Biochemical and Structural Insights into the Mechanism of DNA Recognition by Arabidopsis ETHYLENE INSENSITIVE3

Gaseous hormone ethylene regulates numerous stress responses and developmental adaptations in plants by controlling gene expression via transcription factors ETHYLENE INSENSITIVE3 (EIN3) and EIN3-Like1 (EIL1). However, our knowledge regarding to the accurate definition of DNA-binding domains (DBDs) within EIN3 and also the mechanism of specific DNA recognition by EIN3 is limited. Here, we identify EIN3 82–352 and 174–306 as the optimal and core DBDs, respectively. Results from systematic biochemical analyses reveal that both the number of EIN3-binding sites (EBSs) and the spacing length between two EBSs affect the binding affinity of EIN3; accordingly, a new DNA probe which has higher affinity with EIN3 than ERF1 is also designed. Furthermore, we show that palindromic repeat sequences in ERF1 promoter are not necessary for EIN3 binding. Finally, we provide, to our knowledge, the first crystal structure of EIN3 core DBD, which contains amino acid residues essential for DNA binding and signaling. Collectively, these data suggest the detailed mechanism of DNA recognition by EIN3 and provide an in-depth view at molecular level for the transcriptional regulation mediated by EIN3.

EIN3/EIL1 are the critical transcription factors in the ethylene-initiated signaling network.The Ecker Group first identified the loss of ethylene-mediated effects in mutants of the Arabidopsis EIN3 gene [25].In ethylene signaling pathway, EIN3 regulates the expression of its direct target genes, such as ETHYLENE RESPONSE FACTOR1 (ERF1) and HOOKLESS1 (HLS1) [24,26].In addition to its essential roles in the ethylene signaling pathway, EIN3 also participated in the crosstalk among different hormonal pathways (Fig 1A).We previously revealed that EIN3/EIL1 mediate jasmonate-ethylene (JA-ET) signaling synergy in defense against necrotrophic pathogens [27], while EIN3 also plays a key role in the JA-ET antagonism in apical hook development by interacting with MYC2 (JA-activated transcription factor) [28,29].In the transition from skotomorphogenesis to photomorphogenesis, EIN3/EIL1 cooperate with PIF1 to prevent photo-oxidative damage and optimize de-etiolation of Arabidopsis seedlings, and protochlorophyllide oxidoreductase A and B (PORA/B) are direct target genes of EIN3 [30,31].In plant innate immunity, EIN3/EIL1 directly repress the SALICYLIC ACID INDUCTION DEFICIENT2 (SID2) expression to decrease salicylic acid (SA) level and downregulate pathogen/microbe-associated molecular patterns (PAMP) defenses, while EIN3 (potentially EIL1) direct induces FLS2 expression to promote PAMP defenses [32,33].Furthermore, EIN3 directly targets ESE1 (At3g23220), and ESE1 regulates salt-related genes [34].We also found that EIN3 is a senescence-associated gene that advances the progression of age-dependent leaf senescence [35].Moreover, EIN3 is not only degraded through EBF1/2-dependent proteolysis, but also modulates EBF2 gene expression via the elaborate feedback regulation [36].In all, EIN3/EIL1 act as a signaling hub that integrates the plant signals to adapt complex environment.
As a highly important transcription factor, one crucial function of EIN3 is to recognize and bind to specific target genes.The Ecker group first showed that EIN3/EILs family proteins regulate gene expression by binding directly to a primary ethylene response element (17); recently, they further identified the repertoire of EIN3 target genes, using ChIP-Seq and mRNA-Seq during a timecourse of ethylene treatment [26].However, a central, yet unanswered question is how EIN3 recognizes and binds to a wide variety of target DNA.Previous studies have demonstrated that the N-terminal region of EIN3 (for instance, residues 1-359, 1-314, 141-352 used by different studies, and also its homolog in tobacco TEIL 82-302) is responsible for DNA binding [24,30,32,37], while the optimal DNA-binding domain (DBD) and the core DBD of EIN3 are still unknown.In addition, EIN3 was shown to bind specifically to the fragment of ERF1 promoter (−1213 to −1178) which has two palindromic repeats flanking one central EIN3-binding site (EBS) [24]; yet except for ERF1, no other EIN3 target bearing palindromic repeat sequences has been reported.Hence whether or not such palindromic sequence is necessary for DNA binding is unclear.Furthermore, although the Yamasaki et al. have reported the solution structure of EIL3 major DBD [38], EIL3 subgroup are distinct from other EIL family members and may have functions in regulating sulfur responses instead of mediating ethylene responses [39].Thus, a crystal structure of EIN3 would still be desired for the inspection-at atomic level-of residues essential to the function of EIN3 [25,32,40].Consequently, such limited knowledge regarding to the molecular mechanism of the interaction between EIN3 and DNA targets still significantly impede an in-depth view of transcriptional regulation initiated by EIN3.
To better understand the interactions between EIN3 and its DNA targets, we herein perform systematic and detailed biochemical studies and present, to our knowledge, the first crystal structure of EIN3.We identify the optimal and core DBDs of EIN3, and show that both the number of EBSs and the length of spacing between two EBSs have great impact on DNA-binding preference.Furthermore, we show that the palindromic repeats in ERF1 are not necessary for DNA binding.Finally, a 1.78 Å crystal structure of the core DBD of EIN3 [Protein Data Bank (PDB) accession number: 4ZDS] reveals the novel folding of DBD for the EIN3/EIL family proteins and offers insights into the mechanistic details of key amino acid residues involved in the biological functions of EIN3.
For protein expression, Transetta (DE3) competent cells (Beijing TransGen Biotech Co., Ltd) were transformed with the corresponding vectors described above.Cells were grown at 37°C in LB medium with 50 μg/ml kanamycin.Protein expression was induced at 20°C when the culture reached an OD 600 of 0.6 by the addition of isopropyl-β-D-1-thio-galactopyranoside to a final concentration of 0.1 mM and cells were allowed to grow for 16 additional hours.For seleno-methionine (SeMet) derivative of EIN3, cells were grown overnight at 37°C in LB medium, and diluted twice at an 1:100 ratio in M9 medium supplemented with amino acids (L-Lys, L-Phe and L-Thr at 100 mg/l; L-Iso, L-Leu and L-Val at 50 mg/l) and 50 mg/l L-Selenomethionine, and further allowed to grow to OD 600 of 0.5.Next, expression of protein was induced similarly as described above.
Cells were then harvested by centrifugation at 4,000g for 35 min and resuspended in lysis buffer (100 mM Na 2 HPO 4 /NaH 2 PO 4 , pH 6.5, 300 mM NaCl, 2 mM DTT and 1 mM PMSF).All subsequent steps were conducted either on ice or at 4°C.Cells were lysed by sonication and cleared by centrifugation for 30 min at 15,000g.The supernatant was purified by HiLoad HisTrap (Ni-NTA) column (GE Healthcare) and Superdex 75 PG size-exclusion column (GE Healthcare) equilibrated in 40 mM Na 2 HPO 4 /NaH 2 PO 4 , pH 6.5, 90 mM NaCl, and 1 mM DTT, using an ÄKTA pure system (GE Healthcare).Peak fractions were analyzed by SDS-PAGE, pooled, concentrated, flash frozen in liquid nitrogen, and stored at − 80°C.Protein concentrations were determined from UV absorbance at 280 nm by 8453 UV-Vis (Agilent).Numbering of EIN3 residues is based on UniProt accession number O24606.
Size exclusion chromatography of purified EIN3 82-352 protein under different salt concentration were performed on Superdex 75 PG size-exclusion column (GE Healthcare) using an ÄKTA pure system (GE Healthcare).Superdex 75 PG size-exclusion column (GE Healthcare) was equilibrated in 10 mM Tris-HCl, pH 7.5, 1 mM DTT, and three different NaCl concentrations (100 mM/300 mM/500 mM), respectively.All steps were conducted at 4°C.

Mass spectrometry
Purified EIN3 82-352 protein was run on a 12% SDS-PAGE gel and the band corresponding to EIN3 was sliced.Protein was then digested with trypsin and analyzed using Liquid chromatography-tandem mass spectrometry of the Analytical Instrumentation Center (Peking University).

Oligonucleotides
All DNA oligonucleotides were synthesized and further purified with HPLC by the Shanghai Sangon Biotechnology Company (Shanghai, China).DNA concentrations were determined from UV absorbance at 260 nm by 8453 UV-Vis (Agilent).The forward strands were labeled with the Biotin 3' End DNA Labeling Kit (Pierce).The DNA duplexes used for EMSA experiments were generated by heating the mixture of complementary oligonucleotides at 95°C for 5 min and slowly cooling to 20°C over 2h in the annealing buffer of 10 mM Tris-HCl (pH 7.5), 100 mM NaCl.All oligonucleotide sequences are listed in S1 Table .The annealed probes were further purified using Thermo DNAPac PA 200 RS column on Agilent 1260 HPLC system [with a NaCl gradient of 0.4-1.0M in 10 mM Tris-HCl (pH 7.5) buffer over the course of 12.5 min].As the NaCl gradient was applied, 2EBS-S10 reverse strand was first eluted, followed by forward strand and finally the desired 2EBS-S10 dsDNA probe.The corresponding fractions were collected and the buffer was exchanged to the annealing buffer using Bio-Rad Micro Bio-Spin P-6 Gel Columns.Denaturing PAGE and Native PAGE were used to validate the purified single-stranded and double-stranded probes.Next, EMSA experiments were performed for K d determination using both probes before and after purification (so as to compare the probes).The K d values for the two probes were very close (deviations were less than 5%).To further check the efficiency of the annealing process, we analyzed the quality and purity of the resulting double-stranded probes.The result showed the content of annealed dsDNA probe was ~95%.Furthermore, the EIN3 truncation 82-352 failed to display noticeable binding to ssDNA probes.Therefore, double-stranded probes in the following EMSA experiments were used directly after annealing without further purification.

Electrophoretic mobility shift assays
EMSA was performed using the LightShift Chemiluminescent EMSA Kit (Pierce).A typical binding reaction was performed in a total volume of 20 μL by incubation of an appropriate amount of purified EIN3 protein with 1 nM biotin-labeled DNA probe, in 1X binding buffer [10 mM Tris-HCl, pH 7.5, 40 mM KCl, 3 mM MgCl 2 , 1 mM EDTA, 10% glycerol, 1 mM DTT, 1 μg poly(dI-dC)] at room temperature for 30 min.The binding products were resolved on 6% polyacrylamide gel run in 0.5 X Tris-borate-EDTA.Although the absolute amount of probes in EMSA binding systems (20 μl) were the same for different probes (as determined by 8453 UV-Vis), the labeling efficiency may vary for each probe, and hence the intensity on the gel.Signal intensities of bands on the gel were visualized using ChemiDoc MP Imaging System (BIO-RAD).Therefore, DNA binding of EIN3 to different probes was determined by the percentage of the shifted bands, fraction bound, in the equation below.Signal intensities of bands were measured by Image Lab Software (Bio-Rad).

Circular dichroism analysis
Circular dichroism (CD) measurements of purified recombinant 82-352 and 174-306 were performed on a Chirascan spectropolarimeter (Applied Photophysics, Leatherhead, UK).Protein samples were prepared in 10 mM PBS pH 7.0 at a concentration of ~0.2 mg/ml.Spectra were recorded from 190 to 250 nm.The software CDNN [41] was used to calculate alpha-helix content from CD data.

Size exclusion chromatography with multi-angle light scattering and dynamic light scattering
EIN3 truncations were run at 1 ml/min in 10 mM Tris-HCl, pH 7.5, 100 mM NaCl, and 1 mM DTT on a Superdex 200 analytical size-exclusion column (GE Healthcare) attached to a light scattering diode array (Dawn Heleos II, Wyatt Technology, UK) and dynamic light scattering instrument (DynaPro NanoStar, Wyatt Technology, UK).This allows for the accurate measurement of protein molecular mass without bias from its shape by calculating the ratio of light scattering to the differential refractive index of a sample.Additionally, the DLS can measure the sizes of macromolecules as hydrodynamic radius (Rh).The reproducibility of different runs was assessed using a BSA standard before and after each experiment to see if there was variability in either calculated molecular weight or elution volume.No detectable change was observed over these experiments.All steps were conducted at 4°C.Size exclusion chromatography with multi-angle static light scattering (SEC-MALS) experiments of EIN3 82-352 were also performed with an Agilent ProSEC 300S column of 5 μm particle size and a 7.5 X 300 mm column size on Agilent 1260 Infinity Multi-Detector GPC/ SEC System at 1 ml/min in low salt buffer (10 mM Tris-HCl, pH 7.5, 100 mM NaCl, and 1 mM DTT) and high salt buffer (10 mM Tris-HCl, pH 7.5, 300 mM NaCl, and 1 mM DTT), respectively.The samples were preserved at 4°C before injection; the SEC and detector were performed at room temperature.

Protoplast Transfection Assay
The constructs containing 36 bp 1EBS-S10-36, 2EBS-S10-36 or ERF1 sequence respectively in the pGreenII 0800-LUC vector were used as reporter plasmids [42].The renilla luciferase (REN) gene under the control of 35S promoter in the pGreenII 0800-LUC vector was used as internal control.The construct containing coding sequence of EIN3 in pGreen II 62-SK vector was used as effector plasmid [42].All clones were confirmed by Sanger sequencing.The sequences of 36 bp 1EBS-S10-36, 2EBS-S10-36 and ERF1 used in transient transcriptional activity assay are listed in S1 Table .Arabidopsis mesophyll protoplast was prepared as described [43].5 μg of reporter plasmid together with 10 μg of indicated effector plasmids was cotransformed into the protoplast.The Dual-Luciferase Reporter Assay System (Promega) was used to measure the LUC and REN activities after 14 hour incubation.The activity of LUC and REN are measured sequentially from a single sample on a GLO-MAX 20/20 luminometer (Promega).The ratio of LUC/REN is calculated to indicate the final transcriptional activity.The Relative Fluorescence LUC/REN represents the relative LUC/REN ratio normalized by the control group.Data are means (±SD) of three biological replicates.

Crystal growth and data collection
174-306 SeMet derivative and native crystals were grown from a 1+1 μl sitting drop of, respectively, 10 mg/ml protein solution and crystallization buffer (0.1 M HEPES, pH 7.5, 25% w/v Polyethylene glycol 3,350).Needle-shaped crystals grew to their maximum size within one week and were immediately frozen in crystallization buffer supplemented with 20% glycerol.Diffraction data were collected at 100K at the Shanghai Synchrotron Research Facility beamline BL-17U, China.

Structure determination and refinement
All diffraction data were processed using HKL2000 program suite [44].The initial structure was solved using the PHENIX AutoBuild [45] using SAD data measured at the Se K absorption edge on SeMet-EIN3.Native structure was solved by molecular replacement with SeMet-EIN3 as model structure using PhaserMR [46].Refinement was performed using REFMAC [47,48] of the CCP4 program suite [49].Visualization and manual adjustments were done using Coot [50].Molecular graphics figures were generated using PyMOL (PyMOL Molecular Graphics system, Version 1.3 Schrodinger, LLC).Atomic coordinates and structure factors of EIN3 were deposited in the PDB with the code 4ZDS.A full summary of the data processing and refinement statistics is shown in Table 1.
Previous studies have established the highly conserved N-terminal regions of EIN3 (for instance, residues 1-359, 1-314, 141-352 and its homolog in tobacco, TEIL 82-302) to be competent for DNA binding [24,30,32,37].However, the exact location of EIN3 DBD was unclear from its primary structure, since the EIN3/EIL proteins show no sequence similarity to other known DBDs.Therefore, we first performed systematic comparisons among different truncations of EIN3 for accurate definition of its DBD.To explore the optimal and core DBDs of EIN3, we made a series of EIN3 truncations with gradually truncated N-and C-terminus, based on secondary structure prediction of EIN3 [using the PSIPRED [52]   Among the 11 EIN3 truncations we tested, 82-352 contains the most conserved sequences within the EIN3/EIL1 family and has the highest DNA-binding affinity; 174-306 contains the entire continuous conserved amino acid residues in the DBD and also has the ability to bind DNA (Fig 1

and S3 Fig).
Therefore, we define EIN3 82-352 and 174-306 as the optimal DBD and the core DBD of EIN3, respectively.Furthermore, alpha-helix was the major secondary structure for the optimal and core DBDs, with a higher proportion for the latter truncation determined by circular dichroism (CD) spectrum analysis (S3F Fig) , implying that the alphahelix structure was important for EIN3 to bind to its target DNAs.As the optimal DBD of EIN3 has the highest DNA-binding affinity, all experiments below refer to EIN3 82-352 unless otherwise stated.

Specific inverted repeat sequences within DNA share higher affinity with EIN3
It has been shown that EIN3 binds its DNA target as a homodimer [24].However, whether or not EIN3 can form a dimer in the absence of DNA is not known.To address this question, we performed size exclusion chromatography with multi-angle static light scattering (SEC-MALS) analysis and measured the molecular weight (Mw) of EIN3 82-352.MALS results showed that EIN3 82-352 (predicted Mw: ~31.3 kDa) had a measured Mw of ~63.1 kDa in solution, consistent with a dimer that was independent of DNA binding (Fig 2A).This dimer is also stable under higher salt concentrations (S4 Fig) .In addition, EIN3 was a monomer under denaturing condition, suggesting that dimerization of EIN3 was formed through non-covalent interactions.
EIN3 has been reported to recognize EBS and hence binds DNA in a sequence-specific manner [24].To study how EIN3 recognizes its DNA targets, we first performed dynamic light scattering (DLS) analysis of EIN3.Our results showed that the hydrodynamic radius (Rh) of EIN3 82-352 was ~4.40 nm, which appeared to be of similar size of a 26-27 bp B-form duplex DNA (S5A Fig) .Therefore, we designed a series of 28 bp probes to investigate the effect of specific sequences and their relative orientation within DNA to EIN3 binding.As shown in Fig 2B , EIN3 bound readily to 2EBS-S10 and 2EBS-S10-2, which contain two known EBSs (ATGTAT and ATGTAC, respectively) [24,30].However, when we "flipped" the two EBSs, DNA binding was significantly reduced: for instance, binding of EIN3 to 2EBS-S10-R was around 11-fold weaker than 2EBS-S10 [dissociation constants (K d ): ~1.24 μM for 2EBS-S10-R and ~0.11 μM for 2EBS-S10] (Fig 2C and 2D).In addition, removing one of the two inverted EBSs from Probe 2EBS-S10 again significantly decreased binding affinity (K d of ~0.69 μM for 1EBS-S10), while removing both EBSs further decreased DNA-binding ability (K d of ~2.28 μM for 0EBS-S10) (Fig 2C and 2D).This was in contrast to mutations in flanking sequence of two EBSs, where no apparent effect on DNA binding was observed (S5B Fig) .Taken together, these results suggest that EIN3 homodimer specifically recognizes two inverted EBSs within target DNA sequence, and the orientation of such EBSs is also important for optimal binding.

Spacing between EBSs impacts binding affinity of EIN3
The presence of two inverted EBSs have been found for many in vivo targets of EIN3 (PORB01, SID2, EBF2, ESE1, and etc) [30,32,34,53] (Fig 3A), and such organization of EBSs presumably allows optimal binding of EIN3, also in agreement with our biochemical results shown above.Despite the fact that many EIN3 targets contain two inverted EBSs, a closer examination of such natural promoters reveals that the spacing between the two EBSs differ from each other.For instance, ESE1 has an 8 bp spacing between two EBSs while SID2 has a 10 bp spacing (Fig 3A).To investigate the effects of different spacing on EIN3 binding, we performed EMSA experiments using probes that have two invariant inverted ATGTAT EBSs but contain 5-15 bp spacing in between.As shown in Fig 3B and 3C, it was apparent that the spacing between the two EBSs could significantly affect the DNA-binding ability of EIN3: strongest binding showed up when the two EBSs were separated by a 10 bp spacing and gradually decreased when the spacing was either shortened from 10 bp to 5 bp or extended from 10 bp to 15 bp (Fig 3B and  3C; S6 Fig) .In addition, the binding constants could vary up to 15 fold between 2EBS-S10 and 2EBS-S5 under the same experimental conditions, which was even greater than the difference between EIN3 binding to two and one EBS-containing sequences (2EBS-S10 and 1EBS-S10) and comparable to the difference between EIN3 binding to two and none EBS-containing sequences (2EBS-S10 and 0EBS-S10) (Fig 3D and 3E).Even the K d of 2EBS-S8 was 3-fold higher than 2EBS-S10, further indicating the significant impact of spacing on DNA binding.To rule out the potential roles of the spacing sequence, we made mutations in the spacing sequences and found that EIN3 binding was not affected (S6A and S6B Fig) .Moreover, we also showed that the effect of spacing on the preferential binding of EIN3 was independent on the

Palindromic repeat sequence is not necessary for EIN3 binding
Previous studies reported that palindromic repeats could also be important for DNA binding of EIN3 [24].For example, one of the prominent EIN3 target, ERF1 contains two palindromic repeats flanking one central EBS; although in vivo targets of EIN3 can have either one or two EBS, such nearby palindromic repeats appears to be unique for ERF1.To determine whether the palindromic repeats are necessary for DNA binding, we systematically deleted/mutated the palindromic sequences from a 36 bp fragment of ERF1 promoter, which was widely used in previous studies [24,28].Deletions of the 3' palindromic repeat in ERF1 (resulting in a shorter probe) did not affect binding ability (Fig 4A ); additionally, mutations to the 5' palindromic repeat in the DNA fragment did not affect binding ability either (Fig 4B).Moreover, binding affinity of EIN3 to the 36 bp ERF1 was significantly lower than the 36 bp 2EBS-S10-36 probe (which contains two inverted EBS repeats) (Fig 4C).As a result, we conclude that palindromic repeats in ERF1 promoter are not necessary for EIN3 binding; it is the EBS that is important for the interaction with EIN3.
To further validate the correlation between EIN3 DNA-binding affinity and EIN3 transcriptional activity, we performed transient transcriptional activity assays using Arabidopsis protoplast (S7 Fig)  4C), the two EBS-containing plasmid displayed higher LUC/REN ratio than the two one EBS-containing plasmid (S7B Fig) .Hence, higher affinity of EIN3 to DNA also demonstrates higher transcriptional activation.

Crystal structure of the core DNA-binding domain of EIN3
The core DBD (174-306) of EIN3 we identified contains the entire continuous conserved amino acid residues within the EIN3/EIL1 family (Fig 5A ), and also contains BD III, BD IV and the proline-rich region.Moreover, the core DBD is capable of binding to various EIN3 DNA targets, albeit with lower affinity comparing to the optimal DBD (Fig 1E and S6H Fig).
Furthermore, although the solution structure of EIL3 major DBD is solved [38], whether or not EIL3 can bind to EBS-containing DNAs is still under debate [24,38]; more importantly, the physiological function of EIL3 in Arabidopsis is distinct from other EIL family members [39].Thus, we purified and crystallized the core DBD to gain mechanistic insights of EIN3 at atomic resolution.
Firstly, we used the solution structure of EIL3 (PDB code 1WIJ) as the model for molecular replacement, however no solutions could be found.We then solved the structure of the core DBD using single-wavelength anomalous diffraction (SAD) on a seleno-methionine derivative; the resulting model was used for molecular replacement on the native structure, which was solved to 1.78 Å (R work : 17.4% and R free : 21.8%) (Table 1) (PDB code 4ZDS).The structure revealed that the EIN3 core DBD exists as a monomer and is composed of six α-helices (α1-α6) and five short helical turns (η1-η5) (Fig 5A and 5B).In fact, the core DBD was also a monomer in solution in the absence of DNA (S8 Fig), which is in agreement with the previous NMR study [38].Presumably, this is due to the disruption of N-terminal amino acid residues that are important for dimerization.The core DBD structure also contains several amino acid residues (Lys245, Pro216 and Thr174) that are critical for EIN3 function (S9 Fig) .Structure similarity searches using the DALI server [54] showed that EIL3 is the only protein that shares structural homology to the DBD of EIN3, with no other homologs found in the Protein Data Bank.The N-terminal region (1-340) of EIL3 is quite conserved compared to EIN3 1-352 (~52.8%sequence identity), and the overall structure of EIN3 core DBD has a root-meansquare deviation (RMSD) of 3.32 Å with the solution structure of EIL3 [38] (S10 Fig) .Therefore, together with previous observations, we conclude that the structures of EIN3 and EIL3 represent a novel fold of DBD.

Discussion
EIN3/EIL1 proteins act as master regulators in plant signaling network, recognizing and binding to specific promoters (EBS-containing promoters) and thus modulating target genes expression.To date, numerous targets of EIN3 have been identified and majority of them are regulated by ethylene [26].However, the molecular mechanism of the interactions between EIN3 and its targets is still unclear.Using systematically varied recombinant proteins, we identify 82-352 fragment and 174-306 fragment as the optimal and core DBD of EIN3, respectively.We show that EIN3 homodimer binds to DNA probes containing two inverted EBSs; the EBS sequences are critical to allow specific and optimal binding of EIN3.Notably, our results reveal that the length of spacing between two EBSs can clearly impact binding affinity of EIN3, while the sequence of the spacing does not affect binding.And 10 bp spacing is optimal for EIN3 binding.In addition, palindromic repeats appear to be dispensable for DNA binding of EIN3, as demonstrated in the case of ERF1, one well-known target of EIN3.Furthermore, we solved a high resolution structure of the core DBD of EIN3, which provides the structural basis for the mechanistic views of several key amino acid residues involved in the biological function of EIN3.
Our systematic comparisons revealed that DNA-binding affinity decreased gradually from EIN3 1-352 to 35-352 and 46-352, whereas truncations 68-352, 82-352, 91-352, and 98-352 exhibited significantly increased DNA-binding ability (Fig 1C and 1D).Such observations hint that the DNA-binding activity of EIN3 could be autoinhibited by its N-terminal sequence (~1-50).In fact, autoinhibition of DNA-binding capacity has been observed for several other transcription factors: for instance, E26 transformation-specific transcription factor (Ets-1) is autoinhibited by a conformationally disordered serine-rich region (SRR), and phosphorylation of serines within the SRR has been shown to enhance such autoinhibition [55,56].In the case of EIN3, four Ser and one Thr residues are present within the N-terminal sequence (1-50), implying the possibility of being post-translationally modified and utilized for regulation.In addition to putative "self-regulation", DNA binding of EIN3 can also be regulated through interacting with other proteins.For instance, we reported in a previous study that MYC2 interacts with EIN3 and interferes with its DNA binding and transcriptional activities [29].In this study, we showed that EIN3 can form a homodimer in solution and such dimerization, which is independent of DNA binding events, appears to enhance DNA binding since dimerized EIN3 truncations (e.g.82-352) has better affinity than monomeric EIN3 truncations (e.g.174-306).Thus, one possibility is that MYC2 could interfere with the formation of EIN3 homodimer and hence interfere with DNA-binding ability of EIN3.
Previous studies, together with ours, have shown that EBS sequences are essential for DNA binding of EIN3 [24,37].As a matter of fact, among the many known targets of EIN3, both one and two EBS-containing targets have been reported (for example, SID2 and ESE1 have two inverted EBSs, while ERF1 and PORA02 have only one EBS) [24,30,32,34] (Fig 3A).Therefore, it will be interesting to see whether or not different target genes can be preferentially recognized and bound by EIN3 based on the number of EBS present in the promoter region, particularly given the fact that the cellular concentration of EIN3 proteins are constantly subjected to regulation (e.g. by EBF1/2) [18,19].Results from our EMSA experiments revealed that EIN3 bound to DNA probes bearing two EBSs 6-fold tighter than probes with one EBS (Fig 2C and 2D).Moreover, we found that the length, but not the sequence, of the spacing between two inverted EBSs was able to impact the binding affinity of EIN3 (Fig 3).As natural targets of EIN3 have varied length of spacing between two EBSs, such spacing length (and perhaps also the spatial phase between two adjacent EBS) may add one additional layer of complexity to the transcriptional regulation controlled by EIN3.Furthermore, an optimized probe sequence was identified and could be applied to better reflect the transcriptional activity of EIN3, i.e. the tandem copies of 2EBS-S10-36 could be fused with luciferase or GUS genes and be used in transient assay or stable transgenic lines, defined as enhanced EBS reporter system and replacing the previous 5×EBS::GUS version [27,28].In addition, our analyses on EIN3 target sequence and optimal DBD could also help to obtain EIN3/dsDNA complex structures in future.
Our work provides the crystal structure of EIN3 core DBD and reveals the structural basis of key domains and amino acid residues for EIN3 function.The core DBD 174-306 contains two essential basic domains for DNA binding: BD III (K238-K248) and BD IV (K265-K274).BD III and BD IV are highly positively charged, and are located on the surface of the truncation, but nearly on opposing sides of the truncation (Fig 5C).Therefore, it seems unlikely that both basic domains are simultaneously involved in direct DNA binding.Several amino acids that are critical for EIN3 function have been identified previously: for instance, a single mutation K245N shows a loss of ethylene-mediated effects [25]; P216S mutation abolishes regulation of the ethylene induced triple response, while remains fully functional in repressing plant disease resistance [32]; and T174 of EIN3 can be phosphorylated by the MKK9-MPK3/6 modules, although the role of such post-translational modification in ethylene signaling appears to be under debate [21].These amino acid residues are all present in our crystal structure: residue 245 belongs to α3 in the core DBD and falls within BD III, and the side chain of K245 forms two hydrogen bonds with the main chain carbonyl oxygen atoms of Q179 (α1) and L181 (between α1 and α2), respectively (S9A Fig) However, an EIN3/dsDNA o complex structure would be desired to unambiguously establish exact molecular interactions of these key amino acids.
An important question in plant signaling network is how EIN3/EIL1 regulate such a wide range of target genes.In this work, we studied in details the DNA-binding properties of EIN3 and revealed the mechanistic insights into the interplay between EIN3 and different DNA targets.These results, as well as the EIN3 structure presented here, will open new opportunities to the sequence-dependent gene regulation in plant signaling network.

Fig 1 .
Fig 1.The optimal and core DNA-binding domains of EIN3.(A) A Schematic model for EIN3 action in multiple signaling pathways.(B) Schematic structural diagram of full length and truncated EIN3 proteins.Acidic region, coil region, proline-rich region, basic domains, and asparagine-rich domain are represented by different color filled boxes.Highly conserved regions are also indicated by red boxes.(C) Representative EMSA results of different EIN3 truncations binding to a 3'-biotin labeled ERF1 probe.Specific EIN3 truncations used for each lane are shown in the upper panel and the fraction bound is shown in the lower panel.EIN3 114-352 displayed minimal binding under the current protein concentration, yet clearly bound to DNA at higher concentration (S3C Fig).In addition, the same amount of different EIN3 truncations was used in each PAGE-gel in order to have better visual comparisons, and similar results were obtained at other concentrations of proteins (S3A Fig)."B": bound; "F": free.(D) Relative binding data for 0.2 μM truncated EIN3 proteins with ERF1.The data were normalized to the fraction bound of 82-352 with ERF1; error bars represented the standard deviation from three independent experiments.(E) EMSA results of the core DBD (174-306), which retained the ability to bind to ERF1.doi:10.1371/journal.pone.0137439.g001 server] (Fig 1B, S2 Fig and S1 File).We chose a probe originating from the promoter sequence of ERF1 (-1213 to -1178), a well-established EIN3 target, for electrophoresis mobility-shift assay (EMSA).The results showed that EIN3 1-352 was capable of binding to the ERF1 probe, consistent with previous experiments using EIN3 1-359 (Fig 1C and 1D; S3A Fig) [24].The fraction bound decreased from EIN3 1-352 to 35-352 and 46-352, with the proteins expressed and purified to equally-well degree.Surprisingly, as the N-terminus was further truncated, (to residues 68, 82, 91 and 98), DNA-binding ability was fully restored.In fact, 82-352 exhibits the strongest DNA-binding ability among all truncations tested, even stronger than the longer form 1-352, which has an intact N-terminus.Further deletion of the N-terminal region to residue 114, 141 and 174 caused a sharp decline in DNA-binding ability, and shortening of the C-terminus to residue 306 also decreased DNA binding (Fig 1E and S3 Fig).Therefore, basic domains BD III, BD IV, and the proline-rich region in EIN3 are necessary for DNA binding, while the immediate C-terminal region of BD II and region between BD IV and BD V can further enhance the DNA-binding capacity.

Fig 2 .
Fig 2. Specific inverted repeat sequences within DNA share higher affinity with EIN3.(A) Analysis of EIN3 82-352 by SEC-MALS and SDS-PAGE.Measured molecular weight (Mw) of EIN3 82-352 was 63.13 kDa (± 0.474%), suggesting EIN3 existed as a homodimer in solution (Predicted Mw: monomer = 31.29 kDa; dimer = 62.58 kDa).Nevertheless, in SDS-PAGE a band corresponding to monomer was observed.(B) EIN3 binding to a series of DNA probes with varied EBS sequence.The sequences of DNA probes are listed in the lower panel.The (dark) red arrows indicate the position and direction of reported EBSs while the grey arrows represent the mutated EBSs.The names of different probes are given so as to reflect the number of EBS sequences, the length of spacing in between, and the relative orientation of EBSs, if possible.For example, probe "2EBS-S10" contains two inverted EBSs, which is separated by a 10 bp spacing.The fraction bound is shown right under the gel image.(C) Quantified binding data for 4 probes from (B) fitted with specific binding with Hill slope of GraphPad Prism 6.0.(D) Dissociation constants (K d ) for different probes.Results were obtained from three independent replicates, and the relative folds were normalized to the K d of 2EBS-S10.doi:10.1371/journal.pone.0137439.g002

Fig 3 .
Fig 3. Spacing between EBSs constrains specific EIN3 binding.(A) Reported natural promoters of EIN3 targets.EBSs are shown in red.The spacing sequences between two EBSs are underlined and marked with the spacing lengths.All listed EIN3 targets have been proved by EMSA.(B) DNA-binding of EIN3 to probes with varying length of the spacing between EBSs.The sequences of the probes are listed in the lower panel.NC represents "negative control", which is the free probe of 2EBS-S10 without the addition of any protein.(C) Relative binding data for 0.2 μM EIN3 82-352 to the probes in (B).The data were normalized to the fraction bound of 2EBS-S10; error bars represented the standard deviation from three independent experiments.(D) Quantified binding data and (E) dissociation constants for probe 2EBS-S5, 2EBS-S8, and 2EBS-S10.K d values were obtained from three independent experiments.doi:10.1371/journal.pone.0137439.g003 Three 36 bp probes (1EBS-S10-36, 2EBS-S10-36 and ERF1) were selected to drive LUC gene expression (S7A Fig).We monitored the LUC/REN ratio, which reflects in vivo EIN3 transcriptional activity upon EIN3 co-expression.Consistent with the DNA-binding results from our biochemical experiments (Figs 2B and

Fig 5 .
Fig 5. Crystal structure of EIN3 core DNA-binding domain (PDB code 4ZDS).(A) Sequence alignment of the core DBD of EIN3 proteins in different species produced by the Clustal Omega program.Sequences of the EIN3 proteins from Arabidopsis thaliana (AtEIN3, O24606), Phalaenopsis equestris (PeEIN3, Q711J2), Oryza sativa (OsEIN3, Q8W3M0), Triticum urartu (TuEIN3, M8A6N5), Zea mays (ZmEIN3, B6U809), and Nicotiana tabacum (TEIL, Q9ZWK1) are obtained from the UniProt database.Numbers above the sequences are for AtEIN3.Basic (His, Arg and Lys) and acidic (Asp and Glu) residues presented here are colored blue and purple, respectively.Colored boxes above the sequence alignment indicate helical structures of EIN3 core DBD.Identical and similar residues are marked by * and: respectively.BD III, BD IV and proline-rich region defined in Fig 1B are also indicated, and the sequence alignment includes the second highly conserved region in Fig 1B.(B) Crystal structure of EIN3 core DBD in two orientations (90°rotation along the indicated axis), which has the same coloring with the sequence alignment in (A).(C) Electrostatic surface potentials of EIN3 core DBD in the same orientation to (B).The proline-rich region, BD III and BD IV are indicated by blue, green and orange circles, respectively.doi:10.1371/journal.pone.0137439.g005 Proline 216 resides in the proline-rich region of EIN3 and is involved in a hydrophobic interaction network of the loop region (S9B Fig and S11 Fig).When Pro is replaced by the hydrophilic Ser residue, such interaction may be weakened.Interestingly, binding ability of EIN3 to the ERF1 promoter was not affected by the Pro-to-Ser mutation (S9C Fig).T174 belongs to α1 and its hydroxyl group forms a hydrogen bond to the side chain of E295 (α6) (S9D Fig).