Báo cáo khoa học: Autoregulatory binding sites in the zebrafish six3a promoter region define a new recognition sequence for Six3 proteins potx

Autoregulatory binding sites in the zebrafish six3a promoter region define a new recognition sequence for Six3 proteins Clotilde S. Suh, Staale Ellingsen*, Lars Austbø, Xiao-Feng Zhao, Hee-Chan Seo and Anders Fjose Department of Molecular Biology, University of Bergen, Norway Introduction Vertebrate Six3 proteins have important roles during the development of eyes and forebrain, and belong to the Six ⁄ Sine oculis family. This family represents a divergent group of the homeodomain (HD) superfam- ily of transcription factors [1,2]. The 60 amino acid HD, which is a DNA-binding domain, has a conserved global fold consisting of three a-helices and a flexible N-terminal arm that becomes more ordered upon DNA binding [3–6]. Binding to specific DNA sequences is mediated by interactions between particu- lar amino acids in the ‘recognition helix’ and bases in the major groove, and specific contacts between the N-terminal arm and the minor groove [4,7,8]. Specific base contacts in the minor groove involve the first two nucleotides in the TAAT core, and are achieved through interactions with residues at posi- Keywords chromatin; eye development; homeobox; transcription factor; transgenic Correspondence A. Fjose, Department of Molecular Biology, University of Bergen, PO Box 7803, N-5020 Bergen, Norway Fax: +47 555 89683 Tel: +47 555 84331 E-mail: anders.fjose@mbi.uib.no *Present address National Institute of Nutrition and Seafood Research, NIFES, PO Box 2029 Nordnes, N-5817 Bergen, Norway (Received 11 September 2009, revised 22 December 2009, accepted 29 January 2010) doi:10.1111/j.1742-4658.2010.07599.x The homeodomain (HD) transcription factor Six3, which is a member of the Six ⁄ Sine oculis family, is essential for development of the eyes and fore- brain in vertebrates. It has recently been claimed that the HDs of Six3 and other members of the Six family have a common recognition sequence, TGATAC. However, a different recognition sequence including the typical TAAT core motif, which has not yet been fully defined, has also been pro- posed for the Six3 HD in mice. Our study of the zebrafish orthologue six3a, which has an identical HD, shows that it binds in vitro to multiple TAAT-containing sites within its promoter region. Comparison of the dif- ferent binding affinities for these sequences identifies three high-affinity sites with a common TAATGTC motif. Notably, this new recognition sequence, which is supported by our analysis of the influence of single- nucleotide substitutions on the DNA-binding affinity, is distinct from all of the DNA-binding specificities previously described in surveys of HDs. In addition, our comparison of Six3a HD binding to the novel TAATGTC motif and the common recognition sequence of Six family HDs (TGATAC) shows very similar affinities, suggesting two distinct DNA- binding modes. Transient reporter assays of the six3a promoter in zebrafish embryos also indicate that the three high-affinity sites are involved in auto- regulation. In support of this, chromatin immunoprecipitation experiments show enrichment of Six3a binding to a six3a promoter fragment containing two clustered high-affinity sites. These findings provide strong evidence that the TAATGTC motif is an important target sequence for vertebrate Six3 proteins in vivo. Abbreviations ChIP, chromatin immunoprecipitation; EGFP, enhanced green fluorescent protein; EMSA, electrophoretic mobility shift assay; GFP, green fluorescent protein; GST, glutathione-S-transferase; HD, homeodomain; hpf, hours postfertilization. FEBS Journal 277 (2010) 1761–1775 ª 2010 The Authors Journal compilation ª 2010 FEBS 1761 tions 2, 3 and 5–8 in the N-terminal arm. Also, an arginine at position 5 is important in most HDs [4,5,9]. Similarly, the recognition helix makes specific contacts with several nucleotides in the core motif, and its residues at positions 47, 50 and 54 also specify two adjacent nucleotides 3¢ of the TAAT core [5,10]. For example, HDs containing Lys50 and Gln50 have been shown to bind specifically to TAATCC and TAATGG, respectively [11–13]. Recent studies indicate that the sequence recognition also depends on a few additional flanking nucleotides, and this variation in specificity may include more than 60 distinct DNA- binding activities [14]. The Six ⁄ Sine oculis family proteins also have a con- served Six domain of 115–119 amino acids involved in protein–protein interactions [1,15], and can be subdi- vided into three subfamilies, Six1 ⁄ 2, Six4 ⁄ 5, and Six3 ⁄ 6, on the basis of their HD sequence divergence and characteristic tetrapeptides in the N-terminal arm [16]. The absence of Arg5 in their N-terminal arms may explain, in part, why regulatory DNA sequences that bind Six1 ⁄ 2 and Six4 ⁄ 5 do not contain the TAAT core [17]. Although the Six3 ⁄ 6 proteins also lack Arg5, their HDs are more distinct from those of the members of the other two subfamilies, and various studies have suggested that their DNA-binding specificity is differ- ent [1,16,18]. In a previous investigation of murine Six3, a common TAAT core motif was identified by in vitro binding site selection from a randomized pool of oligonucleotides, and autoregulatory binding sites containing the TAAT core were also identified in the promoter of the Six3 gene [18]. However, more recent studies have provided evidence that Six3 ⁄ 6 proteins have similar in vitro DNA-binding specificities to those of the other Six family members [10,14]. Further analy- sis of the binding affinities of functional Six3 target sites and how they function in vivo may help to clarify uncertainties regarding the recognition sequences of these HD proteins. Two orthologues of the murine Six3 gene, six3a and six3b, are present in the zebrafish, Danio rerio, owing to the extra genome duplication that occurred before the teleost radiation [15,19]. An additional Six3-like gene in zebrafish, six7, was probably generated by an independent gene duplication event [20]. Several stud- ies of Six3 homologues in mouse, fish and Xenopus have demonstrated that these genes are essential for forebrain and eye development, and their importance is also reflected in human mutant phenotypes [21–25]. In these processes, Six3 proteins have been shown to act both as transcriptional activators and repressors, and as regulators of cell proliferation through interac- tions with the cell cycle inhibitor Geminin [26,27]. Studies on Six3 proteins in zebrafish have contrib- uted to our understanding of their functional roles in forebrain and eye development [22,28–30], and how they can act as transcriptional repressors through interactions with members of the Groucho family of corepressors [31]. Relatively little is known about the regulation of zebrafish six3 genes during development [32,33], but essential cis-regulatory elements have been identified in one of the gene homologues in medaka fish [34]. We have investigated the significance of the high density of TAAT sequences present in the zebrafish six3a promoter region. Our comparison of the relative binding affinities of these potential target sites for the Six3a HD identified several strong binding sites that defined the sequence TAATGTC as a recognition motif. Results from chromatin immunoprecipitation (ChIP) experiments and transient reporter assays of the six3a promoter in zebrafish embryos supported the functional role of these high-affinity sites in mediating autoregulation. Hence, it is also likely that many of the target genes of vertebrate Six3 proteins are recog- nized on the basis of high-affinity binding to sequence elements containing this motif. Results A 3.6 kb promoter region of six3a recapitulates early embryonic expression The genomic region upstream of the translational start site in zebrafish six3a is syntenic with a 4.5 kb pro- moter region of the orthologous medaka (Oryzia latipes) gene olSix3.2, which contains cis-regulatory elements responsible for its spatiotemporal regulation in embryos [34]. Additional evidence that the corresponding promoter region of zebrafish six3a contains cis-acting elements required for early expression in the eyes and forebrain was obtained from transient expression assays with injected reporter constructs [33]. To analyse the significance of the six3a promoter region, we fused a 3.6 kb genomic fragment to the ORF of an enhanced green fluorescent protein (EGFP) reporter gene in a Tol2 vector (Fig. 1A), and used this construct to establish transgenic lines of zebrafish (see Experimental procedures). From 53 founders crossed to wild-type fish, we identified three transgenic lines with EGFP expression comparable to that of the endogenous six3a gene (data not shown). The trans- genic line Tg(3.6S3a:EGFP) was chosen for direct comparison of EGFP expression with the spatial distri- bution of endogenous six3a transcripts by in situ hybridization. At 12 h postfertilization (hpf), EGFP New recognition sequence for Six3 C. S. Suh et al. 1762 FEBS Journal 277 (2010) 1761–1775 ª 2010 The Authors Journal compilation ª 2010 FEBS expression was detected in the optic vesicles and ros- tral brain, where six3a transcripts were also shown to accumulate (Fig. 1B). These results confirm that the 3.6 kb promoter region included in the six3a:EGFP transgene contains regulatory sequences sufficient to drive expression mimicking early six3a endogenous expression. Differences in Six3a HD binding to TAAT core motifs within its promoter region The promoters of murine Six3 and human SIX3 contain autoregulatory binding sites [18,35]. In the case of the murine Six3 promoter, it has been shown that negative autoregulation involves clustered TAAT core motifs and interaction with Groucho-related corepres- sors [18]. Sequence analysis of the 3.6 kb promoter region of six3a revealed enrichment and clustering of the same sequence motif (Fig. 2). Within this promoter region, the common core motif is present at 43 posi- tions in both orientations (TAAT or ATTA). The ratio between TAAT and ATTA on the coding strand is 25 : 18 (Fig. S1). In initial studies of several of the 18 ATTA sites by electrophoretic mobility shift assays (EMSAs), we observed the strongest shift for the a1 site (data not shown). Therefore, a1 was selected as a reference for comparisons of the binding affinities of the 18 different ATTA-containing sites. In this study, we used a biotin-labelled 27 bp DNA fragment con- taining a1 as a probe in EMSAs, and tested the influ- ence of this on HD complex formation in the presence of excess amounts of unlabelled fragments representing the individual a1–a18 sites (Fig. 2B). Whereas a 200- fold excess of unlabelled a1 competitor almost completely prevented the formation of probe–HD complexes in EMSAs, only a few of the other sites were able to compete significantly under the same conditions. Notably, the competitor representing a2, which is located a short distance ( 10 bp) upstream of a1 (Fig. S1), also caused a strong reduction in formation of the probe–HD complex. In addition to their clustering and high binding affinities for the Six3a HD, the flanking nucleotides of the core ATTA motifs in a1 (G 1 T 2 C 3 A 4 T 5 T 6 A 7 G 8 G 9 ) and a2 (G 1 A 2 T 3 A 4 T 5 T 6 A 7 T 8 G 9 ) have common Gs at positions upstream (G 1 ) and downstream (G 9 ). Taking into consideration that the binding specificities of HDs have been shown to depend mainly on the two nucleo- tides 5¢ to the ATTA core motif [10–14], the common G 1 was likely to be important. However, three addi- tional sites (a6, a9, and a11), which have a G in the same 5¢-position relative to the ATTA core, showed much weaker binding, indicating that other nucleotide positions also influence the binding affinity (Fig. 2B). To address the functional importance of the G 1 nucle- otide, an inspection of the 25 TAAT sites was per- formed, and this identified four sites (t2, t13, t15, and t17) containing a G in this 5¢-position. Among these sites, only t15 showed similar binding affinity to the Six3a HD as a1 and a2 (Fig. 2C). This further indi- cated that flanking nucleotides other than G 1 have sig- nificant influence on the binding affinity. Notably, three ATTA sites (a10, a14, and a18) without the G 1 flanking nucleotides also bound quite strongly to the Six3a HD (Fig. 2B). However, among the 21 TAAT sites lacking the G 1 nucleotide, none showed significant competition with the a1 probe in EMSAs (Fig. S2). Hence, these comparative analyses showed that the frequency of high-affinity sites was significantly higher among the G 1 -containing sites. An additional compari- son of the relative strengths of the various high-affinity binding sites conducted with lower amounts of competitor still showed strongest binding for the three G 1 -containing sites, a1, a2, and t15 (Fig. S3). six3a promoter region ATG EGFP 730 bp 3635 bp A BCD Tg (3.6S3a:EGFP) six3a EGFP Fig. 1. A 3.6 kb promoter region of the six3a gene is sufficient to recapitulate its early expression. (A) Schematic representa- tion of the six3a promoter region including the 5¢-UTR, fused to the coding region of EGFP. Arrows indicate the position of the transcription start site and the initiation codon. (B) Lateral view of EGFP expression in a Tg(six3a:EGFP) embryo at 12 hpf. (C) Detection of EGFP transcripts in a Tg(six3a:GFP) embryo by in situ hybridization (lateral view of 12 hpf stage). (D) Detection of endogenous six3a transcripts at 12 hpf (lateral view). C. S. Suh et al. New recognition sequence for Six3 FEBS Journal 277 (2010) 1761–1775 ª 2010 The Authors Journal compilation ª 2010 FEBS 1763 Therefore, we aimed to investigate the functionality of these sites in vivo and the relative importance of the different nucleotide positions flanking their ATTA core motifs. Deletion analysis of the six3a promoter indicates autoregulatory binding sites To determine whether any of the strong Six3a HD-binding sites identified by EMSA might have a function in vivo, we made several promoter–reporter constructs with small deletions of regions containing particular sites. These constructs, which were made from the construct pS3aPG used to make the trans- genic line (Fig. 1A; see Experimental procedures), were tested in transient reporter assays based on microinjec- tion into fertilized eggs and measuring the number of EGFP-expressing cells at 12 hpf (Fig. 3). Notably, when six3a mRNA and pS3aPG were coinjected, we observed a more than two-fold increase in the number of EGFP-expressing cells as compared with injection of pS3aPG alone. This indicated that overexpression of Six3a caused an increase in EGFP expression through binding to one or more sites within the promoter region in the pS3aPG reporter construct. Consistent with the expression pattern of the endoge- a1 a11 a12 a13 a14 a15 a16 a17 a18 x 200 a1 a2 t6 a9 a11 t2 t13 t15 t17 x 200 a1 a2 a3 a4 a5 a6 a7 a8 a9 a10 x 200 +++++ ++++ +/– +/– +/– – +/– +/– +/– ++++ +/– – +/– +++ + – +/– ++ – +/– +++++ +/– a1 a2 a3 a4 a5 a6 a7 a8 a9 a10 a11 a12 a13 a14 a15 a16 a17 a18 t2 t13 t15 t17 Relative competition Motif 18 A B C D 17 16 15 14 13 12 1110 9 8 7 6 5 4 3 2 t2 1 * * t13t15t17 * six3a promoter region (pS3aP) Fig. 2. Distribution and relative binding affinities of potential Six3a HD target sites within the six3a promoter region. (A) Distribution of ATTA motifs within the 3.6 kb promoter region of zebrafish six3a (pS3aP). Vertical bars (numbered 1–18) indicate ATTA motifs in the forward strand. Vertical bars indicate 25 ATTA motifs in the reverse strand, and the four GNNATTA sites in the reverse strand are labelled (t2, t13, t15, and t17). Stars indicate the three high-affinity GNNATTA sites (a1, a2, and t15). An arrow indicates the transcription start site. (B) EMSAs with the Six3a HD and biotin-labelled a1 probe. Competition was performed using a · 200 molar excess of unlabelled fragments containing the ATTA motifs a1–a18 (forward strand). The left lane shows the control (labelled probe together with Six3a HD). (C) Competitive EMSA with biotin-labelled a1 probe and a · 200 excess of unlabelled fragments representing all GNNATTA sites (a1, a2, a6, a9, a11, t2, t13, t15, and t17). (D) Table showing the relative competition of the different sites as compared to a1, from highest competition (+++++) to lack of competition ()). New recognition sequence for Six3 C. S. Suh et al. 1764 FEBS Journal 277 (2010) 1761–1775 ª 2010 The Authors Journal compilation ª 2010 FEBS nous six3a gene [15], we observed EGFP-positive cells mainly in the rostral region of the head at 12 hpf (Fig. S5). Among the seven deletion constructs coinjected with six3a mRNA, we observed significant reductions in the numbers of EGFP-expressing cells for three constructs in which the deletions included specific high-affinity sites (Fig. 3). The deletion construct lacking the two strongest sites, a1 and a2 (pS3aPGDa1), showed a reduction of about 50% relative to coinjections of pS3aPG and six3a mRNA. The three additional binding sites (t1, t2, and t3) that were deleted in this construct showed only weak affinity for the Six3a HD (Fig. 2A), suggesting that reduction of EGFP expression could be due to the loss of the two G 1 -containing sites (a1 and a2). For another reporter construct (pS3aPGDt15) in which the deletion included two clustered high-affinity sites (a10 and t15) and five weak binding sites (a8, a9, a11, t14, and t16), the reduction in the number of EGFP-expressing cells was somewhat lower ( 35%). Similarly, when a fragment containing one high-affin- ity site (a18) and eight additional ATTA core motifs (a16, a17, and t20–t25) was deleted (pS3aPGDa18), the reduction in EGFP expression was  45%. How- ever, reporter gene expression was not significantly reduced for one of the constructs (pS3aPGDa14) in which a relatively strong binding site (a14) was deleted together with two low-affinity sites (a13 and a15). Notably, the effects on EGFP expression were also weak or insignificant for the three constructs (pS3aPGDa3, pS3aPGDa6, and pS3aPGDa12) in which none of the high-affinity sites had been deleted. Hence, we observed a correlation between the presence of strong Six3a HD-binding sites, particu- larly the three ATTA motifs with a flanking G 1 nucleotide (a 1 ,a 2 , and t 15 ), and the ability to respond to six3a overexpression in vivo. Relative influence of ATTA core flanking nucleotides on Six3a HD binding To analyse the contribution of individual flanking nucleotides and their importance relative to specific positions within the ATTA core, we investigated how single-nucleotide substitutions influenced the binding affinity of a1. Modified a1 sites with single-nucleotide mutations in nine positions (G 1 T 2 C 3 A 4 T 5 T 6 A 7 G 8 G 9 ) were compared for their ability to compete with unmodified a1 in EMSAs (Fig. 4A). This in vitro anal- ysis showed that substitution of G 1 by T had moderate effects on the binding affinity for the Six3a HD (G 1 T in Fig. 4A). However, changes introduced at posi- tions 2 (T 2 C) and 3 (C 3 G) affected the binding more severely (Fig. 4A). Hence, these two 5¢-nucleotides appeared to be essential for Six3a HD-binding affinity. In fact, these substitutions resulted in less competition than observed for individual mutations in three of the positions within the ATTA core (A 4 C, T 5 G, and T 6 G). Only a substitution at the fourth position (A 7 C) of this core motif showed a similar deleterious effect on the binding. In addition, we observed that mutation of the 3¢-flanking G 8 nucleotide (G 8 A) had a moderate effect, similar to the G 1 T substitution. By contrast, a change of G 9 (G 9 T) did not cause any reduction of the bind- ing affinity, suggesting an insignificant role of this 3¢-flanking position. Overall, these results showed that positions 2 and 3 immediately 5¢ of the ATTA core are EGFP pS3aPG Δ a1 pS3aPG Δ a3 pS3aPG Δ a6 pS3aPGΔ t15 pS3aPGΔa12 pS3aPGΔa14 pS3aPGΔa18 pS3aPG 0.5 –++++++ + + 1.0 1.5 2.0 2.5 EGFP-expressing cells Six3a mRNA * * * Δ386 bp Δ406 bp Δ368 bp Δ470 bp Δ424 bp Δ444 bp Δ497 bp Fig. 3. Deletion of particular fragments within the six3a promoter region affects reporter gene expression in embryos. Schematic represen- tation of promoter–reporter deletion constructs and their corresponding activity in vivo. The pS3aPG vector (shown in Fig. 1A) and deletion constructs derived from it were coinjected with six3a mRNA into one-cell embryos. EGFP-expressing cells from trypsinated embryos (12 hpf) were detected by flow cytometry. The ratio between the EGFP expression from the reporter constructs coinjected with six3a mRNA and from those without six3a mRNA was calculated. Standard deviations were calculated from three repeated experiments. Asterisks indicate significant difference from the unmodified construct (pS3aPG) at P < 0.05. PS3aPGDa1, pS3aPGDt15 and pS3aPGDa18 showed significantly less increase in EGFP expression when coinjected with six3a mRNA. C. S. Suh et al. New recognition sequence for Six3 FEBS Journal 277 (2010) 1761–1775 ª 2010 The Authors Journal compilation ª 2010 FEBS 1765 particularly important for binding affinity for the Six3a HD. It also seemed that the Gs at positions 1 and 8 contributed moderately. Comparisons of the three high-affinity sites (a1, a2, and t15) showed considerable sequence identity at 11 positions, including the ATTA core, and a consensus sequence could be defined (Fig. 4B). By contrast, all of the low-affinity sites containing a G 1 nucleotide showed less identity with this consensus sequence (Fig. 4B). In particular, differences were detected for the critical nucleotides at positions 2 and 3. To inves- tigate the significance of the consensus sequence, we also determined its binding affinity for the Six3a HD. This analysis clearly showed that the consensus sequence bound more strongly than a1 (Fig. S3B). Therefore, we used the high-affinity consensus sequence site, which differed from a1 in only a single position (A 2 ) within the 27 bp probe fragment, as a reference in further studies of the relative importance of the flanking nucleotides. Our analysis tested the effects of other substitutions for the flanking nucleo- tides, and also included assays for the two additional positions (10 and 11) in the consensus sequence (Fig. 4C). The assays for G 1 substitutions in the consensus sequence site showed that a change from G to T had a similar effect (Fig. 4C) to the same substitution in a1 (Fig. 4A). Stronger reduction of the binding affinity was observed when G 1 was changed to C, but a change to A had no detectable effect (Fig. 4C). Consis- tent with the results obtained with the a1 probe (Fig. 4A), A 2 substitutions in the consensus sequence caused a strong reduction in the binding affinity (Fig. 4C). However, it seemed that the presence of A 2 in the consensus sequence caused a reduction in the relative importance of the C 3 nucleotide (Fig. 4C). Whereas two of the substitutions tested for G 8 and G 9 caused some reduction in the binding affinity, a1 G 1 T T 2 C C 3 G A 4 C T 5 G T 6 G A 7 C G 8 A G 9 T x 200 A B C Relative competition Motif sequence Motif name CON x 200 CON x 200 Relative competition Motif sequence Motif name G 1 A G 1 T G 1 C C 3 G C 3 A A 2 G A 2 C G 8 A G 8 C G 9 C G 9 T C 10 G C 10 T G 11 C G 11 T G 11 A Fig. 4. Identification of ATTA core flanking nucleotides critical for Six3a HD binding. (A) EMSAs with the Six3a HD and biotin- labelled a1 probe. Unlabelled fragments with a single point mutation in a1 (table) were used as competitors (· 200 molar excess) and compared with a1 for their ability to compete for Six3a HD binding. The two left lanes show the controls (labelled probe and labelled probe together with Six3a HD). The table shows the mutated sites and their rel- ative competition as compared with a1, from highest competition (+++++) to lack of competition ()). (B) Nucleotide similarities in the GNNATTA sites. The three high-affinity sites (a1, a2, and t15) are aligned at the top. Identical nucleotides flanking the ATTA core in the high-affinity sites are shown in red, and are represented in the consensus sequence (CON). GNNATTA sites with low affinity are aligned below the consensus sequence. (C) EMSAs with the Six3a HD and biotin-labelled consensus sequence probe. Unlabelled fragments with a single point mutation in the consensus sequence site (table) were used as competitors (· 200 molar excess) and compared with consen- sus sequence for their ability to compete for Six3a HD binding. The two left lanes show the controls (labelled probe and labelled probe together with Six3a HD). The table shows the mutated sites and their relative competition as compared with consensus sequence, from highest competition (+++++) to lack of competition ()). New recognition sequence for Six3 C. S. Suh et al. 1766 FEBS Journal 277 (2010) 1761–1775 ª 2010 The Authors Journal compilation ª 2010 FEBS single mutations of the other 3¢-nucleotides in the consensus sequence (C 10 and G 11 ) did not have detect- able effects (Fig. 4C). These results suggested marginal roles for each of the four 3¢-flanking positions within the consensus sequence. Information obtained from investigations of the effects of swapping flanking regions with sequences from a low-affinity binding site also supported this conclusion (Fig. 5A; see below). Our identification of similar high-affinity binding sites within the promoter regions of the zebrafish six3b and six7 genes is also consistent with the recognition sequence defined by these analyses (Fig. S4). Hence, the three Six3-like genes in zebrafish, which have par- tially overlapping expression domains [15,20], may be able to cross-regulate each other. Evaluation of the integrity and relative binding affinity of the recognition motif Regulatory transcription factors generally bind to short target sequences independently of the properties of the adjacent flanking regions. To investigate whether the Six3 HD recognition motif displays such integrity, we compared the relative binding affinities of hybrid sites, in which the a1 regions located 5¢ of G 1 and ⁄ or 3¢ of G 8 in the optimized consensus sequence site were replaced by completely different sequences from the corresponding parts of the low-affinity bind- ing site a9. This analysis showed that the strength of binding was only moderately influenced by changes in the sequences surrounding the recognition motif (Fig. 5A). Hence, when flanked by a9 sequences, the recognition motif displayed the same binding affinity as in the context of a1. This suggests that the identified recognition sequence may occur within different geno- mic contexts, where it can function as a target site for Six3a in vivo. Conversion of the most essential part of the consen- sus sequence, which includes the first seven positions (GACATTA), to its reverse complementary sequence (TAATGTC) facilitates direct comparisons with the recently reported recognition motifs of Six3 and other Six family proteins [10,14]. The two recognition sequences determined by studies of Six family proteins in Drosophila (TGATAC) and mice (TGATACC) [10,14] show a difference of only one nucleotide, and do not contain the TAAT core previously reported for murine Six3 by Zhu et al. [18]. We made direct com- parisons of the binding affinities of these recognition motifs by replacing the seven nucleotide core (TAATGTC) of the consensus sequence probe with the four alternatives of TGATACN. The competitive assays conducted for these fragments showed that the Six3a HD binding affinity for these motifs is compara- ble to the binding affinity for a1 (Fig. 5B). However, consistent with the proposed importance of the consen- sus, which was derived from the three high-affinity sites a1, a2 and t15, we observed strongest binding to the consensus sequence fragment. Using the same probe fragments, we also tested the binding properties of the two related zebrafish proteins Six3b and Six7. The HDs of these two proteins [15,20], which differ from Six3a in one and four residues, a1 Con a9 a1 Con a1 a9 Con a1 a9 Con a9 a9 LP-Con A B Con - TAATGTC a1 - TAATGAC brG - TGATACG nsA - TGATACA nsT - TGATACT brC - TGATACC LP-a1 Fig. 5. Influence of surrounding DNA sequences and comparison with other recognition motifs. (A) EMSAs with the Six3a HD and consensus sequence (CON) as a labelled probe (LP-CON) to ana- lyse how changes in the surrounding sequences (from a1) influence binding to the Six3a HD. Unlabelled fragments, in which the a1 sequences 5¢ to G 1 and 3¢ to G 8 in the consensus sequence were replaced with the corresponding parts from the low-affinity site a9, were used as competitors (· 100 excess). The two left lanes show the controls (labelled probe and labelled probe together with Six3a HD). Competition with unlabelled a9 fragment was also included as a reference (right lane). Replacement of the surrounding sequences had very limited effects on competition relative to the complete consensus sequence site (a1CONa1). (B) Comparison with the binding affinities of previously defined recognition sequences of Six3 and Six family proteins [10,14]. Using a1 as a labelled probe (LP-a1), competition was conducted with a · 100 excess of unla- belled fragments containing the a1 site, the consensus sequence, and the previously reported recognition motifs. The reverse com- plementary sequences are shown, and all fragments included the same additional flanking sequences (from a1). C. S. Suh et al. New recognition sequence for Six3 FEBS Journal 277 (2010) 1761–1775 ª 2010 The Authors Journal compilation ª 2010 FEBS 1767 respectively, showed very similar binding affinities to that of Six3a (Fig. 6). These findings indicate a high degree of overlap in binding sites for Six3a, Six3b, and Six7, and that they might compete for the same recognition motifs in regulatory networks. In vivo effects of eliminating particular high-affinity sites in the six3a promoter Transient reporter assays suggested that deletion con- structs lacking strong Six3a HD-binding sites responded less to the increased amount of Six3a pro- vided by coinjection of six3a mRNA (Fig. 3; see above). However, because of the relatively large dele- tions ( 450 bp) in these constructs, which included several putative Six3a-binding sites and ⁄ or other potential control sequences, it could not be excluded that the reductions in reporter expression reflected other effects. To determine more directly whether par- ticular high-affinity sites mediated responses to increased levels of Six3a, we investigated reporter con- structs with smaller deletions. One of these constructs, pS3aPGDa1.2, in which a deletion of 29 bp removed the closely spaced high-affinity sites a1 and a2, showed  40% reduction in the number of EGFP-expressing cells as compared with the control (Fig. 7A). Similarly, a construct with a 25 bp deletion, pS3aPGDt15.2, which eliminated a10 and t15, reduced expression of the reporter gene by  30%. Notably, these reductions were almost as strong as for the corresponding con- structs in which the deletions were larger (Fig. 3; see above). Hence, the clear effects of deleting the two pairs of high-affinity sites strongly indicated that they were important in mediating responses to six3a overex- pression. However, the possible existence of other con- trol sequences within these small deletions could not be completely ruled out. To affect the individual binding sites more directly, we made single-nucleotide substitutions in the most critical position within each of the two clustered high- affinity sites a1 (T 2 C) and a2 (A 2 C). We also mutated this position in t15 (A 2 C), and made a small deletion to eliminate its adjacent high-affinity site (a10). The oligonucleotides designed to make these site-directed mutations in the pS3aPG reporter construct were first assayed for binding to the Six3a HD. When the short double-stranded fragments generated by base pairing of these oligonucleotides were used as unlabelled com- petitors in EMSA, we observed very poor competition (Fig. 7B). This demonstrated that the mutations to be introduced in the reporter construct would severely affect the binding affinities of the targeted sites. The new reporter construct, which contained the mutated a1 and a2 sites, showed the same reduced ability to respond to six3a overexpression as the con- struct with the a1 ⁄ a2 deletion (Fig. 7A). We also observed similar effects for the construct in which the other two high-affinity sites, t15 and a10, had been mutated and deleted, respectively (Fig. 7A). These results provided strong evidence that the clustered high-affinity sites can mediate positive feedback from Six3a in vivo. However, the transient reporter assay in CON G1C G1A G1T A2G A2C C3G C3A CON Six3a ABC a1 brC brG nsA nsT Six3b Six3b a1 brC brG nsA nsT x 200 x 200 x 150 x 150 Six7 Fig. 6. Analysis of Six3b and Six7 HD binding to different recognition motifs. (A) EMSAs with the Six3b HD and biotin-labelled consensus sequence (CON) probe. Unlabelled fragments with a single point mutation in the consensus sequence site (see table in Fig. 4B) were used as competitors (lanes 4–11, · 200 molar excess) and compared with the consensus sequence for their ability to compete for Six3b HD bind- ing. Six3a HD was used as an internal control (lanes 1 and 2). Lane 3 shows the labelled probe together with Six3b HD. (B) Six3b HD bind- ing to a1 and previously defined recognition sequences of Six family proteins [10,14]. (C) Six7 HD binding to a1 and the previously defined recognition sequences of Six family proteins [10,14]. Using the consensus sequence as a labelled probe, competition was conducted with a · 150 excess of unlabelled fragments containing the a1 site and the previously reported recognition motifs. New recognition sequence for Six3 C. S. Suh et al. 1768 FEBS Journal 277 (2010) 1761–1775 ª 2010 The Authors Journal compilation ª 2010 FEBS injected embryos may not reflect the regulation of six3a gene expression under normal conditions. To gain further information about possible autoregulation of the six3a gene during early embryonic development, we investigated whether Six3a actually binds to some of the high-affinity sites in the endogenous six3a promoter (see below). In vivo validation of Six3a-binding sites in the six3a promoter region The transient assays of reporter constructs, which were conducted at the 12 hpf stage, did not provide any information regarding the possible binding of Six3a to the endogenous six3a promoter region in its natural context. To facilitate such an investigation, which required a specific antibody, we tagged Six3a with EGFP and used an antibody against green fluorescent protein (GFP) for ChIP assays on extracts from this embryonic stage. Following injection of six3a–EGFP mRNA into fertilized eggs (see Experimental proce- dures), we detected high EGFP levels during embryo- genesis (data not shown), and chromatin fragments isolated from these embryos were immunoprecipitated with the antibody against GFP. Using PCR primers designed to amplify specific regions within the six3a promoter (Fig. 8A), we investigated whether fragments containing high-affinity sites were selectively amplified (Fig. 8B). Consistent with the results obtained from the transient reporter assays (Figs 3 and 7), a short fragment (181 bp) containing the clustered a1 and a2 sites was amplified. In addition, we detected selective amplification of a distal fragment, which included a relatively strong binding site (a18) and five ATTA ⁄ TAAT core motifs. This finding also correlates well with results from transient reporter assays, where elim- ination of a18 and eight adjacent core motifs by a deletion caused a significant reduction in the response to six3a overexpression (Fig. 3). Furthermore, cloning and sequencing of the two amplified fragments con- firmed their origins from the six3a promoter region (data not shown). Notably, amplification was not detected with the primer pairs for regions containing only low-affinity TAAT ⁄ ATTA core motifs or lacking these motifs completely. However, we also observed no amplification 0.5 1.0 1.5 2.0 2.5 –+++++ EGFP-expressing cells Six3a mRNA EGFP Δ29 bp pS3aPGΔΔ ΔΔ a1.2 pS3aPG a1T 2 C, a2A 2 C Δ25 bp pS3aPG ΔΔ ΔΔ t15.2 pS3aPG Δ Δ Δ Δ a10, t15A 2 C a1/a2 a10/t15 Mut1 Mut2 x 200 pS3aPG A B Fig. 7. Mutation of high-affinity sites within the six3a promoter region affects reporter gene expression in embryos. (A) Schematic represen- tation of the in vivo activity of promoter–reporter constructs containing deletions (Da1 ⁄ a2 and Da10 ⁄ t15) or mutations of clustered high-affin- ity sites. The plasmid constructs were coinjected with six3a mRNA into one-cell embryos. EGFP-expressing cells from trypsinated embryos (12 hpf) were detected by flow cytometry. The ratio between the EGFP expression from the reporter construct coinjected with six3a mRNA and from that without six3a mRNA was calculated. Standard deviations were calculated from three repeated experiments. Significant differ- ence from the complete promoter–reporter construct (pS3aPG) was calculated at P < 0.05. (B) Double-stranded oligonucleotides with point mutations at position 2 of the Six3a HD-binding sites were used as cold probes in competitive EMSA to confirm loss of binding affinity. Mut1 represents a1T 2 C,a2A 2 C, and Mut2 represents Da10,t15A 2 C. C. S. Suh et al. New recognition sequence for Six3 FEBS Journal 277 (2010) 1761–1775 ª 2010 The Authors Journal compilation ª 2010 FEBS 1769 of the fragment containing the high-affinity sites a10 and t15, despite strong evidence from the reporter assays (Figs 3 and 7). Possibly, an unfavourable state of the chromatin within this particular region of the endogenous six3a promoter may prevent Six3a from binding at this developmental stage (12 hpf). Hence, it remains a possibility that the two sites may have autoregulatory functions at later stages of development. Discussion In this study, we show that binding of zebrafish Six3a to high-affinity sites in the promoter region of its own gene may contribute to autoregulation. The Six3a HD, which is identical to the corresponding DNA-binding domain of orthologous Six3 proteins in other verte- brates, displayed differential binding affinities for vari- ous sites containing a TAAT core motif, and this defined a specific recognition sequence (TAATGTC). When we compared this with a previously reported Six3-binding sequence (TGATAC), we observed simi- lar binding affinity, suggesting that the Six3 HD has several binding modes with distinct DNA-binding specificities. Using several experimental approaches, previous investigations of the DNA-binding preferences of the three Six protein subfamilies (Six1 ⁄ 2, Six3 ⁄ 6, and Six4 ⁄ 5) have found a number of different target sequences. Whereas analyses of native binding sites in putative target genes identified a variety of sequences [17,18,30,36–38], two independent studies based on dif- ferent in vitro selection procedures established a com- mon recognition sequence of the three subfamilies [10,14,39]. These partially conflicting results may reflect the fact that the HDs belonging to different subfami- lies have almost identical recognition helices but diver- gent N-terminal arms [16]. The HDs in the Six3 ⁄ 6 subfamily proteins show a higher degree of divergence relative to the other two subfamilies [16]. Consistent with this sequence diver- gence, early studies of the Six3 ⁄ 6 proteins indicated clear differences in DNA-binding specificity relative to the Six1 ⁄ 2 and Six4 ⁄ 5 family members [1]. Hence, murine Six3 did not bind to the ARE regulatory element of the Na + ⁄ K + -ATPase a1-subunit gene, GGTGTCAGGTTGC, which showed specific binding to Six4 and other members of the Six1 ⁄ 2 and Six4 ⁄ 5 subfamilies [17]. In further investigations of the DNA- binding properties of murine Six3, in vitro binding site selection demonstrated high-affinity binding to sequences containing the TAAT core motif, and clus- ters of these tetranucleotide sites were found in regula- tory elements within the promoter regions of both Six3 and Wnt1 [18,37]. Although these analyses dem- onstrated an involvement of the clustered TAAT motifs in mediating negative regulation by Six3, –3579 –3290 –2893 –2704 –2093 –1872 –1183 –980 –656 –477 –253 –72 six3a promoter region (pS3aP) IP (+Ab, injected) Input control IP (+Ab, uninjected) 300 bp 200 bp 300 bp 200 bp 300 bp 200 bp 18 17 A B 16 15 14 13 12 11 10 9 8 7 6 5 43 2 t2 1 * * t13 t15 t17 * Fig. 8. Detection of Six3a binding to high-affinity sites in the six3a promoter in vivo. (A) Schematic representation of the promoter region of six3a (see legend to Fig. 2A) and the fragments that were selected for PCR amplification. Filled boxes indicate the regions that were ampli- fied, and empty boxes represent regions that showed no amplification. (B) ChIP PCR assay on 12 hpf embryos. The upper panel represents PCR amplification of different regions (indicated by the boxes) within the six3a promoter following immunoprecipitation (IP) of DNA bound to Six3a–EGFP fusion protein (Experimental procedures). The middle panel represents PCR amplification from uninjected zebrafish embryos immunoprecitated with GFP antibody (Ab). The lower panel represents PCR amplification from total embryonic DNA. New recognition sequence for Six3 C. S. Suh et al. 1770 FEBS Journal 277 (2010) 1761–1775 ª 2010 The Authors Journal compilation ª 2010 FEBS [...]... region showing the positions of sites containing ATTA (and TAAT) core sequences Fig S2 Sites within the six 3a promoter containing oppositely oriented TAAT cores that showed weak binding as compared with the a1 site in competition EMSAs Fig S3 Comparison of relative binding affinities of the consensus sequence (CON) and the strongest binding sites located within the six 3a promoter region Fig S4 Comparison... high-affinity GNNATTA sites located within the promoter regions of the zebrafish genes six 3a, six3b and six7 indicates a common recognition motif Fig S5 Autoregulatory activity of the six 3a promoter assayed in zebrafish embryos Table S1 Primer sequences used in ChIP assay This supplementary material can be found in the online version of this article Please note: As a service to our authors and readers,... of mRNA The Six 3a coding sequence was amplified from zebrafish cDNA (10–12 hpf) and cloned into the pEGFP-N1 vector (Clontech, Mountain View, CA, USA) PstI and XbaI were used to cut out the six 3a EGFP fusion and subclone it into the expression vector pCS2+ [46] For six 3a mRNA transcription, the six 3a coding sequence was cloned directly into the pCS2+ vector The six 3a GFP and six 3a mRNAs were generated... et al additional features specifically favouring binding to Six3 proteins were not identified Our examination of the interaction of zebrafish Six 3a with 43 TAAT-containing sequences, distributed within the promoter region of its own gene, identified three high-affinity sites that defined the recognition sequence TAATGTC Notably, this sequence preference includes three adjacent nucleotides 3¢ to the TAAT... transgenic lines Cloning, expression and purification of glutathione-S-transferase (GST)–Six3aHD fusion protein The HD coding sequence of six 3a was amplified by PCR, using 5¢-TCAGGTCGGATCCATGGTTTTCAGA-3¢ as forward primer and 5¢-CTGTGTGGAATTCATACGTCG CATTC-3¢ as reverse primer, cloned in- frame into the BamHI and EcoRI site of the pGEX-2T expression vector (Pharmacia Biotech Inc., Uppsala, Sweden), and transformed... were made: pS3aPG, pS3aPGDa1 ,a2 ,t2 (or pS3aPGDa1), pS3aPGDa3 a5 (or pS3aPGDa3), pS3aPGDa6 ,a7 ,t13 (or pS3aPGDa6), pS3aPGDa8 a1 1,t15 (or pS3aPGDt15), pS3aPGDa12,t17 (or pS3aPGDa12), pS3aPGDa13 a1 5 (or pS3aPGDa14), pS3aPGDa16 a1 8 (or pS3aPGDa18), pS3aPGDa1.2, and pS3aPGDat15.2 Following PCR amplification of the whole plasmid except the region to be deleted, the PCR product was digested with XmaI and DpnI... of the cofactors that may have contributed, the activation was dependent on the Six 3a HD -binding sites shown to bind strongly in vitro Other members of the Six3 ⁄ 6 family, which have similar or identical DNA -binding specificity to that of Six 3a, have partially overlapping expression at early stages of eye and forebrain development [15,20] Hence, transcriptional control of these genes is likely to involve... particular, variations in the position of the N-terminal arm have been shown to in uence its recognition properties [6,10] Investigations of the binding properties of the Ultrabithorax protein have also demonstrated that the N-terminal arm can strongly affect the topological binding mode of the HD and in uence the nucleotide contacts made by the recognition helix [6] Notably, these different binding modes... same strong binding affinity Furthermore, major changes of the surrounding sequences did not affect the binding affinity of the recognition sequence, indicating that it is generally independent of the sequence context Importantly, the TAATGTC motif is also distinct from all the different DNAbinding specificities that were recently described in surveys including the majority of the Drosophila and mouse HDs... (Carlsbad, CA, USA) Briefly, 12 hpf embryos were crosslinked with formaldehyde, and chromatin was isolated The isolated chromatin was sonicated to an average size of about 300 bp Incubation with normal goat IgG precleared the chromatin Protein G magnetic beads were incubated with antibody against GFP at 4 °C overnight Immunoprecipitation reactions were performed in duplicate by incubating 10 lg of antibody . Autoregulatory binding sites in the zebrafish six 3a promoter region define a new recognition sequence for Six3 proteins Clotilde S. Suh, Staale Ellingsen*, Lars Austbø, Xiao-Feng Zhao, Hee-Chan. similar binding affinities to that of Six 3a (Fig. 6). These findings indicate a high degree of overlap in binding sites for Six 3a, Six3b, and Six7, and that they might compete for the same recognition. Six 3a actually binds to some of the high-affinity sites in the endogenous six 3a promoter (see below). In vivo validation of Six 3a -binding sites in the six 3a promoter region The transient assays of