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

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 bindi

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.

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

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 orientaposi-tions (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 presinflu-ence

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 (G1T2C3A4T5T6A7G8G9) and a2 (G1A2T3A4T5T6A7T8G9) have common Gs at positions upstream (G1) and downstream (G9) 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

G1 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 G1 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 G1have sig-nificant influence on the binding affinity Notably, three ATTA sites (a10, a14, and a18) without the G1 flanking nucleotides also bound quite strongly to the Six3a HD (Fig 2B) However, among the 21 TAAT sites lacking the G1nucleotide, 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 G1-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 G1-containing sites, a1, a2, and t15 (Fig S3)

six3a promoter region

ATG

EGFP

730 bp

3635 bp A

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).

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

x 200

x 200

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

t2

1

*

*

t13 t15

t17

*

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 ( )).

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 G1-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 G1 nucleotide (a1, a2, and t15), 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 (G1T2C3A4T5T6A7G8G9) were compared for their ability to compete with unmodified a1 in EMSAs (Fig 4A) This in vitro anal-ysis showed that substitution of G1by T had moderate effects on the binding affinity for the Six3a HD (G1T

in Fig 4A) However, changes introduced at posi-tions 2 (T2C) and 3 (C3G) 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 (A4C, T5G, and T6G) Only a substitution at the fourth position (A7C) of this core motif showed a similar deleterious effect on the binding In addition, we observed that mutation of the 3¢-flanking G8nucleotide (G8A) had a moderate effect, similar to the G1T substitution By contrast, a change

of G9 (G9T) did not cause any reduction of the bind-ing affinity, suggestbind-ing 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 Δa1

pS3aPG Δa3

pS3aPG Δa6

pS3aPG Δt15

pS3aPG Δa12

pS3aPG Δa14

pS3aPG Δa18

pS3aPG

0.5

1.0 1.5 2.0 2.5

EGFP-expressing cells

*

*

*

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.

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 G1 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 (A2) 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 G1 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 G1 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), A2 substitutions in the consensus sequence caused a strong reduction in the binding affinity (Fig 4C) However, it seemed that the presence of A2

in the consensus sequence caused a reduction in the relative importance of the C3nucleotide (Fig 4C) Whereas two of the substitutions tested for G8 and

G9 caused some reduction in the binding affinity,

T 2

C 3

A 4

T 5

T 6

A 7

G 8

G 9

x 200

A

B

C

Relative competition Motif

sequence Motif

name

x 200

x 200

Relative competition Motif

sequence Motif

name

G 1

G 1

G 1

C 3

C 3

A 2

A 2

G 8

G 8

G 9

G 9

C 10

C 10

G 11

G 11

G 11

Fig 4 Identification of ATTA core flanking nucleotides critical for Six3a HD binding (A) EMSAs with the Six3a HD and biotin-labelled a1 probe Unbiotin-labelled 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 ( )).

single mutations of the other 3¢-nucleotides in the

consensus sequence (C10and G11) 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 G1 and⁄ or 3¢ of G8 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).

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 concon-structs,

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 (T2C) and a2 (A2C) We also mutated this position in t15 (A2C), 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

x 200

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 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 bindbind-ing 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.

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

EGFP

Δ29 bp pS3aPG ΔΔΔΔa1.2

pS3aPG a1T 2 C, a2A 2 C

Δ25 bp pS3aPG ΔΔΔΔt15.2

pS3aPG ΔΔΔΔa10, 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 murepresen-tations 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 a1T2C,a2A2C, and Mut2 represents Da10,t15A2C.

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 DNA-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

300 bp

300 bp

18 17

A

B

16 15 14 13 12 1110 9 8 7 6 5 4 3 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.