AP2/ERF Transcription Factor Regulatory Networks in Hormone and Abiotic Stress Responses in Arabidopsis

Dynamic environmental changes such as extreme temperature, water scarcity and high salinity affect plant growth, survival, and reproduction. Plants have evolved sophisticated regulatory mechanisms to adapt to these unfavorable conditions, many of which interface with plant hormone signaling pathways. Abiotic stresses alter the production and distribution of phytohormones that in turn mediate stress responses at least in part through hormone- and stress-responsive transcription factors. Among these, the APETALA2/ETHYLENE RESPONSIVE FACTOR (AP2/ERF) family transcription factors (AP2/ERFs) have emerged as key regulators of various stress responses, in which they also respond to hormones with improved plant survival during stress conditions. Apart from participation in specific stresses, AP2/ERFs are involved in a wide range of stress tolerance, enabling them to form an interconnected stress regulatory network. Additionally, many AP2/ERFs respond to the plant hormones abscisic acid (ABA) and ethylene (ET) to help activate ABA and ET dependent and independent stress-responsive genes. While some AP2/ERFs are implicated in growth and developmental processes mediated by gibberellins (GAs), cytokinins (CTK), and brassinosteroids (BRs). The involvement of AP2/ERFs in hormone signaling adds the complexity of stress regulatory network. In this review, we summarize recent studies on AP2/ERF transcription factors in hormonal and abiotic stress responses with an emphasis on selected family members in Arabidopsis. In addition, we leverage publically available Arabidopsis gene networks and transcriptome data to investigate AP2/ERF regulatory networks, providing context and important clues about the roles of diverse AP2/ERFs in controlling hormone and stress responses.


INTRODUCTION
Abiotic stresses such as water scarcity, extreme temperature and high salinity lead to arrested plant growth and ultimately result in massive agricultural losses (Fahad et al., 2017). Plants have developed sophisticated regulatory mechanisms to respond to external stress signals in a timely manner to ensure optimal growth and stress tolerance. Among these, hormonal signaling pathways and stress responsive transcriptional factors function together to form an interconnected network Ye et al., 2017). In addition, environmental changes are often multifactorial, with several stresses occurring simultaneously. Instead of the linear stress signaling pathways, these regulatory components lead to more complex responses . Under abiotic stress conditions, stress hormones such as abscisic acids (ABA) and ethylene (ET) are induced, whereas the production and distribution of growth promoting hormones, such as gibberellins (GAs), brassinosteroids (BRs), and cytokinins (CTK) are also altered to enable optimal responses (Verma et al., 2016). The regulation of plant hormone signaling during abiotic stresses is partially mediated by hormone-and stressresponsive transcription factors Bechtold and Field, 2018).
APETALA2/ETHYLENE RESPONSIVE FACTOR (AP2/ERF) family transcription factors (AP2/ERFs) have emerged as key regulators of several abiotic stresses and respond to multiple hormones (Dietz et al., 2010;Mizoi et al., 2012;Chandler, 2018). Numerous AP2/ERFs mutants with altered abiotic stress responses and hormone sensitivity have been identified, positioning this family of transcription factors as promising candidates to study the interactions between abiotic stresses and hormones. Several properties of AP2/ERFs, such as induction upon specific stresses and diverse DNA binding preferences, enable these transcription factors to integrate responses of multiple stimuli and participate in different regulatory processes. In this review, we focus on Arabidopsis thaliana to summarize the regulation and function of AP2/ERFs in hormone and abiotic stress responses. The function of AP2/ERFs in crops has been extensively reviewed elsewhere (Abiri et al., 2017;Kulkarni et al., 2017;Phukan et al., 2017). Thus, only a few examples of AP2/ERFs in other species will be discussed. Finally, we investigate AP2/ERFs regulatory networks using publicly available transcriptome data to verify known and uncover novel roles of AP2/ERFs in hormone and stress responses.

Transcriptional Regulation of AP2/ERFs
AP2/ERFs expression is tightly regulated to enable proper stress responses. Gene expression profiling studies have shown that most AP2/ERFs are expressed at low levels under normal conditions, whereas the expression can be induced or repressed at certain growth stages, by hormones and stress stimuli (Feng et al., 2005;Li et al., 2017a;Owji et al., 2017). In many cases, AP2/ERFs expression is regulated by the conserved cis-elements present in their promoter regions, or a combination of multiple responsive elements (Figure 1 left). For example, Arabidopsis DREB2A is highly induced under dehydration and heat conditions (Liu et al., 1998). The expression of DREB2A is controlled by HEAT SHOCK FACTOR1 (HSFA1) and ABRE-BINDING PROTEINs/ABRE-BINDING FACTOR 3 (AREB1/AREB2/ABF3) through binding to Heat Shock Element (HSE) and ABA Response Element (ABRE) motifs on its promoter, respectively (Kim et al., 2011;Liu et al., 2011). Moreover, by aligning 20 Arabidopsis DREB gene promoters to the motifs related to abiotic stresses, it was found that DREB1A/CBF3, DREB1C/CBF2, DREB2C, DREB2G, and DEAR3 have the most types of these motifs including HSE motif to which heat shock factors bind, Low-Temperature Responsive element (LTR) that is important for the induction of cold regulated genes, and ABRE motif that responds to ABA (Sazegari et al., 2015). AP2/ERFs whose promoters contain ethylene-responsive (EBS) motifs are also likely to respond to ET signaling .
Moreover, AP2/ERFs expression might be affected by histone modifications. Histone modifications such as phosphorylation, ubiquitination, methylation, and acetylation can either activate or repress transcription by creating more "open" or "closed" chromatin configurations (Pfluger and Wagner, 2007). A study in peanut (Arachis hypogaea L.) showed that the general inhibition of histone deacetylases (HDACs) and Polyethylene Glycol (PEG) treatment induced the acetylation around peanut DREB1 (AhDREB1) promoter region and resulted in increased AhDREB1 expression (Zhang B. et al., 2018). Given that chromatin features such as histone variants and post-translational histone modifications are altered by abiotic stresses, and these influences could be inherited to the next generation (Asensi-Fabado et al., 2017), the study of epigenetic regulation of AP2/ERFs will advance our understating the mechanisms controlling AP2/ERFs.
Another point of AP2/ERF regulation occurs at the level of splicing. Alternative splicing is required to create a functional isoform for some AP2/ERFs in rice (Oryza sativa) (OsDREB2A/2B) (Matsukura et al., 2010), maize (Zea mays) (ZmDREB2A) (Qin et al., 2007), wheat (Triticum aestivum) (WDREB2) (Egawa et al., 2006), and barley (Hordeum vulgare) (HvDRF1) (Xue and Loveridge, 2004). In these cases, plants produce an inactive AP2/ERF form containing stop codons before the DNA binding domain during normal conditions, while under stress conditions, the exon with a premature stop codons is excluded to generate a functional transcription factor. In addition, microRNA (miRNA)-mediated RNA silencing and translation repression was shown to regulate AP2/ERFs. FIGURE 1 | General regulatory mechanisms of AP2/ERF family transcription factors. AP2/ERFs are regulated by multiple stresses and stimuli at transcription, translation and protein modification levels. Upon stresses, AP2/ERFs are induced though cis-elements presented in their promoter regions (left bottom). These cis-elements include HSE, EBS, LRT, ABRE and many other unknown binding sites that respond to HSFAs, EIN3, ICE, AREBs and other transcription factors, respectively. The transcription of AP2/ERFs is also regulated by HDACs. Alternative splicing helps generate AP2/ERFs functional mRNA (left). Under normal condition, AP2/ERFs have adverse effect on plant growth and development and thus need to be eliminated. miroRNA mediated AP2/ERF silence is one of the ways to inhibit AP2/ERFs translation (left). E3 ubiquitin ligases involved proteasome degradation and phosphorylation mediated by kinases provide additional ways to regulate AP2/ERFs protein levels and activity (middle). These E3 ligases include DRIP1/2, RGLG1/2, SINA2, and PRT6, which mediate DREB2A, ERF53, RAP2.2, and ERF-VIIs degradation, respectively. ERF-VIIs undergo N-end rule mediated degradation under normal condition. BPM provides RAP2.4 docking adaptor for CUL3-E3 complex. MPKs and SnRKs mediated phosphorylation activates and represses ERF104 and RAV1 function, respectively. Under stresses, AP2/ERFs including CRFs and ERF-VIIs translocate into nucleus, bind to conserved or diverged DNA binding sites, interact with many other transcription factors and histone modification complex to either activate or repress stress responsive genes expression (right). Among stress induced genes, RbohD is responsive for ROS and H 2 O 2 generation. Induced H 2 O 2 serves as a signal messenger to active ROS scavenging enzyme genes and stress responsive genes expression. Figure is created with BioRender. miRNA172 targets Arabidopsis AP2 messenger RNA to inhibit its translation (Chen, 2004). Several other miRNAs such as miRNA156 and miRNA838 have also been shown to regulate AP2/ERFs in common bean (Phaseolus vulgaris) (Kavas et al., 2015).

Post-translational Modifications of AP2/ERFs
In addition to transcriptional control, post-transcriptional regulation modulates the activity and abundance of AP2/ERF proteins (Figure 1 middle). One of the modifications, phosphorylation, affects AP2/ERF protein transactivity and stability. In Arabidopsis, SNF1-RELATED PROTEIN KINASES (SnRKs), a positive regulator in ABA signaling pathway, interacts with and phosphorylates RAV1 to inhibit its transcriptional repression function (Feng et al., 2014), while ERF104 and ERF6 are phosphorylated by mitogen-activated protein kinases (MPKs) to positively regulate pathogen responses (Bethke et al., 2009;Meng et al., 2013). The stability of Arabidopsis DREB2A is also affected by phosphorylation. DREB2A is destabilized by phosphorylation on its negative regulatory domain (NRD), which is probably mediated by Casein Kinase 1 (CK1) (Mizoi et al., 2018). Indeed, many kinases such as GSK3-like kinase BR-INSENSITIVE 2 (BIN2) and SnRKs have been reported to mediate abiotic stresses . However, information about how these kinases regulate AP2/ERFs is limited. Therefore, future studies such as phosphoproteomics under abiotic stresses and upon perturbation of specific kinases could help identify AP2/ERF protein modifications, leading to potential strategies to modify the protein activity and stability of AP2/ERFs to confer abiotic stress tolerance.

Transcriptional Regulation of Target Genes by AP2/ERFs
Through directly binding to target gene promoters, AP2/ERFs can either activate or repress target gene expression (Figure 1 right). Besides an N-terminal DNA binding domain, the C-terminal activation domain of AP2/ERFs mediates the activation of target gene expression in Arabidopsis and rice (Nakano et al., 2006). Recently, an activation EDLL motif was identified from Arabidopsis ERF98 and AP2 subfamily. The EDLL motif is strong enough to override the repression effect mediated by ERF-associated amphiphilic repression (EAR) motif, which emphasizes the transactivation activity of AP2/ERFs (Tiwari et al., 2012). However, AP2/ERFs containing an EAR motif with the consensus sequence LxLxL or DLNxxP, or B3 repression domain (BRD) with R/KLFGV sequence exhibit a repressive effect on target genes (Ikeda and Ohme-Takagi, 2009;Kagale and Rozwadowski, 2011). The EAR motif containing AP2/ERFs also interact and recruit transcription co-repressors like TOPLESS (TPL) and TOPLESS-RELATED (TPR) (Causier et al., 2012) or histone modifiers to suppress target gene expression (Song et al., 2005;Song and Galbraith, 2006). In Arabidopsis, ERF7 interacts with a human global co-repressor SIN3 homolog (ATSIN3) that in turn interacts with Histone Deacetylase 19 (HDA19) (Song et al., 2005). Similarly, ERF3 interacts with SIN3 Associated Polypeptide P18 (SAP18) and then recruits HDA19 to repress gene expression (Song and Galbraith, 2006). The BRD motif containing Arabidopsis RAV1 and RAV2 also display repressive activities (Ikeda and Ohme-Takagi, 2009). There is a report demonstrating that genes co-regulated with corresponding AP2/ERFs were enriched with the AP2/ERFs targets. Therefore, by analyzing the enrichment of cognate motifs in AP2/ERFs coregulated genes, it is possible to identify putative target genes of transcription factors and predict their biological functions (Franco-Zorrilla et al., 2014).

AP2/ERF REGULATORY NETWORKS IN ABIOTIC STRESSES
AP2/ERFs regulate numerous abiotic stresses such as cold, drought, heat, salt, and freezing (Lata and Prasad, 2011;Mizoi et al., 2012;Licausi et al., 2013;Phukan et al., 2017). Although many AP2/ERFs are proposed to form an abiotic stress-specific regulatory network, the ability of AP2/ERFs to respond to multiple stimuli and regulate different stresses enable them to  Kagaya et al., 1999;Woo et al., 2010;Feng et al., 2014 form a more complex stress response network. In this network, AP2/ERFs also respond to abiotic stresses with varying dynamic patterns: some AP2/ERFs are induced quickly and continuously, whereas others are regulated by prolonged stress, which indicates they might have mutual influence on each other's function . However, the detailed mechanisms of how different AP2/ERFs cooperate or antagonize with each other are yet to be established. Therefore, in addition to studying the function of individual transcription factors in this family, it is also necessary to study the relationship between different AP2/ERFs in abiotic stress responses. Here we summarize recent works on the DREBs and ERFs regulation in cold, drought, heat and salt stress responses particularly in Arabidopsis, and we also discuss the mutual regulation of different AP2/ERFs.

DREBs: Major Regulators in Cold, Drought, Heat, and Salt Stress Responses
DREBs have been extensively examined in abiotic stresses, where they respond to and positively regulate cold, drought, heat and salt tolerance by directly regulating stress-responsive genes (Figure 2). Among these, DREB1s (DREB-A1 subgroup) containing several C-Repeat-Binding Factors (CBFs) play major roles in acquisition of freezing tolerance (Chinnusamy et al., 2003). CBFs together with another major cold responsive transcription factor, Inducer of CBF Expression (ICE), establish a central cold response pathway to activate a majority of DRE containing Cold Responsive Genes (CORs) in Arabidopsis Liu J. et al., 2018). CORs encode Late Embryogenesis Abundant (LEA) proteins and enzymes for sugar metabolism and fatty acid desaturation that provide the protection for plants from cold stress (Maruyama et al., 2009). The roles of Arabidopsis CBFs in stress responses have been characterized by genetics using cbf mutants generated by CRISPR/Cas9 genome editing techniques and CBFs overexpression lines, as well as transcriptome analysis (Park et al., 2015;. CBFs are also reported to positively regulate plant drought and salt tolerance, which might due to a common set of stress responsive genes (Kasuga et al., 1999;. Despite having a clear function in cold response pathways, how CBFs regulate different stresses and the mechanisms by which they confer stress tolerance are still unknown. Environmental changes are usually multifactorial and several stresses often occur simultaneously. Therefore, the multiple roles of CBFs in abiotic stresses might be necessary for plants to overcome stresses and it will be interesting to examine how CBFs regulate genes under different stress conditions. In addition to CBFs, transcriptome profiling in Arabidopsis also identified several other regulators that are activated during cold acclimation, including members in the DREB-A5 group (Fowler and Thomashow, 2002). The DREB-A5 group contains six members with EAR motifs (DEARs) acting as transcriptional repressors on DRE motif containing genes (Nakano et al., 2006), which provides a repression effect in the DREB-regulated abiotic stress network. Among these, DEAR1 likely acts upstream of CBFs, while DEAR6/RAP2.1 acts downstream of CBFs (Tsutsui et al., 2009;Dong and Liu, 2010). Overexpressed DEAR1 suppressed the cold induced expression of CBFs and displayed reduced plant freezing tolerance (Tsutsui et al., 2009). RAP2.1 is induced by cold, as well as in plants constitutively overexpressing CBFs, but the induction of RAP2.1 by cold has a later onset than the induction of CBF2 (Dong and Liu, 2010). RAP2.1 was first identified as a downstream CBFs subregulon and negatively modifies plants cold tolerance (Fowler and Thomashow, 2002). The presence of the DRE motif in the RAP2.1 promoter region also suggests that RAP2.1 might be a direct target of CBFs (Dong and Liu, 2010). Although the integration of DEAR1 and RAP2.1 in the CBF pathway still needs to be examined using genetics, this negative regulation mechanism might provide checks and balances that minimize the adverse effects of prolonged stress responses. In addition to the negative role of RAP2.1 in cold stress, it also negatively regulates drought tolerance, in that overexpression of RAP2.1 resulted in sensitivity to drought (Dong and Liu, 2010). However, how RAP2.1 regulates drought response, whether it regulates other abiotic stresses and what other DEARs function in abiotic stresses are interesting questions to answer in the future. The checks and balances of DREBs' function in stress were also revealed in Brassica napus, where two groups DREBs (Group I and Group II) regulate cold stress responsive genes sequentially. The early induced Group I DREBs activate cold stress response pathways, whereas the Group II DREBs which are expressed later competitively inhibited Group I DREBs function (Zhao et al., 2006).
DREB2s from DREB-A2 are mostly involved in plant drought and heat tolerance, which has been reviewed in detail elsewhere (Mizoi et al., 2012). Briefly, DREB2s are induced upon drought and heat, and positively regulate DRE containing drought responsive genes such as LEAs, and heat responsive genes such as heat chaperons (Maruyama et al., 2009). Moreover, members in DREB-A4 family such as HARDY (HRD) and in DREB-A6 family such as ERF53, RAP2.4, and TG/RAP2.4A also positively regulate drought and salt tolerance (Karaba et al., 2007;Lin et al., 2008;Cheng et al., 2012;Zhu et al., 2014). Overexpression of HRD in Arabidopsis or rice remarkably improved plant drought and salt tolerance (Karaba et al., 2007). Overexpression of TRANSLUCENT GREEN (TG) resulted in vitrified leaves with increased water content in cells, leading to increased drought tolerance (Zhu et al., 2014). TG regulates cell water homeostasis mainly by directly activating several aquaporin genes (Rae et al., 2011;Zhu et al., 2014), as well as Ascorbate Peroxidases (APx) genes encoding chloroplast peroxidases that functions to protect against photo-oxidative stresses caused by Reactive Oxygen Species (ROS) (Rudnik et al., 2017).
Beside Arabidopsis, DREBs' improvement in stress tolerance has been extensively studied in various crops such as OsDREB1s and OsDREB2s in rice (Dubouzet et al., 2003;Matsukura et al., 2010), ZmDREB2A in maize (Qin et al., 2007), TaDREB1 in wheat (Shen et al., 2003), and HvDRF1 in barley (Xue and Loveridge, 2004). These DREBs provide abundant candidate genes for the engineering of stress tolerance in crops. However, given that DREBs are a major family of AP2/ERF transcription factors that integrate multiple abiotic stress signals, cooperate or antagonize one another, and modulate downstream stress responsive genes, studying the DREB gene regulatory network will provide a platform for a more comprehensive understanding of abiotic stress responses and guide the genetic engineering of crops.

ERF, AP2, and RAV Subfamily Members in Freezing, Hypoxia, and Salt Stress Responses
Members in ERF subfamily also contribute to abiotic stress responses (Licausi et al., 2013). Recently, two groups of ERFs have emerged as central players of abiotic stress regulation in Arabidopsis (Figure 2). CYTOKININ RESPONSE FACTORS (CRFs) in ERF-VI subfamily are induced by CTK as well as multiple abiotic stresses to positively regulate osmotic and freezing tolerance (Rashotte et al., 2006;Rashotte and Goertzen, 2010). CRF6, whose induction is dependent on the perception of CTK, alleviated the H 2 O 2 damage on plants to positively regulate oxidative response (Zwack et al., 2016b). CRF4, one of several CRFs not transcriptionally regulated by CTK, positively regulates freezing tolerance by promoting CORs expression (Zwack et al., 2016a). However, the mechanisms by which CRFs confer stress tolerance remain to be determined.
Members of ERF-VII subfamily in Arabidopsis as well as rice have been demonstrated to play major roles in flooding, low oxygen (hypoxia) and submergence tolerance and their redundant function in hypoxia responses has been reviewed (Bailey-Serres et al., 2012;Bui et al., 2015;Gibbs et al., 2015). For example, rice SUBMERGENCE 1A (SUB1A), and SNORKEL1/2 (SK1/SK2) positively regulate flooding tolerance by two opposite mechanisms: SUB1A mediates a quiescence strategy associated with reduced growth and respiration whereas SK1/SK2 promote a deep-water escape strategy allowing rapid growth of petioles, stems, and vascular changes (Hattori et al., 2009;Locke et al., 2018).
RESPIRATORY BURST OXIDASE HOMOLOG D (RbohD), a NADPH oxidase, helps to generate ROS burst (Yao et al., 2017b). It is reported that RAPs regulate abiotic stresses via an RbohD-dependent mechanism. Apart from being a toxic by-product of biochemical processes, ROS serve as signaling molecules to trigger stress responses and transduce signals crossing cells according to its lower molecular weight and fast cell diffusion (Qi et al., 2018). First it was found that the ROS production and RbohD expression were compromised in single erf74 and double erf74 erf75 mutants at an early stage, which resulted in compromised stress responsive gene expression and stress tolerance. Given that ERF74 binds to RbohD promoter to activate its expression, the RbohD dependent ROS activation was essential for ERF74 and ERF75 mediated hypoxia resistance. However, too much ROS can cause cell injury and cell death. To overcome adverse effects of ROS, overexpression of ERF74 promoted increased ROS scavenging enzymes and stress responsive genes at later stage. Therefore, ERF74 acts as an on-off switch to control RbohD-dependent ROS burst in response to different stresses in Arabidopsis (Yao et al., 2017b). This newly identified mechanism provides more details and divides the stress response into early and later stages, as well as ROS balance. One example that rice ERFs OsLG3 induced ROS scavenging to positively regulate stress tolerance was reported recently, suggesting the similar mechanism exists in rice (Xiong et al., 2018).
Additionally, many other Arabidopsis ERFs also regulate abiotic stresses. ERF1 and Ethylene-and Salt-inducible ERF genes (ESEs) in ERF-IX group positively regulate plant salinity tolerance by promoting salt responsive gene expression . ERF6, another member in ERF-IX group, triggers growth inhibition to confer longterm osmotic stress tolerance (Dubois et al., 2013). RAP2.6L from ERF-X subgroup improves drought and salt tolerance (Yang et al., 2009;Liu et al., 2012). Additionally, Arabidopsis RAVs, especially AP2s, play central roles in developmental processes, such as organ number and size control, shoot and root meristem maintenance, flower initiation and growth (Osnato et al., 2012;Horstman et al., 2014). Members in these subfamilies are also reported to mediate diverse abiotic stress responses. AINTEGUMENTA (ANT) controls organ cell number and size throughout shoot development. ANT also negatively regulates salt tolerance by repressing SOS3-LIKE CALCIUM BINDING PROTEIN 8 (SCABP8/CBL10), a putative Ca 2+ sensor that protects Arabidopsis shoots against salt stress and maintains ion homeostasis (Meng et al., 2015a).
Overexpression of Arabidopsis RAV1 and RAV2 in cotton increased fiber length and even obtained the same yield under drought stress compared with control conditions (Mittal et al., 2015).
Similar as DREBs, ERFs in other plant species like rice, wheat and tomato are also involved in a broad range of abiotic stresses (Abiri et al., 2017;Phukan et al., 2017). Overall, these findings provide the potential of engineering high-efficiency crops under stress conditions. In summary, ERFs function to receive multiple stress signals and control a diverse set of stress responsive genes, where many ERFs have cooperative or antagonistic regulation on stress responses. Therefore, constructing ERFspecific gene regulatory networks would be interesting to provide insight as to how ERFs function as a unit to regulate common downstream genes.

INTEGRATION OF AP2/ERFS WITH HORMONE RESPONSES
In addition to directly regulating abiotic stresses, AP2/ERFs are also involved in hormone signaling and hormone mediatedstress responses. Plant hormones affect abiotic stresses by triggering a wide range of physiological processes (Kazan, 2013(Kazan, , 2015Colebrook et al., 2014;Khan et al., 2015;Muller and Munne-Bosch, 2015;Tao et al., 2015;Sah et al., 2016;. ABA and ET are major stress hormones that are induced under abiotic stress conditions and regulate stress responses associated with AP2/ERFs (Kazan, 2015;Sah et al., 2016). GAs, CTK, and BRs are growth-related hormones that promote cell growth, proliferation and differentiation. It is becoming increasingly evident that these growth-related hormones also have direct and/or indirect effects on abiotic stresses (Kazan, 2013;Colebrook et al., 2014;. Because stress tolerance is often associated with tradeoffs between growth and stress, abiotic stress-tolerant plants usually have lower growth rates and productivity (Bechtold and Field, 2018). Numerous studies in Arabidopsis have shown AP2/ERFs confer plant stresses tolerance associated with growth losses (Figure 2), such as CBFs, DREB2A, HARDY, TG, ERF6, and RAP2.6 (Kasuga et al., 1999;Sakuma et al., 2006;Karaba et al., 2007;Sharabi-Schwager et al., 2010;Krishnaswamy et al., 2011;Dubois et al., 2013;Zhu et al., 2014). However, the growth loss can be minimized by driving expression via the stress-induced RD29A promoter (Kasuga et al., 1999;Sakuma et al., 2006;Matsukura et al., 2010). Thus, in addition to understanding the basis of AP2/ERFs in abiotic stresses, it is also important to explore their roles in the hormone-regulated stresses responses. AP2/ERFs are involved in plant hormone-mediated abiotic stresses through the following aspects: (1) AP2/ERFs transcripts are regulated by hormones; (2) AP2/ERFs alter hormone sensitivity and gene expression by cooperating or antagonizing various hormone signaling components; (3) AP2/ERFs regulate hormone biosynthesis or metabolism via feedback regulation (Figure 3). In the following sections we emphasize how abiotic stress-induced (ABA and ET) and growth-promoting (GAs, CTK, and BRs) hormones are regulated by AP2/ERFs in Arabidopsis through these mechanisms.

AP2/ERFs in ABA-Mediated Stress-Response
The plant hormone ABA is a pivotal hormone that regulates abiotic stress responses including drought, salinity, cold and heat stresses. ABA exerts a protective function through inducing stomata closure, modulating root architecture, and promoting the synthesis of osmolytes (Cutler et al., 2010;Sah et al., 2016). During stress conditions such as water deprivation and osmotic pressure, the rate-limiting ABA biosynthetic enzyme Nine-cis-Epoxycarotenoid Dioxygenase (NCED) is rapidly up-regulated to promote ABA biosynthesis. ABA is then sensed by a large family of PYRABACTIN RESISTANCE1 /PYR1-LIKE/REGULATORY COMPONENTS OF ABA RECEPTORS (PYR/PLY/RCAR) receptors that interact with Protein Phosphatase 2C (PP2Cs) as a ternary complex to release PP2Cs' inhibition on SnRK2 kinases (SnRK2s). The active SnRK2s phosphorylate downstream substrate proteins, including AREBs/ABFs, ion channels, and enzymes such as NADPH oxidases, thereby inducing ABA responses (Finkelstein, 2013;Sah et al., 2016). Generally, transcription factors regulate abiotic stresses through ABAdependent or ABA-independent pathways.
A great deal of studies have shown that DREBs regulate ABAindependent abiotic stresses by directly binding to DRE/CRT motifs on stress responsive genes (Gilmour et al., 2004;Matsukura et al., 2010;Lata and Prasad, 2011;Mizoi et al., 2012;Zhu et al., 2014). However, AP2/ERFs are indispensable for ABA-dependent stress responses as well. ANT (Meng et al., 2015b), ERF53 (Hsieh et al., 2013), ERF-VIIs (Papdi et al., 2015;Yao et al., 2017a), RAP2.6L , and RAP2.6  in Arabidopsis are induced by ABA to up-regulate DRE-and ABRE-element containing genes. Rice OsERF71 positively regulates ABA signaling to alter root architecture and confer drought tolerance (Lee et al., 2017;. The combination of abiotic stresses and ABA also led to the further activation of DREB2s and stress inducible genes . ABA INSENSITIVE 4 (ABI4), a unique one in the DREB-A3 group is a key component of the ABA signaling pathway. Upon ABA and ROS accumulation under stress conditions, ABI4 represses CCAAT Binding Factor A (CBFA) (Zhang et al., 2013). CBFA is a subunit of the trimeric transcription complex of Heme Activator Proteins (HAPs). Repression of CBFA then allows other transcription factors to enter the transcription complex and improves the efficiency of stress responsive gene transcription (Zhang et al., 2013).
Apart from the positive effect of AP2/ERFs in ABAmediated stress responses, many Arabidopsis studies have shown that AP2/ERFs also interrupt ABA signaling, resulting in reduced sensitivity on root growth inhibition and stomata closure (Figure 3). ERF18/ ORA47 activated the PP2C family phosphatase gene ABI2. At the same time ABI1 acted upstream of ORA47 to activate ORA47, leading to an ABI1-ORA47-ABI2 regulation loop that inhibits ABA signaling as well as drought tolerance . RAV1 inhibites ABA sensitivity on root growth by repressing ABI3, ABI4 and ABI5 expression (Feng et al., 2014). SnRK2.2, SnRK2.3, SnRK2.6 also interact with and phosphorylate RAV1 to inhibit RAV1's transcriptional repression of targets genes (Feng et al., 2014).

AP2/ERFs in Ethylene-Mediated Stress-Response
ET is also reported to regulate abiotic stress responses including salt, cold, and flooding (Kazan, 2015). ET is synthesized from the rate limiting enzymes ACC Synthases (ACSs), a major target for the regulation for ET production under stresses (Tao et al., 2015). The binding of ET with its receptor ETHYLENE INSENSITIVE 1 (ETR1) deactivates CONSTITUTIVE TRIPLE RESPONSE 1 (CTR1) kinase activity to release CTR1's inhibition on ETHYLENE INSENSITIVE 2 (EIN2). Then the C-terminal of EIN2 translocates into the nucleus to activate ETHYLENE INSENSITIVE 3 (EIN3) as well as the transcriptional cascade of ethylene-regulated genes (Qiao et al., 2012;Muller and Munne-Bosch, 2015).
In stress responses, AP2/ERFs, especially the ERF subfamily, are the major downstream regulatory factors of the ET signaling pathway (Licausi et al., 2013;Gibbs et al., 2015;Kazan, 2015;Muller and Munne-Bosch, 2015). ET regulates several aspects of Arabidopsis abiotic stress responses, including inhibiting CBFs to negatively regulate cold stress (Figure 3). Conversely, ET positively regulates flooding and submergence mediated by ERF-VIIs, and improves salt tolerance by activating ERF1 and ESEs.
For ET-regulated cold response, the production of ET is inhibited after exposure to cold, which results in compromised cold tolerance (Shi et al., 2012). Consistently, ET insensitive mutants etr1-1, ein2-5, and ein3-1 displayed increased freezing tolerance. EIN3 inhibits the expression of CBFs by directly binding to their promoters. In addition, ET plays crucial roles on plant survival and recovery from flooding, especially in rice. Flooding causes oxygen deficiency, which promotes ET production (Yang et al., 2011) and activates the expression of a set of ERF-VIIs, whose function in hypoxia is discussed above. ET also promotes RAP2.3 nuclear localization and advances ORA59 mediated ethylene responses which is dependent on RAP2.3 (Kim et al., 2018). However, ERF-VIIs regulate hypoxia response partially through ET-independent pathways. The induction of ERF73/HRE1 under hypoxia was not completely abolished in ethylene-insensitive mutants or in the presence of ethylene biosynthesis inhibitors. ERF-VIIs also negatively regulate ET signaling and homeostasis probably via feedback regulations (Hinz et al., 2010;Yang et al., 2011). For instance, HRE1-RNAi seedlings displayed exaggerated triple responses; ACSs was decreased in RAP2.2 overexpression plants, but up-regulated in rap2.2-2 knockout mutant (Hinz et al., 2010). However, how ERF-VIIs control ET homeostasis via negative feedback mechanism under stresses needs further investigation.
ET also has complex regulation in salt stress, which has been extensively discussed (Kazan, 2015;Muller and Munne-Bosch, 2015;Tao et al., 2015). In Arabidopsis, ET signaling is required for plant tolerance to salinity stress as EIN3 activates ERF1 and ESEs to activate downstream stress-related genes and promote salinity tolerance. However, knockout mutants of ACSs also led to salt tolerance, leading to an opposite conclusion in terms of ET signaling and salt tolerance. These different conclusions might be due to the different mutants, growth conditions or experimental setups used. Future investigation is necessary to further our understanding of the role of ET in plant salinity response.

AP2/ERFs in GA-Mediated Stress-Response
The plant hormone GAs is known to promote plant growth and development. GAs have also been shown to regulate abiotic stresses, as reduced GA content slows down plant growth upon exposure to several abiotic stresses including cold, salt, and osmotic stresses (Claeys et al., 2012;Colebrook et al., 2014). GAs are synthesized through several key oxidases including GA 20-oxidases (GA20oxs) and GA 3-oxidases (GA3oxs), and catabolized by GA 2-oxidase (GA2ox) that depletes pools of GA precursors to maintain GA homeostasis (Phillips et al., 1995;Rieu et al., 2008). In the absence of GAs, a group of DELLA proteins (DELLAs) inhibit GA response, and this inhibition can be released by the degradation of DELLAs in the presence of GAs (Claeys et al., 2012). Generally, abiotic stresses cause reduction of GA content and signaling through the inhibition of GA biosynthesis enzymes mediated by Arabidopsis AP2/ERFs (Figure 3). DREB1E and DREB1F confer salt stress-induced growth retardation mostly through the repression of GA20oxs (Magome et al., 2004). CBF1 and ERF6 overexpression plants were sensitive to stress-induced growth retardation because of increased GA2oxs expression as well as the accumulation of DELLAs (Achard et al., 2008;Dubois et al., 2013). Conversely, ERF11 promotes plant internode elongation by activating GA biosynthesis, and expression of GA3ox1 and GA20oxs are increased in ERF11 overexpression plants (Zhou et al., 2016). Nevertheless, ERF11 and ERF6 show antagonistic regulation on stress-induced growth inhibition. ERF11 suppresses the extreme dwarf phenotype of ERF6 overexpression plants and represses ERF6-induced gene expression (Dubois et al., 2015). The opposite regulation by ERF6 and ERF11 reveals that dynamic mechanisms must exist in plants to fine-tune and maintain the balance between plant growth and stress responses.
In addition to GA regulation in Arabidopsis abiotic stresses, rice regulates flooding coping submergence tolerance by two opposite GA regulations. The first quiescence strategy was that SUB1A increased the accumulation of SLENDER RICE1 (SLR1) and SLENDER RICE1 LIKE1 (SLRL1) (DELLA like proteins in rice) to restrict GA signaling and sensitivity, which resulted in inhibition of plant internode elongation and reduced carbohydrate consumption (Fukao and Bailey-Serres, 2008;Locke et al., 2018;Perata, 2018). The second deep-water escape strategy involves SK1/SK2, which lead to up-regulated GA20oxs and promoted internode elongation to escape submergence in water (Hattori et al., 2009;Ayano et al., 2014).

AP2/ERFs in CTK-Mediated Stress-Response
The plant hormone CTK not only plays diverse roles in plant growth and development, but also has been reported to regulate plant abiotic stresses (Zwack and Rashotte, 2015), one of which is mediate by CRFs (Figure 3). CRFs are essential for CTK-mediated embryo, cotyledon, and leaf development, as both single and multiple CRF1/2/3/5/6 mutants displayed cell proliferation deficient phenotypes (Rashotte et al., 2006). The roles of CRF's regulation on CTK-mediated development were further confirmed by the transcriptome analysis of crf 1,2,5 and crf2,3,6 mutants, with or without CTK treatment in Arabidopsis. About 60% of the CTK responsive genes were regulated by both CRFs and type-B ARRs (the typical cytokininresponsive transcription factors), suggesting a model that CRFs acted tandemly with type-B ARRs to mediate CTK response. CRF6 also cooperated with CTK signaling to inhibit stressinduced leaf senescence through a common subset of CTKregulated genes (Zwack et al., 2013). Apart from the CRFs positive effect in the CTK pathway, CRF6 also represses CTKassociated target genes involved in CTK biosynthesis, signaling and transport, to alleviate the adverse effect of CTK on abiotic stress (Zwack et al., 2016b). The opposite regulation between CRF6 and CTK on stresses and similar regulation on senescence suggest that CRF6 regulates CTK signaling through two subsets of genes: one set of genes alleviate the negative effect of CTK on abiotic stresses, while the other set helps CTK to promote plant development. The detailed mechanisms of CRF regulation in these processes remain to be determined. Identification of CRF target genes and the upstream signaling could allow for a better understanding about ERF-VIs function and how CTK regulates abiotic stresses.
BR regulates cold and drought responses through several pathways. For example, BR positively regulates cold tolerance partially though CBF-mediated cold response pathway, where BZR1 binds and promotes the expression of CBF1/CBF2 in response to cold. Cold stress also promotes accumulation of the unphosphorylated and active form of BZR1 by unknown mechanisms (Li et al., 2017b ; Figure 3). However, BR negatively regulates drought tolerance via antagonizing with drought induced transcription factor RD26 on drought responsive genes FIGURE 4 | AP2/ERFs regulatory network in abiotic stresses and hormones. (A) Arabidopsis AP2/ERFs in abiotic stresses and hormone responsive gene co-expression networks. Networks from Chockalingam et al. (2016) and Chockalingam et al. (2017) were used to investigate the functions of AP2/ERFs. The color legend indicates normalized p-value for the enrichment of the indicated hormone-or stress-responsive gene set by Fisher's exact test. (B) Venn diagram showing overlaps among CBFs-regulated genes compared with cold-regulated, ET-regulated and BR-regulated genes (left). Venny (http://bioinfogp.cnb.csic.es/tools/venny/index.html) was used to perform the comparisons. The significance of overlapping genes regulated by CBFs versus cold, ET and BR is calculated by hypergeometric test (right). (C) CBFs regulated cold response pathway with integration of ET and BR signaling components. Panel (C) is created with BioRender. . BR also antagonizes with ABA pathway from receptors to transcription factors and regulates the trade-off of plants growth under stress conditions . Although there are no reports of AP2/ERFs in BR mediated drought response, BES1 and BZR1 target genes include numerous AP2/ERFs, implying that AP2/ERFs have a potential function to integrate the BR pathway with abiotic stresses (Sun et al., 2010;Yu et al., 2011). Future studies in this area will shed light on the mechanisms and roles of AP2/ERFs in BR and stress responses.
In addition, AP2/ERFs regulate the BR pathway through different mechanisms. ERF72/RAP2.3 antagonizes BZR1 and AUXIN RESPONSIVE FACTOR 6 (ARF6) to inhibit hypocotyl elongation, while its role in BR regulated stresses response is unknown (Liu K. et al., 2018). The role of ERF72 in controlling growth implies that ERF72 might be a candidate for the study of the integration of BR and stresses. In fact, rice SUB1A mediates GA and BR cross-talk to control submergence tolerance. SUB1A activates BR biosynthesis and signaling, which in turn induces GA catabolic gene GA2ox7 to lower GAs content, and ultimately promotes rice DELLA protein accumulation (Schmitz et al., 2013).

AP2/ERF TRANSCRIPTION FACTORS AND ABIOTIC STRESSES REGULATORY NETWORK
Based on AP2/ERFs roles in abiotic stresses and hormone signaling, AP2/ERFs function through complicated regulatory networks. These networks are influenced by diverse environment stimuli and plant hormones. Regulatory mechanisms such as protein-protein interactions and cooperative or antagonistic regulation of target genes are involved in dictating the output of AP2/ERF networks for plant growth, development and abiotic stresses. The large number of the AP2/ERF transcription factors coupled with functional redundancy and their diverse roles have made it difficult to fully understand AP2/ERFs networks. A promising approach is to apply computational tools to dissect AP2/ERFs function. These include analyzing promoter sequences for abiotic stress-related motifs (Sazegari et al., 2015), and constructing stress response gene regulatory networks under different stress conditions Van den Broeck et al., 2017). For instance, analysis of 20 Arabidopsis DREB genes promoters demonstrated that they contained most types of HSE LTR and ABRE promoter elements. The multiple stress responsive motifs on one hand help explain the induction of DREBs under various environment stimuli, and on the other hand, they imply that these DREBs might form a central response network to control diverse abiotic stresses (Sazegari et al., 2015).
In addition to promoter analysis, Van den Broeck et al. examined a complex, highly interconnected network of 20 Arabidopsis transcription factors (more than half are AP2/ERFs) to illustrate how stress inhibits plant growth ( Van den Broeck et al., 2017). Under stresses condition, plants limit growth to promote survival (Claeys et al., 2014;Bechtold and Field, 2018). In this case, members from ERF-VIIIs (ERF8, ERF9, ERF11), ERF-IXs (ERF-1, ERF2, ERF5, ERF6, ERF59, ERF98) and ERF-X (RAP2.6L) were significantly up-regulated in proliferating and expanding tissues upon short-term mannitol exposure. Among these, ERF6 (Dubois et al., 2013) and RAP2.6L (Krishnaswamy et al., 2011;Liu et al., 2012) have already been shown to positively regulate stress tolerance and inhibit growth, implying they have potential role in growth and stress trade-offs. This work also illustrated that the activation of these ERFs was sequential. ERF5, ERF6, ERF11, and ERF98 showed a fast, strong and continuous induction, while other ERFs were regulated mildly and slowly. Transcriptomic profiling of each ERF using inducible overexpression plants demonstrated that they formed a highly interconnected gene regulation network consisting of redundant regulation mediated via interaction, activation, repression, and internal regulation. By combining osmotic stress and each individual ERFs transcriptiomic data it was possible to generate a network that simplifies the signaling cascades into linear pathways, although further genetic confirmation is needed.
In order to gain additional insight into AP2/ERF function in abiotic stresses and hormone response pathways, we investigated Arabidopsis AP2/ERFs in gene co-expression networks (Chockalingam et al., 2016;Chockalingam et al., 2017) that were generated by processing and classifying 1000s of public microarray datasets into tissue and process specific categories. If AP2/ERFs are involved in hormone-and stress-responsive gene regulation, then these transcription factors would be expected to connect to hormone/stress regulated genes in the network. We used transcriptome data from various stress and hormone treatments (Maruyama et al., 2009;Bai et al., 2012;Park et al., 2015;Song et al., 2016;Hickman et al., 2017;Albihlal et al., 2018;de Zelicourt et al., 2018;Guo et al., 2018;Xie et al., 2018;Zhang F. et al., 2018) to test this idea and found that many AP2/ERFs connect to more hormone/stress regulated genes in the network than would be expected by chance (Figure 4A, Fisher's exact test). We further divided AP2/ERFs into four clusters based on enrichment for hormone or stress responsive genes in the network. The role of CBFs in stress and hormone responses was investigated as an example. CBFs are distributed into four clusters that include AP2/ERFs enriched for cold, drought, salt and heat responsive genes. These clusters also include AP2/ERFs enriched for ABA, ET, and BR responsive genes. In line with these network predictions, the function of CBFs in ET and cold stress (including those mediated by BRs) has been confirmed by genetics or transcriptomic studies (Shi et al., 2012;Li et al., 2017b). We further compared genes differentially expressed in cbf123 triple mutants after cold treatment, BR and ET responsive genes ( Figure 4B; Nolan T. M. et al., 2017;de Zelicourt et al., 2018;Zhang F. et al., 2018). The comparison indicates that CBFs share a significant amount of genes with cold, ET and BR. For example, cold regulates ∼4000 genes, ∼68% of which are regulated by CBFs, supporting CBFs' major contribution in cold tolerance. Among these, ∼20% of CBF regulated cold responsive genes are regulated by ET and BR, indicating that ET and BR regulated cold stress are dependent on the CBF pathway. ET and BR negatively and positively regulated CBFs through their specific transcription factors EIN3 (Shi et al., 2012) and BES1/BZR1 (Li et al., 2017b), respectively ( Figure 4C). Taken together, these examples of network and transcriptome comparisons can (1) predict AP2/ERFs functions, (2) study AP2/ERFs function in specific stress/hormone responses; (3) examine the cross-talk of hormones in controlling growth and stress tolerance; (4) help clarify their upstream regulators and downstream targets.

SUMMARY AND FUTURE DIRECTIONS
In summary, AP2/ERFs are regulated by stresses and control multiple plant stress responses to coordinate plant growth under stress conditions. AP2/ERFs are both regulated by plant hormone pathways and modulate plant hormone biosynthesis and signaling. Considering the complex roles of AP2/ERFs in abiotic stress and hormonal responses, future studies are required to fully understand this unique family of transcription factors. The regulation of AP2/ERFs by various hormone signaling pathways should be examined at transcriptional and post-transcriptional levels. The genome-wide identification of AP2/ERF target genes should help understand their functions as well as gauge the scope of their actions. Establishing gene regulatory networks using gene expression data and computational tools will be crucial to understand the large family of genes with 100s-1000s of target genes. Considering the complexity, the functions of AP2/ERFs should be examined in different tissues/cell types and in a temporal manner. Many AP2/ERFs are potential candidates for crops stress tolerance engineering. Thus, a full understanding of AP2/ERFs can guide crop engineering to achieve optimal plant growth and crop production under changing environmental conditions.

AUTHOR CONTRIBUTIONS
ZX, TN, HJ, and YY conceived the topic. TN and HJ collected the gene regulation data. TN performed the network analysis. ZX wrote the manuscript with edits from other co-authors.

FUNDING
The work was supported by grants from NSF (MCB-1818160) and Plant Science Institute at Iowa State University. ZX was partially supported by a China Scholarship Council fellowship.