SOX10, in combination with Sp1, regulates the endothelin receptor type B gene in human melanocyte lineage cells

Authors

  • Satoru Yokoyama,

    1. Department of Molecular Biology and Applied Physiology, Tohoku University School of Medicine, Seiryo-machi, Aoba-ku, Sendai, Miyagi, Japan
    Search for more papers by this author
  • Kazuhisa Takeda,

    1. Department of Molecular Biology and Applied Physiology, Tohoku University School of Medicine, Seiryo-machi, Aoba-ku, Sendai, Miyagi, Japan
    Search for more papers by this author
  • Shigeki Shibahara

    1. Department of Molecular Biology and Applied Physiology, Tohoku University School of Medicine, Seiryo-machi, Aoba-ku, Sendai, Miyagi, Japan
    Search for more papers by this author

K. Takeda, Department of Molecular Biology and Applied Physiology, Tohoku University School of Medicine, 2–1 Seiryo-machi, Aoba-ku, Sendai, Miyagi 980–8575, Japan
Fax: +81 22 7178118
Tel: +81 22 7178114
E-mail: ktakeda@mail.tains.tohoku.ac.jp

Abstract

Waardenburg syndrome (WS) is an auditory–pigmentary disorder that exhibits varying combinations of sensorineural hearing loss and abnormal pigmentation of the hair and skin. WS type 4 (WS4), a subtype of WS, is characterized by the presence of the aganglionic megacolon and is associated with mutations in the gene encoding either endothelin 3, endothelin receptor type B (EDNRB), or Sry-box 10 (SOX10). Here, we provide evidence that SOX10 regulates the expression of EDNRB gene in human melanocyte-lineage cells, as judged by RNA interference and chromatin immunoprecipitation analyses. Human melanocytes preferentially express the EDNRB transcripts derived from the conventional EDNRB promoter. SOX10 transactivates the EDNRB promoter through the cis-acting elements, the two CA-rich sequences and the GC box. Moreover, a transcription factor Sp1 enhances the degree of the SOX10-mediated transactivation of the EDNRB promoter through these cis-acting elements. Furthermore, we have shown that the EDNRB promoter is heavily methylated in HeLa human cervical cancer cells, lacking EDNRB expression, but not in melanocytes and HMV-II melanoma cells. The expression of EDNRB became detectable in HeLa cells after treatment with a demethylating reagent, 5′-aza-2′-deoxycytidine, which was further enhanced in the transformed cells over-expressing SOX10. We therefore suggest that SOX10, alone or in combination with Sp1, regulates transcription of the EDNRB gene, thereby ensuring appropriate expression level of EDNRB in human melanocytes.

Abbreviations
ChIP

chromatin immunoprecipitaion

EDN

endothelin

EDNRA

endothelin receptor type A

EDNRB

endothelin receptor type B

EMSA

electrophoretic mobility shift assay(s)

ENS

enteric nervous system

HMG

high mobility group

MITF

microphthalmia-associated transcription factor

PAX3

paired box gene 3

siRNA

small interfering RNA

SOX10

Sry-box 10

WS

Waardenburg syndrome

YDBS

Yemenite deaf–blind hypopigmentation syndrome

Waardenburg syndrome (WS) is an auditory–pigmentary disorder, which is characterized by varying combinations of sensorineural hearing loss, heterochromia iridis, and patchy abnormal pigmentation of the hair and skin [1]. WS is associated with the deficiency of neural crest-derived melanocytes, and is classified into four types, depending on the presence or absence of additional symptoms [2–9]. WS type1 (WS1) and type 2 (WS2) are distinguished by the presence or absence of dystopia canthorum, respectively. The presence of limb abnormalities distinguishes WS type 3 (WS3) from WS2. WS type 4 (WS4), referred to as Hirschsprung's disease type 2 or Shaa–Waardenburg syndrome, is characterized by the presence of the aganglionic megacolon. WS1 and WS3 are caused by mutations in the PAX3 gene [2], and some cases of WS2 are associated with mutations in the microphthalmia-associated transcription factor (MITF) gene [3] or SLUG (SNAI2) gene [4]. WS4 is due to mutations in the endothelin receptor type B (EDNRB) gene [5,6], the endothelin 3 (EDN3) gene [7,8], or the Sry-box 10 (SOX10) gene [9].

Several lines of evidence have suggested the functional relationships between these WS-related transcriptional factors. Expression of Mitf, a basic helix-loop-helix-leucine zipper protein, is critical for the development of melanocytes [10] and is preceded by the expression of Pax3 and Sox10 in melanoblasts located dorsal to the neural tube and migrated along the dorsolateral pathway [11,12]. PAX3 affects the development of melanocytes in culture by regulating MITF expression [13]. Likewise, SOX10 activates the MITF gene promoter [14–17].

Recently, Sox10 has been shown to regulate Ednrb expression in the precursors of the enteric nervous system (ENS) through the ENS enhancers, which contain the Sox10-binding sites, in the mouse Ednrb gene [18]. In fact, Sox10 mRNA and Ednrb mRNA exhibit overlapping expression patterns in neural crest derivatives in wild-type mice [19], whereas the Ednrb expression is reduced in the dominant megacolon (Dom) mouse, which carries the truncated mutation of Sox10 gene [19]. The Dom mouse represents a model for human congenital megacolon [19,20].

The SOX genes encode transcription factors with a high-mobility group box (HMG box) as a DNA-binding motif [21]. SOX10 is defective in some cases of WS4 [9,22,23] and in patients with Yemenite deaf–blind hypopigmentation syndrome (YDBS) [24]. YDBS is a rare disorder characterized by severe early hearing loss, microcornea and colobomata, and cutaneous pigmentation abnormalities. These two syndromes exhibit a remarkable difference in the phenotype of the enteric nervous system; namely, aganglionic megacolon is associated with WS4, but not with YDBS.

The mutations in the EDNRB gene are associated with WS4, which is inherited in a dominant [6] or recessive [5] mode. EDNRB belongs to a superfamily of G protein-coupled receptors [25]. Its ligand, endothelin (EDN), is a highly potent vasoconstricting peptide of 21 amino acid residues and consists of three subtypes EDN1, EDN2, and EDN3 [26]. EDNRB has a high affinity for all three EDNs, whereas endothelin receptor type A (EDNRA), a subtype of EDNR, possesses a higher affinity for EDN1 and EDN2 than EDN3 [25,27,28]. The Ednrb gene is expressed postnataly in various tissues, including the myenteric plexus, mucosal layer, ganglia, and blood vessels of the submucosa of the colon [29,30]. Importantly, the Edn3-Ednrb signaling is required for the terminal migration of melanoblasts and the precursors of the ENS [31,32]. However, little is known about the regulatory mechanism of Ednrb expression in melanocytes.

We have investigated the hierarchy between SOX10 and EDNRB in melanocyte-lineage cells. EDNRB mRNA consists of at least four transcripts in human melanoma cells, which are derived from different promoters [33], termed conventional EDNRB, EDNRBΔ1, EDNRBΔ2, and EDNRBΔ3. Each of the conventional EDNRB, EDNRBΔ1 and EDNRBΔ2 transcripts encodes the same EDNRB protein of 442 amino acids, while the EDNRBΔ3-encoded protein contains additional N-terminal amino acids (89 residues), followed by the EDNRB protein [33]. Here, we show that the conventional EDNRB mRNA is preferentially expressed in melanocytes, and that SOX10 activates the conventional EDNRB promoter, alone or in combination with a transcription factor Sp1. Furthermore, the EDNRB promoter is heavily methylated in HeLa human cervical cancer cells, which do not express the EDNRB gene, but not methylated in melanocytes. Notably, enforced expression of SOX10 induces the expression of EDNRB in HeLa cells only when HeLa cells were treated with a demethylating reagent, 5′-aza-2′-deoxycytidine. The present study suggests that SOX10 is responsible for appropriate expression of the EDNRB gene in human melanocyte-lineage cells.

Results and discussion

SOX10 is required for the expression of EDNRB gene

To investigate the expression profiles of SOX10 and EDNRB mRNAs in normal human epidermal melanocytes (NHEM) and human melanoma cells, we carried out northern blot analysis (Fig. 1A). SOX10 and EDNRB mRNAs are coexpressed in NHEM and four human melanoma cell lines. EDNRB mRNAs were detected as two major bands of about 4300 and 1800 nucleotides, which are generated by the use of alternative polyadenylation sites [34]. To explore the hierarchy among SOX10, EDNRB, and MITF, we carried out the RNA interference analysis against SOX10 in HMV-II melanoma cells (Fig. 1B). We initially confirmed that the expression of SOX10 protein was reduced by the small interfering RNA (siRNA) against SOX10 (65% reduction), but not by the LacZ siRNA (Fig. 1B). Because SOX10 has been known as a transactivator for MITF gene [14–17], MITF could be used as a positive control for the SOX10 siRNA. The expression of MITF protein was reduced by the SOX10 siRNA (69% reduction), but not changed by the LacZ siRNA. These results confirm the regulatory role of SOX10 in the expression of MITF, thereby indicating that the SOX10 siRNA worked properly in HMV-II cells. Likewise, EDNRB protein was reduced by the SOX10 siRNA (46% reduction). Subsequently, we examined the expression of EDNRB mRNA in those HMV-II cells by northern blot analysis, showing that the expression of EDNRB mRNA was reduced in HMV-II cells when the SOX10 siRNA was transfected (upper band 15% and lower band 35% reduction) (Fig. 1C). We repeated a similar experiment with the SOX10 siRNA and confirmed the reduced expression of EDNRB mRNA (upper band 10%, lower band 24% reduction) (data not shown). These results suggest that SOX10 is required for the expression of EDNRB gene in human melanocyte-lineage cells. Under the conditions used, there were no noticeable changes in the viability of the cells transfected with the SOX10 siRNA, despite that SOX10 is required for melanocyte survival. It is conceivable that the reduced level of SOX10 in the experiment does not affect cell survival. Alternatively, certain genes may compensate for the down-regulation of SOX10 expression.

Figure 1.

MITF and EDNRB genes are regulated by SOX10. (A) Northern blot analysis of SOX10 and EDNRB mRNA. Total RNA was prepared from the indicated cell lines at the top of the panel. Autoradiograms of the RNA blots hybridized with the indicated 32P-labeled cDNA probes are shown. The bottom panel shows the expression of 28S ribosomal RNA visualized by ultraviolet transilluminator (internal control). EDNRB mRNA is detected as two bands, about 4300 nucleotides and 1800 nucleotides as described previously [34]. (B) Effects of siRNA against SOX10 on the expression of SOX10, MITF and EDNRB protein. Whole cell extracts were prepared from HMV-II human melanoma cells transfected with each siRNA against SOX10 (siSOX10) or LacZ (siLacZ), or untransfected cells (–). The bottom panel shows α-tubulin as an internal control. (C) The effect of siRNA against SOX10 on the expression of EDNRB mRNA. Total RNA was prepared from the indicated cells same as (B). β-actin mRNA was used as an internal control.

Identification of a major species of EDNRB gene transcripts

EDNRB mRNA consists of at least four transcripts with different 5′-ends (conventional EDNRB, EDNRBΔ1, EDNRBΔ2, and EDNRBΔ3), which are derived from three promoters of the human EDNRB gene [33](Fig. 2A). Thus, EDNRB mRNA may consist of eight isoforms, because each EDNRB transcript may have the two different 3′-ends (see Fig. 1), generated by the use of alternative polyadenylation sites [34]. However, because of small differences in the size of each transcript, it is practically impossible to identify the four transcripts by northern blot analysis. We therefore performed S1 nuclease mapping analysis to identify EDNRB transcripts expressed in NHEM and HMV-II melanoma cells (Fig. 2B). Three protected fragments of 436, 407, and 228 bases were detected in NHEM and HMV-II cells but not in HeLa cervical cancer cells. The two fragments of 436 and 407 bases are preferentially detected and consistent with the conventional EDNRB mRNA, transcribed from the two adjacent transcriptional initiation sites [35]. The faint signal of 228 bases represents the expression of EDNRBΔ2 mRNA or EDNRBΔ3 mRNA. In contrast, the signal for the EDNRBΔ1 transcript of 997 bases was undetectable. These results indicate that the conventional EDNRB mRNA is abundantly expressed in human melanocytes. Furthermore, the alternative promoters have not been reported in the mouse Ednrb gene. In the present study, we thus focused on the conventional promoter of the EDNRB gene, which is termed, the EDNRB promoter, unless otherwise specified.

Figure 2.

Identification of the major transcripts of the EDNRB gene in human melanocytes. (A) Schematic representation of the promoters of the EDNRB gene. The EDNRB gene is shown as a line and four arrows show the transcriptional initiation sites of each transcript. The open boxes of each transcript show the 5′-untranslated region. The shadow boxes and dotted boxes indicate the protein-coding region of each transcript. The transcripts derived from the conventional EDNRB, EDNRBΔ1 and the EDNRBΔ2 promoters code for the same protein. The protein encoded by EDNRBΔ3 mRNA has Δ3-specific N-terminal amino acids (dotted boxes), followed by the common coding region (shaded boxes). The splicing sites are shown by a gap. The number shown on the transcripts represents the position from the transcriptional initiation site (+1) of the conventional EDNRB mRNA. The conventional EDNRB mRNA has two transcriptional initiation sites (+1 and +30) [35]. The S1 probe, shown at the bottom, contains the EDNRB gene sequences (line) and the vector sequence (dotted line). The end-labeled site of S1 probe is the XcmI site (position +436 in the antisense strand of EDNRB cDNA), and is indicated with an asterisk. The predicted fragments are shown below the transcripts (boxes), together with each predicted size. (B) S1 nuclease mapping analysis of EDNRB transcripts. An autoradiogram of the S1 nuclease mapping analysis is shown. Total RNA was prepared from the indicated cell lines at the top of panel. The S1 probe and four predicted fragments are indicated as arrows. Size markers were end-labeled with 1-kb DNA Ladder and 100-bp DNA Ladder (New England Biolabs, Beverly, MA, USA).

Functional analysis of the EDNRB gene promoter in melanocyte-lineage cells

We first analyzed the promoter activity of the EDNRB gene by transient transfection assays in HMV-II melanoma cells and HeLa cervical cancer cells (Fig. 3A). The deletion study showed that the promoter activities of the EDNRB reporter constructs were higher in HMV-II cells than in HeLa cells by about twofold, except for a construct pGL3-E (−12), carrying the 12-base pairs promoter region. Likewise, the promoter activity of the EDNRB promoter was higher in normal human epidermal melanocytes and other human melanoma cell lines, 624mel (Fig. 3B), G361 and SK-MEL-28 (data not shown) than that detected in HeLa cells (Fig. 3B). Thus, the 5′-flanking region between −105 and −12 is required for the basal promoter activity of the EDNRB gene in melanocytes-lineage cells and may confer the marginal cell specificity on the ENDRB promoter. There was noticeable difference in the promoter activities in melanoma cells between pGL3-E (−3002) and pGL3-E (−105) (Fig. 3A), which suggests the presence of the negative elements for the promoter activity in the deleted region. Such a difference in the promoter activity was also detected in HeLa cells. On the other hand, the EDNRB promoter contains the three potential SOX10 sites (Fig. 3A), which correspond to the enteric nervous system (ENS) enhancer in the mouse Ednrb gene [18]. The expression levels of pGL3-Em, containing the mutations at the three potential SOX10 sites, were lower than those of a wild-type construct, pGL3-E (−3022), but the expression level of pGL3-Em is significantly higher in HMV-II cells than that in HeLa cells. This observation was also seen in normal human epidermal melanocytes, and 624mel human melanoma cell lines examined (Fig. 3B), and other human melanoma cell lines, G361 and SK-MEL-28 (data not shown). These results suggest that the potential SOX10 sites are dispensable for melanocytes lineage-specific expression of EDNRB, which is consistent in part with the report in the transgenic mouse analysis of the Ednrb gene [18].

Figure 3.

The EDNRB gene promoter shows the melanocyte lineage cell-specific activity. (A) Functional analysis of the EDNRB gene promoter in melanoma cells. The left panel shows the reporter plasmids used. The number shown at the 5′- or 3′-end of each construct represents the position from the transcriptional initiation site (+1). The open circles on the EDNRB gene show the potential SOX10 sites, which correspond to the enteric nervous system (ENS) enhancers in mice [18]. The internal deletion site is shown as a gap. X shows the mutation of each potential SOX10 site. Relative luciferase activity in transient transfection assays is shown on the right. The cell lines used are HeLa cells (open bars) and HMV-II cells (filled bars). Luciferase activity was normalized by each internal control activity (pRL-TK). Relative luciferase activity is shown as the ratio to the normalized luciferase activity obtained with pGL3-E (−3022) in HeLa cells. Data are mean ±sd of at least three independent experiments. The activity with * or ** is significantly higher in HMV-II cells than the value obtained with each reporter plasmid in HeLa cells, P < 0.01 or P < 0.05. (B) Functional analysis of the EDNRB gene promoter in melanocytes-lineage cells. The left panel shows the reporter plasmids used. Relative luciferase activity in transient transfection assays is shown on the right. The cell lines used are HeLa cells, HMV-II and 624mel melanoma cells (dotted bars), and normal human epidermal melanocytes (NHEM) (shadow bars). Luciferase activity was normalized by each internal control activity (pRL-TK). Relative luciferase activity is shown as the ratio to the normalized luciferase activity obtained with pGL3-E (−3022) in HeLa cells. Data are mean ±sd of at least three independent experiments. The activity with * is significantly higher in melanoma cells and NHEM than the value obtained with each reporter plasmid in HeLa cells, P < 0.01 or P < 0.05.

Transactivation of the EDNRB gene promoter by SOX10 and Sp1

We then analyzed the effect of SOX10 on the promoter function of the EDNRB gene by transient cotransfection assays (Fig. 4A). SOX10 significantly increased the expression levels of reporter constructs, containing the promoter region between −105 and −12. The localized promoter region (−105 and −12), which is also responsible for the marginal melanocyte-lineage specificity of the EDNRB promoter, contains the GC box, a consensus sequence of the binding site for Sp1 (Fig. 4B). Moreover, Sp1 has been reported to interact with SOX10 [36,37]. We also confirmed the interaction of SOX10 and Sp1 by in vitro pull-down assay (data not shown). We therefore analyzed whether Sp1 influences the function of the EDNRB promoter. An Sp1 expression plasmid was coexpressed with SOX10 expression plasmid in HeLa cells, which endogenously express Sp1 protein [38]. We thus confirmed the over-expression of Sp1 protein in HeLa cells, when transfected with Sp1 expression plasmid (Fig. 4C). Sp1 or SOX10 alone increased the promoter activity of pGL3-E (−3022) 1.9- or 3.9-fold, respectively (Fig. 3C). The combination of SOX10 and Sp1 led to an 8.0-fold increase, suggesting that SOX10 and Sp1 synergistically transactivate the EDNRB promoter. Furthermore, SOX10, alone or in combination with Sp1, significantly increased the expression of pGL3-Em, containing the mutations at the three potential SOX10 sites in the putative ENS enhancer. It is therefore conceivable that these potential SOX10 sites may be dispensable for the SOX10-mediated transactivation. Taken together, the localized promoter region (−105 and −12) is responsible not only for the marginal melanocyte-lineage specificity of the EDNRB promoter but also for the SOX10-mediated transactivation.

Figure 4.

SOX10 and Sp1 synergistically transactivate the EDNRB promoter. (A) Effects of SOX10 on the EDNRB promoter. The right panel shows the result of the transient transfection assay in HeLa cells. The cotransfection with empty vector (–) or SOX10 expression vector (SOX10) is shown as a white bar or a shadow bar, respectively. Relative luciferase activity is shown as the ratio to the normalized luciferase activity obtained with cotransfection of pGL3-E (−3002) and empty vector. Data are mean ±sd of at least three independent experiments. The activity with * is significantly higher than the value obtained with cotransfection of each reporter plasmid and empty vector, P < 0.01. (B) Effects of Sp1 on the SOX10-mediated transactivation of the EDNRB promoter. Shown is the nucleotide sequence of the localized promoter region, in which a GC box is underlined. HeLa cells were cotransfected with each reporter shown to the left, and an empty vector (open bars), Sp1 expression vector (dotted bars), SOX10 expression vector (shadow bars), or both of Sp1 and SOX10 expression vectors (filled bars). Relative luciferase activity is shown as the ratio to the normalized luciferase activity obtained with cotransfection of pGL3-E (−3002) and the empty vector. Data are mean ±sd of at least three independent experiments. The activity with * is significantly higher than the value obtained with each reporter plasmid and empty vector, P < 0.01. (C) Over-expression of Sp1 and SOX10 in HeLa cells. Enforced-expression of Sp1 and/or SOX10 was assessed in HeLa cells transfected with Sp1 and/or SOX10 expression vector by western blot analysis. Note that the amount of Sp1 protein was increased in the cells transfected with Sp1 expression vector compared to that in mock-transfected HeLa cells shown as (–).

Identification of the cis-acting elements responsible for the SOX10-mediated transactivation of the EDNRB promoter

Within the region between −105 and −12 of the EDNRB promoter, there is no consensus sequence, 5′-(A/T)(A/T)CAA(A/T)-3′, for SOX10 binding. Instead, there are the two CA-rich sequences (CA1 and CA2) and the GC box (5′-CCGCCC-3′) (GC1). Notably, the CA2 is overlapping with the GC box (Fig. 5A). It has been reported that SOX10 binds the CA-rich sequence in the neuronal nicotinic acetylcholine receptor β four subunit gene promoter [39]. We therefore examined whether CA1, GC1, or CA2 is involved in the basal promoter activity and/or the SOX10-mediated transactivation of the EDNRB promoter (Fig 5B,C). Each base change at CA1, GC1 or CA2 resulted in the significant decrease in the promoter activity in normal melanocytes and HMV-II melanoma cells (Fig. 5B). The GC box is especially important for the basal promoter activity. Likewise, each base change abolished the SOX10-mediated transactivation of the EDNRB promoter (Fig. 5C), indicating that the three elements, CA1, GC1 and CA2, are responsible for the transactivation of the EDNRB promoter by SOX10.

Figure 5.

Identification of the cis-acting elements responsible for the SOX10-mediated transactivation of the EDNRB promoter. (A) Schematic representation of the EDNRB promoter. The number represents the position from the transcriptional initiation site (+ 1) of the EDNRB gene. The two CA-rich sequences and the GC box are marked. The CA1 probe (line) and the GC1/CA2 probe (dotted line) are used in electrophoretic mobility shift assay of Fig. 4(D).(B) Requirement of CA1, GC1, and CA2 for the melanocyte lineage cell-specific activity. Base changes were introduced into CA1, GC1, or CA2 of a parent plasmid, pGL3-E (−3022) reporter vector. X shows the mutation introduced into CA1, GC1, or CA2. HeLa cells (open bars), HMV-II cells (filled bars), or normal human epidermal melanocytes (NHEM) (shaded bars) were transfected with each reporter shown to the left. Other conditions are represented as indicated inFig. 3(A, C). Effects of the base change at CA1, GC1, or CA2. The transactivation of the EDNRB promoter by SOX10 was assessed in the case of each mutation shown as X. Other conditions are represented as indicated in Fig. 3(B). (D) Electrophoretic mobility shift assays (EMSA) for binding of SOX10 to the EDNRB promoter. Shown are the autoradiographs of EMSA with the CA1 probe (left) or the GC1/CA2 probe (right). Lanes shown as (–) indicate no protein or no competitor. The mock shows the lysate containing a parent vector, pIVEX3.2-MCS. SOX10 shows the lysate containing SOX10 protein synthesized by the in vitro transcription/translation Escherichia coli lysate system. The competitors used are wild-type probe or the mutated probe (shown as Fig. 5A). The consensus SOX10-binding site of human MITF gene promoter is used as a competitor (cSOX10). The arrows show the SOX10–DNA complexes.

SOX10 binds to the CA-rich sequences and the GC box of the EDNRB promoter in vitro

We carried out electrophoretic mobility shift assay (EMSA) using the labeled probes, which include CA1 (CA1 probe) or GC1 and CA2 (GC1/CA2 probe) (Fig. 5A,D). These oligonucleotide probes were incubated with the lysates containing SOX10 protein synthesized by in vitro transcription/translation system. The CA1 probe was specifically bound by recombinant SOX10 (Fig. 5D, left panel). The formation of the SOX10–DNA complexes was inhibited by CA1 probe, but the degree of inhibition with a competitor containing mutated CA1 (mCA1) was lower than that of CA1 probe. When we used the consensus SOX10-binding site (cSOX10) as a competitor, which contains the SOX10-binding site in human MITF gene promoter [14–17], the formation of the SOX10-DNA complexes was reduced. The SOX10–DNA complexes were not detected with the lysates in the case of an empty vector as a negative control. Thus, SOX10 binds to CA1 in the EDNRB promoter. Unexpectedly, the degree of competition with the mutated CA1 was similar to that with cSox10 competitor, which may be due to additional SOX10 binding sites in the region of CA1 probe. Likewise, we showed that SOX10 specifically bound to GC1 and CA2, as the SOX10–DNA complexes were competed by the GC1/CA2 probe or the cSOX10 probe, but not by a mutated GC1 and/or CA2 (mGC1/CA2, GC1/mCA2, or mGC1/mCA2) (Fig. 5D, right panel). Taken together, these results suggest that SOX10 binds to the three cis-acting elements, CA1, GC1, and CA2, in the EDNRB promoter, which is consistent with the results of the cotransfection assays.

Synergistic activation of the EDNRB promoter by SOX10 and Sp1 through the two CA-rich sequences and the GC box

We analyzed the functional significance of CA1, GC1, and/or CA2 in the EDNRB promoter activity (Fig. 6A). Each base change at CA1 and/or GC1 significantly reduced the degree of activation caused by SOX10 and Sp1, compared to the parent reporter plasmid, pGL3-E (−3022). The base change at CA2 showed 35% reduction compared to E (−3022) (P =0.006). These results suggest that SOX10 and Sp1 synergistically transactivate the EDNRB promoter through the two CA-rich sequences and the GC box. To examine whether Sp1 binds to GC1, we performed EMSA using the labeled GC1/CA2 probe and recombinant human Sp1 protein (Fig. 6B). The Sp1-DNA complexes were detected, and their formation was completely competed by a wild-type GC1/CA2 probe, mutated CA2 probe (GC1/mCA2), or a consensus Sp1-binding sequence (cSp1). Interestingly the complex formation was competed by mGC1/CA2 probe, but its competition ability was lower than that of the GC1/CA2 probe. Furthermore, mGC1/mCA2 probe did not inhibit the complex formation. These results suggest that Sp1 recognizes the region containing GC1 and CA2. The binding of Sp1 to GC1 may be influenced by the overlapping CA2.

Figure 6.

Sp1 is involved in the SOX10-mediated transactivation of the EDNRB promoter. (A) Effects of Sp1 on the transactivation of the EDNRB promoter. HeLa cells were cotransfected with each reporter shown on the left, and an empty vector (open bars), Sp1 expression vector (dotted bars), SOX10 expression vector (shadow bars), or both of Sp1 and SOX10 expression vectors (filled bars). Relative luciferase activity is shown as the ratio to the normalized luciferase activity obtained with cotransfection of pGL3-E (−3002) and the empty vector. Data are mean ±sd of at least three independent experiments. The activity with # or ### is significantly lower than the value obtained with cotransfection of pGL3-E (−3002), SOX10, and Sp1, #P < 0.01 or ###P < 0.001. (B) EMSA for the in vitro binding of Sp1 to the EDNRB promoter. Shown is the autoradiograph of the EMSA by using the GC1/CA2 probe. The recombinant Sp1 was used. The competitors used are wild-type probe or the mutated probe (shown as Fig. 4A). The consensus Sp1 binding site is used as a competitor (cSp1) [56]. The arrows show the Sp1–DNA complexes.

We then examined the simultaneous binding of SOX10 and Sp1 proteins to the GC1/CA2, but were unable to detect the complexes, containing both SOX10 and Sp1 proteins (data not shown). It is conceivable that the in vitro binding conditions are not suitable for simultaneous binding of the two proteins to the GC1/CA2.

SOX10 and Sp1 bind to the EDNRB promoter in vivo

To investigate whether SOX10 and/or Sp1 bind to the EDNRB gene in vivo, chromatin immunoprecipitaion (ChIP) assay was performed in HMV-II cells (Fig. 7A). The PCR primer sets were designed to amplify the DNA segments containing CA1, GC1, and CA2 of the EDNRB promoter, which is responsible for the SOX10-mediated transactivation. ChIP assay revealed that the DNA segments of the EDNRB promoter were amplified when precipitated with rabbit anti-SOX10 IgG or goat anti-Sp1 IgG in HMV-II cells, but not noticeably amplified when precipitated with normal rabbit IgG (negative control) or normal goat IgG (negative control) (Fig. 7B). These results indicate that SOX10 and Sp1 bind to the region containing CA1, GC1, and CA2 in the EDNRB promoter in vivo. We could not detect the amplified fragments of GAPDH gene as a negative control, when precipitated with rabbit anti-SOX10 IgG or goat anti-Sp1 IgG.

Figure 7.

Binding of SOX10 and Sp1 to the EDNRB promoter in vivo. (A) Strategy for chromatin immunoprecipitation assay. Arrows indicate the PCR primers used for the EDNRB gene promoter. (B)  In vivo binding of SOX10 and Sp1 to the EDNRB promoter. HMV-II cells were fixed with formaldehyde, precipitated with anti-SOX10, anti-Sp1, normal rabbit IgG, or normal goat IgG, and subjected to PCR for the EDNRB promoter (top panel) or GAPDH gene (bottom panel) as a negative control. The amplified DNA segments were visualized with ultraviolet transilluminator.

SOX10 activates the EDNRB promoter in the demethylation status

It has been reported that the EDNRB promoter is located in the CpG islands, which are target sites of DNA methylation [40–42]. In fact, the EDNRB promoter is DNA-methylated in several types of tumor cells, leading to gene silencing [40–43]. Moreover, the GC1, identified as one of the key regulatory elements, includes the CpG dideoxynucleotides, which are potential targets of DNA methylation (Fig. 8A). We therefore investigated the DNA methylation status of the EDNRB promoter in normal human epidermal melanocytes (NHEM), HMV-II melanoma cells, and HeLa cells. The EDNRB promoter is heavily methylated in HeLa cells, but not in NHEM and HMV-II cells (Fig. 8A). Subsequently, we examined whether enforced expression of SOX10 induces the endogenous EDNRB expression in HeLa cells. We established FLAG-tagged SOX10 (F/SOX10)-expressing stable transformants from HeLa cells, and then chose the stable transformants #6 and #8, which appear to express F/SOX10 more abundantly than other transformants (Fig. 8B). Expression of EDNRB mRNA was undetectable in cells transformed with the empty vector (Mock) and the F/SOX10-expressing cells, as judged by RT-PCR analysis (Fig. 8C). However, in the mock-transformed cells, expression of EDNRB mRNA became detectable after treatment with a demethylating reagent, 5′-aza-2′-deoxycytidine (5′-aza-dC) (Fig. 8C). Importantly, the expression levels of EDNRB mRNA were higher in F/SOX10-expressing cells after treatment with 5′-aza-dC (Fig. 8C). These results indicate that SOX10 transactivates the endogenous EDNRB promoter in the demethylation status. The induction of the EDNRB expression by 5′-aza-dC alone in mock-transformed cells may be directed by Sp1, which is expressed in HeLa cells [38].

Figure 8.

SOX10 activates transcription of the EDNRB gene in the demethylation status. (A) Methylation status of the EDNRB promoter. DNA methylation status was determined in normal human epidermal melanocytes (NHEM), HMV-II melanoma cells, and HeLa cervical cancer cells. Note that the EDNRB promoter is heavily methylated in HeLa cells, lacking EDNRB mRNA expression. (B) Isolation of stable transformants, which express SOX10. HeLa cells were transfected with FLAG-tagged SOX10 (F/SOX10) or the empty vector (Mock), and four independent clones were isolated, named #1–4 for Mock-transformed cell lines and #5–8 for F/SOX10. The western blot analysis with anti-FLAG IgG (top panel) or anti-α-tubulin IgG (bottom panel) is shown. (C) Demethylation induces expression of the endogenous EDNRB gene. Indicated stable transformants (#1, 2, 6, and 8) were untreated or treated with a demethylating reagent, 5′-aza-dC for 7 days. The stained gels of the RT-PCR products, obtained at logarithmic phase, are shown. The induction of EDNRB mRNA was enhanced in F/SOX10-expressing cells.

Thus, the degree of the methylation in the EDNRB promoter determines the transcription levels of the EDNRB gene. However, we were unable to assess the contribution of the methylation status to the promoter activity by the transient expression assay, which could account in part for a small difference in the melanocyte lineage-specific promoter activity. Alternatively, the human EDNRB gene contains an additional melanocyte-enhancer, which is however, not carried by the reporter constructs used, containing the 3-kb length of human EDNRB gene promoter, although the 1.2-kb length of mouse Ednrb gene promoter is sufficient for melanocyte-specific activity [18].

Implications

SOX10 interacts with Sp1 and activates the EDNRB promoter. Sox10 also interacts with Pax3 and activates the c-RET promoter [44,45]. The mutation in the c-RET gene is responsible for the pathogenesis of aganglionic megacolon [46]. Ubiquitously expressed Sp1 may affect the pathogenesis of WS4 by cooperating directly with SOX10 or by influencing the interaction of SOX10 with PAX3.

It should be noted that WS4 shows the phenotypic variability [9,22]. In one family case of WS4, for example, the proband and his sister are heterozygous for the Q377X mutation of SOX10, but only the proband has an aganglionic megacolon [22]. These observations suggest the presence of modifier genes for the EDNRB gene, and the expression of the modifier genes may be influenced by the environmental factors, thereby leading to the phenotypic variability of WS4. One of such modifiers might be Sp1, the function of which is modulated growth factors [47] or metals [48].

In summary, we have provided evidence that SOX10, alone or in combination with Sp1, may activate transcription of the human EDNRB gene, which contributes to melanocyte lineage cell-specific expression, and that the regulation of EDNRB expression by SOX10 requires the demethylation status of its promoter. The regulatory network involving SOX10 and Sp1 may ensure the fine-tuning of EDNRB expression, which contributes the homeostasis of human melanocytes. Future research on the genetic network of WS genes will help clarify the pathogenesis of WS.

Experimental procedures

Cell culture

Human normal epidermal melanocytes (NHEM) were obtained from Kurabo (Kurabo, Osaka, Japan) and cultured in Medium 154S (Kurabo) containing human melanocyte growth supplement (Kurabo) at 37 °C under 5% CO2/95% room air. HMV-II human melanoma cells were obtained from RIKEN Cell Bank and cultured in nutrient mixture Ham's F12 medium containing 10% fetal bovine serum at 37 °C under 5% CO2/95% room air. Human melanoma cell lines, G361 and SK-MEL-28, were obtained from Cell Resource Center for Biomedical Research (Institute of Development, Aging and Cancer, Tohoku University, Miyagi, Japan). 624mel human melanoma cells were provided by Y. Kawakami (Division of Cellular Signaling, Institute for Advanced Medical Research, School of Medicine, Keio University, Tokyo, Japan). G361 and 624mel cells were cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum. SK-MEL-28 cells were cultured in minimum essential medium Eagle supplemented with 10% fetal bovine serum. HeLa human uterine cervical cancer cells and COS-7 monkey kidney cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum.

RNA interference

The small interfering RNA (siRNA) against SOX10 (927–1753 numbered in accession number NM_006941) or LacZ were generated by BLOCK-iTTM Complete Dicer RNAi Kit protocol (Invitrogen, Carlsbad, CA, USA). HMV-II melanoma cells were cultured for 24 h after plating in 6-cm dishes and then transfected with each siRNA against SOX10 or LacZ by LipofectamineTM2000 protocol (Invitrogen). The cells were incubated for 3 days and then each siRNA was transfected again. After further 3 days, the cells were harvested for northern blot analysis and western blot analysis.

Northern blot analysis

Total RNA was prepared by TRIzol reagent (Invitrogen) and subjected to northern blot analysis as described previously [49]. The probes used were the human SOX10 cDNA fragment (278–1753 numbered in accession number NM_006941) and the human EDNRB cDNA fragment (550–1268 numbered in accession number NM_003991) and the human β-actin cDNA (124–1050 numbered in accession number X00351). These DNA fragments were labeled with [α-32P]dCTP using a BcaBEST labeling kit (Takara biomedical, Shiga, Japan) and were used as hybridization probes. Radioactive signals were detected by exposing the filters to X-ray films (X-AR5; Kodak) or with a Bioimage Analyzer (BAS1500; Fuji Film). The exposure time to X-ray films varied depending on the experiments. The intensities of hybridization signals were determined by photo-stimulated luminescence with a Bioimage Analyzer.

Western blot analysis

Whole cell extracts were prepared from the cells transfected with siRNA or the expression plasmid for SOX10 or Sp1 by the method of Schreiber et al. [50] and then subjected to western blot analysis using anti-SOX10 IgG (Sigma-Aldrich Japan, Tokyo, Japan), anti-MITF polyclonal IgG [51], anti-Sp1 IgG (Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-FLAG IgG (Invitrogen), or anti-α-tubulin IgG (NeoMarkers, Fremont, CA, USA), as described previously [51]. The relative amount of protein was determined by NIH Image 1.63 f software. The amount of SOX10, MITF, or EDNRB protein is normalized by that of α-tubulin protein.

S1 nuclease mapping analysis

Total RNA was isolated from NHEM, HMV-II cells, and HeLa cells with TRIzol reagent. An S1 probe used was the fragment of the EDNRB gene (1–1438 numbered in accession number D13162), end-labeled with [γ-32P] ATP at XcmI site, located at the position 1438 in the antisense strand of the EDNRB gene. The S1 probe contains the pGEM-T Easy vector (Promega, Madison, WI, USA) sequence, the 5′-flanking region and exon 1 of the EDNRB gene (Fig. 2A). S1 nuclease mapping analysis was performed with 20 µg of total RNA from NHEM, HMV-II, or HeLa cells, or yeast tRNA (negative control), as described previously [52,53].

Plasmids construction

A human SOX10 expression vector, pcDNA3-SOX10, was described previously [17]. A human Sp1 expression vector, pCIneo-Sp1, was a gift from Dr Kojima (The Institute of Physical and Chemical Research, Tsukuba Life Science Center, Tsukuba, Ibaraki, Japan) [54]. To construct the FLAG-tagged SOX10 expression vector, p3XFLAG-SOX10, the HindIII/XbaI fragment of pcDNA3-SOX10 was inserted into p3XFLAG-CMV7.1 vector (Sigma-Aldrich Japan). The 5′-flanking region (position −3022 to +160 from the transcription initiation site of the conventional EDNRB mRNA) of the human EDNRB gene was generated by genomic polymerase chain reaction (PCR) and was inserted into the luciferase reporter plasmid pGL3-Basic (Promega), yielding pGL3-E (−3022). Likewise, deletion constructs, pGL3-E (−1002), pGL3-E (−678), pGL3-E (−493), pGL3-E (−105), and pGL3-E (−12) were generated. The internal deletion plasmid, pGL3-EΔ, was constructed by deleting the XmnI/MscI fragment from pGL3-E (−3022). Base changes were introduced into pGL3-E (−3022) using QuickChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA) according to manufacturer's instruction. The resulting mutant constructs, pGL3-EmCA1, pGL3-EmGC1, pGL3-EmCA2, and pGL3-Em3, carry a single or a combination of the following base changes; CA1 (CACACCCC) was changed to mCA1 (CtCgagCC), GC1 (CCGCCC) to mGC1 (CCGaaC), and CA2 (CCACTGC) to mCA2 (CCcCgGg). The mutant construct, pGL3-Em, carries the mutation of three potential SOX10 sites; TTCAAT, ATTGAT, and GTTGAA were changed into cTCgAg, ATTaAT, and GTcGAc, respectively.

Transient transfection assays

The activities of the human EDNRB gene promoters were assessed by transient expression of firefly luciferase genes in HeLa cervical cancer cells, or human melanoma cells HMV-II, G361, SK-MEL-28 or 624mel, and normal human epidermal melanocytes as described previously [53]. Cells used were cultured for 24 h after plating in 12-well dishes and then transfected with each reporter plasmid (120 ng), pRL-TK (20 ng), and SOX10 cDNA (50 ng), Sp1 cDNA (50 ng), or an empty expression vector by FuGENE 6 (Roche Diagnostics, Mannheim, Germany). Total amount of plasmids transfected was 500 ng per well. pRL-TK contains the herpes simplex virus thymidine kinase promoter region upstream of Renilla luciferase (Promega). The cells were harvested 24 or 48 h after transfection, and then luciferase activity was measured by Dual-Luciferase Reporter Assay System (Promega). Firefly luciferase activity was normalized with each Renilla luciferase activity.

EMSA

SOX10 cDNA was subcloned in the pIVEX2.3-MCS (Roche Diagnostics), which was used for the production of recombinant SOX10 protein by the RTS100 (Roche Diagnostics) in vitro transcription/translation Escherichia coli lysate system according to manufacturer's instruction (Roche Diagnostics). A parent vector pIVEX2.3-MCS was also used as a negative control. The recombinant human Sp1 protein was purchased from Promega. EMSA was performed as previously described [55]. In brief, double-stranded oligonucleotides, CA1 probe (5′-CATTCCCTCCCTGGCACACCCCTTCCAGAACGCCCC-3′) and GC1/CA2 probe (5′-GATCCCCTTCCAGAACGCCCCGCCCCACTGCATATTATTTACCCCTCCA-3′) were end-labeled with [γ-32P] ATP and used as probes. The SOX10-containing lysates were preincubated for 30 min on ice with or without a competitor in the absence of a probe and then incubated with the probe for 30 min on ice in 20 µL solution containing 1 mg·mL−1 bovine serum albumin (BSA), 0.1 µg poly (dI-dC), 10 mm 4-(2-hydroxyethyl) piperazine-1-ethanesulfonic acid (Hepes) (pH 7.5), 50 mm NaCl, 5 mm MgCl2, 0.5 mm ethylenediamine-N,N,N′,N′-tetraacetic acid (EDTA), 0.1 mm dithiothreitol (DTT), and 5% glycerol. The reaction mixture was loaded on 5% polyacrylamide gels (PAGE) and run in 0.5 × TBE buffer (50 mm Tris, 1 mm EDTA, 48.5 mm boric acid). The gels were dried and autoradiographed to detect mobility shift. Competitors used were mCA1 (5′-CATTCCCTCCCTGGCtCgagCCTTCCAGAACGCCCC-3′), mGC1 (5′-CCCCTTCCAGAACGCCCCGaaCCACTGCATATTATTTAC-3′), mCA2 (5′-GAACGCCCCGCCCCcCgGgATATTATTTACCCC-3′), mGC1/ mCA2 (5′-CCCCTTCCAGAACGCCCCGaaCCcCgGgATATTATTTACCCC-3′), a consensus Sp1 binding site (cSp1) (5′−ATTCGATCGGGGCGGGGCGAGC-3′) [56], and a consensus SOX10 binding site (cSOX10) (5′-AGAGAAATACCATTGTCTATTAATACT-3′).

Chromatin immunoprecipitation (ChIP) assay

ChIP was performed according to the methods, reported previously [51,57]. Briefly, formaldehyde was added to the culture medium of HMV-II cells at the final concentration of 1%; after 10 min the cells were washed with phosphate-buffered saline, collected, and sedimented at 5000 g for 5 min. The pellet was resuspended in 5 mm of 1,4-piperazinediethanesulfonic acid (pH 8.0), 85 mm KCl, 0.5% Nonidet P-40, and protease inhibitor cocktail (Sigma-Aldrich Japan) for 5 min. Nuclei were collected by centrifugation at 5000 g for 5 min, then resuspended in 50 mm Tris/HCl (pH 8.1), 10 mm EDTA, and 1% sodium dodecyl sulfate (SDS), and incubated on ice for 10 min. The chromatin was fragmented by sonication to 1000–600 basepairs. For immunoprecipitation, the solution was diluted with immunoprecipitation dilution buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mm EDTA, 16.7 mm Tris/HCl, pH 8.1, and 167 mm NaCl) and incubated with anti-SOX10 IgG (Sigma-Aldrich Japan), anti-Sp1 IgG (Santa Cruz Biotechnology), normal rabbit IgG (Santa Cruz Biotechnology), or normal goat IgG (Santa Cruz Biotechnology) at 4 °C overnight. Protein G MicroBeads (Miltenyi Biotec, Gladbach, Germany) were then added to the mixture and rotation was continued for 2 h. After immune complexes were washed, cross-link reversal step was performed by addition of 300 mm NaCl and heating at 65 °C overnight. DNA was ethanol precipitated, treated with proteinase K, extracted with phenol/chloroform, and finally ethanol precipitated, which was subjected to PCR for the amplification of the EDNRB gene promoter region. In the case of the EDNRB promoter region, the primers used were 5′-CATTCCCTCCCTGGCACACCCCTTCCAGAACGCCCC-3′ as a forward primer and 5′-CGAGCCAAGTCGCTGCAAACGCTAATACCG-3′ as a reverse primer. GAPDH gene was amplified as a negative control. The primers used were 5′-TGGTATGACAACGAATTTG-3′ as a forward primer and 5′-TCTACATGGCAACTGTGAGG-3′ as a reverse primer. The PCR condition was 42 cycles of 98 °C for 20 s, 55 °C for 30 s, and 72 °C for 1 min. The identity of the amplified segment was confirmed by sequencing.

Methylation analysis

In order to analyze the methylation status in 5′-flanking region of the EDNRB gene, we prepared genomic DNA from NHEM, HMV-II, and HeLa cells. Methylation status was determined by the method of bisulfite-modified DNA [41]. This technique depends on chemical modification of the genomic DNA with sodium bisulfite, where unmethylated cytosine residues are converted to uracil residues, but methylated cytosines remain unchanged. The PCR primers used for the EDNRB gene promoter were 5′-GGAGTTTTGTTTGGGATTTTTATT-3′ (sense) and 5′-AAACTCCTTCCTAATACCCT-3′ (antisense). The amplified segments were subcloned into the pGEM-T Easy vector (Promega), and the nucleotide sequence of the insert fragment was determined with ABI3100 analyzer (Applied Biosystems, Foster, CA, USA.).

5′-Aza-2′-deoxycytidine treatment

For construction of another FLAG-tagged SOX10 expression vector, pIRES-F/SOX10, the SacI/SacI and SacI/XbaI fragments of p3XFLAG-SOX10 were inserted into pIRESneo vector (Clontech, Mountain View, CA, USA). The stable-transformed HeLa cells with pIRESneo (Mock) or pIRES-F/SOX10 (F/SOX10) were selected with G418 (350 µg·mL−1) (Clontech).

HeLa transformants (Mock or F/SOX10) were seeded in a 60-mm dish 24 h prior to treatment with 5′-aza-2′-deoxycytidine (5′-aza-dC) (1 µm) (Sigma-Aldrich Japan). Culture medium and drug were replaced every 24 h. After 7 days, cells were harvested for RNA extraction. The total RNA was used for RT-PCR. The primers used were as follows:

EDNRB: 5′-TTGTGTCCTGCCTTGTGTTC-3′ (sense) and 5′-CAGAATCCTGCTGAGGTGAA-3′ (antisense); SOX10: 5′-CCCAAGCTTATGGCGGAGGAGCAGGATCTATCGGAGGTG-3′ (sense) and 5′-GCTCTAGATGAGGTGGGCAAGGAACAGGGCACACAGGCT-3′ (antisense); and GAPDH: 5′-ACCACAGTCCATGCCATCAC3′ (sense) and 5′-TCCACCACCCTGTTGCTGTA-3′ (antisense). The PCR condition was 42 cycles (SOX10), 45 cycles (EDNRB) or 30 cycles (GAPDH) of 94 °C for 30 s, 55 °C for 30 s, and 72 °C for 1 min. The identity of the amplified segment was confirmed by sequencing.

Statistical analysis

All data are mean ±sd of at least three independent experiments. Two-tailed Student's t-test was used for comparison between the two groups. Differences between mean values were considered significant when P < 0.05.

Acknowledgements

This work was supported in part by Grants-in-Aid for Scientific Research (B) (to S.S.), for Scientific Research (C) (to K.T.), for Scientific Research on Priority Area (to S.S.), and for Exploratory Research (to S.S.) from the Ministry of Education, Science, Sports, and Culture of Japan, and by the 21st Century COE Program Special Research Grant ‘the Center for Innovative Therapeutic Development for Common Diseases’ from the Ministry of Education, Science, Sports and Culture of Japan.

Ancillary