EgJUB1 and EgERF113 transcription factors as master regulators of defense response in Elaeis guineensis against the hemibiotrophic Ganoderma boninense

Hemibiotrophic pathogen such as the fungal pathogen Ganoderma boninense that is destructive to oil palm, manipulates host defense mechanism by strategically switching from biotrophic to necrotrophic phase. Our previous study revealed two distinguishable expression proles of oil palm genes that formed the basis in deducing biotrophic phase at early interaction which switched to necrotrophic phase at later stage of infection. The present report is a continuing study from our previous published transcriptomic proling of oil palm seedlings against G. boninense. We focused on identifying differentially expressed genes (DEGs) encoding transcription factors (TFs) from the same RNA-seq data resulting in 106 upregulated and 108 downregulated TFs being identied. DEGs involved in four established defense-related pathways are presented which are responsible in cell wall modication, reactive oxygen species (ROS)-mediated signaling, programmed cell death (PCD) and plant innate immunity. We discovered upregulation of JUNGBRUNNEN 1 (EgJUB1) during fungal biotrophic phase while Ethylene Responsive Factor 113 (EgERF113) demonstrated prominent upregulation when the palm switches to defense against necrotrophic phase. EgJUB1 was shown to have binding activity to a 19 bp palindromic SNBE1 element, WNNYBTNNNNNNNAMGNHW found in the promoter region of co-expressing EgHSFC-2b. Further in silico analysis of promoter regions revealed co-expression of EgJUB1 with TFs containing SNBE1 element with single nucleotide change at either the 5th or 18th position. Meanwhile, EgERF113 binds to both GCC and DRE/CRT elements promoting plasticity in upregulating the downstream defense-related genes. Both TFs were proven to be nuclear localized based on subcellular localization experiment using onion epidermal cells. results propose key The ethylene-responsive factor like protein 1 (CaERFLP1) of hot pepper (Capsicum annuum L.) interacts in vitro with both GCC and DRE/CRT sequences with different binding anities: Possible biological roles of CaERFLP1 in response to pathogen infection and high salinity conditions in transgenic tobacco plants.


Abstract
Background Hemibiotrophic pathogen such as the fungal pathogen Ganoderma boninense that is destructive to oil palm, manipulates host defense mechanism by strategically switching from biotrophic to necrotrophic phase. Our previous study revealed two distinguishable expression pro les of oil palm genes that formed the basis in deducing biotrophic phase at early interaction which switched to necrotrophic phase at later stage of infection.

Results
The present report is a continuing study from our previous published transcriptomic pro ling of oil palm seedlings against G. boninense. We focused on identifying differentially expressed genes (DEGs) encoding transcription factors (TFs) from the same RNA-seq data resulting in 106 upregulated and 108 downregulated TFs being identi ed. DEGs involved in four established defense-related pathways are presented which are responsible in cell wall modi cation, reactive oxygen species (ROS)-mediated signaling, programmed cell death (PCD) and plant innate immunity. We discovered upregulation of JUNGBRUNNEN 1 (EgJUB1) during fungal biotrophic phase while Ethylene Responsive Factor 113 (EgERF113) demonstrated prominent upregulation when the palm switches to defense against necrotrophic phase. EgJUB1 was shown to have binding activity to a 19 bp palindromic SNBE1 element, WNNYBTNNNNNNNAMGNHW found in the promoter region of co-expressing EgHSFC-2b. Further in silico analysis of promoter regions revealed co-expression of EgJUB1 with TFs containing SNBE1 element with single nucleotide change at either the 5th or 18th position. Meanwhile, EgERF113 binds to both GCC and DRE/CRT elements promoting plasticity in upregulating the downstream defense-related genes. Both TFs were proven to be nuclear localized based on subcellular localization experiment using onion epidermal cells.

Conclusion
Our ndings demonstrated unprecedented transcriptional reprogramming of speci c TFs potentially to enable regulation of speci c set of genes during different infection phases of this hemibiotrophic fungal pathogen. The results propose EgJUB1 and EgERF113 as key TFs in orchestrating the defense mechanisms during biotrophic and during the subsequent transition to necrotrophic phase, respectively.
Identi cation of these phase-speci c oil palm TFs is important in designing strategies to tackle or attenuate the progress of infection. Background vary depending on the intensity and intricacy of multiple stresses [25]. There are six major TF families involved in plant defense response including MYB, bHLH, AP2/ERF, NAC,bZIP and WRKY [26]. Regulation by TFs is crucial to mediate the transcriptional reprogramming which include induced expression of defense-related genes responsible in the production antifungal proteins or the antimicrobial secondary metabolites known as phytoalexins [27]. Intriguingly, plants also display cross-tolerance phenomena in which single type of stress may trigger multitude tolerant levels to different stresses [28]. For instance, heat stress transcription factors (HSFs) play important role as master regulator of defense response under multiple stresses [29,30]. Over-expressed HSFs confer resistance against dehydration, bacterial and oomycetes infections and improve yield under water-limited conditions [31].
TFs bind to cis-acting elements located in promoters of either other mediator TFs or downstream target genes which results in up-or down-regulation of their expression [32,33,34]. NAC TFs are recognized as master regulator for secondary cell wall biosynthesis mediated by MYB TF through formation of NAC-MYB-CESA signaling cascade [35]. In addition, bHLH-MYB association was involved in the biosynthesis of avonoids secondary metabolites under phytohormones signaling, wounding and fungal interaction [36,37]. Interaction to speci c DNA sequences (binding motifs) is dependent on the DNA binding domain (DBD) of TFs [38]. binding preferences of TF during biotic or abiotic stress such as ERFs have been suggested to correlate with the composition of amino acid sequences in DBD [38,39]. The presence of multiple cis-acting elements in the promoter region contribute to overlapping roles in development and/or defense against multiple stresses of the expressed proteins [26].
In addition to biotic or abiotic stresses [40], in this study we are revealing that the regulation of TFs is dependent on modes of pathogen's infection (biotrophic and necrotrophic). Only EgUNE10 TF and a few TF families have been reported to regulate defense against later stage of G. boninense infection [4,41,42]. However, there is no comprehensive report on transcriptomic pro ling of defense-related TFs during different infection phases of hemibiotroph in plants. Thus, this study is the rst attempt to recognize speci c TFs as 'key' biomarkers involved in transcriptional switching from biotrophic to necrotrophic infection phase based on early oil palm-G. boninense interaction. Their potential targeted defense response pathways that distinguished the two phases are discussed based on the identi cation of speci c motifs interacting with the newly discovered TFs. The ndings might allow a more effective disease management strategy to attenuate the progress of G. boninense infection of oil palm and prevent the spread of the disease.

Results
Elaeis guineensis defense-related transcription factors and biomarkers of biotrophy-necrotrophy switch In order to identify the TF families involved in early defense of oil palm against. G. boninense, transcriptomics analysis of Ganoderma-treated root tissues at 3, 7 and 11 d.p.i was carried out. We identi ed 106 of upregulated and 108 of downregulated TFs upon early interaction with G. boninense ( Fig. 1a and b). Two distinct phases can be observed having profound expression at 3 d.p.i (biotrophic) and 11 d.p.i (early necrotrophic) with intermediate at 7 d.p.i. We have previously reported the transition of biotrophic to necrotrophic defense mechanism, based on preliminary screening using defense-related molecular biomarkers EgPR1 (biotrophic) and EgMYC2 (necrotrophic) [5]. The pairwise comparison was constructed between control against time course treatments using stringent cut-off values of log 2 fold change (FC) ≥ |1.0| and P-value < 0.01. We found that the expression patterns of the genes showed either decreasing or increasing over time. Major TF families upregulated during early interaction with G. boninense were mainly bHLH > MYB > AP2/ERF, followed by bZIP > MADS > TCP > OFP > NAC > GATA > HSF > NFY > E2F > WRKY > EIN/EIL. On the other hand, highest downregulated families of TF were found to be mainly AP2/ERF > bHLH > MYB followed by MADS > CAMTA > NAC > TCP > GATA > bZIP > HSF > NFY > E2F > WRKY > OFP. The families of oil palm's TFs involved during defense response against G. boninense were found to be the same but involving different members in both upregulated and downregulated DEGs. Distinctively, EIN/EIL and CAMTA TF family was only found in upregulated and downregulated DEGs, respectively. AtCAMTA3 and AtCAMTA4 were reported to negatively regulates plant defense response under SA-mediated signaling pathway against obligate biotroph [43,44].
To further understand plant response against G. boninense interaction, common pathways regulating defense mechanisms were selected from the RNA-seq data (Fig. 1c). Four main defense-related pathways are presented. Reported genes involved in cell wall modi cation, ROS-mediated signaling, PCD and plant innate immunity were all differentially expressed indicating active regulation of defense response in E. guineensis against hemibiotroph G. boninense. The expression patterns of selected biotic stress-related genes were important to distinguish the defense mechanisms executed by plant during two different phases of biotrophic and necrotrophic infections.
Six genes of Cellulose synthase (CESA) and Cellulose synthase-like D (CSL) which were reported in cell wall modi cation including EgCESA2, EgCESA4, EgCESA5, EgCESA9, EgCSLD2 and EgCSLD5 demonstrated similar expression patterns with high upregulation at 3 d.p.i that successively decreased across time. Three out of six genes involved in ROS-mediated signaling including Superoxide dismutase (EgSOD), Glutathione S-transferase (EgGSTF12) and Respiratory burst oxidase homolog (EgRBOHA) were increasingly expressed from 3 d.p.i to 11 d.p.i, whilst EgGSTU17, EgGST3 and EgRBOHB demonstrated decreasing expression patterns. Two genes reported in PCD, Metacaspase 9 (EgMC9) and Bifunctional nuclease (EgBFN) showed antagonistic expression patterns of increasing and decreasing regulation, respectively. Five genes involved in plant innate immunity of PTI/ETI signaling were found to be differentially expressed. Three of the genes including NONEXPRESSOR OF PATHOGENESIS-RELATED GENE 1 (EgNPR1), Calmodulin-like 3 (EgCML3) and Calcium dependent protein kinase 7 (EgCDPK7) were upregulated highest at 3 d.p.i and 7 d.p.i before subsequently declined. The other genes which were EgCML7 and EgCDPK28 were successively upregulated with highest expression at 11 d.p.i. The reliability of RNA-seq data has been validated in our previous report [5] using representative upregulated and downregulated DEGs with different levels of fold change compared to untreated seedlings (control).
Expression patterns of transcription factors regulating defense response against Ganoderma boninense qPCR analysis was able to distinguish TFs regulated under biotic and/or abiotic stresses by comparing Ganoderma-treated (GT) with mock-treated (MT) samples, respectively at designated time points (Fig. 2). Biotrophic-associated TFs, EgJUB1 and EgTCP15 were highly upregulated at 3 and 7 d.p.i in GT samples. However, EgTCP15 also exhibited signi cant upregulation in MT samples at 11 d.p.i. TCP15 was reported in both developmental and stress response pathways under SA-mediated signaling, enhanced by interaction with NPR1 [45,46]. The expression levels in MT samples at all time points were signi cant for all analyzed genes except for EgJUB1, which indicates its speci c role during biotic stress.
Meanwhile, the qPCR analysis on candidates of necrotrophic-associated TFs revealed that EgERF113, EgEIN3 and EgMYC2 were highly upregulated at 11 d.p.i on GT samples. Both EIN3 and MYC2 are JAdependent which were upregulated and downregulated in the RNA-seq data, respectively. The expression of these genes is known to be mutually exclusive whereby MYC2 relies on co-actions of JA-abscisic acid (ABA) while EIN3 regulates plant defense response through JA-ET signaling [47,48]. EgERF113 which demonstrated non-signi cant expression on MT samples (abiotic stress) at all time points were selected for further characterization as novel necrotrophic-speci c TF.

EgJUB1 binds to novel SNBE motif during biotrophic infection
Characterization of EgJUB1 TF against G. boninense infection was carried out via Y1H assay and EMSA. Three potential binding motifs; one NAC binding site (NACBS) and two secondary wall NAC binding elements (SNBEs) with respective mutants were tested on EgJUB1 (Fig. 3). NACBS is the established binding motif for JUB1 TF during abiotic stress while SNBEs are the novel target motifs tested in the present study. The SNBE consensus sequence was identi ed as WNNYBTNNNNNNNAMGNHW, whilst NACBS was RRYGCCGT. Y1H assay demonstrated positive interaction with SNBE1 motif but not to NACBS and SNBE2. The results indicate regulation of alternative pathway by JUB1 in enhancing resistance during biotic stress. The positive binding with one of SNBE1 motif indicates that the binding a nity is dependent on the core motifs of SNBE consensus sequence, whereby four nucleotides change on the core motif of SNBE1 resulted in no interaction with SNBE2 ( Fig. 3a). Colony PCR was carried out on ve positive clones and sequencing was performed using T7 promoter primer of pGADT7-Rec vector to con rm genuine positive interaction. p53-AbAi yeast reporter vector was used as positive control in the Y1H assay. We further con rmed the protein-DNA interaction via EMSA by testing the binding of nuclear protein extracted from positive clone of Y1H assay on biotinylated DNA probes. Shifted band was observed upon testing with biotinylated DNA probe. Unlabeled probe (molar excess 200-fold) was able to compete effectively for binding with biotinylated target DNA probe. The inability of mutated fragment to compete with biotinylated target DNA probe revealed speci c binding of EgJUB1 to SNBE1 (Fig. 3b). SNBE consensus sequence is the predicted DNA binding motif of EgJUB1 (Fig. 3c), based on Plant Transcription Factor Database (PlantTFDB), but thus far there is no reported evidence to support this.
Here, we listed SNBE motifs identi ed in the 1.5 kb promoter regions of TFs co-expressing with EgJUB1 during biotrophic phase (Fig. 3d). The core motif of SNBE1 sequence was identi ed in the promoter region of EgHSFC-2b, suggesting direct regulation of EgJUB1 with the EgHSFC-2b. We highlighted the promoter regions of a few TFs co-expressing with EgJUB1 with single nucleotide change in the SNBE1 core motif tested in the present study at the 5th or 18th position, including EgHSFB-4b, EgGAMYB-X2, EgERF003, EgKAN1-like-X3, EgILI-5-like, EgERF086-like, EgPIF3-X1 and -X2. It is plausible that changes of single nucleotide on SNBE1 core motif, still maintain the ability of EgJUB1 to activate these TFs. Changes of two and more nucleotides in the core motif of SNBE1 is expected to result in signi cant decline in binding a nity [49], which merit analysis in more detail.

EgERF113 binds to both GCC and DRE/CRT motifs during necrotrophic infection
EgERF113 TF was tested via Y1H assay and EMSA (Fig. 4) on two AP2/ERF DNA binding preferences which were GCC-box also known as Ethylene-Response Element (ERE) and Dehydration Response Element/C-Repeat (DRE/CRT). The GCC-box and DRE/CRT motifs in the study share 6-bp core sequence of GCCGMC. Our ndings revealed recognition of EgERF113 by both motifs in the Y1H assay. The binding a nity of EgERF113 with both binding motifs was supported by the EMSA results. Unlabelled probes of GCC-box and DRE/CRT (molar excess 200-fold) were able to compete to respective biotinylated target DNA probes. No shifted band was observed on mutated fragments which proved binding speci city of EgERF113 with both GCC-box and DRE/CRT motifs. The ndings suggest that EgERF113 can regulate stress-related genes harboring GCC-box and/or DRE/CRT in their promoter region.

EgJUB1 And EgERF113 Are Localized In The Nucleus
Analysis of the deduced amino acid sequence revealed the NAC binding domain of EgJUB1 with ve highly conserved subdomains A to E at the N-terminal region. The C-terminal region showed highly conserved domain between different species which suggested its possible role as a transcriptional activator, repressor or in binding to other protein. Putative nuclear localization signals (NLS), KKSLVYYLGSAGKGTKT was identi ed in subdomain D (Fig. 5a).
The DBD of EgERF113 TF consists of three-stranded ß-sheets and one α-helix running almost in parallel. Analysis of the deduced amino acid sequence of EgERF113 proved the presence of tryptophan (W) amino acid at 14th position of the DBD which explained the binding speci cities to both GCC-box and DRE/CRT motifs. The extra amino acid at 24th position of DBD implies GCC-box binding speci city of ERF TF is 61 amino acids long [39]. However, AP2/ERF DBD of EgERF113 lacks this additional amino acid. Putative NLS PWGKWAAEIRDPRKAIRV was identi ed in ß-sheets (Fig. 5b).
To determine the subcellular localization of EgJUB1 and EgERF113 proteins, we utilized the Agrobacterium-mediated transformation for transient expression of green uorescent protein (GFP) in onion epidermal cells. As shown in Fig. 5c, both GFP-labelled EgJUB1 and EgERF113 could coaccumulate in the nucleus. The observation is consistent with the putative role of these proteins acting as TFs.

Discussion
Upon assaulted by pathogens, plants respond by activation of intricate defense systems. Depending on the nature of pathogens, biotrophic and necrotrophic infections are fundamentally different in terms of their infection approach, effector proteins, and the host defense response [50]. Thus, tackling the disease based on infection stage, hypothetically should be able to save or at least prolong the life span of Ganoderma-infected palms. Identifying the infection at biotrophic phase may help planters to take suitable disease management strategy to prevent the disease from transition to the more chronic necrotrophic phase. Meanwhile, infected palms at necrotrophic phase may be treated with more intensi ed practice such as using chemical fungicide. Differentiation of biotrophic and necrotrophic TFs were based on established defense-related biomarkers in known defense mechanisms such as plant innate immunity and HR that can distinguish these two phases. In this study, we demonstrated the key roles of EgJUB1 as a biotrophic-speci c while EgERF113 as a necrotrophic-speci c transcriptional regulator, respectively during early oil palm-G. boninense interaction.
EgJUB1 mediates defense response during biotrophic phase through SNBE motif Expression pro le of EgJUB1 observed in RNA-seq was validated via qPCR which con rms its role in host defense regulation during biotrophic phase of G. boninense infection independent of abiotic stress. JUB1 was rst linked to homeostasis of oxidative stress particularly related to hydrogen peroxide (H 2 O 2 ) signaling. It binds to cis-element that serves as the NACBS in the promoter region of DREB2A TF for tolerance to abiotic stresses [51]. DREB2A TF binds directly to DRE sequence of drought-stress responsive genes, including HSFs but the mechanisms are still unclear [52]. AtJUB1 (ANAC042) was also reported as a key TF that induces camalexin expression, a major phytoalexin of Arabidopsis against bacterial pathogen [53]. A more recent study revealed induced expression of NAC042_5, an orthologue of AtJUB1 in response to biotrophic fungus Erysiphe necator [54]. The JUB1 which acts independent of SA was induced speci cally during pathogen colonization. Intriguingly, oil palm EgJUB1 was found to co-express with two SA-dependent genes of EgTGA1 and EgNPR1. Our results were more in line with the SAdependent master regulator of NPR1, a cofactor of TGA1 reported by [55], which induces Pathogenesis related (PR) genes [56] during biotrophic phase.
We are reporting for the rst time induced expression of EgJUB1 under pathogen challenge, regulating defense-related gene(s) harbouring SNBE binding motif. SNBE motif is composed of an imperfect palindromic 19-bp sequence which can be present in various targets' promoters including TFs and downstream genes involved in secondary cell wall biosynthesis, cell wall modi cation as well as PCD [57]. Binding to the 19 bp SNBE (A/T)NN(C/T)(C/G/T)TNNNNNNNA(A/C)GN(A/C/T)(A/T) consensus sequence is critical at the 9 core nucleotides, regardless of mutation on the other nucleotides [49]. They reported that mutation(s) on these 9 core nucleotides causes reduced and/or elimination of the transcriptional activation, on the contrary changes in the other non-critical nucleotides enhance the binding a nity.
It was discovered from our study that EgJUB1 directly regulates EgHSFC-2b to promote resistance against biotrophic phase through SNBE1 motif. The SNBE1 motif tested in this study consists of the nucleotides G and T at the 5th and 18th position of the core motif, respectively. Based on the report by Zhong et al. [49], the binding a nity can still be maintained if a single nucleotide in the core motif of SNBE consensus sequence is changed, however, changes of two and more nucleotides may reduce the binding a nity signi cantly. Thus, promoters of the listed oil palm TFs (Fig. 3D), EgHSFB-4b, EgGAMYB-X2, EgERF003, EgKAN1-like-X3, EgILI-5-like, EgERF086-like, EgPIF3-X1 and -X2, which harbour single nucleotide change at either the 5th or the 18th position of the SNBE1 core motif most likely are still able to bind to SNBE1.
Our ndings are in line with a recent study which observed high up-regulation of HSF and heat shock proteins (HSPs) against biotrophic fungus [58,59]. HSF was also reported during bacterial infection which directly regulated Enhanced Disease Susceptibility 1 (EDS1) and PR4 under SA-mediated signaling [60].
Proposed defense-related pathways regulated by EgJUB1 co-expressing genes Here, we report that high expression of EgCESA4, EgCESA9 and EgCSLD2 correlates with the expression of EgGAMYB-X2 TF during biotrophic phase (3 and 7 d.p.i) before a subsequent decline in expression. GAMYB TF interacts with GAMYB binding motif to activate downstream genes [61]. The GAMYB motif was found in the promoter region of CESA responsible in secondary cell wall cellulose biosynthesis [35,62]. Consistently, CESA4, CESA7 and CESA9 were reported as regulators of secondary cell wall cellulose synthesis [63]. In addition, local cell wall reinforcement by CSLD2 has been proven under biotroph challenge of powdery mildew fungus (Douchkov et al., 2016). Thus, it is strongly postulated that EgJUB1 binding to SNBE1 motifs in the promoter regions of EgGAMYB-X2 activates oil palm defense response through regulation of secondary cell wall biosynthesis.
Increased production of ROS accompanied with PCD [7], provide evidence on the occurrence of HR. In the current study, genes regulating antioxidant enzymes EgSOD, EgGSTU17, EgGST3 and EgGSTF12 were highly upregulated during early interaction with G. boninense. We also observed the expression of EgNPR1 and EgTGA1 which have been recognized to be overexpressed exclusively during biotrophic attack under SA-mediated signaling pathway [65,66]. The co-actions of EgNPR1 and EgTGA1 results in upregulation of EgPR1 which has been proven as a biotrophic marker in our previous report [5].
EgERF113 mediates defense response during necrotrophic phase through GCC-box and DRE/CRT motifs MYC2 relies exclusively on co-actions of JA-ABA branch response to regulate resistance against insects and wounding by repressing the JA-ET branch [48]. Although EgMYC2 was preliminarily screened as a necrotrophic biomarker [5], the TF is categorized as a downregulated DEG in our RNA-seq data. This can be explained by elevated expression of ethylene insensitive 3 (EgEIN3) which then activates EgERF113 through ET regulation. Binding of ET to its receptors inactivates CONSTITUTIVE TRIPLE RESPONSE 1 (CTR1) which in turn release repression on EIN2 activity and stabilize EIN3 and EIN3-LIKE1 (EIL1) within the nucleus [67]. As a result, ERFs are activated and the ERF TFs modulate transcriptional activity of developmental as well as stress-induced responsive genes [68]. Master regulator of ethylene signaling pathways, EIN3 and EIN3-like (EIL) have been proven to modulate multitude cascades of downstream transcriptional responses [47,69]. De-repression of EIN3/EIL from JA-ZIM domain (JAZ) activates JA-ET signaling that positively regulate transcriptional activations of development as well as defense response against necrotrophic pathogens [47,70]. The ndings demonstrated that oil palm establishes resistance against early necrotrophic through co-actions of JA-ET, rather than JA-ABA.
Here, we report regulation of defense response against necrotrophic phase of G. boninense through multicascades activation of JA-ET branch leading to overexpression of EgERF113. ERF113, also recognized as RELATED TO APETALA2.6-LIKE (RAP2.6L) which is closely related to ERF108 (RAP2.6) was found to be responsive to abiotic stresses (salinity, heat and drought) as well as stress hormones, particularly JA and/or ET [71]. ERF108 has been reported in JA-induced defense response against wounding and pathogens [72,73]. Similarly, ERF113 was shown to be induced upon wounding [74,75], as well as pathogens infection [76,77].
We are the rst reporting on EgERFF113 with binding preferences for GCC-box and DRE/CRT motifs. AP2/ERFs are known to have multiple conserved DNA binding preferences [78]. However, DREBs are typically known to recognize DRE/CRT conferring resistance against abiotic stresses whilst ERFs bind to GCC-box promoting defense against biotic stresses. ERF113 was reported to bind to the GCC-box [77], but not tested on DRE/CRT motif. To date, only a few ERFs were reported with binding preferences on both GCC-box and DRE/CRT conferring resistance against pathogens attack [79,80,81,82]. Binding to GCC-box and DRE/CRT motifs that are present in plant defensin (e.g PDF1.2) as well as PR genes, activates defense-related genes [83,84]. Kaur et al. [84], also reported DRE/CRT element in the promoter region of calcium-responsive genes. Calcium ion (Ca 2+ ) signaling plays paramount importance in defense mechanisms of plant in perceiving invading pathogens [85]. Phukan et al. [39] has studied the divergent of AP2/ERF TF DNA-binding speci cities based on sequence characterization. Their ndings suggested glutamic acid (E) at the 20th and alanine (A) at the 48th positions as identi ed in EgERF113 in the 60 amino acids long DBD denote binding speci city to DRE/CRT motif. EgERF113 also showed conservation of amino acids for GCC-box binding at the 10th, 18th, 20th, 37th and 59th positions. Although the speci c amino acid in DBD that regulates binding of TFs to both GCC-box and DRE/CRT motifs has not yet been con rmed, replacing phenylalanine (F) at the 14th position to tryptophan (W) in EgERF113 changes the binding speci city from GCC-box only to binding both GCC-box and DRE/CRT motifs, as suggested by Phukan et al. [39]. An additional amino acid (basic polar) at the 24th position of DBD which is lacking in EgERF113 is essential for speci c binding to GCC-box only. Thus, EgERF113 with 60 amino acids DBD opposes the classi cation of speci c GCC-box binding DBD of 61 amino acids in length.
Proposed defense-related pathways regulated by EgERF113 co-expressing genes Two RBOH genes, EgRBOHA and EgRBOHB have been previously reported in our study [5], and reduced EgRBOHB expression during early necrotrophic phase may suggest plant's response in delaying the progression of disease. This is in line with the report on the response against hemibiotroph Macrophomina phaseolina [86], wherein increase expression of RBOH leads to increase susceptibility of plants to necrotrophic infection.
Another less studied TF, AtMYB122 was reported to regulate genes involved in camalexin biosynthesis [87]. Increased accumulation of indolic glucosinolates during induction of glucose signaling was evidently regulated by AtMYB122 [88]. Glucosinates [89,90]. Based on the expression patterns of EgMYB122, the TF channeled its regulation from responding to abiotic stress at 3 d.p.i. into defensing against Ganoderma attack and it is expected that the gene regulation will be more extensive at later stage of the necrotrophic phase.
Rapid and transient increase of cytosolic Ca 2+ particularly during pathogens interaction results in activation of both PTI and ETI signaling [85]. CMLs and CDPKs are known as Ca 2+ sensor proteins which are responsible in perceiving the transduction during plant innate immunity [91,92]. Based on our RNAseq data analysis, both EgCML7 and EgCDPK28 were upregulated during necrotrophic phase. Consistent with the analysis, EgERF113 is suggested to orchestrate defense mechanisms through regulation of PR and calcium-responsive genes.

Other transcription factors differentially regulated under biotrophic or necrotrophic phase
The Calmodulin-binding transcription activator (CAMTA) gene was rst discovered in Nicotiana tabacum, regulating senescence and cell death [93], and was recently comprehensively studied by Kakar et al. [94]. We discovered members of the novel CAMTA TF family, namely EgCAMTA3 and EgCAMTA4 which were downregulated at 3 d.p.i. The suppression of both genes was later reduced across time, which might be the result of infection phase transition from biotrophic to early necrotrophic phase. CAMTA was proven to regulate responses under Ca 2+ signaling during both abiotic and biotic stresses [95,96].
In general, most downregulated DEGs of TFs showed de-repression across time. It is reasonable to postulate that changes of expression patterns might be the results of plant immunity interplay against biotrophic and necrotrophic infection phases. For instance, downregulation of EgMYB108 was highest at 7 d.p.i before reducing at 11 d.p.i. Coherently, MYB108 was proven to positively regulate defense mechanism against hemibiotroph Verticillium dahliae in the presence of calmodulin and Ca 2 , antagonistic to the regulation of CAMTA3 [97,98]. In contrast, EgERF9 was found to be downregulated exclusively during necrotrophic phase at 11 d.p.i. The result provides agreement with other study which reported repression activity of AtERF9 in enhanced resistance against necrotrophic Botrytis cinerea [83]. Likewise, expression patterns of TFs in upregulated DEGs were higher at early interaction against G. boninense before decreasing over time. The plant-speci c EgTCP15 demonstrated upregulation during biotrophic phase before declining but showing an opposite expression pattern under abiotic stress.

Conclusions
Together from the results presented above, we were able to recognize TF genes that were regulated during switching of the fungal mode of infection. With this rst analysis of oil palm RNA-seq data encoding TFs, six common major families of TFs were identi ed responsible in promoting oil palm defense response against G. boninense attack which include MYB, bHLH, AP2/ERF, NAC, bZIP and WRKY. Reported genes involved in cell wall modi cation, ROS-mediated signaling, PCD and plant innate immunity were all differentially expressed indicating active regulation of oil palm defense response against the hemibiotrophic G. boninense. The biotrophic and necrotrophic infection phases of G. boninense was further con rmed through gene expression of biotrophy-speci c, EgJUB1 and necrotrophy-speci c, EgERF113. Our nding is the rst reporting EgJUB1 as a potent master regulator based on its positive interaction with the imperfect palindromic SNBE consensus sequence which may promote branches of biotrophy-associated defense mechanisms including cell wall strengthening and HR-mediated defense responses. In addition, EgERF113 is the rst AP2/ERF TF reported to modulate multifaceted defense mechanisms through binding to GCC-box and DRE/CRT motifs during necrotrophic phase. Binding to these motifs result in transcriptional upregulation of PR and calcium-responsive genes. Based on our ndings, a proposed defense mechanism inferring oil palm against hemibiotroph G. boninense during biotrophic and necrotrophic infection phases is illustrated as in Fig. 6. The information presents promising rst step in recognizing the downstream target defense-related genes regulated by the infection phase-speci c TFs.

Plant materials and fungal treatment
Four-months old of oil palm seedlings (Elaeis guineensis Jacq. Dura x Pisifera) were purchased from Sime Darby Plantations, Banting, Selangor, Malaysia. Pathogenic Ganoderma boninense strain PER71 was obtained from GanoDROP Unit, Biology Division, Malaysian Palm Oil Board (MPOB). Arti cial infection of oil palm seedlings with G. boninense using rubber wood blocks (RWBs) was carried out following previous study [5]. Control (C) was set as seedlings without treatment. Two different treatments were carried out; mock treatment (MT) consisted of oil palm seedlings with bare RWBs while Ganoderma treatment (GT) involved oil palm seedling treated with Ganoderma-inoculated RWBs. Destructive sampling was performed on two pooled biological replicates of oil palm seedlings at different days post inoculation (3, 7 and 11 d.p.i). Each biological replicate consisted of pooled RNA provided equally from six constituent seedlings. Instead of mathematical averaging of individual sample, biological averaging is more cost e cient and commonly practiced in the attempt of reducing high biological variability among samples in RNA-seq studies [99]. Pooling bias can be reduced by using three to eight biological individual samples per pool with two pools per treatment group [100].

RNA Extraction And DEGs Analysis Of TFs
Total RNA of all samples was extracted following the method reported in Bahari et al. [5]. The extracted RNA was used in all subsequent experiments. A high-throughput NGS data analysis was performed as described in Bahari et al. [5]. Sequences that are reproducible in both pooled biological replicates were chosen to eliminate biased pro ling of transcripts due to manipulations stages during library construction [101]. Alteration of gene expression pro le was analyzed by comparing genes expressed from control with GT samples. DEGs were further evaluated following stringent cut-off values of log 2 FC ≥ |1.0| (corresponding to 2-fold or more upregulation/downregulation) and P-value < 0.01 [5,102]. Comparative analysis was conducted between two biological replicates of control and GT samples at all time points. DEGs of TFs that met the cut-off values were clustered according to upregulated and downregulated genes. Transcripts that were identi ed in both pooled biological replicates were further analyzed for identi cation of DEGs. DEGs of a few stress-related genes involving cell wall modi cation, ROS production and PCD were mined from NGS data analysis.
Validation By Quantitative Real Time Pcr (qPCR) qPCR was performed using qPCR Green Master Mix LRox (2X) (Biotechrabbit GmbH, Germany). Stability of ve endogenous controls; EgGAPDH2, EgNADH5, EgMSD, EgUBQ, and Egß-actin were tested across treated and control samples. The qPCR analysis was performed using Bio-Rad CFX Manager™ Software version 3.1. Expression levels of all target genes were normalized with the expression level of three most stable reference genes which were EgGAPDH2, EgNADH5 and Egß-actin. Real-time PCR was carried out on control (C), mock-treated (MT) and Ganoderma-treated (GT) oil palm root samples at 3, 7 and 11 d.p.i. The data included three biological replicates of root samples of oil palm seedlings. Comparative analysis of expression levels was expressed as fold change ± standard error of mean (SEM) of three individual technical replicates at P-value < 0.01. Signi cant differences of expression levels between test groups were determined using one-way analysis of variance (ANOVA) analysis followed by Tukey's test. The primers used for the qPCR are listed in Table 1. Yeast One-hybrid (Y1H) Assay Coding regions of EgJUB1 and EgERF113 (500 ng total RNA) anking SMART sequences were ampli ed from GT samples at 3 and 11 d.p.i, respectively, using Q5® Hot Start High-Fidelity 2x Master Mix (New England Biolabs). Puri ed SMART-EgJUB1 and -EgERF113 were further ampli ed by long-distance PCR using Advantage® 2 PCR Mix (Takara Bio). Putative tandem repeats of target baits fragments (NACBS, SNBE1, SNBE2, GCC-box and DRE/CRT) as well as the respective mutated fragments were cloned into pAbAi vector and integrated into yeast genome. Minimal inhibitory concentration (MIC) of Aureobasidin A (AbA) for each bait-and mutant-reporter yeast strain was determined.
Transformation control, p53 was plated on SD/-Leu/AbA 200 . Plates were incubated at 30 °C for 3 days.
Conformation of positive clones was performed by colony PCR using Matchmaker Insert Check PCR Mix 2 (Takara Bio) and sent for sequencing using T7 promoter primer. Single colony of positive clone was cultured in synthetic dropout (SD) medium lacking leucine (SD/-Leu) broth to an optical density (OD 600 ) of 0.1 and diluted in ten-fold dilution series. From each dilution, 10 µL of yeast culture was spotted on SD/-Leu and SD/-Leu/AbA plates. Oligonucleotides and primers used in Y1H assay are listed in Table 2.

Isolation Of EgJUB1 And EgERF113 Nuclear Protein
Single colony of positive clones from Y1H assay was incubated in 5 mL SD/-Leu broth with overnight shaking at 30 °C. Yeast culture was transferred into 45 mL SD/-Leu broth and further grown with overnight shaking(18-20 hrs) at 30 °C until OD 600 reached 0.8-1.0. Yeast cells were pelleted by centrifugation at 1,000 g for 5 min at 4 °C. Pellet was washed with 30 mL ice-cold sterile ultrapure water and centrifuged at 1,000 g for 5 min at 4 °C. Pellet was immediately ash-frozen in liquid nitrogen and ground into ne powder. Extraction of nuclear protein was carried out using NE-PER™ Nuclear and Cytoplasmic Extraction Reagents (Thermo Scienti c). EgJUB1 and EgERF113 nuclear extracts were stored at -80 °C.

Electrophoretic Mobility Shift Assay (EMSA)
Double-stranded of sense and anti-sense oligonucleotides were biotin-labelled using Biotin 3' End DNA Labeling kit (Thermo Scienti c). Oligonucleotides used are listed in Table 2. Binding reaction system of EMSA was prepared using LightShift EMSA Optimization and Control kit (Thermo Scienti c). The binding mixture was resolved in 6% non-denaturing polyacrylamide gel in 0.5X Tris-borate-EDTA (TBE) buffer and transferred to a positively charged nylon membrane. The membrane was cross-linked at 120 mJ/cm 2 for 1 min. The protein-DNA complexes were visualized by LightShift® Chemiluminescent EMSA kit (Thermo Scienti c) according to manufacturer's protocol.

Subcellular Localization
The open reading frames of EgJUB1 and EgERF113 lacking stop codon were ampli ed using KAPA HiFi HotStart ReadyMix PCR kit (Thermo Scienti c). The gene-speci c primers with CACC anking at 5'end of forward primer are listed in Table 3. The PCR products were transferred into the gateway pENTR/D-TOPO entry vector using the pENTR™ Directional Topo Cloning Kit (Thermo Fisher Scienti c). TOPO Cloning Reaction was transformed into One Shot® TOP10 Chemically Competent cells (Thermo Fisher Scienti c) to generate entry clone and sequence veri ed. The expression clones for EgJUB1 and EgERF113 were generated using Gateway LR Clonase II Enzyme Mix (Thermo Fisher Scienti c) by recombination of entry clones into Gateway-compatible destination vector consisting double CaMV 35S promoter (2 × 35S) of pMDC85, respectively. The construction of pMDC85 vector without ccdB gene and insert was used as a negative control. The LR reaction was transformed into One Shot™ OmniMAX™ 2 T1® Chemically Competent cells (Thermo Fisher Scienti c).

Statistical Analysis
For RNA-seq data analysis, DEGs were determined following cut off-values of log 2 FC ≥ |1.0| and P-value < 0.01. Expression levels of each gene from qPCR analysis were normalized by three reference genes; EgGAPDH2, EgNADH5 and Egß-actin. Data was presented as mean ± standard error of mean (SEM) of three independent technical replicates. Differences of expression level between samples at different time points to control and between group of treatments (MT and GT) were determined by using one-way ANOVA analysis followed by Tukey's test. Signi cantly different expression levels as compared to the control were measured according to **P < 0.01, ***P < 0.001 and ****P < 0.0001. ns is de ned as not signi cant. All graphs were generated and analyzed using GraphPad Prism version 5.0 (GraphPad software Inc., USA). Authors' contributions SNA conceived and designed the research plan. NMS performed most of the experiments and prepared the manuscript. MNB contributed in RNA-seq data analysis. AMA and NAS provided technical assistance in analyzing assays, Y1H and EMSA. IAS provided fungal material and assisted in arti cial inoculation.
SNA complemented in the writing of the manuscript. All authors have read and approved the nal version of the submitted manuscript.  (0) to blue (4). For RNA-seq data analysis, two biological replicates which each consisted of pooled RNA provided equally from six constituent seedlings were used. Each heatmap data was constructed using average of pooled biological replicates. Pairwise comparison of RNA-seq data was evaluated according to cut-off values of log2 fold change (FC) ≥ |1.0| and P-value < 0.01.

Figure 2
Expression patterns of transcription factors in response to Ganoderma boninense infection at different time points. The expression patterns of each gene was normalized by three most stable reference genes; EgGAPDH2, EgNADH5 and Egß-actin expression levels. Real-time PCR was carried out on control (C), mock-treated (MT) and Ganoderma¬-treated (GT) oil palm root samples at 3, 7 and 11 days post inoculation (d.p.i). The data included three biological replicates of root samples of oil palm seedlings.
Error bars represent the mean ± SEM of three technical replicates of each sample. The statistical analyses were performed by comparing expression levels of different treatments at all time points to control using one-way ANOVA analysis followed by Tukey's test. Signi cantly different expression level as compared to control are measured according to **P < 0.01, ***P < 0.001 and ****P < 0.0001. ns is de ned as not signi cant. Proposed defense mechanisms of oil palm seedlings against hemibiotroph Ganoderma boninense during biotrophic and necrotrophic infection phases. Based on the NGS data on differentially expressed genes encoding TFs, two oil palm transcription factors (TFs), EgJUB1 and EgERF113 were discovered to regulate speci cally under biotic stress during biotrophic and necrotrophic phases, respectively. The EgJUB1 TF binds to SNBE motif and directly regulates EgGAMYB and EgHSFs. The EgERF113 TF binds to GCC-box and DRE/CRT motifs which promotes PR and calcium responsive genes. The oil palm defense pathways are identi ed based on established defense mechanisms and new ndings from this study. Black arrows represent direct regulation by encoded proteins/genes; blue arrows with broken line suggest highly probable defense regulatory pathways; brown arrows indicate activation of downstream targets through binding to respective motifs. The question marks highlight the gaps in oil palm defense mechanism that need to be further elucidated. Ca: calcium; CESA: cellulose synthase; CSLD: cellulose synthase-like D; DAMPs: damage-associated molecular patterns; DRE/CRT: dehydration response element/C-repeat; EIN3: ethylene insensitive 3; ET: ethylene; ETI: effector-triggered immunity; HR: hypersensitive response; HSFs: heat stress transcription factors; HSP: heat shock protein; JA: jasmonic acid; NPR1: NONEXPRESSOR OF PATHOGENESIS-RELATED GENE 1; PR: pathogenesis-related; PRRs: pattern recognition receptors; PTI: PAMP-triggered immunity; secondary wall NAC binding element (SNBE); SA: salicylic acid; TSS: transcription start site.