The Arabidopsis KH-Domain RNA-Binding Protein ESR1 Functions in Components of Jasmonate Signalling, Unlinking Growth Restraint and Resistance to Stress

Glutathione S-transferases (GSTs) play important roles in the protection of cells against toxins and oxidative damage where one Arabidopsis member, GSTF8, has become a commonly used marker gene for early stress and defense responses. A GSTF8 promoter fragment fused to the luciferase reporter gene was used in a forward genetic screen for Arabidopsis mutants with up-regulated GSTF8 promoter activity. This identified the esr1-1 (enhanced stress response 1) mutant which also conferred increased resistance to the fungal pathogen Fusarium oxysporum. Through positional cloning, the ESR1 gene was found to encode a KH-domain containing RNA-binding protein (At5g53060). Whole transcriptome sequencing of esr1-1 identified altered expression of genes involved in responses to biotic and abiotic stimuli, hormone signaling pathways and developmental processes. In particular was an overall significant enrichment for jasmonic acid (JA) mediated processes in the esr1-1 down-regulated dataset. A subset of these genes were tested for MeJA inducibility and we found the expression of some but not all were reduced in esr1-1. The esr1-1 mutant was not impaired in other aspects of JA-signalling such as JA- sensitivity or development, suggesting ESR1 functions in specific components of the JA-signaling pathway. Examination of salicylic acid (SA) regulated marker genes in esr1-1 showed no increase in basal or SA induced expression suggesting repression of JA-regulated genes is not due to antagonistic SA-JA crosstalk. These results define new roles for KH-domain containing proteins with ESR1 unlinking JA-mediated growth and defense responses.


Introduction
Plants are subject to constant changes in their environment and rapid molecular responses to these are necessary for plant survival. Upon detection of abiotic or biotic stress a series of induced signalling pathways are activated, mediated by key signalling hormones such as salicylic acid (SA), jasmonic acid (JA) and abscisic acid (ABA), culminating in the expression of plant protectant and defense genes (reviewed by [1][2][3][4][5][6][7][8]). However, as multiple abiotic and biotic stresses can take place at the same time, a complex interplay of signalling pathways and responses can manifest resulting in opposing reactions. One mechanism to rapidly modify opposing stress-induced transcriptomes is to control the stability, degradation or turnover of specific transcripts at a post-transcriptional level through RNA-binding proteins.
RNA-binding proteins are mostly characterised by the presence of one or more RNA-binding domains. In addition to mRNA stability and decay, RNA-binding proteins are involved in diverse post-transcriptional processes including the maturation of mRNA through splicing, capping, polyadenylation and export from the nucleus [9,10]. Along with plant specific processes such as flowering, the sessile nature of plants and a necessity to adapt quickly to changing environmental conditions may be why plants encode many RNA-binding, with over 200 RNA-binding proteins predicted in Arabidopsis [11]. Interestingly though, very few RNAbinding proteins have been functionally characterised in plants.
One group of genes expressed in response to biotic and abiotic stress are those belonging to the ubiquitous GLUTATHIONE S-TRANSFERASE (GST) enzyme family [12][13][14][15]. Plant GSTs protect tissues against oxidative damage or from toxic products typically by catalyzing the conjugation of glutathione to a variety of electrophilic substrates of endogenous or exogenous origin, rendering the substrate less toxic (reviewed by [13,15]). Expression of the Arabidopsis GSTPHI8 (GSTF8) gene is induced rapidly by diverse biotic and abiotic elicitors including pathogen attack, phytohormones, herbicides, heat and high-light stress, and as such has become a marker gene for early stress and defense responses [12,[16][17][18][19][20][21][22][23].
Using the GSTF8 promoter to control the expression of a Firefly Luciferase reporter gene (GSTF8:LUC), we have been able to non-invasively monitor the plant's stress status, primarily within root tissues where GSTF8 is predominantly expressed [17,19,21,24]. To identify mechanisms controlling GSTF8:LUC activity we conducted a forward genetic screen using mutagenized populations of plants containing GSTF8:LUC [23]. One mutant isolated from this screen, designated disrupted in stress responses (dsr1), exhibited loss of SA inducible GSTF8:LUC activity and increased susceptibility to several fungal and bacterial pathogens [23]. The dsr1 mutation was mapped to a single amino acid change in a subunit of the mitochondrial energy machinery (complex II subunit SDH1-1), causing a reduction in induced reactive oxygen species production (ROS) from mitochondria and identifying mitochondrial derived ROS as a critical component of plant defense [23].
To complement the dsr1 study, we screened for mutants with enhanced GSTF8:LUC expression in the aim of identifying mutants with increased tolerance to biotic stress. We identified several alleles of a mutant called enhanced stress response 1 (esr1) encoding a K homology (KH) domain containing RNA-binding protein (At5g53060). The esr1 mutants confer constitutive GSTF8:LUC expression and increased resistance to the root-infecting fungal pathogen Fusarium oxysporum. Detailed analysis of the esr1-1 allele also identified significant down-regulation of genes enriched for involvement in JA-mediated responses. Other mutants of At5g53060 are reported to confer altered tolerance to abiotic stress [25][26][27] such as heat stress which we also established for esr1-1. While many Arabidopsis mutants conferring increased resistance to specific pathogens have been identified, these are commonly associated with a consequential decrease in tolerance to abiotic stress, or fitness costs such as poor growth or yield [2,6,28,29].
By contrast, esr1-1 displays increased F. oxysporum resistance, heat tolerance, and lacked observable defects in growth or development. These results define new roles for ESR1/At5g53060, functioning in biotic stress responses, JA-signalling, and unlinking growth restraint and resistance to stress.

Plant material and growth conditions
Unless otherwise specified, all experiments were conducted with the Arabidopsis thaliana Columbia-0 transgenic line (JC66/GSTF8:LUC) containing 791 bp of the GSTF8 promoter fused to a luciferase reporter [17,24]. Seeds were surface-sterilized, stratified at 4°C, and plated onto 100-mm square agar plates containing Murashige and Skoog (MS) salts as described previously [17]. Plates for luciferase assays were supplemented with 50 uM luciferin (Biosynth AG). Plate and soil grown plants were incubated under a 16-h light/8-h dark cycle at 22°C. The T-DNA insertion mutant [30] used to generate esr1-2 (SALK_095666) was obtained from the Arabidopsis Biological Resource Centre (ABRC). For generation of plants expressing candidate ESR1 genes the At5g53060, At5g53150 and At5g52860 CDS were amplified using primers listed in S1 Table. The resulting amplicons were cloned into BamHI/EcoRI digested binary vector pKEN [31] and confirmed by sequencing. 35S:At5g53060 pKEN, 35s:At5g53150 pKEN and 35S: At5g52860 pKEN were mobilized into Agrobacterium tumefaciens AGL1 and transformed into esr1-1 as per [31]. Transgenic plants were selected based on resistance to 10 ug/mL glufosinate ammonium (Fluka). To generate esr1-2, SALK_095666 and wild-type GSTF8:LUC lines were crossed and F 3 seedlings homozygous for the T-DNA and GSTF8:LUC selected.

Bioluminescence and luciferase assays
Seedling bioluminescence was captured and quantified by imaging in an EG & G Berthold molecular light imager as previously described [17,21]. Biochemical luciferase assays were performed as described by [24]. For 1 mM SA (Sigma) or temperature (45°C) treatments, 7-day old seedlings grown on square 100-mm Petri dishes were either covered with the liquid treatment at room temperature or incubated in temperature controlled cabinets for 40 min. After this time, excess liquid was discarded from relevant plates and the plates imaged such that, after acquiring the 0 hour bioluminescence image, 1 hour had elapsed.

Developmental and MeJA root elongation inhibition assays
Seeds of wild-type GSTF8:LUC, esr1-1 and esr1-2 were surface sterilized and plated onto MS media with germination rates measured as a percentage of total seeds plated (n = 60-70). For root length and MeJA root elongation inhibition assays, seeds were sterilized as above and plated onto MS media in either the presence or absence of 25 or 50 μM MeJA. Root length was measured on 7-day old seedlings using ImageJ [38]. Flowering time assays were conducted under long day conditions16-h light/8-h dark cycle at 22°C (n = 10).

Pathogen assays
For F. oxysporum inoculations the isolate Fo5176 was used. Root-dip inoculations on 4-weekold plants with a 1x10 6 cell/mL spore suspension were performed as described previously [39][40][41]. A. brassicicola assays were performed with isolate UQ4273 as described by [23]. A 5 ul portion of a 1x10 6 cell/mL spore suspension was applied to leaves of 3-to 4-week-old plants. Mock treatments with potato dextrose broth (PDB) were also conducted. Lesion size was measured with ImageJ [38]. For R. solani inoculations the strains AG2 or AG8 were used as previously described [23,42]. 7-day old seedlings were sown into vermiculite and inoculated with 1 mL of 1x10 6 cell/mL mycelium suspension.

RNA isolation
For qRT-PCR and RNAseq experiments on untreated plants, tissue was collected from 4-, 7-or 14-day old seedlings grown vertically on MS agar plates. For gene expression under MeJA or SA treatment, 12-d-old seedlings germinated on MS plates were gently lifted into a mock medium (MS broth), 100 uM MeJA medium (MS broth plus MeJA), or 1 mM SA (MS broth plus SA) such that the roots were submerged, and left for 6 or 24 h before harvesting. Three separate biological replicates were taken for all experiments consisting of whole seedlings pooled from 20-30 seedlings grown at the same time in the same environment, then frozen in liquid nitrogen, and stored at 80°C. RNA isolation was performed using the Qiagen RNeasy Plant Mini Kit (Qiagen). DNase treatment was performed after RNA isolation using TURBO DNase followed by treatment with the DNase Inactivation reagent (Ambion).
(Promega) with 1ug of input RNA. qRT-PCR was performed using SsoFast EvaGreen Supermix (Bio-Rad) on a CFX384 (Bio-Rad) system. Thermoycling and melt-curve conditions are described by [43]. Absolute gene expression levels relative to the previously validated [41,44,45] reference gene mix β-actin2, β-actin7, and β-actin8 (At1g49240, At3g18780, and At5g09810, respectively) were used for each complementary DNA sample using the equation: relative ratio gene of interest/actin = (Egene -Ct gene )/(Eactin -Ct actin ) where Ct is the cycle threshold value. The β-actin mix contains reverse primers for each of the three β-actin genes and a universal forward primer. The mean expression range of the reference gene was found to be within ±1 Ct across all samples. Several gene-specific primer sequences are previously published [45,46] and are also listed in S1 Table. RNA-seq library construction, Illuminia sequencing and identification of differentially expressed genes Following RNA isolation and DNase treatment of 14-day old wild-type or esr1-1 samples, Illumina TruSeq libraries were generated from 1 μg of total RNA and sequenced on a HiSeq1000 platform (Illumina). RNA-seq paired-end reads were trimmed for low-quality base-calls and Illumina adapter sequences via Cutadapt (v1.1, parameters: -quality-cutoff 30 -overlap 10 -times 3 -minimum-length 25) [47]. Reads trimmed to less than 25 bp were discarded and remaining reads sorted into pairs and singleton reads. RNA-seq reads were mapped to the TAIR10 Arabidopsis genome reference [33] via Tophat (v2.0.9, parameters: -b2-very-sensitive-r 50 -mate-std-dev 100-i 20-I 4000-g 20 -report-secondary-alignments -report-discordant-pairalignments-m 0 -min-coverage-intron 20 -coverage-search -microexon-search -library-type frunstranded) [48]. RNAseq read alignments were supplied to Cuffdiff (Cufflinks v2.1.1) [49] to calculate normalised expression within the TAIR10 annotated genes as fragments per kilobase of transcript per million mapped fragments (FPKM) (parameters: -frag-bias-correct -min-fragsper-transfrag 4 -multi-read-correct). Significantly differentially expressed transcripts between wild-type and esr1-1 were detected from 3 biological replicates using Cuffdiff with default Benjamini-Hochberg correction for multiple-testing, based on a False Discovery Rate 0.05. Functional annotations of genes and AGI symbols were sourced from TAIR10 datasets. RNA-seq reads have been deposited in the NCBI Sequence Read Archive under BioProject ID SRP056904.

Results
Identification of the constitutive GSTF8:LUC mutant esr1-1 To identify negative regulators of root stress responses we screened mutants from an ethyl methansulfonate (EMS) mutagenised GSTF8:LUC population [23] for enhanced basal luciferase expression. Over 50 mutants with constitutive GSTF8:LUC expression were identified and termed enhanced stress response (esr) mutants. One of the mutants with the highest basal GSTF8:LUC expression (esr1-1) was further analysed and its phenotype confirmed in the M 3 generation (Fig 1a and 1b). Quantitative real-time RT-PCR (qRT-PCR) was performed to confirm LUCIFERASE (LUC) gene expression and determine endogenous GSTF8 expression. While LUC expression was up-regulated (5.6-fold greater than wild-type), GSTF8 expression was unaltered (Fig 1c and 1d), suggesting the esr1-1 mutation may only affect the GSTF8:LUC transgene and not endogenous GSTF8 expression.
For cloning and heritability studies, we out-crossed esr1-1 to the Landsberg erecta ecotype (Ler). All F 1 plants showed the wild-type phenotype, and F 2 plants displayed a~3:1 segregation (59:21, χ 2 test p = 0.8) suggesting the esr1-1 phenotype is due to a recessive mutation in a single nuclear gene.
GSTF8:LUC activity and endogenous GSTF8 expression is up-regulated in esr1-1 following SA treatment To further characterise esr1-1, we monitored GSTF8:LUC expression following SA treatment, known to rapidly induce GSTF8 promoter activity in wild-type plants [17,24]. GSTF8:LUC activity increased more rapidly in esr1-1 following SA treatment where it plateaued at 6-7 hours post treatment compared to wild-type seedlings where this occurred at 8-9 hours (Fig 2a). Expression of the endogenous GSTF8 gene in esr1-1 under SA-inducing conditions was also significantly higher in esr1-1 compared to wild-type (Fig 2b). Combined with the lack of increased basal GSTF8 expression in esr1-1 (Fig 1d), these results suggest regulation of basal but not stress inducible GSTF8 promoter:LUC activity differs from the context of the endogenous GSTF8 gene, possibly due to regulatory components beyond the promoter fragment used in this study. ESR1 is a negative regulator of resistance to the fungal pathogen Fusarium oxysporum We previously identified a mutant from the same screen as esr1-1 but with loss of SA inducible GSTF8:LUC activity. This mutant termed disrupted in stress responses 1 (dsr1) exhibits increased susceptibility to several pathogens [23]. As esr1-1 exhibits increased root localised GSTF8:LUC expression (Figs 1 and 2), we hypothesized esr1-1 may confer increased resistance to root pathogens. To test this, we first inoculated wild-type and esr1-1 plants with the root-infecting fungal pathogens Rhizoctonia solani and Fusarium oxysporum [19,50]. While no significant difference in disease symptom development or survivorship was observed between wild-type or esr1-1 plants inoculated with R. solani (strains AG2 or AG8) (Fig 3a), esr1-1 did exhibit increased resistance to F. oxysporum (Fig 3b-3d). This was observed through both a delay in disease symptom development and increased survival. While jasmonate (JA)-mediated defences are required for resistance to most fungal necrotrophic pathogens (e.g. Botrytis cinerea, Alternaria brassicicola, [51]), JA-signalling confers susceptibility to F. oxysporum with mutants compromised in JA-dependant responses exhibiting resistance to this pathogen [41,45,52].
The increased resistance to F. oxysporum prompted us to determine if esr1-1 conferred increased susceptibility to A. brassicicola. Larger A. brassicicola induced lesions were observed on esr1-1 leaves compared to wild-type (Fig 3e and 3f), however not a statistically significant level. This phenotype was observed over several independent experiments suggesting ESR1 contributes a small affect to inhibition of A. brassicicola lesion development. As could be hypothesized from the F. oxysporum results, increased A. brassicicola induced lesions may be due to reduced JA-responses in esr1-1.

ESR1 encodes a K homology (KH) domain RNA-binding protein
To determine the causal esr1-1 mutation, map based cloning was initiated using F 2 seeds of the esr1-1 and Ler cross. Genetic mapping was conducted on 1040 homozygous F2 plants.
The mutation was narrowed down to a region on Chromosome 5 spanning~200 Kb across three Bacterial Artificial Chromosomes (BACs); MXC20, MNB8 and MFH8 (Fig 4a). Whole genome sequencing of the GSTF8:LUC wild-type parent and esr1-1 and alignment to the TAIR10 genome identified five single nucleotide polymorphisms (SNPs) within the mapped loci with three causing non-synonymous mutations in gene coding regions. Molecular complementation assays were conducted by individually introducing the three candidate genes under the control of the 35S promoter into the esr1-1 background. The 35S:At5g53060 construct eliminated the esr1-1 constitutive GSTF8:LUC expression and restored the wild-type phenotype, suggesting a G300A nucleotide change in At5g53060 and resulting stop codon substitution of W100 Ã causes the esr1-1 phenotype (Fig 4b, 4c and 4f). At5g53060 encodes a heterogenous nuclear riboprotein (hnRNP) K homology (KH) domain containing RNAbinding protein. The protein contains five KH-domains and the esr1-1 mutation disrupts the first of these domains. KH-domains are found in many RNA-binding proteins and are associated with transcriptional and post-transcriptional processes where they can bind RNA or single stranded DNA [11,53,54]. In all subsequent experiments the esr1-1 line used is a line backcrossed twice to the wild-type parent (to remove other EMS induced mutations) and has the same GSTF8:LUC phenotype as the M 3 esr1-1 line (data not shown).
To confirm a mutation in At5g53060 is responsible for the esr1-1 increased GSTF8:LUC and Fusarium resistance phenotypes, a T-DNA insertion line in At5g53060 (SALK_095666) was crossed with wild-type GSTF8:LUC. We generated F 3 seedlings homozygous for the T-DNA and GSTF8:LUC (subsequently named esr1-2) which displayed both the constitutive esr1-1 luciferase phenotype and increased resistance to F. oxysporum, further supporting that the esr1-1 mutant phenotypes result from disruption of At5g53060 (Figs 4d-4f, 5a and 5b). As with esr1-1, esr1-2 plants exhibited increased basal expression of LUCIFERASE but not GSTF8 (Fig 5c). To rule out the possibility that increased transgene activity in the esr1 mutants contributes to Fusarium resistance, we inoculated SALK_095666 in the absence of the GSTF8:LUC transgene, and included control Col-0 plants. As with both esr1-1 and esr1-2 mutants, the SALK_09566 line also showed significantly increased resistance to Fusarium (Fig 5d and 5e). These results therefore confer a new role for At5g53060/ESR1, in mediating responses to biotic stress.
Of the other 50 constitutive GSTF8:LUC mutants isolated from our initial screen, we identified two further recessive alleles of esr1 using the process of Next Generation Mapping (NGM) [37]. Using the NGM tool, SNPs were called against the TAIR10 genome reference and identified a region on chromosome 5 low in heterozygosity, incidentally mapping to the ESR1/ At5g53060 loci (S1 Fig). We found both mutants conferred increased resistance to F. oxysporum and through genetic complementation confirmed both mutants were indeed alleles of esr1, subsequently designating them as esr1-3 and esr1-4 (Fig 5f-5h). esr1-3 confers a G2744A change at the second last exon/intron boundary (TGAGCAgtaagtt>TGAGCAataagtt), while esr1-4 confers a G690A change at the last codon of the first exon causing a synonymous change (CAG>CAA, Q>Q) (Fig 5i). Processing of these mutated genes through the splice site prediction software NetGene2 (version 2.4; [55]) revealed that both mutants would encode missspliced transcripts with esr1-3 losing the last donor splice site and esr1-4 losing the first donor splice site. Three other mutant alleles of At5g53060 have been independently identified through abiotic stress screens using salt, desiccation or cold-inducible promoters and confer increased or reduced tolerance to specific abiotic stresses. These are Regulator of CBF gene expression 3 (rcf3-1, [26]), Shiny1 (shi1, [27]), and High Osmotic Stress Gene Expression 5 (hos5-1, [56]). To determine if esr1-1 exhibited altered tolerance to abiotic stress we conducted heat tolerance assays described by [26] and found esr1-1 also conferred increased tolerance to heat stress (S2 Fig). This was also confirmed for the insertional knockout mutant esr1-2 (S2 Fig). To further characterise the role of At5g53060/ESR1 in temperature stress, we monitored GSTF8:LUC activity in wild-type and esr1-1 seedlings over a 12 h time-course following heat stress. As with results following SA treatment (Fig 2), GSTF8:LUC activity increased more rapidly in esr1-1 compared to wild-type seedlings (S2 Fig) suggesting ESR1 contributes to the repression of SA and heat induced stress responses.
Whole transcriptome sequencing of esr1-1 reveals altered expression of genes involved in biotic and abiotic stress responses To uncover the possible direct or indirect targets of At5g53060/ESR1, we performed wholetranscriptome sequencing (RNA-seq) on three biological replicates of esr1-1 and wild-type seedlings using the Illuminia HiSeq platform. We used un-treated seedlings as GSTF8:LUC is constitutively up-regulated in esr1-1 under normal growing conditions (Fig 1a). Between 18.5 and 21.4 million paired-end reads (100 bp in length) per sample were mapped to the Arabidopsis TAIR10 exome reference sequence.
Using the Cuffdiff tool within Cufflinks [49] we identified 1176 significantly differentially expressed genes between wild-type and esr1-1 (Benjamini-Hochberg correction for multipletesting based on a False Discovery Rate 0.05). Based on significant FPKMs (Fragments Per Kilobase of exon model per Million mapped fragments) fold changes, more transcripts were down-regulated in esr1-1 (873) compared to up-regulated (303) (S2 and S3 Tables). To gain insight into the functions of these genes we performed Gene Ontology (GO) term enrichment analysis using agriGO v1.2 [57] with the default False Discovery Rate (p 0.05) determined pvalue significance. Among the most significantly enriched biological processes from genes down-regulated in esr1-1 were those involved in response to stress, biotic and abiotic stimuli, defense responses, wounding responses, and response to other organisms (bacteria and fungi) (S4 Table). Among the most significantly enriched biological processes from genes up-regulated in esr1-1 were those involved in response to light, abiotic and hormone stimulus, cell death, signaling pathways and developmental processes (S5 Table). While esr1-1 displays these significant changes in stress and defense responsive gene expression, interestingly neither esr1-1 or esr1-2 show obvious deleterious growth or developmental defects (Fig 6a-6d) often exhibited by Arabidopsis mutants with similar expression profiles [2,58].
To consolidate a smaller list of differentially expressed genes for follow up qRT-PCR studies we sorted the gene lists for those with significantly altered expression greater than 2-fold from wild-type levels. This identified 48 genes up-regulated in esr1-1 and 174 genes down-regulated (Tables 1 and 2). Although analysis of the 48 up-regulated genes did not reveal any specific GO term enrichment for biological processes, there was molecular function enrichment for nucleoside and nucleotide binding. The gene list included C-Terminal Domain Phosphatase-Like1 (CPL1) which can physically interact with At5g53060 [27,56,59] and was one of the most significantly up-regulated genes in esr1-1. Other genes with strong up-regulation were At3g54160 an uncharacterised F-box/RNI-like protein, SYNAPTOTAGMIN 2 (SYTB) a calcium-dependent lipid-binding protein involved in protein secretion, and SUGAR TRANSPORTER 11 (STP11). Examination of these three genes and CPL1 using qRT-PCR within a developmental time-course comparing wild-type versus esr1-1 seedlings, including the 14-day old seedlings that were sampled for the RNAseq analysis, revealed their constitutive up-regulation at all time-points (S3 Fig). We next examined the list of 174 genes that were significantly down-regulated in esr1-1 >2-fold over wild-type (Table 2) for GO term enrichment. Amongst the most significantly     enriched biological process GO terms were those involved in responses to stress, defense, biotic stimulus, other organisms, jasmonic acid (JA) including JA-biosynthesis and-signalling, fungus, wounding, SA, chemical stimulus and responses to starvation and nutrient levels (Fig 6e).
As with the up-regulated esr1-1 dataset, to confirm our RNAseq data we examined the expression of several down-regulated genes over a developmental time-course. This included the highly down-regulated genes DARK INDUCIBLE 11 (DIN11) encoding a 2-oxoglutarate (2OG) and Fe(II)-dependent oxygenase, At2g39030/NATA1 encoding an Acyl-CoA N-acyltransferase (NAT) superfamily protein with roles in pathogen resistance [60], as well as CHLOROPHYLLASE 1/CORONATINE-INDUCED PROTEIN 1 (CLH1/CORI1) which has roles in several of the enriched biological process GO categories including defense responses, response to fungus and JA-signalling. No significant difference in DIN11, NATA1 or CLH1 expression was observed in 4-or 7-day old seedlings however, as in the RNAseq dataset they were highly down-regulated in 14-day old seedlings (Fig 7a).

At5g53060/ESR1 affects basal JA-mediated responses involved in defense but not growth and development
A role for At5g53060 in JA-responses to our knowledge has not been described before, and as the down-regulated esr1-1 gene list was enriched for genes with roles in these processes including defense and biotic stimulus (response to fungus and wounding), we were interested to dissect this further. We first examined the expression of representative JA-biosynthesis, signalling, and JA-regulated defense genes. Using qRT-PCR, the LIPOXYGENASE 3 (LOX3) and ALLENE OXIDE CYCLASE 1 (AOC1) genes involved in JA-biosynthesis, and JASMONATE-ZIM-DO-MAIN PROTEIN 10 (JAZ10) involved in repression of JA-responses were down-regulated in esr1-1 in both 7-and 14-day old seedlings (Fig 7b) and were identified in the RNAseq dataset as down-regulated genes ( Table 2). The down-regulation of these genes suggests an overall downregulation of JA-signalling processes in esr1-1 as their expression is in part regulated through JA-feedback loops [61][62][63]. In 14-day old seedlings the JA-regulated defense and wound marker genes analysed were all down-regulated in esr1-1 compared to wild-type seedlings (Fig 7c). The expression of these marker genes in 4-or 7-day old seedlings was either lowly expressed or not detectable by qRT-PCR. Overall expression patterns in wild-type seedlings highlighted a trend in increasing expression from 4-to 14-days. Examination of these genes in publically  available, developmental series transcriptome datasets (Genevestigator; [64]) also revealed similar gene expression profiles in wild-type plants (data not shown). GSTF8:LUC activity also increases in esr1-1 seedlings over this timeframe (Fig 7e). Together, these results suggest At5g53060/ESR1 has a negative effect on GSTF8:LUC activity and a positive effect on the regulation of JA-mediated genes during early development.
It is well known that antagonistic interactions occur between some aspects of JA and SA signalling (reviewed in [6,8,29,70]. We therefore analysed expression of the SA-marker genes PR1 and PR2 in wild-type and esr1-1 seedlings to determine if repression of JA-regulated genes in esr1-1 was due to up-regulated SA-mediated signalling as is suggested by increased GSTF8:LUC activity and GSTF8 expression in esr1-1 following SA treatment (Fig 2). There was no significant difference in PR1 or PR2 expression between wild-type and esr1-1 at 4 and 7 days of age, but their expression was significantly reduced in esr1-1 at 14 days as was also detected by RNAseq (Fig 7d, Table 2). PR1, but not PR2 expression, was also down-regulated in esr1-1 following SA treatment (Fig 7f). To determine if other aspects of SA-signalling where altered in esr1-1, we determined expression of the ISOCHORISMATE SYNTHASE1 (ICS1) and PHENYLALANINE AMMONIA LYASE (PAL1) genes involved in SA-biosynthesis (reviewed by [71]). Neither of these genes were significantly altered in expression suggesting ESR1 functions specifically in JAsignalling and down-regulation of PR1 expression is due to non-SA-mediated processes.

At5g53060/ESR1 is required for full activation of a subset of JAregulated genes
Other mutants with reduced basal JA-biosynthesis or JA-regulated defense gene expression and exhibiting increased resistance to Fusarium oxysporum include coi1 (coronatine insens-tive1) and pft1/med25 (phytochrome and flowering time1) [41,45]. Expression of JA-regulated genes in these two mutants are also reduced following MeJA treatment. To determine if At5g53060/ESR1 affected the JA-inducibility of JA-regulated genes and other genes down-regulated in esr1-1, we examined the expression of Thi2.1, PDF1.2, JAZ10, NATA1, CLH1 and DIN11 in esr1-1 and wild-type plants following MeJA or a mock treatment. As expected, MeJA treatment strongly induced Thi2.1, PDF1.2 and JAZ10 expression in wild-type plants relative to the mock-treated wild-type plants (Fig 8a). Expression of these genes was also induced by MeJA in esr1-1 however, Thi2.1 and JAZ10 expression was 5-fold and 2-fold less respectively in esr1-1 compared to wild-type levels at 6 and 12 hours post treatment. PDF1.2 expression was also reduced in esr1-1 at 6 hours but increased above wild-type levels at 24 hours. We next examined NATA1, CLH1 and DIN11 expression and found esr1-1 had reduced induction of NATA1 and CLH1, but did not affect the MeJA-induced expression of DIN11 (Fig 8b). We also found ESR1 expression is MeJA-inducible (Fig 8c). Combined, these results suggest ESR1 affects components of JA-signalling.

Discussion
In a forward genetic screen using the defense and stress responsive GSTF8 promoter, we isolated several alleles of the constitutive GSTF8:LUC expression mutant esr1 encoding the KH-domain containing RNA-binding protein At5g53060. We identify At5g53060 as a susceptibility gene for F. oxysporum disease symptom development and a requirement for full-activation of components of JA-mediated gene expression. Four independent mutants of At5g53060 termed esr1-1, esr1-2, esr1-3 and esr1-4 displayed increased resistance to F. oxysporum and define new roles for plant KH-domain containing proteins, linking At5g53060 to biotic stress and JA-mediated defense responses.
In plants the most widely spread RNA-binding domains are the RNA Recognition Motif, the heterogeneous nuclear ribonucleoprotein K (hnRNP K) homology (KH), and Pentatricopeptide Repeat (PPR) [10,72]. Most plant RNA-binding proteins contain one or more of these domains, often combined with multiple auxiliary domains involved in protein-protein interactions or protein targeting, or other RNA-binding domains. An InterPro Scan of At5g53060 for known protein signatures only identified its five KH-domains (data not shown). KH-domain proteins typically contain more than one KH-domain where they can function independently or co-operatively to bind RNA or ssDNA [54]. The KH-domain was first identified in the human hnRNP K protein and is characterised by a conserved VIGXXGXXI sequence in the middle of the~60 amino acid domain [11,54]. In addition to At5g53060, only three other of the 26 predicted Arabidopsis KH proteins have been functionally characterised and these have roles in flowering, floral morphogenesis, and vegetative and reproductive development [73][74][75]. Unlike mutants of these KH genes, we found neither esr1-1 nor esr1-2 exhibited observable differences in flowering, growth or development (Fig 6a-6d).
In plants, RNA-binding proteins have been identified as regulators of floral transition, floral patterning, circadian rhythm, chromatin modification, ABA signalling and mediators of abiotic stress responses such as to dehydration, drought, flooding, salinity, cold and heat ([9, 10] and references within). However, few RNA-binding proteins have been characterized for roles in plant immunity [76,77]. Examples include the RNA Recognition Motif containing proteins ATBRN1/ATRBP-DR1 and GLYCINE RICH PROTEIN 7 regulators of resistance against Pseudomonas syringae pv. tomato DC3000 [78,79], and the double stranded RNA-binding domain proteins DICER LIKE2 and DICER LIKE4 involved in viral defense ( [76,80] and references within). Some RNA-binding proteins directly target pathogen RNA to control infection. For example PATHOGENESIS RELATED PROTEIN 10 members such as the cotton PR10 have ribonuclease activity [1,81]. In addition to our findings on At5g53060, the only other plant KH-domain containing protein characterised for a role in plant immunity is BINDING TO TOMV RNA 1 (BTR1, At5g04430) [82] which functions by directly binding to Tomato Mosaic Virus (TOMV) RNA and preventing viral multiplication.
Through non-biased whole transcriptome sequencing we found esr1-1 plants exhibited significant down-regulation of genes involved in responses to defense, biotic stimulus, fungus, wounding, and JA including JA-biosynthesis and-signalling (Fig 6). JA has roles in defense against specific microbial pathogens and insect pests, wounding responses, and roles in developmental processes such as fertility and root growth [65,[67][68][69]. The esr1 mutants were not affected in latter responses but were affected in pathogen defense. Further, in addition to repression of basal JA-mediated gene expression, the MeJA inducibility of most JA-regulated genes tested were also repressed in esr1-1 (Fig 8). Unlike the other repressed genes, PDF1.2 expression in esr1-1 was reduced at 6 hours but increased above wild-type levels at 24 hours. This suggests, in addition to activation of components of early JA-regulated gene expression, ESR1 may have roles in repression at later stages. We also found repression of JA-regulated genes in esr1-1 was not due to antagonistic SA-JA crosstalk as SA-regulated marker genes in esr1-1 showed no increase in their basal or SA induced levels (Fig 7).
The down-regulation of JA-signalling in esr1-1 likely contributes to its enhanced resistance to F. oxysporum and increased susceptibility to A. brassicicola as this pathway confers susceptibility and resistance to these pathogens respectively. For example, the coi1 and pft1/med25 mutants, which are highly resistant to F. oxysporum and susceptible to A. brassicicola, also exhibit reduced expression of JA-biosynthesis (e.g. LOX3, OPR), JA-signalling (e.g. JAZ9, JASMONIC ACID CARBOXYL METHYLTRANSFERASE) and JA-regulated defense/senescence associated genes (e.g. Thi2.1, PDF1.2, CLH1) under mock, F. oxysporum or MeJA induced conditions [41,45,51]. While defensive components of JA-signalling do contribute positively to F. oxysporum resistance (e.g. increased PDF1.2, Thi2.1 expression), global up-regulation of JA-signalling including its non-defensive components promote susceptibility [31,44,46,83]. For example, senescence is proposed to strongly contribute to F. oxysporum disease symptom development [41]. Unlike the coi1 and pft1 mutants which have impairments in growth or development (coi1 is male sterile with insensitivity to MeJA inhibition of root growth, while pft1 is delayed in flowering [45,84]), esr1-1 is indistinguishable from wild-type plants (Fig 6a-6d and S4 Fig). We also inoculated esr1-1 with another root-infecting necrotrophic fungal pathogen, R. solani (isolates AG8 or AG2-1) but found no difference in disease phenotypes compared to wild-type plants (Fig 3a). It is suggested that neither JA or SA signalling pathways contribute to R. solani resistance or susceptibility in Arabidopsis [42].
Through independent forward genetic screens utilizing abiotic stress inducible promoters, At5g53060 was shown to be nuclear localised and have roles in diverse transcriptional processes including mRNA capping efficiency, polyadenylation site selection, mRNA export, and in the regulation of expression and alternate splicing of some stress inducible genes ( [26,27,56] this work). Some of these processes, including its nuclear localisation, are mediated through At5g53060 interactions with the RNA PolII CTD interacting phosphatase protein CPL1 [25,27,56,59] whose expression is up-regulated in esr1-1 ( Table 1,  We characterised four mutants of At5g53060. The esr1-1 mutant confers a Tryptophan to STOP codon substitution (W100 Ã ) in the first of five At5g53060 KH-domains, esr1-2 is a null T-DNA insertional inactivation line, while esr1-3 and esr1-4 harbour mutations at splice site junctions (Figs 4 and 5). Three other independently identified At5g53060 alleles confer other mutations. The rcf3-1 (regulator of CBF gene expression 1) mutant isolated through a cold responsive CBF2 promoter screen confers a Glycine to STOP codon substitution (G344 Ã ) within the third At5g53060 KH-domain and displays increased heat tolerance [26]. The shi1 (shiny1) mutant isolated through the salt inducible sulfotransferase AtSOT12 promoter confers a Glutamic Acid to Lysine substitution (E389K) also within the third KH-domain [27]. The shi1 mutant is more resistant to ABA during germination and has increased sensitivity to cold stress [27]. Unlike these mutants of At5g53060, the hos5-1 mutant has increased sensitivity to ABA and salt stress, although tolerance/sensitivity to other abiotic stresses has not yet been tested for this At5g53060 mutant allele [56]. The hos5-1 (high osmotic stress gene expression 5) mutation confers a Glycine to Serine change (G233S) within the second At5g53060 KH-domain. Interestingly, esr1-1, rcf3-1 and shi1 mutations (Fig 9) disrupt either the first or third KHdomains, both of which along with the fourth domain can interact with CPL1 ( [27,56,59] this work). Although the shi1 mutant confers only an amino acid substitution, this change disrupts the CPL1 interaction [27]. The second KH-domain and location of the hos5-1 mutation (Fig 9) does not interact with CPL1 but may affect RNA binding [56]. This may explain why hos5-1 is more sensitive to ABA while shi1 is more tolerant. As with our esr1-1 and esr1-2 findings, rcf3-1 and shi1 exhibit increased expression of their Promoter:LUC transgenes but not of their endogenous stress-inducible genes under basal conditions [26,27]. We did however find GSTF8 expression was significantly higher than wild-type in esr1-1 following SA treatment (Fig 2b) suggesting regulation of GSTF8:LUC promoter and endogenous GSTF8 differ under basal conditions. Indeed, [27] suggest under basal conditions the ESR1/SHI1-CPL1 complex may associate with other repressors on general transcriptional machinery targeting stress responsive promoters and upon stress inducing conditions this complex is modified.
In summary, we identified roles for the KH-domain RNA-binding protein At5g53060 in JA-and biotic induced stress responses, and define new functions for KH-domain proteins in plants. Further research of interest will be determining At5g53060/ESR1 direct RNA targets, which are yet to be identified, and other proteins it interacts with under specific biotic stresses, in particular those with JA-involvement.  Fig. Basal esr1-1 up-regulated genes are similarly expressed in wild-type and esr1-1 when treated with MeJA or SA. Fold changes in relative transcript abundance of RNA-seq identified genes in wild-type (WT) and esr1-1 seedlings 6 and 24 hours post MeJA or SA treatment. Shown are values from 12 day old seedlings (values are averages ± SE of 3 biological replicates consisting of pools of 20-30 seedlings). Transcript levels of each gene of interest following MeJA or SA treatment were normalised against the internal control β-actin genes and expressed relative to the normalised levels in mock-treated WT or esr1-1 seedlings. The numbers on each bar show fold increase or fold decrease caused by each treatment relative to mock-treated plants. PDF1.2 and PR1 were used as marker genes for MeJA and SA treatment respectively. (TIF) S1