APETALA2/Ethylene Responsive Factor (AP2/ERF) transcription factors: mediators of stress responses and developmental programs



  • Transcription factors belonging to the APETALA2/Ethylene Responsive Factor (AP2/ERF) family are conservatively widespread in the plant kingdom. These regulatory proteins are involved in the control of primary and secondary metabolism, growth and developmental programs, as well as responses to environmental stimuli. Due to their plasticity and to the specificity of individual members of this family, AP2/ERF transcription factors represent valuable targets for genetic engineering and breeding of crops. In this review, we integrate the evidence collected from functional and structural studies to describe their different mechanisms of action and the regulatory pathways that affect their activity.


During evolution, DNA binding proteins colonized eukaryotic genomes and generated new gene families of transcription factors (TFs) (Yamasaki et al., 2012). The APETALA2/Ethylene Responsive Factor (AP2/ERF) superfamily represents a paradigm of these events in plants. Initially defined as plant specific, the ERF/AP2 domain was also found in proteins of protists, cyanobacteria and phagi (Wessler, 2005). Conservation in the ability of these proteins to bind DNA was also observed (Magnani et al., 2004) although, in nonplant organisms, AP2 domains have been found associated to the His- and Asn-rich HNH class of homing endonucleases (Wessler, 2005). Among the various hypotheses of gene exchange between plant and nonplant organisms that have been formulated in the past, the most plausible scenario is represented by the lateral transfer of AP2-containing proteins from bacteria or viruses to plants. The presence of tandem repetitions of ERF-factors in specific chromosomal regions in woody species led to speculatation over whether some ERF retained potential as transposable elements (Licausi et al., 2010a).

Classification and structural features of AP2/ERF proteins

Historically, AP2/ERF proteins containing at least one DNA binding domain – named the AP2 domain – have been divided into three separate families, namely the ERF, AP2 and RAV families (Fig. 1). Most proteins with a single AP2 domain and whose genomic sequence contains a small amount of introns are assigned to the ERF family (Nakano et al., 2006). The AP2 family consists of members characterized by a tandem repetition of two AP2 domains and a small number of proteins with a single AP2 domain that shows higher similarity to the one contained in double-AP2 proteins than to the AP2 domain of the ERF proteins' (Fig. 1). The AP2 family was further subdivided into the AP2 and ANT groups according to the amino acid sequence of the double AP2 domain and the nuclear localization sequence (Fig. 1) (Shigyo & Ito, 2004). A third class of proteins possesses an ERF domain associated with a B3 DNA-binding domain. They constitute the RAV family (Swaminathan et al., 2008) (Fig. 1). Conserved in all the plant genomes sequenced so far is an additional sequence belonging to the AP2/ERF family: Soloist (Nakano et al., 2006; Zhuang et al., 2008; Licausi et al., 2010a). Although this protein contains a single AP2 domain, its sequence and gene structure strongly diverge from those of the ERF transcription factors.

Figure 1.

Structure of the APETALA2/Ethylene Responsive Factor (AP2/ERF) superfamily. The AP2 superfamily is composed of single-AP2 domain proteins (ERF family), single or double ERF domain proteins (AP2 family), proteins containing one AP2 domain plus a B3 DNA binding domain (RAV family). Soloist, an ERF-related protein that appears in single copy in most of the plant genomes studied so far. ERF members have been subdivided into groups (Nakano et al., 2006) or subfamilies (Sakuma et al., 2002).

Although the original acronym ERF has been maintained, responsiveness to the growth regulator ethylene is not a universal feature of this protein superfamily, and the conserved DNA-binding element is not directly affected by ethylene signaling. The nomenclature of the ERF transcription factors is based on two major efforts of organizing these proteins into phylogenetic groups which allow easy functional recognition (Sakuma et al., 2002; Nakano et al., 2006). In 2002, Sakuma et al. subdivided 121 ERF proteins from Arabidopsis thaliana according to the similarity of their AP2 domain into two main groups, the Dehydration-Responsive Element Binding-proteins (DREBs) and the ERFs, each further separable into six subgroups. Four years later, Nakano et al. (2006) exploited the increase in genome annotation for rice (Oryza sativa) and Arabidopsis to refine Sakuma's classification taking into account the intron–exon structure of the ERF genes and the occurrence of additional motifs. In this way, the ERF proteins were subdivided into twelve groups that substantially matched the classification provided by Sakuma et al. (2002). The advantages brought by the new classification of the ERF family by Nakano et al. (2006) were two-fold: it recapitulated the phylogenetic history of this class of transcription factors and additionally allowed proteins possessing similar regulatory features to be grouped. Indeed, in the following years, the function of several conserved motifs of the ERF subgroups has been defined. Nowadays, both nomenclatures are still used independently, but it would be advisable to integrate them in order to avoid ambiguities and facilitate memorization. In this review we follow the nomenclature suggested by Nakano et al. (2006). With regard to the remaining families of the AP2/ERF superfamily, we attempted a systematic nomenclature for their members following the structural features definied by previous authors (Shigyo & Ito, 2004) (Supporting Information Table S1).

The conserved DNA binding domain characteristic of the AP2/ERF superfamily is composed of c. 60 amino acid residues that confer a typical three-dimensional conformation organized into a layer of three antiparallel beta-sheets followed by a parallel alpha-helix (Allen et al., 1998). Within the ERF family, the AP2/ERF DNA binding domain has been distinguished between DREB and ERF domains depending on the identity of residues at specific positions (Sakuma et al., 2002). The differences in amino acid sequence reflect in the DNA affinity and specificity of the two subfamilies. Many DREB proteins have been shown to bind to an A/GCCGAC element, which is often associated with ABA, drought and cold responsive genes (Stockinger et al., 1997). Conversely, members of the ERF subfamily specifically bind in vitro an AGCCGCC element, named the GCC-box often found in the genomic regions upstream of genes that respond to ethylene, pathogens and wounding (Ohme-Takagi & Shinshi, 1995). Despite this generalization, single members of both families have been reported to bind DRE and GCC elements (Sun et al., 2008) or even novel DNA elements that diverge significantly from these two (Welsch et al., 2007; Shaikhali et al., 2008). A high diversification of DNA-binding affinities may be required to modulate the response to different stimuli according to stress-specific modules (Mizoi et al., 2012). Additionally, the wide variety of oligonucleotides being specifically bound by proteins that harbor similar DNA recognition domains can be explained by taking into account the plethora of interactions with transcription factors belonging to different protein families. Surprisingly, members of the AP2 family that possess a double AP2 domain do not recognize a tandem repetition of ERF- or DREB-binding elements (Dinh et al., 2012). The Arabidopsis AP2-13 (APETALA2) binds as monomer to a motif which includes a T/A rich element that is directly contacted by the second AP2 repeat (Dinh et al., 2012). Information is lacking about the DNA binding properties of AP2 proteins, although they interact in vivo with the promoter of flowering-regulatory genes (Mathieu et al., 2009). The AP2 domain in RAV1 binds a CAACA motif (Kagaya et al., 1999).

Activation and repression: protein motifs involved

Following a general rule, AP2-containing TFs can be roughly classified as activators or as repressors depending on whether they activate or suppress transcription of specific target genes. Transcriptional repressors are further classified as active or passive repressors. Active repressors contain a distinct repression domain (RD), which confers repressive activity to a TF or DNA binding domain; when bound to the promoter region, these proteins can thus actively prevent transcription of a target gene. By contrast, passive repressors do not possess an RD; rather, they suppress transcription by competing with transcriptional activators for binding of the target sequence.

In general, activation domains identified in plant transcription factors do not have distinct sequence motifs but tend to be rich in acidic amino acids. For example, acidic N-terminal and/or C-terminal regions of tobacco ERF2 and ERF4 act as activation domains in protoplasts (Ohta et al., 2000). Recently, the EDLL motif found in the AtERF98/TDR1 transcription factor was shown to be a strong activation domain and a useful tool to confer transcriptional activation potential to heterologous DNA-binding proteins (Tiwari et al., 2012).

In contrast to the activation domain found in plant transcriptional activators, plant active repressors usually contain a distinct RD. Three RDs, namely the ERF-associated amphiphilic repression (EAR: LxLxLx or DLNxxP) motif (Ohta et al., 2001; Hiratsu et al., 2003), the TLLLFR motif (Matsui et al., 2008) and the B3 repression domain (BRD: RLFGV) (Ikeda & Ohme-Takagi, 2009), are plant-specific. Among 147 AP2/ERF proteins in Arabidopsis, 23 contain either an EAR (or EAR-like) motif or a BRD (Nakano et al., 2006) (Table 1). ERFs that contain an EAR- or BRD-motif appear to be transcriptional repressors and suppress the transcription of target genes by interacting with the transcriptional corepressors TOPLESS (TPL) and TOPLESS-RELATED (TPR) (Causier et al., 2012). The RDs in these ERF proteins are often conserved in their orthologs in other plant species (Table 1). In the AP2 family, the Arabidopsis proteins AP2-13 (AP2), AP2-14 (TARGET OF EAT 1, TOE1), AP2-8 (ANT-LIKE 6, AIL6) and AP2-12 (ANT-LIKE7, AIL7) possess EAR-like motifs. Although AP2-13 is a positive regulator of floral development, it acts as negative regulator of the AGAMOUS gene (Drews et al., 1991).

Table 1. APETALA2/Ethylene Responsive Factor (AP2/ERF) proteins that contain a repression domain
LocusCommon nameRD motifOrthologueSpecificityReference
  1. Os, Oryza sativa; Po, Populus trichocarpa; Ph, Physcomitrella patens; Se, Selaginella moellendorffii.

AT5G60120TOE2KLDLNLOs, PoAngiospermKagale et al. (2010)
AT2G39250SNZDLELSLPoEudicotKagale et al. (2010)
AT3G54990SMZDLELSLOs, PoAngiospermKagale et al. (2010)
AT2G28550TOE1, RAP2.7MLDLNLOs, PoAngiospermKagale et al. (2010)
AT5G65510PLT7, AIL7KLKLAL ArabidopsisKagale et al. (2010)
AT1G13260EDF4, RAV1RLFGVOs, PoAngiospermIkeda & Ohme-Takagi (2009)
AT1G53170ERF8DLDLNLOs, Po, PhPlantOhta et al. (2001)
AT5G10510PLT3, AIL6RLKLSLOs, Po, PhPlantKagale et al. (2010)
AT3G15210ERF4LDLDLNLP ArabidopsisOhta et al. (2001)
AT1G28370ERF11LDLDLNFPOs, PoAngiospermOhta et al. (2001)
AT3G20310ERF7FDLNFPPOs, Po, PhPlantOhta et al. (2001)
AT1G50640ERF3FDLNFPPOs, Po, PhPlantOhta et al. (2001)
AT1G28360ERF12LDLNHLPOs, Po, PhPlantOhta et al. (2001)
AT1G03800ERF10LDLNASPOs, PoAngiospermOhta et al. (2001)
AT3G25730EDF3RLFGVOs, PoAngiospermKagale et al. (2010)
AT4G36920AP2NLDLSLOs, PoAngiospermKagale et al. (2010)
AT5G67180TOE3NLGLSLOs, PoAngiospermKagale et al. (2010)
AT5G44210ERF9ELNLDLOs, Po, PhPlantOhta et al. (2001)
AT1G25560EDF1, TEM1RLFGVOs, PoAngiospermIkeda et al. (2009)
AT1G68840EDF2, RAV2RLFGVOs, PoAngiospermIkeda et al. (2009)

Post-transcriptional regulation of ERF transcription factors

Post-transcriptional control of ERF protein activation is emerging as a key feature of several physiological processes in plants (Fig. 2). Alternative splicing has been reported to play a major role in the fast accumulation of DREB2-like sequences in grass species. Barley (DRF1), wheat (WDREB2) and maize (DREB2A) orthologs show accumulation of an mRNA isoform characterized by a STOP-codon before the DNA binding domain under nonstress conditions, thus producing a nonfunctional protein (Xue & Loveridge, 2004; Egawa et al., 2006; Qin et al., 2007). When stress stimuli occur, alternative splicing takes place excluding the exon that contains the premature stop codon, rapidly generating a functional isoform. Alternative splicing has also been reported for members of the ERF-VII group in Arabidopsis and tomato (Solanum lycopersicum) (Pirrello et al., 2006; Licausi et al., 2010b). In these cases, the removal of the first exon would cause the release of protein-destabilizing elements, although the existence of the corresponding stabilized proteins has still to be demonstrated.

Figure 2.

Regulatory mechanism affecting Ethylene Responsive Factor (ERF) protein activity, localization and abundance. The transcription of ERF genes can be regulated in response to biotic and abiotic stresses or in the frame of growth and development programs. Hormones are often mediators of this regulatory step. ERF mRNAs can be alternatively spliced to produce functional or nonfunctional proteins. Once translated, ERF transcription factors can be controlled by protein modifications that enhance or hamper their activity, restrict their localization or promote their stability/degradation. An example for each regulatory process is provided in parentheses.

Regulation of the abundance and activity of ERF transcription factors takes place also at the protein level. Nakano et al. (2006) identified several motifs related to putative phosphorylation sites in ERF proteins belonging to groups VI, VII and IXb in Arabidopsis and rice. For some ERF proteins the regulation imposed by phosphorylation has also been characterized. The ERF-VI cytokinin responsive factors (CRFs) were suggested to require phosphorylation mediated by the protein histidine kinases and histidine-containing phosphotransfer proteins to translocate from the cytosol into the nucleus (Rashotte et al., 2006). In Arabidopsis, fifteen members of the ERF family have been shown to act as substrates of mitogen activated protein kinases (MPKs) (Popescu et al., 2009). In particular, AtERF104 (ERF-IXb group), a positive regulator of the pathogen response in Arabidopsis, is stabilized by its interaction with MPK6 (Bethke et al., 2009). Several pieces of evidence point at phosphorylation acting as negative regulator of the transcriptional activity of DREB2A homologs (Sakuma et al., 2006a; Agarwal et al., 2007). Following an opposite trend, phosphorylation appears to improve the transcriptional activity of ERF-proteins in tomato, rice and tobacco (Nicotiana tabacum) (Gu et al., 2000; Cheong et al., 2003; De Boer et al., 2011).

The stability of ERF proteins is also regulated by several pathways that involve the 26S proteasome. Once more, the ERF-VII group provides examples of the involvement of ubiquitin ligases: in Arabidopsis, RAP2.2 interacts with the RING-finger SEVEN IN ABSENTIA OF ARABIDOPSIS 2 (SINAT2) and the whole group is subjected to an oxygen-dependent degradation mediated by the N-end rule pathway that involves PROTEOLYSIS6 (PRT6) (Welsch et al., 2007; Gibbs et al., 2011; Licausi et al., 2011). Another example of proteasomal regulation of AP2/ERF proteins comes from the DREB family. In fact, the abundance of the Arabidopsis DREB2A is restrained by the RING-E3 ligases DRIP1 and DRIP2 under nonstress conditions (Qin et al., 2007). By a similar mechanism the drought responsive Arabidopsis ERF53 is degraded by the proteasome when it is ubiquitinated by RGL1 and RGL2 (Cheng et al., 2012). In some cases the interaction between ERF proteins and the ubiquitin ligase is not direct but rather mediated by an adaptor. This applies to the Arabidopsis RAP2.4a, which is targeted to the 26S proteasome through the interaction with the MATH domain of BTP/POZ proteins (Weber & Hellmann, 2009).

Other post-transcriptional modifications such as acetylation, nitrosylation, glutathionylation and sumoylation are likely to regulate the activity or abundance of ERF transcription factors. However, so far the techniques applied to identify proteins carrying these modifications have not been sensitive enough to capture scarce proteins such as transcriptional regulators.

Protein–protein interactions involving ERF proteins

Apart from the temporary interaction with enzymes involved in post-transcriptional modification, ERF proteins can form stable complexes with other transcriptional regulators and structural proteins (Table 2). In most cases, these interactions determine the localization of ERF-proteins, their stability and abundance, their transcriptional activity and their target specificity. Acyl-coA binding proteins (ACBPs), for instance, have been shown to bind to ERF-VII transcription factors via an ankyrin-domain (Li et al., 2004). The Arabidopsis ACBP1 and ACBP2, in particular, are associated with the plasma membrane by means of a lipophilic N-terminal motif and they have been suggested to retain ERF-VII RAP2.12 at the membrane site, thus preventing it from entering the cytosol and the nucleus (Licausi et al., 2011). Because ERF-VII proteins are susceptible to fast proteasomal degradation following the N-end rule, ACBPs ensure the maintenance of a reservoir of RAP2.12 and potentially other ERF-VII proteins that allow a fast activation of their targets when required by the cell (Licausi et al., 2011). In a similar fashion, Arabidopsis ACBP4 has been shown to interact with RAP2.3 in the cytosol and at the periphery of the nucleus (Li et al., 2004). However, the dynamic of this interaction has not yet been described.

Table 2. Interaction partners of Arabidopsis ERF proteins
 Protein-protein interactions involving ERFs
Affecting localizationRAP2.12AT1G53910ACBP1AT5G53470Lipid metabolismLicausi et al. (2011)
RAP2.12AT1G53910ACBP2AT4G27780Lipid metabolismLicausi et al. (2011)
RAP2.3AT3G16770ACBP4AT3G05420Lipid metabolismLi et al. (2004)
RAP2.3AT3G16770ACBP2AT4G27780Lipid metabolismLi et al. (2004)
Affecting activityERF3AT1G50640SAP18AT2G45640Salt stress responseKagale & Rozwadowski (2011)
ERF3AT1G50640HDA19AT4G3813Histone deacetylationKagale & Rozwadowski (2011)
ERF7AT3G20310SIN3AT1G24190Response to ABASong et al. (2005)
RAP2.4aAT1G22190RAP2.4aAT1G22190Response to oxidative stressShaikhali et al. (2008)
Affecting target recognitionDRN/DRNLAT1G12980/AT1G24590PHAVOLUTAAT1G30490Growth and developmentChandler et al. (2007)
DRN/DRNLAT1G12980/AT1G24590PHABULOSAAT2G34710Growth and developmentChandler et al. (2007)
DRN/DRNLAT1G12980/AT1G24590REVOLUTAAT5G60690Growth and developmentChandler et al. (2007)
DRN/DRNLAT1G12980/AT1G24590HB8AT4G32880Growth and developmentChandler et al. (2007)
RAP2.3AT3G16770TGA4AT5G10030Plant defenseBüttner & Singh (1997)

Interaction partners can also affect the activity of ERF transcription factors. This is exemplified by the complex composed of ERF3, SAP18 and HDA19 in Arabidopsis. In this complex, SAP18 acts as a bridging factor that physically links the histone deacetylating enzyme HDA19 to the EAR of ERF3 (Kagale & Rozwadowski, 2011). This complex behaves as an active repressor by deacetylating histones located on ethylene response gene, thereby repressing their activation (Song & Galbraith, 2006). For other ERF repressors, such as ERF7, the connection to the deacetylating activity is provided by SIN3 (Song et al., 2005). Similarly, the transcriptional corepressors TPL and TPR are able to interact with the EAR and BRD-containing ERFs in Arabidopsis (Causier et al., 2012). Homodimerization has also been shown to have an effect on the activity of ERF transcription factors. For instance, the Arabidopsis RAP2.4a was shown to dimerize upon oxidation of specific cysteine residues in vitro and be able to bind, as a dimer, the promoter of 2-Cys peroxiredoxin-A. When oxidation proceeds in vivo, further RAP2.4a subunits have been proposed to associate with the dimer, thus causing the dissociation from the promoter (Shaikhali et al., 2008).

The target specificity of ERF transcription factor also strongly depends on interaction partners. This was observed both for developmental programs and stress responses. For instance Arabidopsis AP2-18 (DORNROSCHEN, DRN) and AP2-19 (DNR-LIKE, DRNL) interact with class III HD-ZIP transcriptional regulators PHAVOLUTA, PHABULOSA, REVOLUTA and ATHB8. These interactions involve the PAS-like domain of the HD-ZIP proteins and the AP2 domain of AP2-18 and AP2-19 (Chandler et al., 2007). It has been suggested that these higher order protein complexes may act as a transcriptional unit in the control of embryo patterning defining the targets in specific cell types. Also in the case of transcriptional responses to stress conditions ERF proteins interact to activate specific subsets of targets. This is the case for RAP2.3 and the bZIP transcription factor TGA4, whose interaction has been suggested to enhance or specify activation of genes whose promoter contains binding elements for both transcription factors (Büttner & Singh, 1997).

A number of proteins able to interact with ERF transcription factors have been identified, although knowledge is lacking about their biological function and their role in the ERF-mediated regulation of transcription. For instance, the Arabidopsis WWE and PARP-like domain containing proteins SRO1 and RCD1 have been suggested to be involved in regulating protein stability or activity of members of the DREB-group (Vainonen et al., 2012).

ERFs involved in developmental processes

In the AP2/ERF superfamily, AP2 TFs act primarily in the regulation of developmental programs, while ERF proteins mainly affect these processes in the frame of responses to environmental stimuli or hormones. Among the ERFs that regulate ethylene-dependent transcription, ERF members positively or negatively regulate the expression of ethylene inducible genes downstream of EIN3 (Solano et al., 1998; Yang et al., 2005). Some ERF-VI proteins mediate the transcriptional response to cytokinins and were therefore named Cytokinin Responsive Factors (CRFs) (Rashotte et al., 2006). The abscisic acid (ABA)-insensitive abi4 mutant is defective in a gene encoding an AP2/ERF protein (Finkelstein et al., 1998) and AP2-like ABA repressor 1 (ABR1) represses ABA-responsive gene expression in Arabidopsis (Pandey et al., 2005).

Several AP2/ERF protein genes affect plant morphology when expressed ectopically. For example, ectopic expression of TINYT, ERF1 or DREB1 results in a dwarf phenotype in Arabidopsis and expression of DREB2 retards growth (Wilson et al., 1996; Liu et al., 1998; Solano et al., 1998). These dwarf and retardation phenotypes may result from over-expression of defense- or stress-related genes that are activated by ectopically expressed ERFs.

In Arabidopsis, AP2-13 (AP2), AP2-05 (AINTEGUMENT, ANT) and AP2-09 (ANT-LIKE1) genes regulate floral growth and ovule development, respectively (Jofuku et al., 1994; Elliott et al., 1996; Klucher et al., 1996; Mizukami & Fischer, 2000; Krizek, 2009). FRIZZY PANICLE (FZP) is necessary to suppress formation of axillary meristems and to establish floral meristem identity in rice spikelets (Komatsu et al., 2003). The maize ANT-like AP2/ERF protein Glossy15 regulates leaf epidermal cell identity (Moose & Sisco, 1994). In rice, SNORKELs, the key regulators of internode elongation of deepwater rice, were identified to be the AP2/ERF transcription factors that trigger submergence-induced growth in response to ethylene (Hattori et al., 2009). Instead, OsEATB restricts internode elongation by downregulating a gibberellin biosynthetic gene (Qi et al., 2011). CROWN ROOTLESS is involved in crown root initiation in rice through the induction of a type-A response regulator of cytokinin signaling (Kitomi et al., 2011).

Several AP2/ERF proteins regulate the development of shoot meristems. For instance, ectopic expression of DRN induces cytokinin-independent shoot formation in Arabidopsis (Banno et al., 2001) while the Brassica napus ANT orthologue, BABY BOOM (BBM), causes the spontaneous formation of somatic embryos (Boutilier et al., 2002). The ANT-like proteins, AP2-06 (PLETHORA1, PLT1) and AP2-07 (PLT2), act as key effectors for establishment of the stem cell niche during embryonic pattern formation in Arabidopsis (Aida et al., 2004). Another interesting finding is that WOUND INDUCED DEDIFFERENTIATION 1 (WIND1) controls cell dedifferentiation in Arabidopsis (Iwase et al., 2011). During organ regeneration after wounding, adult somatic cells often dedifferentiate to reacquire cell proliferation potential. WIND1 promotes cell dedifferentiation and subsequent cell proliferation to form a mass of pluripotent cells (callus). Ectopic expression of WIND1 is sufficient to establish and maintain the dedifferentiated status of somatic cells without exogenous auxin and cytokinin; therefore, WIND1 appears to be involved in plant totipotency. Another AP2/ERF protein, RAP2.6, is essential for tissue reunion in wounded Arabidopsis stems (Asahina et al., 2011). In addition, AP2/ERF transcription factors determine stem cell identity in Physcomitrella patens (Aoyama et al., 2012). Four Physcomitrella proteins orthologous to Arabidopsis ANT, PLT and BBM were shown to be indispensable for the formation of gametophore apical cells from protonema cells.

ERFs involved in the regulation of primary and secondary metabolism

ERF proteins belonging to the group I and V have been shown to be especially involved in the regulation of genes coding for enzymes involved in the biosynthesis of lipids and cell wall components. The analysis of Arabidopsis plants over- or under-expressing SHINE (SHN)-like genes indicated that this clade of transcription factors in involved in the regulation of cutin-, suberin- and wax- related protein families in Arabidopsis (Shi et al., 2011). Interestingly, wax biosynthesis in Medicago truncatula is under the control of an ERF-V, WXP1 (Zhang et al., 2005). The ERF-VII proteins RAP2.12, HRE1 and HRE2 are involved in the activation of the fermentative pathway under oxygen limitations (Licausi et al., 2010b, 2011). These TFs were predicted to bind a canonical GCC-box and in vitro assays confirmed this possibility. However, the ERF-VII RAP2.2 was shown to bind an ATCTA element by south-western assay (Welsch et al., 2007) and indeed the promoters of anaerobiosis-related genes possess such element. The constitutive activation of fermentative pathways under aerobic conditions by expressing stabilized RAP2.12 caused aberrations in the phenotype of the transgenic plants, highlighting the importance of restricting their expression only when required (Licausi et al., 2011).

The involvement of ERF transcription factors in the regulation of pathways belonging to secondary metabolism derives mostly from studying plants producing compounds of utmost interest for pharmaceutical applications. A common feature of these secondary metabolites is jasmonate responsiveness. For instance, the ERF-IX Artemisia annua AaERF1 and AaERF2 regulate the expression of artemisin biosynthetic genes (Yu et al., 2012). Similarly, the jasmonate-responsive Catharantus roseus ORCA2 and ORCA3 control the expression of streptosidine synthase, required for the biosynthesis of terpenoid indole alkaloids (Van Der Fits & Memelink, 2001). Moreover in tobacco, a cluster of ERF-IX located at the NIC-2 locus was demonstrated to be required to induce the jasmonate-mediated activation of nicotine biosynthesis (De Boer et al., 2011). All these ERFs belong to the group IX (B3) suggesting that ERF IX ancestors were recruited early in the evolution of distinct plant lineages to regulate the synthesis of jasmonate-inducible metabolites (De Boer et al., 2011).

ERFs involved in biotic and abiotic stress responses

ERF proteins were originally isolated as transcription factors that bind to the promoter regions of stress-responsive genes. The ERF genes examined to date are induced by biotic and abiotic stresses, including pathogen infection, salt stress, osmotic stress, wounding, drought, hypoxia, temperature stress and the stress-related hormones ethylene, jasmonic acid and ABA.

A number of ERF genes confer tolerance to various biotic stresses when expressed ectopically in various plants and in many cases. For example, several ERFs activate the transcription of basic type defense-related genes, pathogenesis-related (PR) genes, osmotin, chitinase and β-1,3-glucanase. However, the set of target genes regulated by each ERF has not been completely elucidated. ERF1 and its homologs belonging to the ERF-IX group are probably the ERF transcription factors whose involvement in pathogen response in Arabidopsis thaliana has been most extensively characterized (Lorenzo et al., 2002; Zarei et al., 2011; Moffat et al., 2012). Also, several ERF proteins regulate ethylene biosynthesis (Zhang et al., 2009; Li et al., 2011). Tobacco OPBP1 enhances resistance to pathogens when expressed ectopically in transgenic rice (Chen & Guo, 2008). Arabidopsis RAP2.2 plays a role in plant resistance to Botrytis cinerea and ethylene responses (Zhao et al., 2012) and rice OsERF922 negatively regulates resistance to Magnaporthe oryzae (Liu et al., 2012).

The most intensely studied ERF in abiotic stress responses are the DREBs proteins. Members of the DREB1/CBFs subfamily are rapidly induced in response to cold stress and, when ectopically expressed, improve tolerance to freezing (Jaglo-Ottosen et al., 1998; Liu et al., 1998; Kasuga et al., 1999). By contrast, Arabidopsis RNAi lines for CBF1 or CBF3 have reduced freezing tolerance (Novillo et al., 2007). A number of cold-inducible genes are upregulated in Arabidopsis plants ectopically expressing DREB1/CBF members. The upregulated genes include those for late embryogenesis abundant (LEA) proteins and enzymes for sugar metabolism and fatty acid synthesis; these genes are considered to be important for survival at low temperatures, including genes involved in fatty acid desaturation (Fowler & Thomashow, 2002; Maruyama et al., 2004). Metabolomic analyses show that many of the metabolites that accumulate during cold stress also accumulated in Arabidopsis plants ectopically expressing DREB1/CBF (Cook et al., 2004; Maruyama et al., 2009). The fact that the quantitative trait locus for freezing tolerance is associated with CBF2 (Alonso-Blanco et al., 2005), indicates that the cold-inducible DREB1/CBF genes are likely to be major regulators of the response to cold stress in Arabidopsis. Additionally, ectopic expression of CBF2 has been shown to delay leaf senescence in Arabidopsis suggesting that CBFs promote cold-endurance by slowing down growth and postponing flowering in winter until temperatures increase in spring (Sharabi-Schwager et al., 2010).

The DREB2 subgroup has eight members in Arabidopsis and homologs in the genomes of many angiosperm species. Among the members of the DREB2 subfamily, Arabidopsis DREB2A and DREB2B are induced by dehydration, high salinity and heat in an ABA-independent manner (Liu et al., 1998; Nakashima et al., 2000; Sakuma et al., 2006a). The ectopic expression of a constitutively active form of DREB2A exhibits improved tolerance to drought, high salinity and heat stresses (Sakuma et al., 2006a,b). By contrast, mutants in DREB2A are more sensitive to heat shock (Sakuma et al., 2006a).

Many ERF-VII genes from Arabidopsis or rice have been shown to be involved in the response to submergence and hypoxia (Xu et al., 2006; Hattori et al., 2009; Hinz et al., 2010; Licausi et al., 2010b). Constitutive ERF-VII factors, such as RAP2.12 in Arabidopsis, have been suggested to act as primary triggers for the molecular response to oxygen deficiency (Gibbs et al., 2011; Licausi et al., 2011), which is then sustained by the hypoxia-inducible ERFs HRE1 and HRE2 (Licausi et al., 2010b). In submerged deepwater rice, SK1and SK2 stimulate internode elongation to effect clearance of the water level (Hattori et al., 2009), whereas Sub1A promotes a quiescent strategy that allows carbohydrate saving and improves tolerance after flash-flood (Xu et al., 2006).

A number of ERF genes from various plants have been shown to confer multiple stress tolerance when expressed ectopically (Yi et al., 2004; Seo et al., 2010; Fukao et al., 2011; Mito et al., 2011). This ‘unspecific’ effect can be explained by the activation of tolerance pathways that alleviate a generic stress status, such as oxidative bursts produced as consequences of the primary stresses. Alternatively, constitutive ERF expression could set the plant in a general alert state which fasten or magnify the response when a specific stress is applied.

The applicability of ERF genes in genetic engineering/breeding approaches in agriculture

The increase in world population, estimated to reach 9 billion by 2050, poses a serious challenge for crop production (Xu et al., 2011). It demands a concomitant increase in food production that is unlikely to be achieved only by improving agricultural practices. Yield is strongly affected by environmental cues such as water deficit or excess, high soil and water salinity, cold and drought stresses, and hence it is of the utmost importance to develop crop varieties able to withstand such adverse conditions but at the same time limiting yield decreases. Transcription factors represent ideal targets for traditional or genetic engineering-assisted breeding of plants with specific traits related to stress tolerance or higher yield. In particular, ERF genes are among the most interesting TFs because they have been selected through evolution to regulate a series of stress-response pathways. Records of successful attempts to improve crop tolerance via targeted approaches have been reported. For example, the ERF-VII Sub1a gene from wild Oryza sativa var. indica was introgressed in cultivated rice to obtain a flood-resistant variety (Xu et al., 2006). Also transgenic approaches to express foreign ERF genes or to overexpress endogenous ones have been attempted in different plant species with a focus on cereals and solanaceae (extensively reviewd by Xu et al., 2011). As observed for the ectopic expression of transcriptional regulators, ERF overexpression led to growth impairment (Pré et al., 2008), which would limit its biotechnological application. However, in several cases, the transgenic plants obtained did not show an altered phenotype as compared to the wild-type under nonstress conditions but exhibited better performances in terms of survival and yield when subjected to stress treatments (Oh et al., 2005, 2009; Qi et al., 2011). However, most of these reports are based on small-scale tests and controlled conditions whereas glasshouse or field trials would be required to confirm results. The appreciated feature of the substantial equality between transgenic and wild-type plants under nonstress condition is likely to derive from the requirement of further regulatory steps activated only under stress conditions. In this way, the transgenic plants do not waste resources when the stress is not present, but are constantly primed to respond rapidly and efficiently at the onset of the stress. When the constitutive expression of a transcription factor has a negative impact on yield, the use of host-specific stress-inducible promoters is preferable to constitutive sequences.

Perspectives in AP2/ERF studies

The importance of AP2 and ERF proteins in the regulation of physiological processes in plants has been already established by several authors. In general, following a common ancestral origins, AP2 and ERF genes seem to have diverged to orchestrate developmental programs and responses to environmental factors, respectively (Nakano et al., 2006). With the advancement of molecular techniques and increased sensitivity of proteomic assays, the increment or reduction of transcription factor activity and abundance can also be assessed (Kaufmann et al., 2010; Smaczniak et al., 2012). At the same time, the occurrence of post-transcriptional regulation of the ERF transcription factor is emerging as a crucial factor in the control of their activity (Qin et al., 2007; Bailey-Serres et al., 2012; Cheng et al., 2012). Systematic analyses of secondary modifications coupled to activity assays are likely to shed new light on this aspect. Moreover, the comparison of members of the same group or orthologs in different species will lead to the identification of the structural requirements that act as substrates of these regulatory cascades. Concomitantly, the identification of interaction partners of the ERF/AP2 proteins and the factors affecting these interactions will be required to understand the role of transcription factors in conferring protection against different stress conditions.