Activation of Krox20 gene expression by Sox10 ... - Wiley Online Library

JOURNAL OF NEUROCHEMISTRY

| 2010 | 112 | 744–754

doi: 10.1111/j.1471-4159.2009.06498.x

Institut fu¨r Biochemie, Emil-Fischer-Zentrum, Universita¨t Erlangen-Nu¨rnberg, Erlangen, Germany

Abstract The high-mobility group domain transcription factor Sox10 is believed to influence myelination in Schwann cells by directly activating myelin genes and by inducing Krox20 as a pivotal regulator of peripheral myelination. Krox20 induction at this stage is thought to be mediated by the myelinating Schwann cell element 35 kb downstream of the Krox20 transcriptional start site and requires cooperation with Oct6. Here, we prove for the first time in vivo that Schwann cell-specific Krox20 expression indeed depends on Sox10. We also provide evidence that Sox10 functions through multiple, mostly monomeric binding sites in the myelinating Schwann cell element in a manner that should render the enhancer exquisitely sensitive to Sox10 levels. Synergistic activation of the enhancer by Sox10 and Oct6 furthermore does not involve cooperative binding to closely spaced binding sites in defined composite

elements. Nevertheless, the POU domain of Oct6 and the high-mobility group domain of Sox10 as the two DNA-binding domains were both essential indicating that each transcription factor has to bind independently to DNA. Whereas the POU domain was the only important region of Oct6, two further Sox10 domains were required for synergistic Krox20 activation. These were the carboxyterminal transactivation domain and the conserved K2 domain in the central portion of Sox10. All required regions are conserved in several closely related POU and Sox proteins thus explaining why Oct6 and Sox10 can be replaced by their relatives during Krox20 induction in myelinating Schwann cells. Keywords: glia, high-mobility group, neural crest, peripheral nervous system, POU, transcription factor. J. Neurochem. (2010) 112, 744–754.

Schwann cells represent a major population of glial cells in the PNS. They can either be myelinating or non-myelinating. To become myelinating, Schwann cells have to progress through several consecutive stages. Prior to the onset of myelination, Schwann cells establish a one-to-one contact with large caliber axons and start to wrap the axons for one and a half turns with their plasma membrane (Jessen and Mirsky 2005). To exit in time from this promyelin stage, Schwann cells have to express the Pit1-Oct1/2-Unc86 (POU) family transcription factor Oct6 and its close relative Brn2 (Jaegle et al. 2003). These transcription factors induce expression of the zinc-finger protein Krox20 (Ghislain and Charnay 2006) which in turn activates expression of many myelin genes and functions as the master regulator of PNS myelination (Topilko et al. 1994; Nagarajan et al. 2001). We have previously shown that the high-mobility group (HMG)-box containing transcription factor Sox10 is expressed during all stages of Schwann cell development and continues to be present in myelinating Schwann cells of the adult PNS (Kuhlbrodt et al. 1998a). Targeted deletion in the mouse had confirmed the importance of Sox10 for Schwann

cell development as Schwann cell precursors failed to develop from neural crest cells in Sox10-deficient mice (Britsch et al. 2001). Although the early loss of Schwann cells precluded an in-depth analysis of Sox10 functions during later phases of Schwann cell development, other evidence points to such a role. This includes later Schwann cell defects in mice carrying hypomorphic Sox10 mutations (Schreiner et al. 2007) and evidence from tissue culture that expression of several peripheral myelin genes (for instance the Mpz, the Connexin-32 and the Mag genes) is controlled by Sox10 (Peirano et al. 2000; Bondurand et al. 2001; Jones et al. 2007; LeBlanc et al. 2007). Myelin genes furthermore

744

Received September 4, 2009; Revised manuscript received November 10, 2009; Accepted November 10, 2009. Address correspondence and reprint requests to Dr Michael Wegner, Institut fu¨r Biochemie, Emil-Fischer-Zentrum, Universita¨t Erlangen, Fahrstrasse 17, 91054 Erlangen, Germany. E-mail: [email protected] Abbreviations used: EMSA, electrophoretic mobility shift assays; HEK293, human embryonic kidney cells; MSE, myelinating Schwann cell element; POU, Pit1-Oct1/2-Unc86.

2009 The Authors Journal Compilation 2009 International Society for Neurochemistry, J. Neurochem. (2010) 112, 744–754

Glial Krox20 induction by Sox10 | 745

appear to be synergistically activated by Krox20 and Sox10, thus establishing the combination of both transcription factors as essential for myelination. Recent evidence suggests an additional link between Krox20 and Sox10. Ghislain and colleagues identified a 1.3 kb region 35 kb downstream of the Krox20 transcriptional start site that is responsible for Krox20 expression in myelinating Schwann cells (Ghislain et al. 2002), and later showed that this MSE is synergistically activated by Oct6 and Sox10 (Ghislain and Charnay 2006). The same enhancer also responds to nuclear factor of activated T cells (NFAT) and c4 in Schwann cells, again in synergy with Sox10 (Kao et al. 2009). Sox10 therefore also seems to be a central factor in Krox20 induction and in this way induces the partner with which it later activates myelin gene expression. Here, we studied the mechanism by which Sox10 activates Krox20 expression and determined the requirements for synergism with Oct6 during this process.

Materials and methods Plasmids Eukaryotic pCMV expression plasmids for rat Sox10, Sox10 DK2, Sox10 aa1, Sox10 WS95, Sox10 MIC, Sox10 Q377X, Sox9, Sox8, Sox2, Oct6, Oct6DN, Oct6DC, Oct6 POU, Oct6DPOU, Oct6DPOUS, Oct6DPOUHD, Oct6 WF-CS, Brn2, Brn5, and Oct1 have been described before (Renner et al. 1994; Sock et al. 1996, 2003; Schreiber et al. 1997; Kuhlbrodt et al. 1998a,b; Schlierf et al. 2002; Schreiner et al. 2007). For reporter gene assays, the pTATAluc-Krox20MSE plasmid was used. It contained the luciferase reporter gene under control of the 1.3 kb MSE of the mouse Krox20 gene (position 9361594902 of AC153379) (Ghislain and Charnay 2006) upstream of the b-globin minimal promoter. Single and multiple binding site mutations were generated using the QuickChange XL SiteDirected Mutagenesis Kit (Stratagene, La Jolla, CA, USA). At least the 5¢-CAA-3¢ core of the heptameric Sox consensus binding motif was altered. For bacterial expression of glutathione-S-transferase (GST) fusion proteins, sequences corresponding to amino acids 233–306 (K2 domain), or 1–230 of Sox10 were inserted in frame into pGEX-KG after amplification and introduction of restriction sites by PCR. Cell culture and luciferase assays Human embryonic kidney cells (HEK293) and S16 cells were maintained in Dulbecco’s Modified Eagle’s Medium containing 10% fetal calf serum. For luciferase assays, S16 cells were transfected in duplicates with 4.5 lL Superfect reagent per 35 mm plate according to the manufacturer’s instructions (Qiagen, Hilden, Germany). To analyze the impact of Sox10 alone, S16 cells were transfected with 1.5 lg luciferase reporter plasmid and 0.5 lg empty pCMV5 plasmid or pCMV5-Sox10 plasmid. To analyze synergistic activity, S16 cells were transfected with combinations of 750 ng luciferase reporter plasmid and 750 ng effector plasmid per plate. Expression plasmids for Sox and POU proteins were employed in a 10 : 1 ratio.

The total amount of expression plasmid was kept constant with empty pCMV5 plasmid. Cells were harvested 48 h post-transfection, and extracts were assayed for luciferase activity. Luciferase activities obtained with the different effectors were normalized to luciferase activity observed with pCMV5 alone. Bar graphs usually result from three duplicates and the SEM is shown as error bar. Statistical analysis was performed using GraphPad Prism 4 software (GraphPad Software Inc., La Jolla, CA, USA). Proteins, cell extracts, western blots, and electrophoretic mobility shift assay Whole cell extracts from transfected HEK293 cells were prepared 48 h post-transfection. Cells from two 100 mm plates were lysed in the presence of 10 lg/mL aprotinin, 10 lg/mL leupeptin, and 2 mM dithiothreitol in 10 mM Hepes pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 0.5% NP40 (Applichem, Darmstadt, Germany), 170 mM NaCl, and 10% glycerol. Extracts were cleared by centrifugation and checked for expression of Oct6 or Sox10 MIC (containing amino acids 1–189 of Sox10) in western blots using polyclonal rabbit antisera directed against Oct6 (Sock et al. 1996) or Sox10 (Kuhlbrodt et al. 1998a). For pulldown experiments, GST or GST fusion proteins with amino acids 1–230 or 233–306 of Sox10 were expressed in Escherichia coli strain BL21 DE3. Harvested bacteria were resuspended in 50 mM NaH2PO4, 300 mM NaCl, 10 lg/mL leupeptin, 10 lg/mL aprotinin, 10 lg/mL DNaseI, 2.5 lg/mL lysozyme, and 0.1% Triton X-100. After sonification and centrifugation, protein was coupled to GSH-Sepharose beads in 4.3 mM Na2HPO4, 1.47 mM KH2PO4, 137 mM NaCl, and 2.7 mM KCl. In the same buffer, beads were incubated with extracts from Oct6-transfected HEK293 cells. After centrifugation and washing, bead-bound proteins were analyzed by western blot. For electrophoretic mobility shift assays (EMSA), 32P-labeled oligonucleotides spanning one or more neighboring Sox consensus motifs (for sequences see Figs 1a and 2a) were incubated with protein extracts from Sox10 MIC-transfected HEK293 cells for 20 min on ice in the presence of poly(deoxy-guanylate-deoxycytidylate) as unspecific competitor (Kuhlbrodt et al. 1998a). Samples were loaded onto native 5% polyacrylamide gels and electrophoresed in 0.5 · TBE (45 mM Tris/45 mM boric acid/ 1 mM EDTA, pH 8.3) at 120 V for 1.5 h. Gels were dried and exposed for autoradiography. Tissue preparation and immunohistochemistry Embryos homozygous for the Sox10DK2 allele were generated as described from heterozygous parents (Schreiner et al. 2007). At 18.5 days post coitum, mutant embryos and their wild-type littermates were recovered by Cesarean section and processed for immunohistochemistry (Stolt et al. 2003). Immunohistochemistry was performed on 10-lm-thick cryotome sections using anti-Sox10 guinea pig antiserum (1 : 2000 dilution, Maka et al. 2005) and antiKrox20 rabbit antiserum (1 : 500 dilution; Covance, Berkeley, CA, USA) as primary antibodies, as well as Cy2 and Cy3 immunofluorescent dyes (Dianova, Hamburg, Germany) as secondary antibodies. Samples were analyzed and documented using a DMIRB inverted microscope (Leica, Bensheim, Germany) equipped with a cooled MicroMax CCD camera (Princeton Instruments, Trenton, NJ, USA).


746 | S. Reiprich et al.

(a)

(b)

Fig. 1 Identification of Sox10 binding regions within the Krox20 MSE. (a) Sequence of the Krox20 MSE from mouse. Asterisks indicate those nucleotides that are fully conserved among mouse, human and chicken. The previously identified Oct6 binding sites are shown as grey boxes (Ghislain and Charnay 2006). Putative Sox10 binding sites are marked by Roman numerals and arrows. Grey arrowheads mark previously noted putative Sox10 binding sites, black arrowheads newly identified ones. (b) EMSA with radiolabelled double-stranded oligo-

nucleotides encompassing regions I-XI. Oligonucleotides were incubated in the absence ()), or presence (C, Sox10) of protein extracts before gel electrophoresis as indicated above the lanes. Extracts were from mock-transfected 293 cells (C) or 293 cells expressing Sox10 MIC (Sox10). Oligonucleotides with sites B and C/C¢ from the Mpz promoter (Peirano et al. 2000) served as control for high-affinity monomeric and dimeric Sox10 binding. All oligonucleotides were sizematched. m, bound monomer; d, bound dimer.

Results

by sequence inspection, but not functionally tested. Sox binding sites, however, adhere to a fairly loose heptameric consensus 5¢-(A/T)(A/T)CAA(A/T)G-3¢ (Harley et al. 1994). DNA binding of Sox proteins is furthermore substantially influenced by sequences flanking the heptamer, and cooperation between two Sox10 molecules allows closely spaced elements to be recognized despite significant deviations from the consensus sequence in one or both elements (Wegner

Sox10 binds to multiple regions within the 1.3 kb MSE In their original analysis of the MSE, Ghislain and Charnay (2006) had focused on POU binding sites in this Schwann cell-specific enhancer of the Krox20 gene and had mapped four such sites in the second half of the MSE (Fig. 1a). Potential binding sites for Sox proteins were also identified



(a)

Fig. 2 Identification of the exact high-affinity Sox10 binding sites within the Krox20 MSE. (a) Sequence of wildtype and mutant oligonucleotides used to identify Sox10 binding sites within regions I, II, VI, VII, VIII, IX, and XI of the MSE. Arrowheads indicate the location of the putative binding sites in wildtype oligonucleotides. Names of wildtype oligonucleotides consist of a Roman numeral, mutant versions carry an additional small letter suffix which is indicative of the remaining potential binding site. VIIa, for instance, still contains site ‘a’, whereas site ‘b’ is mutated. For wildtype oligonucleotides, the exact position within the MSE according to accession number AC153379 is given. Putative Sox10 binding sites are highlighted by capital letters and base changes introduced in the mutant oligonucleotide indicated by the loss of capital letters. Bold face is used to mark the experimentally confirmed Sox10 binding sites. (b) EMSA with radiolabelled doublestranded oligonucleotides carrying the different mutations in regions I, II, VI, VII, VIII, IX, and XI. Oligonucleotides were incubated in the absence ()) or presence (C, Sox10) of protein extracts before gel electrophoresis as indicated above the lanes. Extracts were from mock-transfected 293 cells (C) or 293 cells expressing Sox10 MIC (Sox10).

(b)

2009). Sox protein binding is therefore difficult to predict bioinformatically and must be validated by DNA-binding assays. To identify the relevant binding sites for Sox10 in the MSE, we first searched for sequences that had less than two mismatches compared with the heptameric Sox consensus. Additionally, sites with two mismatches were included when present in close proximity to sites with zero or one mismatch.

This analysis revealed 11 regions within the MSE that carried at least one or several closely spaced potential binding sites for Sox10 (Fig. 1a). These regions comprised all the sites previously identified by Ghislain and Charnay (2006) (Fig. 1a, indicated in grey), and revealed additional potential binding sites (Fig. 1a, indicated in black). Double-stranded oligonucleotides were synthesized from all 11 regions and analyzed in EMSA for their ability to bind



the MIC variant of Sox10 which contains amino acids 1–189 and exhibits the same binding specificity as the holoprotein (Kuhlbrodt et al. 1998b). Two Sox10 binding sites from the promoter of the Mpz gene served as positive control for sites that allow binding of a single Sox10 molecule (site B) or cooperative binding of two Sox10 molecules (site C/C¢) (Fig. 1b). Of all MSE regions tested, only regions I, VI, VII, VIII, IX, and XI bound Sox10 in amounts comparable with the controls (Fig. 1b). All of these regions bound a single Sox10 molecule despite the fact that several contained more than one potential Sox binding site. The only region that allowed cooperative Sox10 binding was region II (Fig. 1b). For further analysis, we concentrated on the six high-affinity monomeric sites and the dimeric site. To further locate the exact binding site in those regions where multiple potential Sox binding sites were present, we mutated at least the 5¢-CAA-3¢ core in every single existing heptameric consensus sequence (Fig. 2a) and analyzed respective oligonucleotides for their binding capacity (Fig. 2b). When only site ‘a’ remained intact in region I, Sox10 bound no longer (Fig. 2a and b). In contrast, Sox10 still bound as long as site ‘b’ was present (Fig. 2a,b). This allowed us to identify site ‘b’ in region I as the relevant one (Fig. 2b). Analogous EMSA helped us to locate binding to site ‘a’ in region VI, site ‘b’ in region VII, site ‘a’ in region VIII, site ‘a’ in region IX, and site ‘b’ in region XI (Fig. 2a and b). In the context of region II, site ‘b’ is essential as its presence allowed continuous Sox10 binding, although it shifted from dimeric to monomeric binding (Fig. 2b). Site ‘a’, in contrast, did not bind Sox10 on its own. A similar binding pattern has previously been observed for other dimeric sites (Peirano et al. 2000; Ludwig et al. 2004). Why site ‘b’ is preferred over site ‘a’ is not fully clear from sequence inspections. Both sites vary from the heptameric Sox consensus, but only site ‘a’ carries a variant base in the especially important central CAA core (Fig. 2a). Spacing and tail-to-tail arrangement of both sites are furthermore not uncommon for dimeric Sox10 binding sites (Wegner 2009). Sox10 binding sites differentially contribute to MSE activation Having identified six high-affinity monomeric and one dimeric binding site, we next asked how these sites contribute to the overall MSE activation by Sox10. We introduced mutations into each of the binding sites in the context of the full-length MSE and analyzed the consequence of each mutation in luciferase reporter gene assays. These assays were performed after transient transfection in the S16 Schwann cell line with reporter plasmids in which the luciferase gene was under control of the Krox20 MSE in combination with the b-globin minimal promoter. All constructs exhibited comparable basal activities except the MSE luciferase reporter construct with a site VIII mutation

Fig. 3 Contribution of Sox10 binding sites to MSE activation. Transient transfections were performed in S16 Schwann cells in the absence ()) or presence (+) of Sox10 with a luciferase reporter under the control of the Krox20 MSE in wild-type or various mutant versions and the b-globin minimal promoter. Mutant MSE either carried mutations in one of the seven binding sites (a) or in successively increasing numbers of binding sites (b). ‘M’ below the bars indicates which binding sites (Roman numerals on the left) are mutated in the particular MSE luciferase reporter. Luciferase activities in extracts from transfected cells were determined in three experiments each performed in duplicates. The luciferase activity obtained for each construct in the absence of Sox10 was arbitrarily set to 1 and fold induction in the presence of Sox10 was calculated. Data are presented relative to transactivation of the wildtype MSE by Sox10 with the SEM shown as error bar. An unpaired, two-tailed Student’s t-test was employed to determine whether Sox10-dependent activation rates of mutant MSE luciferase reporters were significantly different from the wildtype (***p £ 0.001; **p £ 0.01; *p £ 0.05; ns, not significant, p > 0.05).



which had approximately eightfold increased basal activities (data not shown). When the wild-type MSE was used, a robust induction was obtained by co-expressed Sox10 (Fig. 3a). Comparable induction rates were also obtained with an MSE with mutant site XI. In this mutant, we introduced changes in both the ‘b’ and ‘c’ heptamers, as we did not want to ignore the weak binding of Sox10 to site ‘c’ in region XI. All other single site mutations led to significant reductions in luciferase gene expression. However, the remaining rates of Sox10-dependent activation still ranged between one-fifth for the MSE with a site VI mutation and half for the MSE with a site VIII mutation (Fig. 3a). We conclude that Sox10 exerts its effects on the MSE via several binding sites and that the various sites contribute to different extents to the overall activation. None of the sites, however, appears so essential that upon its mutation, Sox10 looses its ability to activate the MSE. Therefore, we generated MSE constructs with increasing numbers of Sox10 binding site mutations (Fig. 3b). Starting with an MSE carrying a mutant site IX, additional mutation of site II already led to a further decrease in Sox10-dependent luciferase gene expression. Introduction of mutations into sites VI and I on top had no significant impact on luciferase activity. The remaining inducibility by Sox10 was only lost upon mutation of site VII. As expected, Sox10-dependent activation of the MSE did not recover upon mutation of the remaining sites XI and VIII. Thus, we had to abolish Sox10 binding to at least five of the seven identified sites to prevent Sox10-dependent activation of the MSE. As previous studies had shown that Sox10 is an even better activator of the MSE in the presence of Oct6 (Ghislain

and Charnay 2006), we also analyzed how the Sox10 binding site mutations affect the synergism between these two transcription factors. For the wild-type MSE, we indeed observed a more than additive increase in luciferase gene expression after co-transfection of both Sox10 and Oct6 (Fig. 4, compare grey striped bar with the empty bar which represents the calculated sum of single activation rates). With increasing number of Sox10 binding site mutations in the MSE, Sox10-dependent activation rates were lost as already observed in Fig. 3b. A parallel decrease in the activation rates obtained after transfection of Oct6 alone is likely caused by the progressive failure of cell-endogenous Sox10 to bind the MSE and contribute to reporter gene activation. In contrast to activation rates obtained with Sox10 or Oct6 alone, joint activation by co-transfected Sox10 and Oct6 remained high and stayed more than additive. Intriguingly, in the presence of Sox10 and Oct6, we still observed a 81-fold activation of an MSE in which all high-affinity Sox10 binding sites were mutated, even though there was virtually no MSE activation by Sox10 alone and only a sevenfold activation by Oct6 (Fig. 4). Although it may seem surprising that synergistic MSE activation is largely refractory to the mutation of high-affinity Sox10-binding sites, it should be kept in mind that further low-affinity Sox10 binding sites still exist.

Fig. 4 Contribution of Sox10 binding sites to the synergistic MSE activation by Sox10 and Oct6. Transient transfections were performed in S16 Schwann cells with luciferase reporters under the control of the b-globin minimal promoter and the Krox20 MSE carrying successively increasing numbers of binding site mutations. ‘M’ below the bars indicates which binding sites (Roman numerals on the left) are mutated in the particular MSE luciferase reporter. Luciferase reporters were transfected alone (black bars), in the presence of Sox10 (striped bars), Oct6 (grey bars), or a combination of both (grey striped bars) as

effectors. Data are presented as fold induction, with the luciferase activity obtained for each construct in the absence of effector arbitrarily set to 1 ± SEM. The empty bars represent the calculated sum from the single transcription factor transfections for each luciferase reporter. A one-sample t-test was employed to determine whether the experimentally observed activation rate obtained by combination of Sox10 and Oct6 was significantly different from the calculated sum of activation rates obtained with each factor alone (***p £ 0.001; **p £ 0.01; *p £ 0.05; ns, not significant, p > 0.05).

Synergism on the MSE is specific for SoxE and class III POU proteins Schwann cells express other POU proteins as well, including Brn2, Brn5, and Oct1. Previous studies had shown that Brn2 activates the Krox20 MSE as efficiently as Oct6 and can also



presence of Sox10 and Brn5 or Oct1 were even lower than the rates obtained with Sox10 alone, but this difference did not reach statistical significance. This argues that not only MSE activation, but also synergism with Sox10 is specific to class III POU proteins. To address the specificity on the side of the Sox protein, we also exchanged Sox10 for other Sox proteins in luciferase reporter gene assays (Fig. 5b). These studies revealed that both Sox8 and Sox9 activated the MSE at least as efficiently as Sox10, whereas Sox2 as a known inhibitor of Schwann cell differentiation and myelination failed to do so (Le et al. 2005). Sox8 and Sox9 were also able to cooperate with Oct6 and synergistically activated the MSE. Sox2, in contrast, did not synergize with Oct6 in agreement with its biological activity in Schwann cells. Sox8 and Sox9 are closely related to Sox10 and belong to the same SoxE subgroup, whereas Sox2 is a member of the more distant SoxB1 subgroup (Wegner 1999). It is thus reasonable to assume that MSE activation and synergism are restricted to SoxE proteins.

Fig. 5 Specificity of the synergistic MSE activation. Transient transfections were performed in S16 Schwann cells with a luciferase reporter under the control of Krox20 MSE and the b-globin minimal promoter to determine the specificity of MSE activation and synergism for the POU protein (a) as well as for the Sox protein (b). Luciferase reporters were transfected alone ()) or in the presence of various effectors including the POU proteins Oct6, Brn5 and Oct1, the Sox proteins Sox10, Sox9, Sox8 and Sox2 and various combinations thereof. The luciferase activity in the absence of effector was arbitrarily set to 1 and fold induction in the presence of effectors was calculated. Data are presented relative to transactivation of the MSE by Sox10 and Oct6 with the SEM shown as error bar. A one-way analysis of variance (ANOVA) with Tukey’s multiple comparison post-test was employed to determine whether transactivation rates obtained for transcription factor combinations were significantly different from the transactivation rate obtained for Sox10 alone (a) or Oct6 alone (b). ***p £ 0.001; **p £ 0.01; *p £ 0.05; ns, not significant, p > 0.05.

synergize with Sox10 (Ghislain and Charnay 2006). Brn2 is highly related to Oct6 and belongs to the same class III of the POU protein family. The more distantly related POU proteins Brn5 and Oct1, in contrast, failed to activate the MSE. Here, we analyzed whether Brn5 and Oct1 are able to synergize with Sox10 on the MSE enhancer (Fig. 5a). Not only did we confirm that Brn5 and Oct1 failed to induce reporter gene expression from the MSE, both proteins also did not cooperate with Sox10. The MSE induction rates in the joint

Specific domains in Oct6 and Sox10 are required for synergistic MSE activation Considering that not only MSE activation but also synergism is specific for Sox10, Oct6 and their respective close relatives, we also analyzed which domain in both proteins was required for synergistic MSE activation. First, we concentrated on Oct6 (Fig. 6a and b). When wild-type Oct6 was replaced by a mutant version in which all residues aminoterminal to the POU domain were deleted (Fig. 6a), synergism with Sox10 was still observed, although activation by Oct6DN alone was strongly reduced (Fig. 6b). The transactivation domain in the aminoterminal region is thus necessary for activation by Oct6 on its own, but dispensable for synergism with Sox10. Similarly, loss of all residues carboxyterminal to the POU domain had no detrimental impact on synergism. However, when the POU domain was deleted (Fig. 6a), synergism was lost (Fig. 6b). Synergism also did not survive the selective loss of either of the two POU subdomains. Neither deletion of the POU-specific domain (Oct6DPOUS) nor the POU homeodomain (Oct6D POUHD) was compatible with synergism (Fig. 6a and b). Differences between the loss of the whole POU domain or its subdomains were statistically not significant. Even the exchange of two amino acids within the POU homeodomain (WF-CS mutant) that prevents DNA-binding abolished synergism with Sox10 completely. This shows that the POU domain and its DNA-binding ability are absolutely necessary for synergistic activation. Whether a DNA-binding POU domain is also sufficient for synergism, was analyzed by co-transfection of the Oct6 POU domain with the Krox20 MSE luciferase reporter in the presence or absence of Sox10 (Fig. 6b). Although the POU domain failed to activate the MSE luciferase reporter on its own because of the lack of a



(a)

(c)

(b)

(d)

Fig. 6 Domain requirements for synergistic MSE activation. (a) Schematic representation of Oct6 mutant proteins. (b) A luciferase reporter under control of the Krox20 MSE and the b-globin minimal promoter was transfected alone ()) in S16 Schwann cells or in the presence of Sox10, various Oct6 mutants and combinations between the two. (c) Schematic representation of Sox10 mutant proteins. (d) The Krox20 MSE luciferase reporter was transfected alone ()) in S16 Schwann cells or in the presence of Oct6, various Sox10 mutants and combinations between the two. The luciferase activity in the absence

of effector was arbitrarily set to 1 and fold induction in the presence of effectors was calculated. Data are presented relative to transactivation of the MSE by Sox10 and Oct6 with the SEM shown as error bar. A one-way analysis of variance (ANOVA) with Tukey’s multiple comparison post-test was employed to determine whether transactivation rates obtained for transcription factor combinations were significantly different from the transactivation rate obtained for Sox10 alone (b) or Oct6 alone (d). ***p £ 0.001; **p £ 0.01; *p £ 0.05; ns, not significant, p > 0.05.

transactivation domain, it readily stimulated Sox10-dependent reporter gene activation indicating that the POU domain is indeed sufficient for synergistic MSE activation. With respect to Sox10, variants were tested in which specific domains were removed or inactivated in their function (Fig. 6c). When DNA-binding of Sox10 was destroyed by in-frame insertion of two amino acids into the DNA-binding HMG domain (WS95 mutant), both MSE activation and synergism with Oct6 were completely lost (Fig. 6d). Similarly, removal of the carboxyterminal transactivation domain of Sox10 (Q377X mutant) interfered with MSE activation and synergism. Activation rates obtained in the presence of Oct6 and one of the two Sox10 mutants were even lower than those obtained for Oct6 alone. In contrast, mutational inactivation of the dimerization domain of Sox10 (aa1 mutant) neither interfered with MSE activation nor

synergism with Oct6. This agrees with the fact that Sox10 dimer binding does not play a major role in the interaction of Sox10 with the MSE. The most interesting Sox10 mutant, however, was the DK2 mutant in which amino acids 233–306 are deleted from the central region of the protein (Fig. 6c). This mutant activated the MSE in a manner very similar to wildtype Sox10. At the same time, it failed to exhibit any synergistic effect in the presence of Oct6 arguing that synergism is selectively impaired in this mutant and that the K2 domain has an essential role in mediating the synergism (Fig. 6d). The K2 domain is specifically required for the synergistic activation of Krox20 Considering the intriguing involvement of the K2 domain in the synergistic activation of the MSE by Sox10 and Oct6, we



(a)

(b)

(c)

Fig. 7 Role of the K2 domain of Sox10 in synergistic MSE activation. (a) The Krox20 MSE luciferase reporter was transfected alone ()) in S16 Schwann cells or in the presence of Brn2, wildtype Sox10, the Sox10 DK2 mutant or various combinations thereof. The luciferase activity in the absence of effector was arbitrarily set to 1 and fold induction in the presence of effectors was calculated. Data are presented relative to transactivation rates by Sox10 and Brn2 with the SEM shown as error bar. A one-way analysis of variance (ANOVA) with Tukey’s multiple comparison post-test was employed to determine whether transactivation rates obtained for transcription factor combinations were significantly different from the transactivation rate obtained for Brn2 alone (***p £ 0.001; **p £ 0.01; *p £ 0.05; ns, not significant, p > 0.05). (b) Extracts from transiently transfected Oct6 expressing 293 cells were incubated with GSH-Sepharose beads carrying immobilized GST or GST fusions with amino acids 1–230 of Sox10 or its K2 domain. Oct6 protein pulled-down with the beads was visualized by western blot with Oct6-specific antibodies. (c) Coimmunohistochemistry was performed on spinal nerves of wildtype (wt) and Sox10DK2/DK2 embryos at 18.5 days post coitum using antibodies against Sox10 (green) and Krox20 (red). The merge is shown in the right panels.

next asked whether it is similarly important for synergistic activation by Sox10 and Brn2. Transient transfections in S16 cells indeed showed that the DK2 mutant also failed to enhance the Brn2-dependent activation of the MSE, whereas wildtype Sox10 synergistically activated the MSE in combination with Brn2 just as well as in combination with Oct6 (Fig. 7a). The easiest mechanistic explanation for the role of the K2 domain in the synergism would be an involvement of this domain in direct protein–protein interaction with Oct6 and

Brn2. GST-pulldown experiments, however, failed to provide any evidence that the K2 domain is able to bind and precipitate Oct6 (Fig. 7b). Such an interaction was only observed in parallel GST-pulldowns between the aminoterminal part of Sox10 and Oct6 (Fig. 7b). As previously shown, this is likely because of direct interaction between the DNA-binding domains of both proteins with the carboxyterminal tail involved on the side of the Sox10 HMG domain (Remenyi et al. 2003; Wißmu¨ller et al. 2006). Even if the K2 domain does not directly interact with Oct6, it could still indirectly strengthen the HMG domain-mediated interaction with Oct6. Therefore, we also performed coimmunoprecipitation experiments. However, we failed to detect any difference between wildtype Sox10 and the DK2 mutant in their respective abilities to co-precipitate Oct6 (data not shown). It thus appears unlikely that the K2 domain exerts its role in synergism through protein–protein interactions with Oct6. Finally, if the K2 domain is required for synergism with both Oct6 and Brn2 and if synergistic activation of the MSE is in turn essential for Krox20 induction in myelinating Schwann cells, we would predict that Krox20 is not induced in Schwann cells in the absence of the K2 domain. We have previously generated a mouse mutant in which we have replaced the wild-type Sox10 by a mutant Sox10 without K2 domain (Schreiner et al. 2007). We had shown that Schwann cells in these mice normally enter the promyelin stage and express Oct6, but fail to turn on myelin gene expression. What we had not analyzed, was whether the DK2 mutant failed to activate Krox20 expression as indicated by the present data. To address this issue, we performed immunohistochemistry on spinal nerves of mouse embryos at 18.5 days post coitum (Fig. 7c). These analyses indeed confirmed that Krox20 is readily detected in the Schwann cells along the spinal nerves of wild-type embryos, but completely missing in Schwann cells of the DK2 mutant, thus impressively confirming in vivo the role of the K2 domain in Krox20 induction.

Discussion Sox10 is already required during the earliest phases of Schwann cell development. Nevertheless, it continues to be expressed even into the myelinating stage (Kuhlbrodt et al. 1998a; Britsch et al. 2001). This has led to the assumption that Sox10 may also be required for the myelination process. Although this has not yet been formally proven in an animal model, evidence has been presented from many groups that Sox10 directly regulates the expression of myelin genes that are strongly up-regulated during myelination and code for structural components of the peripheral myelin sheath (Peirano et al. 2000; Bondurand et al. 2001; Jones et al. 2007; LeBlanc et al. 2007). Sox10 furthermore cooperates with Krox20 during myelin gene induction.



Sox10 may also be involved in Krox20 induction when Schwann cells transit from the promyelin to the myelin stage. This has been deduced from the fact that the MSE which drives Krox20 expression in myelinating Schwann cells contains binding sites for both Oct6 and Sox10, and is synergistically activated by this transcription factor combination in tissue culture (Ghislain and Charnay 2006). Again, there is no formal proof for this epistatic relationship in a mouse model. Such proof is now provided for the first time in a hypomorphic mouse model in which wild-type Sox10 was replaced by a mutant Sox10 without K2 domain (Schreiner et al. 2007). Schwann cells in this mouse mutant progress into the promyelin stage but fail to induce Krox20 despite the presence of Oct6. This verifies the role of Sox10 in Krox20 induction. The MSE as the enhancer responsible for Krox20 expression in myelinating Schwann cells contains multiple, highly conserved binding sites for Sox10 of which six exhibited particularly high affinity for Sox10 monomers and one allowed binding of Sox10 dimers. Mutational analyses of these sites furthermore showed that all contribute, but none is absolutely required for Sox10-dependent MSE activation. Activation was, however, completely lost once five out of seven sites were mutated. This argues that Sox10 is only functional upon binding to at least three of the high-affinity sites in the MSE. It furthermore implies that the MSE can only be activated once Sox10 levels reach a certain threshold and that its activity is sensitive to the actual Sox10 levels. This could be relevant for Krox20 induction at the onset of myelination, if Sox10 levels increase in actively myelinating Schwann cells, but could also help to lower Krox20 levels once the active phase of myelination is over, if Sox10 levels again decline in the steady-state. Thus, it might be worthwhile to quantify Sox10 levels during Schwann cell development. This has not been done so far. The reliance on multiple binding sites also implicates that the synergism between Sox10 and Oct6 is unlikely to be mediated by a single composite element with adjacent binding sites for Oct6 and Sox10 in which DNA binding of one factor strengthens the ability of the other to bind next to it. Such cooperative DNA binding has been shown as the basis of synergistic activation for Sox2 and several of its partner proteins such as Oct4 during the activation of embryonic stem cell-specific genes, and Pax6 in the activation of lens-specific genes (for review, see Kamachi et al. 2000; Wegner 2005). Here, however, we failed to obtain any evidence for cooperative DNA binding of Sox10 and its partner Oct6 in EMSA with MSE regions where Sox binding sites abut the previously mapped Oct6 sites (data not shown). Although the remaining levels of MSE activation may not be sufficient to activate Krox20 expression in vivo, synergistic activation of the MSE was even maintained under conditions where all high-affinity Sox10 binding sites were inactivated.

The fact that several low-affinity binding sites for Sox10 in the MSE can substitute under these circumstances also argues against the essential role of cooperative DNA-binding to composite elements in the observed synergism. Analyses of transcription factor mutants that have lost their DNA-binding ability because of minor changes in their primary structure argue that both Sox10 and Oct6 have to be able to bind independently to DNA for synergism to occur. It is also noteworthy, that the POU domain of Oct6 is sufficient for synergism, and that the transactivation domain in the aminoterminal part of the protein is dispensable. As absolute levels of MSE activation invariably drop in the absence of the Oct6 transactivation domain, it is again difficult to predict whether the remaining synergism would be sufficient for Krox20 induction in vivo. The POU domain of Oct6 is furthermore strongly conserved in other class III POU proteins such as Brn2 and Brn1. This may explain why the closely related Brn2 and Brn1 proteins can substitute for Oct6 both in vitro during synergistic MSE activation and in vivo during Schwann cell development (Jaegle et al. 2003; Friedrich et al. 2005; Ghislain and Charnay 2006). Sox10, in contrast, requires both a functional DNAbinding domain and the transactivation domain in the carboxyterminal part of the protein. Again, both domains are highly conserved in other closely related SoxE proteins such as Sox8 and Sox9 so that it is not unexpected that they can substitute for Sox10 in MSE activation in vitro as well as Krox20 induction in vivo (Kellerer et al. 2006). The requirement for the POU domain may indicate that unlike the situation in EMSA, Sox10 can bind to DNA better if the MSE is already occupied or simultaneously contacted by Oct6. Such an effect may be indirectly mediated through Oct6-dependent changes in the chromatin structure of the MSE or through direct protein–protein interactions between the POU domain of Oct6 and the HMG domain of Sox10 (Wißmu¨ller et al. 2006). Quite intriguingly, there also is a role of the K2 domain of Sox10 in this synergism, although there is no evidence to suggest direct protein–protein interactions between the K2 domain and Oct6. It appears more likely that K2 domain and Oct6 independently recognize a common interaction partner that helps to keep the two proteins in a joint complex. Alternatively, interaction between the POU domain of Oct6 and the HMG domain of Sox10 may lead to conformational changes in Sox10. These conformational changes may eventually be transmitted to the transactivation domain to transform it from an inactive into an active state, and the K2 domain may be involved in the process. It is currently impossible to distinguish between these possibilities. What is clear, however, is the fact that the synergism between Oct6 and Sox10 on the MSE is mechanistically complex, but nevertheless responsible for timely induction of Krox20 at the onset of myelination in Schwann cells. It further highlights the fact that Sox10



influences myelination through at least two different routes, i.e., as an inductor and as an interaction partner for Krox20.

Acknowledgements This work was supported by a grant from the Deutsche Forschungsgemeinschaft to M.W. (We1326/8-1). Steffi Scholz is acknowledged for expert technical assistance.

References Bondurand N., Girard M., Pingault V., Lemort N., Dubourg O. and 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. Britsch S., Goerich D. E., Riethmacher D., Peirano R. I., Rossner M., Nave K. A., Birchmeier C. and Wegner M. (2001) The transcription factor Sox10 is a key regulator of peripheral glial development. Genes Dev. 15, 66–78. Friedrich R., Schlierf B., Tamm E. R., Bo¨sl M. R. and Wegner M. (2005) The class III POU domain protein Brn-1 can fully replace the related Oct-6 during Schwann cell development and myelination. Mol. Cell. Biol. 25, 1821–1829. Ghislain J. and Charnay P. (2006) Control of myelination in Schwann cells: a Krox20 cis-regulatory element integrates Oct6, Brn2 and Sox10 activities. Embo Rep. 7, 52–58. Ghislain J., Desmarquet-Trin-Dinh C., Jaegle M., Meijer D., Charnay P. and Frain M. (2002) Characterisation of cis-acting sequences reveals a biphasic, axon-dependent regulation of Krox20 during Schwann cell development. Development 129, 155–166. Harley V. R., Lovell-Badge R. and Goodfellow P. N. (1994) Definition of a consensus DNA binding site for SRY. Nucleic Acids Res. 22, 1500–1501. Jaegle M., Ghazvini M., Mandemakers W., Piirsoo M., Driegen S., Levavasseur F., Raghoenath S., Grosveld F. and Meijer D. (2003) The POU proteins Brn-2 and Oct-6 share important functions in Schwann cell development. Genes Dev. 17, 1380– 1391. Jessen K. R. and Mirsky R. (2005) The origin and development of glial cells in peripheral nerves. Nat. Rev. Neurosci. 6, 671–682. Jones E. A., Jang S. W., Mager G. M., Chang L. W., Srinivasan R., Gokey N. G., Ward R. M., Nagarajan R. and Svaren J. (2007) Interactions of Sox10 and Egr2 in myelin gene regulation. Neuron Glia Biol. 3, 377–387. Kamachi Y., Uchikawa M. and Kondoh H. (2000) Pairing SOX off: with partners in the regulation of embryonic development. Trends Genet. 16, 182–187. Kao S. C., Wu H., Xie J., Chang C. P., Ranish J. A., Graef I. A. and Crabtree G. R. (2009) Calcineurin/NFAT signaling is required for neuregulin-regulated Schwann cell differentiation. Science 323, 651–654. Kellerer S., Schreiner S., Stolt C. C., Bo¨sl M. R. and Wegner M. (2006) Functional equivalency of transcription factors Sox8 and Sox10 is tissue-specific. Development 133, 2875–2886. Kuhlbrodt K., Herbarth B., Sock E., Hermans-Borgmeyer I. and Wegner M. (1998a) Sox10, a novel transcriptional modulator in glial cells. J. Neurosci. 18, 237–250. Kuhlbrodt K., Schmidt C., Sock E., Pingault V., Bondurand N., Goossens M. and Wegner M. (1998b) Functional analysis of Sox10 mutations found in human Waardenburg-Hirschsprung patients. J. Biol. Chem. 273, 23033–23038.

Le N., Nagarajan R., Wang J. Y., Araki T., Schmidt R. E. and Milbrandt J. (2005) Analysis of congenital hypomyelinating Egr2Lo/Lo nerves identifies Sox2 as an inhibitor of Schwann cell differentiation and myelination. Proc. Natl. Acad. Sci. USA 102, 2596–2601. LeBlanc S. E., Ward R. M. and Svaren J. (2007) Neuropathy-associated Egr2 mutants disrupt cooperative activation of myelin protein zero by Egr2 and Sox10. Mol. Cell. Biol. 27, 3521–3529. Ludwig A., Rehberg S. and Wegner M. (2004) Melanocyte-specific expression of dopachrome tautomerase is dependent on synergistic gene activation by the Sox10 and Mitf transcription factors. FEBS Lett. 556, 236–244. Maka M., Stolt C. C. and Wegner M. (2005) Identification of Sox8 as a modifier gene in a mouse model of Hirschsprung disease reveals underlying molecular defect. Dev. Biol. 277, 155–169. Nagarajan R., Svaren J., Le N., Araki T., Watson M. and 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. and Wegner M. (2000) Protein zero expression is regulated by the glial transcription factor Sox10. Mol. Cell. Biol. 20, 3198–3209. Remenyi A., Lins K., Nissen L. J., Reinbold R., Scho¨ler H. R. and Wilmanns M. (2003) Crystal structure of a POU/HMG/DNA ternary complex suggests differential assemby of Oct4 and Sox2 on two enhancers. Genes Dev. 17, 2048–2059. Renner K., Leger H. and Wegner M. (1994) The POU-domain protein Tst-1 and papovaviral T-antigen function synergistically to stimulate glia-specific gene expression of JC virus. Proc. Natl Acad. Sci. USA 91, 6433–6437. Schlierf B., Ludwig A., Klenovsek K. and Wegner M. (2002) Cooperative binding of Sox10 to DNA: requirements and consequences. Nucleic Acids Res. 30, 5509–5516. Schreiber J., Enderich J., Sock E., Schmidt C., Richter-Landsberg C. and Wegner M. (1997) Redundancy of class III POU proteins in the oligodendrocyte lineage. J. Biol. Chem. 272, 32286–32293. Schreiner S., Cossais F., Fischer K., Scholz S., Bo¨sl M. R., Holtmann B., Sendtner M. and Wegner M. (2007) Hypomorphic Sox10 alleles reveal novel protein functions and unravel developmental differences in glial lineages. Development 134, 3271–3281. Sock E., Enderich J., Rosenfeld M. G. and Wegner M. (1996) Identification of the nuclear localization signal of the POU domain protein Tst-1/Oct6. J. Biol. Chem. 271, 17512–17518. Sock E., Pagon R. A., Keymolen K., Lissens W., Wegner M. and Scherer G. (2003) Loss of DNA-dependent dimerization of the transcription factor SOX9 as a cause for campomelic dysplasia. Hum. Mol. Genet. 12, 1439–1447. Stolt C. C., Lommes P., Sock E., Chaboissier M.-C., Schedl A. and Wegner M. (2003) The Sox9 transcription factor determines glial fate choice in the developing spinal cord. Genes Dev. 17, 1677– 1689. Topilko P., Schneider M. S., Levi G., Baron V. E. A., Chennoufi A. B., Seitanidou T., Babinet C. and Charnay P. (1994) Krox-20 controls myelination in the peripheral nervous system. Nature 371, 796– 799. Wegner M. (1999) From head to toes: the multiple facets of Sox proteins. Nucleic Acids Res. 27, 1409–1420. Wegner M. (2005) Secrets to a healthy Sox life: Lessons for melanocytes. Pigment Cell Res. 18, 74–85. Wegner M. (2009) All purpose Sox: the many roles of Sox proteins in gene expression. Int. J. Biochem. Cell Biol. doi: 10.1016. Wißmu¨ller S., Kosian T., Wolf M., Finzsch M. and Wegner M. (2006) The High-mobility-group domain of Sox proteins interacts with the DNA-binding domains of many transcription factors. Nucleic Acids Res. 34, 1735–1744.
