The rapeseed cultivar (B. napus. L. cv. Huyou15) was used as experiment material in this study. Plants were grown in pots containing soil, vermiculite, and peat moss (1:1:1 by volume) in a growth chamber at 22°C for 20 d under a 16 h day/8 h night cycle. The BnaERF-B3-hy15 gene encoding the AP2/ERF protein was cloned from B. napus L. Huyou15 as previously described, and three mutants were obtained by Jin et al. (2010a). Two-month-old plants of A. thaliana (L.) Heynh. (ecotype columbia) were used to clone AtCBF genes.

The details of the site-directed mutagenesis of the BnaERF-B3-hy15 and CBF genes to generate mutant genes were accorded to the overlap extension PCR strategy (Xiong et al., 2004, 2006; Peng et al., 2006). The strategy for site-directed mutagenesis is shown in Figure 1. Briefly, the upstream product (Product P1) and downstream product (Product P2) were amplified by first-round PCR with two pairs of oligonucleotides, respectively. Then, the full-length mutant fragment was amplified using a second round PCR with outer pair primers. All the mutagenesis products were ligated to the pMD-19 simple-T vector (TaKaRa, Dalian, China) for DNA sequencing identification and enzyme digestion. All the primers used to clone or synthesize mutations were listed in Table S1.

The amplification mutagenesis products were inserted between the Bam HI and Sac I sites of the yeast expression vector pPC87, which was reconstructed from pPC86. The vector of pPC87 hosts the Trp synthesize gene and contains a galactose-inducible (GAL4) protein activating the domain under the control of the yeast alcohol dehydrogenase (ADH1) promoter. A 75-bp fragment containing the GCC element (5′-ACCCTC GAGCGGATAACAATTTCACACAGGGGCGGCTCTTAGGCGGCTCTTATAAGAGCCGCCGGAT CCGGGCCC-3′, GenBank No. AF394909) was inserted into the vector G222, which could synthesize Ura; the vector also contained the reporter gene LacZ under the control of the iso-1-cytochrome C (CYC1) minimal promoter. This recombinant plasmid was used as a bait plasmid in the yeast one-hybrid system (Zhang Y. et al., 2009; Jin et al., 2010a,b).

The bait plasmid was transformed into the yeast strain EGY48. Subsequently, the pPC87 recombinant plasmids were transferred into yeast cells, which could grow on the medium without Ura. The transformants were cultured on SD media without Ura and Trp at 30°C for 3 d. A colony-lift filter assay was used for qualitative analysis of trans-activation activity. After the fusion protein interactions with the GCC element, the lacZ gene was induced and positive cells (blue) were detected through enzymatic staining with X-gal as the substrate. The relative β-galactosidase activity of the transformant colonies were examined with the o-nitrophenyl-β-d-galactopyranoside (ONPG) (Clontech Laboratories, Inc.).

In our previous study, we cloned a member of the ERF family gene, named BnaERF-B3-hy15; the DNA shuffling strategy was used to shuffle the BnaERF-B3-hy15 gene, and three mutants (Mu1, Mu2, and Mu3) of this gene were obtained from variant mutants. The four proteins (BnaERF-B3, Mu1, Mu2, and Mu3) were investigated using yeast one-hybrid assays, and the binding activity to the GCC box relatively varied (Jin et al., 2010a).

The 3D structure of the proteins showed that the AP2-domain consisted of three β-sheets and one α-helix, which functions in DNA binding. Nevertheless, whether the mutant site plays a key role in the binding activity remains unclear. Here, further sequencing analysis of these four genes showed that a total of 26 nucleic acid sites were altered (Figure 2, Table 1), which led to 12 amino acid site substitutions. A subsequent comparison of the amino acid sequences of four proteins showed that these proteins were highly homologous, only one different amino acid (Gly156) was found in the AP2-domain and α-helix, the remaining mutants were located in the periphery of the domain, and one amino acid (site192) differed between Mu2 (Val192) and Mu3 (Ile192) (Figure S1).

To identify the amino acid substitution site that may be essential for the binding activity of the mutants, a series of site-directed mutants of BnaERF-B3 were generated according to the sequence alignment results. The strategy for site-directed mutagenesis is shown in Figure 1. All the 11 mutants (Mu4 to Mu14) were confirmed through sequencing, then cloned into the expression vector pPC87, and transformed to yeast, which hosts another vector with the GCC box sequence. Results of the yeast one-hybrid screening showed that the clones of Mu10, where Gly156 was substituted with Arg, appeared dark blue on X-gal-containing plates, but the 10 other mutants (Mu4 to Mu9 and Mu11 to Mu14) did not appear obviously blue (Figure 3). The β-galactosidase activity assay was used to calculate relative activity. As shown in Figure 4, Mu10 showed a 13-fold increase over BnaERF-B3. The differences in the trans-activation activity of BnaERF-B3 and the mutants showed that the mutation of Gly156 located in the AP2-domain significantly increased the binding activity. However, the other mutations outside the AP2-domain may have minimal or no influence on the binding activity. Our results indicated that the amino acid residues at site 156 may have an important role in binding to the GCC box.

Only one amino acid differed between Mu2 and Mu3, but the binding activity of these mutants varied. We further designed two sets of site-directed mutagenesis (Thr192Val and Thr192Ile) based on Mu10, which hosts Gly156Arg. Yeast one-hybrid assay and β-galactosidase activity were also used to test the activity of Mu15 (Gly156Arg + Thr192Val) and Mu16 (Gly156Arg + Thr192Ile; Figures 3, 4). Although Mu16 showed a slightly higher increase in activity than Mu15, both mutants exhibited higher activity than BnaERF-B3. This result implied that the mutation of T192 may minimally affect the activity, whereas the mutation of Arg156 played a dominant role.

The mutation at site 156 from Gly to Arg greatly increased the activity, but we cannot determine whether the activity will be constant even if the 156th amino acid was mutated to the 18 other kinds of amino acid. To investigate the other amino acid residues that affect the binding activity, we replaced the 156th amino acid with the other 18 remaining amino acids into BraERF-B3 and the corresponding mutants were named Mu17 to Mu34.

Yeast one-hybrid analysis results showed that the transformant cells of various mutants had different activity, in which some clones appeared colorless, weakly stained blue, or stained deep blue (Figure 5). The results of quantitative analysis of β-galactosidase activity showed that Mu30 (Gly156Gln) had higher binding ability, followed by Mu33 (Gly156Lys), Mu34 (Gly156His), and Mu19 (Gly156Leu) (Figure 4). However, the activity of all the mutants was lower than that of Mu10 (Gly156Arg). These results illustrated that the space conformation of different amino acids caused conformational changes in the active domain, and the transcription factor showed different levels of activity. Therefore, the 156th amino acid is essential to maintain protein activity. Arg, Gln, Lys, His, and Leu are candidate amino acids for promoting BraEFR-B3 binding to the GCC box; among which, Arg exhibits the highest potential.

Numerous studies have proved that the structure of amino acids considerably affects the protein structure and conformation. Studies on ERF proteins also showed that sequence diversity in the ERF domain can indirectly affect the binding activity to the GCC box (Ohta et al., 2001; Jin et al., 2010a). Among the 12 amino acid site substitutions, only Gly156Arg is located in the AP2-domain. The mutant hosting Gly156Arg (Mu10) showed the highest binding activity to the GCC box, followed by Gly156Gln (Mu30), Gly156Lys (Mu33), Gly156His (Mu34), and Gly156Leu (Mu19). However, the mutants Gly156Thr (Mu26), Gly156Asp (Mu31), and Gly156Glu (Mu32) showed no detectable binding activity to the GCC box. The 3D structures of the mutant and wild-type BnaERF-B3 were derived using the Swiss–Model server, with AtERF1 as the template (Figure 6). Comparison of the mutated amino acid residue sequences and the structure of BnaERF-B3 indicated that the 3D structures have no visible differences in the three β-sheets. However, an obvious change in the α-helix structure occurred when the amino acid residues were substituted. In the mutants which had high binding activity, more N atom (blue ball) were be found in models. The different amino acids in site 156th represented different conformation, which may lead to activity change.

Previous studies have showed that some residues in AP2-domain of ERFs were indispensable for binding with GCC box (Jofuku et al., 1994; Allen et al., 1998; Cao et al., 2001; Hao et al., 2002; Liu et al., 2006). Sequence alignment revealed that these residues were not found on the BnaERF-B3. The mutant Mu3 of BnaERF-B3 showed highest binding activity to GCC box, it is reasonable to deduce that Arg156 is responsible for the binding activity of GCC box. To further validate the residues in the AP2-domain regions that might be responsible for the activity of ERFs to activate transcription, a series of site-directed mutants (Mu35–Mu42) were further constructed according to the different sites of sequence alignment of Mu3 and other ERFs, which reported by other researchers (Jofuku et al., 1994; Allen et al., 1998; Cao et al., 2001; Hao et al., 2002; Liu et al., 2006). All the mutants were transformed into yeast cells and the interaction with GCC box were quantified with X-gal assay.

Figure 7 showed that the mutants still showed weak or strong binding activity to GCC box, except Mu39. The synthetic mutant Mu43, which contained all the mutation sites of Mu35 to Mu42, exhibited a higher binding activity. On the basis of Gly156 was replaced by Arg, the other amino acid sites have little effect on the binding with GCC box. β-galactosidase activity assay also showed a consistent result (Figure 8). Mu36 (Trp116Ser) had the highest binding activity, followed by Mu43, Mu35 (Pro115Asn). In tobacco, the substitution of Pro9 or Trp10 with Asn or Ser in AP2-domain led to increase of the binding activity of NtERF2-F (Hao et al., 2002). Together, these results strongly suggested that the binding specificity with cis element can be altered by amino acid exchange mutation, and also demonstrated that Arg156 is an important determinant of the Bna-ERF-B3 binding specificity.

As the members of AP2/ERF family, ERFs and DREBs have high similarity on amino acid sequences. Cao et al. (2001) and Hao et al. (2002) have reported that Val14 and Glu19 residues were crucial in the regulation of the binding activity of DREB1A to the DRE cis element, while Ala14 was an important determinant of the ERF binding specificity. Three DREB proteins (CBF1, CBF2, and CBF3) in Arabidopsis have been identified to bind with DRE element (Liu et al., 1998). Amino acid sequence analysis of BnaERF-B3 and these three DREB proteins indicated that the AP2-domains were also highly homologous. We found that in the AP2-domains of CBFs of Arabidopsis, the corresponding amino acids of Gly156 in BnaERF-B3 were all Arg. To explore whether Gly can affect the binding activity of DREB proteins, three mutants MuCBF1 (Arg95Gly), MuCBF2 (Arg98Gly), and MuCBF3 (Arg98Gly) of Arabidopsis were obtained. Yeast one-hybrid (Figure 9A) and β-galactosidase activity (Figure 9B) showed that these three mutants also had the strong binding ability with DRE element. This result showed Gly did not affect the binding affinity of DREBs to DRE cis element.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.