord . Genes Dev. 17, 1677–1689. Sock, E., Schmidt, K., Hermanns-Borgmeyer, I., Bösl, M. R. & Wegner, M. (2001). Idiopathic weight reduction in mice deficient in the high-mobilitygroup transcription factor Sox8. Mol. Cell. Biol. 21, 6951–6959.

Edited by M. Yaniv (Received 4 January 2006; received in revised form 26 May 2006; accepted 30 May 2006) Available online 15 June 2006