Four divergent Arabidopsis ethylene-responsive element-binding factor domains bind to a target DNA motif with a universal CG step core recognition and different flanking bases preference Shuo Yang 1 , Shichen Wang 1 , Xiangguo Liu 1 , Ying Yu 1 , Lin Yue 3 , Xiaoping Wang 1 and Dongyun Hao 1,2 1 Key Laboratory for Molecular Enzymology and Engineering of the Ministry of Education, Jilin University, Changchun, China 2 Biotechnology Research Centre, Jilin Academy of Agricultural Sciences (JAAS), Changchun, China 3 School of Physical Education, Northeast Normal University, Changchun, China Introduction The ethylene-responsive element-binding factor (ERF) gene family of transcriptional factors is one of the largest transcriptional factor gene families in the plant kingdom [1,2]. The ERF domain was first identified as a conserved motif of 58–59 amino acids in four DNA- binding proteins from tobacco and was shown to bind specifically to a GCC box [3]. After the completion of the sequencing of the Arabidopsis genome [4], 124 genes were predicted to encode proteins belonging to the AtERF family [2]. Keywords CG step; DRE motif; ERF domain; homology; universal binding characteristic Correspondence D. Hao, Biotechnology Research Centre, Jilin Academy of Agricultural Sciences (JAAS), Changchun 130033, China Fax: +86 431 87063080 Tel: +86 431 87063195 E-mail: dyhao@cjaas.com (Received 31 August 2009, revised 29 September 2009, accepted 8 October 2009) doi:10.1111/j.1742-4658.2009.07428.x The Arabidopsis ethylene-responsive element-binding factor (AtERF) fam- ily of transcription factors has  120 members, all of which possess a highly conserved ERF domain. AtERF1, AtERF4, AtEBP and CBF1 are members from different phylogenetic subgroups within the family. Electrophoretic mobility shift assay analyses revealed that the ERF domains of these four proteins were capable of binding specifically to either GCC or dehydration-responsive element (DRE) motifs. In vitro and in vivo binding assays of the four AtERFs with the DRE motif showed that the recognition of the CG step was indispensable in all four of the specific binding reactions, implying that there may be a universal binding characteristic of various ERF domains binding to a given con- sensus (e.g. the DRE motif). In addition, the core DNA-binding motifs preferred by the four AtERFs were identified, and all of these motifs contained a conserved CG step core. Thus, conserved recognition of the CG step may be the foundation of the formation of the stable complex by the ERF domain with the DRE motif, which is probably determined by the highly conserved residues presented in the DNA contact surface among the whole AtERF family members. The different preferences at flanking bases of individual ERF domains, which appear to be attrib- uted to the subfamily- or subgroup-specific residues, may be essential discrimination of the target binding motif from various similar sequences by divergent AtERF domains. Abbreviations DBD, DNA binding domain; DRE, dehydration-responsive element; EMSA, electrophoretic mobility shift assay; ERE, ethylene-responsive element; ERF, ethylene-responsive element-binding factor. FEBS Journal 276 (2009) 7177–7186 ª 2009 The Authors Journal compilation ª 2009 FEBS 7177 The AtERF family is further divided into various subgroups according to the homology of ERF domains [5,6]. An ERF domain consists of a three-stranded anti- parallel b-sheet and an a-helix, packed approximately parallel to the b-sheet, with the seven thoroughly con- served amino acids (Arg6, Arg8, Trp10, Glu16, Arg18, Arg26 and Trp28) in the b-sheet contacting uniquely with the bases of the target DNA at the major groove (see Fig. 1A) [7]. Phylogenetic analyses of the ERF domains of all members within the AtERF family show that the residues Arg6, Glu16 and Trp28 are completely conserved among all 124 members, whereas more than 95% of members contain the Arg8, Arg18, Arg26 residues [6]. From the results of the few AtERFs studied, however, the conserved ERF domains do not seem to prefer identical DNA consensus sequences. For ins- tance, some AtERFs have been shown to bind in vitro to the ethylene-responsive element (ERE), a GCCGCC motif designated the GCC motif [3,8–12], to conduct GCC motif-mediated transcription (activation or repres- sion) in leaves of Arabidopsis [12]. This ERE was first reported to be a binding site (referred to as the GCC box) of a number of tobacco ERF proteins [3] and it was later presumed to be the target site of many other ERF proteins [2]. The ERF protein, AtEBP, was also found to protect the GCC box in a DNase I foot-printing analysis [10]. In contrast, the dehydration-responsive element (DRE), with the TACCGACAT motif, in the drought- responsive gene rd29A from Arabidopsis has been proven to be the recognition site of DRE-binding proteins, which are transcription factors that have authentic ERF domains [13] and that are involved in the induction of rd29A expression by low-temperature stress. A similar element to DRE, the C-repeat (TGGCCGAC) has been identified in the cold-induc- ible gene cor15a and it is reported to function in cold-response regulation through binding by another ERF protein, CBF1 [14]. The similarity of these ERF-binding elements and the high similarity of ERF domains among the mem- bers of the entire ERF family have led to speculation that the ERF domains from various subgroups within the AtERF family recognize a certain binding site with universal binding characteristic to a conserved core. The divergent short flanking bases, on the other hand, allow preference to govern differential recognition. We have previously demonstrated that various ERF domains had divergences in their DNA recognition modes [9], but, to date, additional supporting evidence has been lacking. Indeed, little is still known regarding the ways in which these differences are important for the functionalities of members in the AtERF family, the majority of which have not yet been studied. In the present study, we selected four representatives from different functional subgroups of the AtERF family and characterized the in vivo and in vitro bind- ing specificities of the four ERF domains for a sequence containing the DRE motif. In addition, we used a random sequence selection method to identify the core recognition motifs preferred by each of the four domains. A universal binding characteristic was revealed, in addition to the individual features of vari- ous ERF domains involved in recognition of the DRE A B C Fig. 1. (A) Solution structure of AtERF1–GCC box complex (PDB code: 1GCC) [7]. The DNA-binding domain is shown in the sche- matic; DNA is represented by tubes. The b-sheet of the ERF domain is light blue and the seven conserved amino acid residues reported to contact DNA bases directly are red; other conserved amino acid residues that do not directly contact with DNA bases are blue. (B) The DNA base sequence with position numbering along the 16 bp fragment of DREwt. The bases in the core ACC- GAC are in bold and boxed in gray. (C) Sequence alignment of four ERF domains of AtERF1, AtERF4, AtEBP and CBF1. The secondary structure scheme is indicated above the sequence. The conserved amino acid residues that directly contact with DNA bases and the other conserved amino acid residues that do not directly contact with DNA bases are in red and blue, respectively. Arabidopsis ERFs recognize a common CG step core S. Yang et al. 7178 FEBS Journal 276 (2009) 7177–7186 ª 2009 The Authors Journal compilation ª 2009 FEBS motif. The results have important implications for understanding the foundations of recognition of a given binding site by divergent members of the AtERF family. Results and Discussion The members of the ERF family in Arabidopsis can be classified into a number of different phylogentic sub- groups according to the sequence similarity of the ERF domains [6]. We selected four AtERFs – AtERF1, AtERF4, AtEBP and CBF1 – as representa- tives from divergent subgroups (for details, see Figs 1C and 6), to investigate whether the highly homologous ERF domains of different AtERFs have universal binding characteristics for the recognition of a given consensus sequence (e.g. DRE). Binding specificity of AtERFs to the GCC and DRE motifs Having established that CBF1 can specifically recog- nize both GCC and DRE motifs [9], the two most popularly reported ERF-binding sites, we continued to explore the DNA-binding specificity of the other three AtERFs. Table 1 shows that all four ERF domains were capable of binding specifically to the 16 bp frag- ment containing either the GCC or the DRE motif. The equilibrium dissociation constants (K d )of AtERF1, AtERF4 and AtEBP for binding to the DRE motif were within the level of typical monomeric interaction, although the binding activities were in gen- eral lower than those for binding to the GCC motif. CBF1 appeared to bind to the DRE motif more strongly than to the GCC motif, implying CBF1 may prefer the DRE motif over the GCC motif. To further confirm if these variations in binding affinity were caused by binding instability as a result of nonspecific interference, rather than the alternation of a binding site, we carried out the competition binding assay using a nonspecific competitor poly[dA-dT].poly[dA- dT] in an electrophoretic mobility shift assay (EMSA). Figure 2 shows that most of the AtERFs exhibited similar stability in binding to either the GCC or the DRE motif. The most remarkable feature arising from the competition binding assay was the consistency of the binding preference of the AtERFs with the EMSA analysis. The three AtERFs, AtERF1 AtERF4 and AtEBP, with higher sequence similarity to each other than to CBF1, had similar binding preferences in comparison with CBF1. Verification of the binding characteristics of the selected AtERFs with the DRE motif To verify the detailed binding characteristics of the four different AtERFs to a given consensus sequence DRE, EMSAs were carried out with the DRE motif and its mutants possessing single T substitutions (see Fig. 3). Each base in the DRE motif from T5 to C11 was replaced with a T, except that T5 was replaced by A, and the binding free energy changes (DDG) were obtained from quantitative titration analysis. Figure 3 shows that AtERF1 and AtERF4 exhibited the highest specific interactions at C7, C8, G9 or C11, because the Table 1. Binding activities of the selected AtERFs to GCC and DRE motif-containing sequences. Four ERF domains were tested for binding to the 16 bp DRE or GCC motif-containing sequences using quantitative EMSA, as described in Materials and methods. K d values are represented as the mean of three replicates ± stan- dard deviation. The K d value for nonspecific binding was estimated to be  1 l M or higher. ERF fragments GCCwt (n M) DREwt (nM) AtERF1-f 0.17 ± 0.04 2.02 ± 1.44 AtERF4-f 0.38 ± 0.24 31.7 ± 14.3 AtEBP-f 0.34 ± 0.17 1.25 ± 0.62 CBF1-f 5.63 ± 0.61 1.46 ± 0.99 Fig. 2. Competition binding assay of the ERF–DNA complex. The binding complex of the ERFs and their binding DNAs were incu- bated together with 0, 0.001, 0.01, 0.1, 1.0 and 10 lg poly[dA- dT].poly[dA-dT] in a 10 lL volume and analysed by EMSA, as described in the Materials and methods. S. Yang et al. Arabidopsis ERFs recognize a common CG step core FEBS Journal 276 (2009) 7177–7186 ª 2009 The Authors Journal compilation ª 2009 FEBS 7179 base substitution at that position caused the greatest decline in binding activity. AtEBP requested C8, G9 and C11 most frequently and with the moderate requirements of C7. As for CBF1, the prerequisite bases appeared to be C8 and G9, whereas the other bases within the binding motif were only moderately required to varying extents. In the four reactions, bases C8 and G9 in the DRE motif were absolutely requested by all AtERFs for specific binding, indicat- ing that the recognition of the CG step was conserved by various AtERFs and may be the universal binding characteristic of different AtERFs in recognition with the DRE motif. In addition, bases C7 and C11 within the motif were required to different extents by AtERFs from the divergent phylogentic subgroups, implying that the recognition of these bases was the individual feature of distinct AtERFs binding to the DRE motif. In vivo DNA binding specificity of AtERFs by the reporter–effector transient assay To confirm if these binding specificities of AtERFs observed in vitro were also capable of regulating the DRE-mediated transcription within plant tissue, repor- ter effect cotransformation assays were carried out. An effector plasmid possessing the coding region of the full-length AtERF1, AtEBP or CBF1 genes driven by the CaMV 35S promoter, together with the luciferase reporter gene containing four tandem copies of either the DRE motif or its mutants at the upstream regula- tory region, was coexpressed into Arabidopsis leaves by particle bombardment. Figure 4 shows that these three AtERFs were able to transactivate the transcription of a gene carrying the wild-type DRE motif (DREwt), which was represented by an increase in luciferase activity of about four- to seven-fold over the control. No luciferase activity was detected when any of the three ERF effectors was cotransformed with a reporter carrying DREt1, in which the C8 was replaced by T. Although AtERF1 did not activate transcription of the reporter gene carrying either DREt2 or DREt3, the coexpressions of AtEBP and CBF1 activated tran- scription of DREt3 reporter genes to varying degrees. As AtERF4 was a repressor, an extra effector in which the activation domain of viral protein 16 was fused to the yeast GAL4 DNA binding domain (DBD) and then coexpressed with the AtERF4 effector was used to test the in vivo binding specificity of AtERF4. The reporter gene containing multicopies of the GAL4 binding sequence was inserted into the existing lucifer- ase reporter next to the four tandem DRE motifs and the transcription suppression by AtERF4 was assayed. Figure 5 shows that AtERF4 suppressed viral protein 16 activation by more than 50% when cotransformed with the reporter carrying DREwt, whereas no repres- sion was detected with a reporter having mutant DRE motifs in which C8, G9 or C11 were replaced by T. 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 AtERF1 AtERF4 AtEBP CBF1 5 6 7 8 9 10 11 ΔΔG (Kcal·mol –1 ) ΔΔG (Kcal·mol –1 ) ΔΔG (Kcal·mol –1 ) ΔΔG (Kcal·mol –1 ) Fig. 3. Effect of single base substitutions on the relative binding free energy change (DDG) in the binding of the four ERF domains to the DRE motif. The DNA sequence shown at the bottom is the DRE motif in which each base was substituted individually one by one as illustrated. The solid bars indicate the increase in DDG caused by the base substitution at the corresponding position. Posi- tive DDG represents a decreased binding activity; a 10-fold decrease in binding activity increased DDG by  1.3 kcalÆmol )1 . Arabidopsis ERFs recognize a common CG step core S. Yang et al. 7180 FEBS Journal 276 (2009) 7177–7186 ª 2009 The Authors Journal compilation ª 2009 FEBS These observations were consistent with the findings in the in vitro single base substitution binding assays: the substitution at C8 or G9 abolished the specific recogni- tion of the DRE motif by all four of the AtERFs. Random binding site selection reveals the binding characteristics of divergent AtERFs to the DRE motif To clarify the possible existence of the moderately divergent binding motifs of the four AtERFs from the divergent phylogenetic subgroups, randomized oligonu- cleotide selection was performed. The resulting binding motif of hexamers selected by these four ERF domains is shown in Table 2. AtERF1 seemed to prefer the hexamer GCCGCC motif, which is consistent with the results from previous studies [7,8]. Although the AtERF4 required a relatively relaxed G or A at posi- tion 2 of the hexamer G ⁄ aCCGCC, AtEBP selected a binding motif of hexamer GCCGCC. The selected motif of CBF1, AA ⁄ cCGAC, appears to agree with a previous report [14]. Although each ERF domain showed different binding preferences, all of the binding sites selected by the AtERFs from the four subgroups possessed a common CG core in the centre and a con- served C at the last position (position 7). These moder- ately divergent bases existed in the other positions within the binding motifs, discriminating the members from different subgroups. The solution structure of the complex formed by the ERF domain of AtERF1 with the GCC box (1GCC) shows that two categories of residues within the domain are considered to be important for specific DNA bind- ing: one consists of the residues in the b-sheet directly contacting the DNA bases; and the other is made up of the numerous Ala residues in the a-helix and the hydro- phobic residues with larger side chains in the b-sheet (in particular Phe13, Phe32, Val27 and Ile17), which appears to determine the geometry of the a-helix rela- tive to the b-sheet [3–5,7–9,17]. A multiple alignment of Arabidopsis ERF domains (Fig. 6) shows that a series of residues (e.g. Gly4, Arg6, Arg8, Gly11, Glu16, Ile17, Arg18, Arg26, Trp28, Leu29, Gly30, Ala38, Ala39, Asp43 and Asn57) were almost absolutely conserved among all members of the ERF family. Most of these residues are present in the b-sheet, especially Arg6, Arg8, Glu16, Arg18, Arg26 and Trp28 (Fig. 1A), which are reported to contact directly with DNA, suggesting that the conformation of a partial DNA contact surface may be conserved among various ERF domains, which result in the conserved recognition of the CG step in the DRE motif by all four of the different AtERFs. On the other hand, some other residues reported to determine the geometry of the a-helix relative to the b-sheet were not as conserved as these other residues, but instead were subfamily or subgroup specific, e.g. the Ile17 in almost all of the ERF family (V17 in CBF1), V27 in ERF subfamily (Ile27 or Leu27 in the DRE-binding protein subfamily) and Tyr42 in the major ERF family (His42 in the CBF1 and TINY sub- group) (Fig. 6). However, these subfamily- or group- specific residues seem not to be involved in the direct base contact, which may affect the local conformation of the interface by the determination of the geometry of the a-helix relative to the b-sheet. It seems that the Reporters Effectors 4 x DRE TATA LUC Nos Ω CaMV-35S Nos ERF Relative luciferase activity 0 2 4 6 8 10 DREwt 0 2 4 6 8 10 DREt2 DREt1 DREt3 Binding motifs: l o r t n o C 1 F B C 1 F R E t A P B E t A l o r t n o C 1 F B C 1 F R E t A P B E t A Fig. 4. AtERF1, AtEBP and CBF1 activate the transcription of the luciferase reporter gene driven by the DRE motif and its mutants. The luciferase reporter gene contains four copies of the cis-acting binding motif, DREwt, DREt1, DREt2 or DREt3, which are high- lighted and underlined. The effector was constructed with a full length of ERF cDNA that was controlled under the CaMV 35S pro- moter following a translation enhancer (X) from tobacco mosaic virus. These effectors induce transactivation of the reporter gene. The control in the transient assay was the same as the experi- ments without the addition of an effector. The results are shown as relative luciferase activity per control. S. Yang et al. Arabidopsis ERFs recognize a common CG step core FEBS Journal 276 (2009) 7177–7186 ª 2009 The Authors Journal compilation ª 2009 FEBS 7181 flanking positions, as well as the CG step core in the DNA motif, were required to varying extents by diver- gent ERF domains, and may be determined by these subfamily- or group-specific residues. The biological function in DNA binding of individ- ual ERF domains is apparently determined by the primary structures of the divergent DBD and a phylo- genetic classification of the ERF family may partly reflect the features in DNA binding of a certain popu- lation of ERF domains. The observations acquired in the present study imply that the divergent ERF domains from various groups of the family bind to a given consensus sequence by conserved recognition of a CG step core as the universal binding characteristic. This may be the foundation of the formation of a sta- ble ERF–DNA complex and the different flanking position preferences by individual ERF domains may be crucial for the precise regulation of their own target genes by various ERFs. Materials and methods Preparation of ERF domain-containing proteins The coding region of the ERF domain of CBF1 (Uni- ProtKB: P93835) (amino acids 47–142), which contains 10 and 38 amino acids in the N- and C-terminal regions, respectively, was prepared as described previously [9]. The Fig. 5. AtERF4 suppresses the transcription of the luciferase repor- ter gene driven by the DRE motif and it mutants. A multicopy of the GAL4 binding sequence was inserted into the DRE:luciferase reporter next to the 4· DRE motif. An extra effector was con- structed carrying the coding sequences of the activation domain of viral protein 16 and the yeast GAL4 DBD. The reporter and two effectors in a ratio of 1 : 1 : 1 were cotransformed into plant tissue; the remainder was the same as in Fig. 4. Table 2. Selection of binding sites from a random oligonucleotide pool by ERFs. Selections were performed using a 60 bp oligonu- cleotide containing a randomized site of 10 bp. The selected sequences were aligned computationally and the appearance of a base at each position in a motif was presented as a percentage fre- quency of all four kinds of base. The base with a frequency higher than 50% (bold) was defined as the selected site. If the second highest frequency base showed not less than half the highest fre- quency (marked with an asterisk), it was defined as the second possible site and is presented in lower case letter. Proteins Selection position Frequency (%) AC G T Deduced consensus AtERF1 1 38.5 15.4 30.8 15.4 N 2 0.0 11.5 88.5 0.0 G 3 0.0 80.8 15.4 3.8 C 4 11.5 76.9 7.8 3.8 C 5 0.0 0.0 100 0.0 G 6 15.4 69.2 3.8 11.5 C 7 11.5 80.8 3.8 3.8 C AtERF4 1 29.6 33.3 22.2 14.8 N 2 25.9 11.1 51.8 11.1 G ⁄ a* 3 3.7 77.8 14.8 3.7 C 4 22.2 66.7 3.7 7.4 C 5 0.0 0.0 100 0.0 G 6 7.4 81.5 11.1 0.0 C 7 7.4 74.1 18.5 0.0 C AtEBP 1 15.4 57.7 23.1 3.8 C 2 0.0 3.8 96.2 0.0 G 3 3.8 92.3 0.0 3.8 C 4 0.0 92.3 0.0 7.7 C 5 0.0 0.0 100.0 0.0 G 6 11.5 84.6 3.8 0.0 C 7 0.0 100.0 0.0 0.0 C CBF1 1 20.0 40.0 16.0 24.0 N 2 36.0 20.0 36.0 8.0 V 3 8.0 64.0 16.0 12.0 C 4 8.0 60.0 20.0 12.0 C 5 0.0 4.0 88.0 8.0 G 6 56.0 32.0 8.0 4.0 A ⁄ c 7 4.0 68.0 24.0 4.0 C Arabidopsis ERFs recognize a common CG step core S. Yang et al. 7182 FEBS Journal 276 (2009) 7177–7186 ª 2009 The Authors Journal compilation ª 2009 FEBS Fig. 6. Sequence alignment of ERF domains of members of the Arabidopsis ERF family. All ERF domain sequences were aligned and classi- fied according to the results from the phylogenetic tree. The names of the ERF domains are represented by their gene locus numbers except that the names of the four domains used in this study are represented by the transcriptional factor names. The secondary structure indicated above the sequence and the seven conserved amino acid residues reported to contact DNA bases directly [7] are in red; other conserved amino acid residues that do not directly contact DNA bases are in blue. S. Yang et al. Arabidopsis ERFs recognize a common CG step core FEBS Journal 276 (2009) 7177–7186 ª 2009 The Authors Journal compilation ª 2009 FEBS 7183 ERF domains of AtERF1 (UniProtKB: O80337), AtERF4 (UniProtKB: O80340) and AtEBP (UniProtKB: P42736) with 10 and 8 amino acids in the terminal regions, respec- tively, were prepared according to the previous work of Hao et al. [8]. The PCR products were then cloned into the pET16b plasmid (Novagen, Merck, Darmstadt, Germany) Fig. 6. (Continued ). Arabidopsis ERFs recognize a common CG step core S. Yang et al. 7184 FEBS Journal 276 (2009) 7177–7186 ª 2009 The Authors Journal compilation ª 2009 FEBS and the corresponding proteins were expressed in BL21(DE3) pLysS (Merck) Escherichia coli cells and puri- fied using a His-Trap his-tagged protein purification kit (Amersham Pharmacia Biotech, Uppsala, Sweden). The pro- tein concentrations were determined using the bicinchoninic acid protein assay kit (Pierce, Chester, UK) and further confirmed using the method of Gill and von Hippel [18]. EMSA Two 16 bp fragments, EREwt (5¢-CATAAGAGCCGCC ACT-3¢) and DREwt (5¢-ATACT ACCGACATGAG-3¢) (for DNA base sequence and position numbering of DREwt, see Fig. 1B), from the promoter region of the tobacco Gln2 gene [3] and the Arabidopsis rd29A gene [19], respectively, were prepared, together with their mutants, by synthesizing both stands. The EMSA, binding titration analysis and the calculation of the binding free energy change (DDG) were performed as described previously [8,9]. Binding competition assay The binding condition and buffers used in the competition assay were the same as used in the quantitative DNA-bind- ing assay described above. The radioisotope-labelled DNA probe was first mixed with the binding protein at a concen- tration corresponding to its K d . After allowing it to com- plex for 5 min at room temperature, the mixture was distributed into aliquots, to which a poly.[d(A-T)].poly[dA- dT] (Amersham Pharmacia Biotech) gradient of 0.001– 10 lg was added to a final volume of 10 lL of each aliquot. After incubation for a further 10 min, the contents were loaded on to an 8% nondenaturing PAGE and visual- ized as for EMSA. Construction of the reporter and effector genes For the reporter gene constructs, see Fig. 4. The detailed dual-luciferase reporter transient assay was performed as described previously [9]. Selection of the DNA-binding site A 60 bp single-stranded DNA RDM10, with 10 random- ized oligonucleotides in the center, i.e. CTGTCAGTGAT GCATATGAACGAATN 10 AATCAACGACATTAGGATC CTTAGC was synthesized. A 100 ng sample of RDM10 was radiolabelled during synthesis of double-stranded DNA using [ 32 P]dATP[aP] with the E. coli Klenow fragment (New England Biolabs, Ipswich, MA, USA). The selections were performed after incubation with the individual ERF domains (25–100 ng) followed by EMSA. Briefly, each binding reaction was carried out in a 10 lL binding buffer [25 mm Hepes-KOH (pH 7.5), 40 mm KCl, 0.1 mm EDTA, 0.1 mgÆmL )1 BSA, 10% glycerol and 1 lg double-stranded poly(dI–dC)] and 25–100 ng of individual ERF domain. The bound oligonucleotides were gel purified, extracted with phenol ⁄ chloroform and precipitated with ethanol. The puri- fied DNAs were radiolabelled during amplification by PCR using 5¢ and 3¢ primers in the presence of [ 32 P]dATP[aP]. This product was used for the next round of selection follow- ing the same protocol. After seven cycles of selection, the retarded DNA band of the final selection was cut off, puri- fied and then cloned into the pUC119 plasmid (New England Biolabs). Plasmid DNAs from  40–50 independent insert- containing colonies were prepared and the insert fragments were sequenced. At least 25 of the resulting quality sequences containing the randomized 10 bp oligonucleotides were aligned computationally using clustal x [20]. The frequency of each nucleotide appearing in the aligned position of the selected sequences was calculated, leading to the establish- ment of the selected binding site. Phylogenetic analysis The amino acid sequences of all AtERFs were downloaded from the Database of Arabidopsis Transcription Factors (DATF) (http://datf.cbi.pku.edu.cn) [21]. The sequences of all ERF domains were extracted in bulk by a manual pro- gram using Perl script. The sequence alignment was gener- ated using clustal x: Gap at 10; Gap Extension at 0.2; Delay Divergent Sequence at 10%; Negative Matrix Off and Protein Weight Matrix of BLOSUM Series [20]. Acknowledgements The experiments were carried out at the National Insti- tute of Advanced Industrial Science and Technology, Japan. DH was a recipient of a fellowship from the former Agency of Industrial Science and Technology, MITI, Japan, and of an STA fellowship from the Science and Technology Agency of Japan. This study was also supported partially by a grant issued by the National Natural Science Foundation of China (grant no. 30470159 ⁄ C01020304) and the National High- Technology Research and Development Program (‘863’ Program) of China (grant no. 2007AA10Z110). References 1 Okamuro JK, Caster B, Villarroel R, Van Montagu M & Jofuku KD (1997) The AP2 domain of APETALA2 defines a large new family of DNA binding proteins in Arabidopsis. Proc Natl Acad Sci USA 94, 7076–7081. 2 Riechman JL & Meyerowitz EM (1998) The AP2 ⁄ ERE- BP family of plant transcription factors. Biol Chem 279, 633–646. S. Yang et al. Arabidopsis ERFs recognize a common CG step core FEBS Journal 276 (2009) 7177–7186 ª 2009 The Authors Journal compilation ª 2009 FEBS 7185 3 Ohme-Takagi M & Shinshi H (1995) Ethylene-inducible DNA binding proteins that interact with an ethylene- responsive element. Plant Cell 7, 173–182. 4 The Arabidopsis Genome Initiative (2000) Analysis of the genome sequence of the flowering plant Arabidop- sis thaliana. Nature 408, 796–815. 5 Sakuma Y, Liu Q, Dubouzet JG, Abe H, Shinozaki K & Yamaguchi-Shinozaki K (2002) DNA-binding speci- ficity of the ERF ⁄ AP2 domain of Arabidopsis DREBs, transcription factors involved in dehydration and cold-inducible gene expression. Biochem Biophys Res Commun 290, 998–1009. 6 Nakano T, Suzuki K, Fujimura T & Shinshi H (2006) Genome-wide analysis of the ERF gene family in Arabidopsis and rice. Plant Physiol 140, 411–432. 7 Allen MD, Yamasaki K, Ohme-Takagi M, Tateno M & Suzuki M (1998) A novel mode of DNA recognition by a beta-sheet revealed by the solution structure of the GCC-box binding domain in complex with DNA. EMBO J 17, 5484–5496. 8 Hao D, Ohme-Takagi M & Sarai A (1998) Unique mode of GCC box recognition by the DNA-binding domain of ethylene-responsive element-binding factor (ERF domain) in plant. J Biol Chem 273, 26857– 26861. 9 Hao D, Yamasaki K, Sarai A & Ohme-Takagi M (2002) Determinants in the sequence specific binding of two plant transcription factors, CBF1 and NtERF2, to the DRE and GCC motifs. Biochemistry 41, 4202–4208. 10 Buttner M & Singh KB (1997) Arabidopsis thaliana ethylene-responsive element binding protein (AtEBP), an ethylene-inducible, GCC box DNA-binding protein interacts with an ocs element binding protein. Proc Natl Acad Sci USA 94, 5961–5966. 11 Zhou J, Tang X & Martin GB (1997) The Pto kinase conferring resistance to tomato bacterial speck disease interacts with proteins that bind a cis-element of pathogenesis-related genes. EMBO J 16, 3207–3218. 12 Fujimoto SY, Ohta M, Usui A, Shinshi H & Ohme-Takagi M (2000) Arabidopsis ethylene-responsive element binding factors act as transcriptional activators or repressors of GCC box-mediated gene expression. Plant Cell 12, 393–404. 13 Yamaguchi-Shinozaki K & Shinozaki K (1994) A novel cis-acting element in an Arabidopsis gene is involved in responsiveness to drought, low-temperature, or high-salt stress. Plant Cell 6, 251–264. 14 Baker SS, Wilhelm KS & Thomashow MF (1994) The 59-region of Arabidopsis thaliana cor15a has cis-acting elements that confer cold-, drought- and ABA-regulated gene expression. Plant Mol Biol 24, 701–713. 15 Prabakaran P, An J, Gromiha M, Selvaraj S, Uedaira H, Kono H & Sarai A (2001) Thermodynamic database for protein-nucleic acid interactions (ProNIT). Bioinfor- matics 17, 1027–1034. 16 Triezenberg SJ, Kingsbury RC & McKnight SL (1988) Functional dissection of VP16, the trans-activator of herpes simplex virus immediate early gene expression. Genes Dev 2, 718–729. 17 Liu Y, Zhao TJ, Liu JM, Liu WQ, Liu Q, Yan YB & Zhou HM (2006) The conserved Ala37 in the ERF ⁄ AP2 domain is essential for binding with the DRE element and the GCC box. FEBS Lett 580, 1303–1308. 18 Gill SC & von HippelPH (1989) Calculation of protein extinction coefficients from amino acid sequence data. Anal Biochem 182, 319–326. 19 Liu Q, Kasuga M, Sakuma Y, Abe H, Setsuko M, Yamaguchi-Shinozaki K & Shinozaki K (1998) Two transcription factors, DREB1 and DREB2, with an EREBP ⁄ AP2 DNA binding domain separate two cellular signal transduction pathways in drought- and low-temperature-responsive gene expression, respectively, in Arabidopsis. Plant Cell 10, 1391–1406. 20 Thompson JD, Gibson TJ, Plewniak F, Leanmougin F & Higgins DG (1997) The CLUSTAL_X windows interface: flexible strategies for multiple sequence align- ment aided by quality analysis tools. Nucleic Acids Res 25, 4876–4882. 21 Guo A, He K, Liu D, Bai S, Gu X, Wei L & Luo J (2005) DATF: a database of Arabidopsis transcription factors. Bioinformatics 21 , 2568–2569. Arabidopsis ERFs recognize a common CG step core S. Yang et al. 7186 FEBS Journal 276 (2009) 7177–7186 ª 2009 The Authors Journal compilation ª 2009 FEBS . Four divergent Arabidopsis ethylene-responsive element-binding factor domains bind to a target DNA motif with a universal CG step core recognition and different. the DNA- binding site A 60 bp single-stranded DNA RDM10, with 10 random- ized oligonucleotides in the center, i.e. CTGTCAGTGAT GCATATGAACGAATN 10 AATCAACGACATTAGGATC CTTAGC