Arabidopsis ethylene-responsive element binding factors act as transcriptional activators or repressors of GCC box-mediated gene expression

Abstract

Ethylene-responsive element binding factors (ERFs) are members of a novel family of transcription factors that are specific to plants. A highly conserved DNA binding domain known as the ERF domain is the unique feature of this protein family. To characterize in detail this family of transcription factors, we isolated Arabidopsis cDNAs encoding five different ERF proteins (AtERF1 to AtERF5) and analyzed their structure, DNA binding preference, transactivation ability, and mRNA expression profiles. The isolated AtERFs were placed into three classes based on amino acid identity within the ERF domain, although all five displayed GCC box-specific binding activity. AtERF1, AtERF2, and AtERF5 functioned as activators of GCC box-dependent transcription in Arabidopsis leaves. By contrast, AtERF3 and AtERF4 acted as repressors that downregulated not only basal transcription levels of a reporter gene but also the transactivation activity of other transcription factors. The AtERF genes were differentially regulated by ethylene and by abiotic stress conditions, such as wounding, cold, high salinity, or drought, via ETHYLENE-INSENSITIVE2 (EIN2)-dependent or -independent pathways. Cycloheximide, a protein synthesis inhibitor, also induced marked accumulation of AtERF mRNAs. Thus, we conclude that AtERFs are factors that respond to extracellular signals to modulate GCC box-mediated gene expression positively or negatively.

Figure 1.

Alignment of the ERF Domain from Various ERF Proteins. (A) The amino acid sequences of the ERF domain from various ERF proteins are aligned. They include AtERF1 to AtERF5 (this article); tobacco ERF1 to ERF4 (Ohme-Takagi and Shinshi, 1995); Pti4, Pti5, and Pti6 (Zhou et al., 1997); TINY (Wilson et al., 1996); CBF1 (Stockinger et al., 1997); DREB1 and DREB2 (Liu et al., 1998); Arabidopsis ERF1 (Solano et al., 1998); AtEBP (Büttner and Singh, 1997); and ABI4 (Finkelstein et al., 1998). The asterisks represent those proteins that were shown to bind to the GCC box. Filled box areas with dots indicate amino acid identities; dashes indicate gaps introduced to maximize alignment. Numbers at right indicate the amino acid position of the ERF domain in each protein. Amino acid residues identified by nuclear magnetic resonance analysis (Allen et al., 1998) to interact with nucleotides within the GCC box are underlined in the ERF domain consensus sequence. The structure of the ERF domain is shown below the sequence; the bar and black arrows indicate the β sheet and the β strands within the β sheet, respectively. The cross-hatched box indicates the α helix (Allen et al., 1998). (B) Comparison of the ERF domain consensus sequence with the AP2 domain. The consensus amino acid sequence of the ERF domain (ERF cons.) is compared with the D2 (AP2-D2) and D1 (AP2-D1) domains of APETALA2 (Jofuku et al., 1994). The amino acids with asterisks in the ERF consensus indicate residues that interact with nucleotides within the GCC box (Allen et al., 1998). Amino acids identical to the ERF consensus are boxed.

Figure 2.

AtERF cDNAs. A schematic representation of the AtERF1 to AtERF5 cDNAs is shown. The AtERF proteins are grouped into three classes according to sequence similarity within the ERF domain. Boxes indicate the open reading frames, starting from the first ATG codon, and lines show putative untranslated regions. Black boxes indicate the ERF domain; hatched boxes represent acidic domains. Black boxes with a black triangle above them represent putative nuclear localization signals, and plus signs (+) indicate putative MAP kinase target sites. Numbers above the line indicate positions of amino acid residues; numbers below the line refer to nucleotide positions.

Figure 3.

Comparison of Deduced Amino Acid Sequences of AtERF Proteins. Amino acid sequences grouped by class are aligned and compared with tobacco ERFs. Identity within the same class is indicated by shading. The filled bar below the sequences represents the ERF domain; AD indicates putative acidic domains. Carets represent putative nuclear localization signals. Plus signs (+) indicate putative MAP kinase target sites. Dashes indicate gaps used to optimize alignment. The GenBank, DDBJ, EMBL, and NCBI accession numbers of the nucleotide sequences of the AtERF cDNAs are AB008103 (AtERF1), AB008104 (AtERF2), AB008105 (AtERF3), AB008106 (AtERF4), and AB008107 (AtERF5).

Figure 4.

Characterization of AtERF DNA Binding Affinity to the GCC Box. (A) GCC box sequences used in this experiment. Underlining and boldface denote the GCC box. GCC, wild type; mGCC, a double mutant. (B) AtERF possesses GCC box–specific binding activity. Electrophoretic mobility shift assays were performed using MBP–AtERF fusion proteins and a 16-bp wild-type GCC box fragment or a mutated version. MBP was used alone as a control. Numbers indicate MBP–AtERF1 to MBP–AtERF5, respectively. (C) Sequence of the wild-type GCC box (W) and mutants (1 to 7) used for the DNA binding assays. The GCC box sequence, AGCCGCC, is underlined. Positions within the GCC box systematically substituted with a T residue are indicated as A-1, G-2, C-3, C-4, G-5, C-6, and C-7, respectively. Dashes indicate nucleotides identical to the wild-type sequence. (D) Effect of single-base substitutions within the GCC box on DNA binding activity of the AtERFs. The binding activities of each AtERF to the mutated versions relative to the wild-type GCC box are shown graphically. W and the number (1 to 7) indicate the probes shown in (C). (E) Autoradiographs of only the shifted bands from (D) are shown.

Figure 5.

AtERFs Can Transactivate or Repress GCC-Mediated Gene Expression in Arabidopsis Leaves. (A) Schematic diagram of the reporter and effector plasmids used in transient assays. A region of the Arabidopsis HLS1 gene, which contains the GCC box (filled circles), was fused to a minimal TATA box and a firefly LUC gene. Effector plasmids were under the control of the CaMV 35S promoter. Nos denotes the terminator signal of the gene for nopaline synthase. Ω indicates translational enhancer of tobacco mosaic virus. (B) Transactivation of the 4×HLS GCC box–LUC reporter gene by AtERF1, AtERF2, and AtERF5. (C) Transactivation of the reporter gene by AtERF1, AtERF2, and AtERF5 in the ethylene-insensitive mutant ein2. (D) Repression of reporter gene activity by AtERF3 and AtERF4 and suppression of AtERF5-mediated transactivation by AtERF3. (E) Dependence of repression on the amount of AtERF3 effector plasmid. Activation of the reporter gene by AtERF5 is reduced by AtERF3 in a dose-dependent manner. Different concentrations of AtERF3 effector were co-bombarded with AtERF5 effector (from a ratio of 1:0 to 1:1; AtERF3 to AtERF5) and the reporter plasmid. Values shown are averages of results from three independent experiments. Error bars indicate standard deviation. All LUC activities are expressed relative to the reporter construct alone (value set at 1).

Figure 6.

AtERF3 Suppresses Transactivation without Competing for the Same DNA Binding Site. (A) Schematic diagram of the reporter and effector plasmids used. The GAL4 binding site (patterned hatched boxes) and GCC box sequence (filled circles) were fused to a minimal TATA box and the LUC gene. The AtERF3 effector and VP16 effector plasmid containing the GAL4 DNA binding domain (GAL4DBD) fused upstream of the VP16 activation domain (VP16 AD) were under the control of the CaMV 35S promoter. Nos denotes the terminator signal of the gene for nopaline synthase. Ω indicates translational enhancer of tobacco mosaic virus. (B) AtERF3 represses VP16-mediated transactivation without competing for the same DNA binding site. Values shown are averages of results from three independent experiments. Error bars indicate standard deviation. All LUC activities are expressed relative to the reporter construct alone (value set at 1).

Figure 7.

Induction of AtERF mRNA Expression. (A) Expression pattern of AtERF mRNAs after exposure to ethylene and abiotic stress in the wild type (wild) or an ethylene-insensitive mutant, ein2 (ein2). Total RNA was prepared from Arabidopsis leaves treated with ethylene gas for 0, 1, 6, and 12 hr, respectively, or treated with cold (Co), heat (H), NaCl (Na), drought (D), abscisic acid (A), CHX (Cx), or wounded (W) for 6 hr. C, the control for CHX treatment; EtBr, ethidium bromide staining. (B) Expression patterns of the genes encoding PDF1.2 (top) and chitinase (bottom) in wild-type (Wild) and ein2 (ein2) Arabidopsis plants treated with ethylene for 0, 1, 6, and 12 hr. Ten micrograms of RNA was loaded and hybridized with the cDNA encoding AtERF, chitinase (Chen and Bleecker, 1995), or PDF1.2 (Penninckx et al., 1996).

Figure 8.

A Model for GCC Box–Mediated Stress Signal—Dependent Transcription by ERF Proteins in Arabidopsis. After reception of a stress signal by plants, ERF genes may be upregulated via an ethylene-dependent pathway or an ethylene-independent pathway. ERF genes such as ERF1 (Solano et al., 1998) are induced by ethylene and can interact with GCC box–containing stress response genes. The AtERFs are able to interact with the GCC box sequence in an ethylene-independent manner. AtERF1, AtERF2, or AtERF5 may activate a specific subset of GCC box–containing genes, whereas AtERF3 and AtERF4 may repress the expression of these genes. These data suggest that the ERF proteins may act as factors responsive to extracellular signals and that they are involved in regulating a subset of GCC box–containing stress response genes.