AtWRKY22 promotes susceptibility to aphids and modulates salicylic acid and jasmonic acid signalling

Highlight

The transcription factor WRKY22 increases susceptibility to aphids in Arabidopsis via the suppression of salicylic acid signalling.

Abstract

Aphids induce many transcriptional perturbations in their host plants, but the signalling cascades responsible and the effects on plant resistance are largely unknown. Through a genome-wide association (GWA) mapping study in Arabidopsis thaliana, we identified WRKY22 as a candidate gene associated with feeding behaviour of the green peach aphid, Myzus persicae. The transcription factor WRKY22 is known to be involved in pathogen-triggered immunity, and WRKY22 gene expression has been shown to be induced by aphids. Assessment of aphid population development and feeding behaviour on knockout mutants and overexpression lines showed that WRKY22 increases susceptibility to M. persicae via a mesophyll-located mechanism. mRNA sequencing analysis of aphid-infested wrky22 knockout plants revealed the up-regulation of genes involved in salicylic acid (SA) signalling and down-regulation of genes involved in plant growth and cell-wall loosening. In addition, mechanostimulation of knockout plants by clip cages up-regulated jasmonic acid (JA)-responsive genes, resulting in substantial negative JA–SA crosstalk. Based on this and previous studies, WRKY22 is considered to modulate the interplay between the SA and JA pathways in response to a wide range of biotic and abiotic stimuli. Its induction by aphids and its role in suppressing SA and JA signalling make WRKY22 a potential target for aphids to manipulate host plant defences.

Introduction

As plants are sessile organisms in often dynamically changing environments, plasticity is fundamental to survival. Transcriptional regulation plays an important role in how plants cope with environmental stimuli. In Arabidopsis approximately 50 transcription factor families have been identified, accounting for approximately 2000 genes (Guo et al., 2005; Mitsuda and Ohme-Takagi, 2009). Together with signal perception and transduction elements, these transcription factors participate in complex and dynamic networks that regulate developmental processes and responses to (a)biotic stress. Insect infestations are typical situations that require quick transcriptional reprogramming in order to mount an effective defence response. Aphids are phloem-feeding insects that manoeuvre their piercing–sucking mouthparts between cells and reach the vascular bundle without inflicting major physical damage (Minks and Harrewijn, 1989). They are vectors of many plant viruses and deprive the plant of photoassimilates. Aphids cause strong transcriptional perturbations in plants, inducing or repressing up to several thousand genes, whereas other insects such as caterpillars and cell-content feeders alter the expression of only up to several hundred genes (De Vos et al., 2005; Kusnierczyk et al., 2007; Barah et al., 2013; Dubey et al., 2013; Kerchev et al., 2013; Appel et al., 2014; Foyer et al., 2015). An open question is, however, whether these transcriptional changes lead to enhanced resistance to aphids or whether they are unsuccessful or even counter-effective modulations. Aphids are known to secrete effectors via their saliva into the apoplast and the vascular bundle (Rodriguez and Bos, 2012), and might be able to manipulate the host plant physiology for their own benefit. In this study, genome-wide association (GWA) mapping revealed WRKY22 (At4g01250) as one of the candidate genes for affecting feeding behaviour of the generalist aphid Myzus persicae (Sulzer) on Arabidopsis thaliana. WRKY22 is a member of the WRKY transcription factor family, which was discovered in the 1990s and named after its binding affinity to the W-box promoter motif (Eulgem et al., 2000). WRKY22 and its homologue WRKY29 are part of group IIe WRKYs and are both established markers of pathogen-triggered immunity (PTI). Pathogen-associated molecular patterns (PAMPs) such as flagellin, chitin and cellulysin are recognized elicitors of the mitogen-activated protein kinase (MAPK) cascade that induce WRKY22 and WRKY29 within 30min post-inoculation (Asai et al., 2002; Dong et al., 2003; Navarro et al., 2004; Mészáros et al., 2006; Thilmony et al., 2006; Schikora et al., 2011; González-Lamothe et al., 2012; Shi et al., 2015). In general, PTI results in the accumulation of reactive oxygen species and callose deposition and involves salicylic acid (SA), jasmonic acid (JA) and ethylene (ET) signalling (Yi et al., 2014). Although the exact role of WRKY22 and WRKY29 in PTI is unknown, WRKY22 has been shown to be required for resistance to the hemibiotrophic pathogen Pseudomonas syringae (Hsu et al., 2013) and WRKY29 has been described as conferring resistance to P. syringae as well as to the necrotrophic pathogen Botrytis cinerea (Asai et al., 2002). In this study, we assessed the involvement of WRKY22 in plant resistance to M. persicae aphids and its downstream transcriptional effects.

Materials and methods

Plants and insects

A collection of 344 natural accessions of A. thaliana was obtained from the ABRC Stock Center (Baxter et al., 2010). This set was selected in a previous study to represent most intraspecific genetic variation and minimal redundancy (Platt et al., 2010), and was genotyped for 214000 single nucleotide polymorphism (SNPs) with AtSNPtile1 arrays (Atwell et al., 2010; Li et al., 2010; Horton et al., 2012). Transfer (T)-DNA lines SALK_094892 (wrky22-3) and SALK_098205 (wrky22-4), and TRANSPLANTA-inducible overexpression lines TPT_4.01250.1C and TPT_4.01250.1E were obtained from NASC (Coego et al., 2014). Seeds were cold stratified for 72h at 4 °C before they were sown in pots (5cm diameter) with pasteurized (4h at 80 °C) Arabidopsis potting soil (Lentse Potgrond, Lent, The Netherlands) in a climate room at 24±1 °C, 50–70% relative humidity, 8 h–16h light–dark photoperiod, and a light intensity of 200 μmol m⁻² s⁻¹. Homozygous T-DNA plants were selected based on PCR and harvested for seeds for subsequent experiments. The location of the T-DNA insertion was confirmed via sequencing, and abolition of WRKY22 expression was tested with RT-qPCR (Supplementary Table S1 at JXB online). Expression of WRKY22 in the TRANSPLANTA-inducible overexpression lines (Coego et al., 2014) was measured with RT-qPCR 24h after application of 10 μM oestradiol in water to the plant trays (Supplementary Table S1). Green peach aphids, M. persicae, were reared on radish, Raphanus sativus (L.), at 19 °C, 50–70% relative humidity and a 16 h–8h light–dark photoperiod.

Automated video tracking

Aphid behaviour was tracked on 344 natural accessions of Arabidopsis (n=5–6 per accession) according to the methodology of Kloth et al. (2015). One adult, wingless aphid was introduced into a well of a 96-well plate containing a leaf disc of 6mm diameter, abaxial side up, on 1% agar substrate. Wells were covered with cling film to avoid aphid escape, and 20 aphids were recorded on 20 different accessions simultaneously with a camera mounted above the plate, at 22±1 °C. EthoVision^® XT 8.5 video tracking and analysis software (Noldus Information Technology bv, Wageningen, The Netherlands) was used for automated acquisition of aphid position and velocity. The number and duration of probes were subsequently calculated with the statistical computing program R (R Core Team, 2013). Leaf discs were made of intermediately aged leaves of 4- to 5-week-old Arabidopsis plants, one disc per plant. Aphid behaviour was recorded for 85min, starting at 4.5h after inoculation of the aphids. The video-tracking assay was performed in an incomplete block design with each complete replicate consisting of 18 blocks of 20 accessions. Sixty plants were screened each day across three blocks, and one replicate of the complete Hapmap collection was acquired in 6 days. An alpha design was generated with Gendex (http://designcomputing.net/gendex/) to assign accessions to each block. Five to six replicates were acquired per accession.

GWA mapping and haplotype analysis

GWA mapping was performed on the proportion of aphids making long probes (> 25min) with scan_GLS (Kruijer et al., 2015), using a kinship matrix based on all SNPs to account for population structure. SNPs with a minor allele frequency <0.05 were excluded from analysis. Block and replicate were included in the model as covariates. SNPs with −log₁₀(P) value larger than 4 were taken as candidate loci. Generalized heritability was estimated as in Oakey et al. (2006). For haplotype analysis, SNPs with a minor allele frequency above 5% were retrieved from the Arabidopsis 1001 genomes browser for 173 accessions (Cao et al., 2011). For each domain, haplotypes were defined as unique SNP combinations with a frequency above 5%. For exons, only non-synonymous SNPs were included. A promoter region of 1000kb was used and gene domains were obtained from Interpro (Mitchell et al., 2015). Promoter motifs were retrieved from Athamap (Hehl and Bülow, 2014).

RT-qPCR

For each sample, two intermediately aged leaves per 4- to 5-week-old Arabidopsis plant were harvested between 12.00 and 15.00h. Samples were immediately frozen in liquid nitrogen, and stored at −80 °C until processing. RNA was isolated from homogenized leaf material with an InviTrap^® Spin Plant RNA kit, and treated with Ambion^® TURBO DNA-free^TM according to the manufacturer’s instructions. RNA was quantified with a NanoDrop^® ND-1 000 spectrophotometer, and integrity was assessed with gel electrophoresis. DNA-free RNA was converted into cDNA using the Bio-Rad iScript^TM cDNA synthesis kit. Quantitative reverse transcription PCR was carried out on a Bio-Rad IQ^TM5 system using SYBR Green. For each primer combination (Supplementary Table S1), RT-qPCR products were sequenced to validate the region of amplification. To test aphid induction of WRKY22, PR1, VSP2, and PDF1.2, plants were treated with and without aphids (n=4). For infested samples, a Petri dish with indentation for the petiole was used to contain 15 adult M. persicae aphids on the leaf, to inflict as little mechanostimulation as possible. Four biological replicates were collected for three treatments: (1) an empty Petri dish for 48h, (2) a Petri dish for 48h with addition of aphids in the last 6h, and (3) a Petri dish with aphids for 48h.

Electrical penetration graph recording

Feeding behaviour of M. persicae aphids was investigated with electrical penetration graph (EPG) recording on 4- to 5-week-old Arabidopsis plants, using direct current (DC) according to the methodology of ten Broeke et al. (2013). To adjust the radish-reared aphids to Arabidopsis, aphids were transferred to Col-0 Arabidopsis plants 24h before the experiments. EPG recording was performed at 22±2 °C and light intensity of 120 μmol m⁻² s⁻¹, using clean plants and one aphid per plant. An electrode was inserted in the potting soil and a thin gold wire of 1.5cm was gently attached to the dorsum of an adult, wingless aphid with silver glue. The electrical circuit was completed when the aphid’s piercing–sucking stylet mouthparts penetrated the plant cuticle. Electrical signals associated with stylet activities were recorded and annotated with EPG Stylet+ software (http://www.epgsystems.eu) and further processed in R (R Core Team, 2013; Tjallingii, 1988). Between 20 and 24 biological replicates were measured on T-DNA lines (Col-0: n=24; wrky22-3: n=22; wrky22-4: n=20) and between 15 and 19 on overexpression lines (Col-0: n=15; OE.c: n=19; OE.e: n=19). WRKY22 overexpression was induced by supplying 10 μM oestradiol solution to the plants 24h before the experiment. To correct for potential side-effects of oestradiol, the wild-type plants received the same oestradiol treatment as the overexpression lines.

Aphid population development

To assess aphid developmental rate and population size, 2.5-week-old Arabidopsis plants were infested with one M. persicae neonate of age 0–24h and placed in a climate room at 24±1 °C, 50–70% relative humidity, 8 h–16h light–dark photoperiod, 200 μmol m⁻² s⁻¹ light intensity. A soap-diluted water barrier prevented aphids from moving between plants. None of the aphids developed wings. From day 7 onwards, occurrence of the first offspring was checked twice per day using 5× magnification glasses (Col-0: n=18; wrky22-3: n=19; wrky22-4: n=22). The number of aphids per plant was counted at 14 days after infestation. Plants without an adult aphid 8 days after introduction and plants without any adults or neonates 14 days after introduction were excluded from the analysis.

Statistics

Data were tested for a normal distribution and homogeneity of variances using Shapiro’s test and Levene’s test. Non-parametric data sets were assessed with the Mann–Whitney U-test (two groups) or Kruskall–Wallis test (more than two groups). Data sets with a normal distribution were tested with Student’s t-test (two groups) or a one-way ANOVA (more than two groups).

RNA-seq analysis

RNA-seq analysis was conducted on leaves of Col-0 and wrky22-3 with three biological replicates per treatment. Five leaves of five different 4- to 5-week-old plants were pooled per sample. Plants had been exposed to one of three treatments: (1) an empty clip cage for 48h, (2) a clip cage for 48h with addition of 15 aphids in the last 6h, and (3) a clip cage with 15 aphids for 48h. Only fourth-instar nymphs and adult M. persicae aphids were used. Experiments were conducted simultaneously in a climate chamber (24±1 °C, 50–70% relative humidity, 8 h–16h light–dark photoperiod, and a light intensity of 120 μmol m⁻² s⁻¹), but in separate cages with an air circulation system that prevented contamination of plant volatiles between treatments (Menzel et al., 2014). Samples were harvested in two batches between 13.00 and 16.00h, immediately frozen in liquid nitrogen, and stored at −80 °C until processing. RNA was isolated and checked according to the description above (260/280 OD range: 2.0–2.2, 260/230 OD range: 1.9–2.3). Library preparation was performed with a TruSeq™ RNA Sample Prep Kit (Illumina^®) and between 11 million and 24 million single-end 50-bp reads were sequenced per sample with Illumina^® HiSeq^TM 2000 in three lanes, multiplexed with 12 samples per lane. Reads were cleaned from adaptors and trimmed to 51bp using the program Trimmomatic version 0.32 (Bolger et al., 2014). Quality control was performed with FastQC (http://www.bioinformatics.bbsrc.ac.uk/projects/fastqc). Reads were mapped to the TAIR10 Arabidopsis reference genome (https://www.arabidopsis.org/) with Tophat version 2.0.13, intron length 20–2000 (Trapnell et al., 2012; Trapnell et al., 2013). An index file was built with Bowtie 2 (Langmead et al., 2009). Transcript assembly, quantification, normalization and differential expression analysis were performed with Cufflinks, using the bias detection and correction algorithm, multi-read correction for reads mapping to multiple locations, and a minimum alignment count of 10. Treatments were compared both between plant lines (Col-0 versus mutant) and within plant line (empty clip cage versus 6h post-inoculation (hpi), empty clip cage versus 48 hpi, and 6 hpi versus 48 hpi). Only differentially expressed genes (false discovery rate Q-value<0.05) with an absolute fold change ≥2 (log2≥1) were taken into account. Differentially expressed genes were tested for overrepresentation of biological processes against a reference set including all transcripts in the complete data set with at least 1 count, using the application BiNGO in Cytoscape (Maere et al., 2005; Cline et al., 2007). Genes associated with cell-wall processes were selected and classified based upon their TAIR description, and the heatmap was constructed with the R package ‘gplots’ (Warnes et al., 2009).

Results

GWA mapping

To identify genes involved in resistance to the green peach aphid, M. persicae, GWA mapping was performed on 344 natural accessions of Arabidopsis, using a selection of approximately 214 000 SNPs (Atwell et al., 2010; Li et al., 2010; Horton et al., 2012). The behaviour of aphids was screened on these accessions with an automated video-tracking platform (Kloth et al., 2015). The number and duration of plant penetrations was estimated by analysing the location and movement of aphids on single leaf discs. It is known that aphids need on average 25min to penetrate the epidermis and mesophyll before they reach the vascular bundle (van Helden and Tjallingii, 1993; Tjallingii, 1994; Prado and Tjallingii, 2007). Therefore, we used the proportion of aphids making long probes (>25min) as a proxy for the success rate of phloem ingestion. The majority of the Arabidopsis accessions did not show indications of resistance to aphids, but on 10% of the accessions at least half of the aphids were unsuccessful in feeding after 4.5h of infestation (Fig. 1A and Supplementary Table S2). GWA mapping of aphid feeding behaviour revealed seven genomic regions with a −log₁₀(P) value above 4 and a heritability of 10% (Fig. 1B and Table 1). WRKY22 (At4g01250) was identified as a candidate gene in a 40kb region around a polymorphism with a −log₁₀(P) value of 4.7 (chromosome 4 position 543516). Other candidates in the region included a gene with unknown function (At4g01290), a methyltransferase and a gene (At4g01240) with an MYB-like domain (At4g01280). Resequenced data of 173 accessions (Cao et al., 2011) showed that WRKY22 contained one non-synonymous SNP in its coding region, and that most of the polymorphisms were confined to the introns and the promoter region (Fig. 1C). A silent SNP in the last exon and two SNPs in the promoter were correlated with aphid feeding behaviour (Fig. 1D). Both polymorphisms in the promoter coincided with an AT-hook DNA-binding motif of AHL20, a transcription factor involved in plant defence to bacteria (Lu et al., 2010). Because WRKY22 is involved in PAMP-triggered immune responses (Asai et al., 2002; Navarro et al., 2004) and its expression is induced by M. persicae and Brevicoryne brassicae aphids (De Vos et al., 2005; Barah et al., 2013), we conducted further experiments to assess whether WRKY22 is involved in resistance to aphids.

Fig. 1.

Genome-wide association mapping of aphid feeding behaviour. (A) Phenotypic distribution of the proportion of M. persicae aphids making long probes (>25min) during a 1.5h recording on plants from 344 natural Arabidopsis accessions 4.5h post-inoculation. For accessions for which the percentage of long probes was below the dotted line, at least half of the aphids were unsuccessful in feeding. (B) Genome-wide associations with 214 000 SNPs. SNPs in red are positioned in a 40kb region around WRKY22 (highest −log₁₀(P)=4.7). (C) All SNPs in WRKY22 and its 1000kb promoter region according to 173 resequenced Arabidopsis accessions (green: silent; red: non-synonymous). Predicted gene domains are shown in grey, unknown domains in black. Triangles represent T-DNA insertions. (D) One synonymous SNP in the last exon and two SNPs in the promoter had an effect on aphid feeding behaviour (*P<0.05, **P<0.01, Student’s t-test, chromosome 4, positions 523037, 524726 and 525079).

Table 1.

SNPs and corresponding genes associated with the proportion of aphids making long probes (>25min, −log ₁₀(P) value>4)

Only the highest scoring SNP is shown per gene. Genes were grouped in one linkage disequilibrium (LD) region, if they were located within 20kb from each other. Chr.: chromosome.

LD region	Chr.	Position	−log10(P)	AGI code	Description
1	1	28995670	6.6	At1g77160	Protein of unknown function (DUF506)
2	2	10866313	4.3	At2g25530	AFG1-like ATPase family protein
3	3	20709836	4.2	At3g55800	Chloroplast enzyme sedoheptulose-1,7-bisphosphatase (SBPase)
4	4	519513	4.1	At4g01240	S-Adenosyl-l-methionine-dependent methyltransferase superfamily protein
4	4	536493	4.1	At4g01280	Homeodomain-like superfamily protein, SANT DNA-binding MYB-like domain
4	4	543516	4.7	At4g01290	Unknown protein
5	4	6641192	4.5	At4g10790	UBX domain-containing protein
5	4	6644022	4.9	At4g10800	BTB/POZ domain-containing protein
6	5	15927540	4.9	At5g39770	Pseudogene homologous to AtMSU81, restriction Endonuclease
7	5	19854700	4.1	At5g48965	Mutator-like transposase family
7	5	19858466	4.1	At5g48970	Mitochondrial substrate carrier family protein

Mesophyll-located susceptibility to aphids

To validate the previously reported induction of WRKY22 by aphid infestation (De Vos et al., 2005; Barah et al., 2013), RT-qPCR was performed on wild-type plants with aphids and without aphids. WRKY22 expression was unaffected at 6h post-infestation (hpi), and showed a non-significant increase at 48 hpi (Fig. 2A). Two wrky22 transfer (T)-DNA insertion lines and two WRKY22-inducible overexpression lines (Coego et al., 2014) were selected for further experiments. RT-qPCR confirmed that both T-DNA lines were true knockouts, and that the overexpression lines showed a 3- to 5-fold up-regulation of WRKY22 at 24h after induction with oestradiol (Fig. 2B). For a detailed insight into aphid feeding behaviour on knockout and overexpression lines, we used electrical penetration graph (EPG) recordings (McLean and Kinsey, 1964; Tjallingii, 1988). Aphid feeding behaviour was affected on both wrky22 knockout lines; on wrky22-3 aphids spent almost 20% more time on penetrating the epidermis and mesophyll, and on wrky22-4 aphids showed an hour’s delay in reaching the vascular bundle compared with the wild-type (Col-0) (Fig. 3A–C and Supplementary Table S3). One of the WRKY22-inducible overexpression lines showed the opposite trend, with aphids arriving almost an hour earlier at the vascular bundle compared with the wild-type (Fig. 3D–F and Supplementary Table S4). The other WRKY22 overexpression line did not show any differences compared with the wild-type. The total time of phloem ingestion was not affected in any of the (mutant) lines, suggesting that in the first 8h of infestation, the overall effects are small and confined to activities in the epidermis and/or mesophyll. An aphid population development assay on wrky22-3 and wrky22-4 showed that after 2 weeks of infestation, aphid populations were approximately 20% smaller on the knockouts compared with the wild-type (Fig. 4). Both behavioural experiments and population assays indicate that WRKY22 increases susceptibility to M. persicae aphids.

Fig. 2.

WRKY22 expression. (A) WRKY22 expression in the wild-type without M. persicae aphids (control) and after 6 and 48h of aphid infestation. (B) Expression in the wild-type (Col-0), wrky22-3 and wrky22-4 knockout lines, and WRKY22-inducible overexpression lines OE.c and OE.e. Overexpression lines were induced with oestradiol 24h before sampling (one-way ANOVA and Student’s t-test; different letters refer to significant differences).

Fig. 3.

Aphid behaviour on wrky22 knockout lines (upper panels) and WRKY22 overexpression lines (lower panels). (A, D) The total time M. persicae aphids were penetrating the epidermis and mesophyll during 8-h recordings on knockout lines wrky22-3 and wrky22-4 (A), and overexpression lines OE.c and OE.e (D). (B, E) Time between the start of the recording and the first contact with either a phloem or xylem bundle measured on knockout (B), and overexpression lines (E). (C, F) The total time aphids were ingesting phloem on knockout (C), and overexpression lines (F); knockout and overexpression lines were compared with the wild-type with Mann–Whitney U-test (*P<0.05; **P<0.01). To test the effect of overexpression, all plants were induced with oestradiol 24h before the assay.

Fig. 4.

Aphid population size on wild-type and knockout plants. The total number of M. persicae aphids per plant was counted 2 weeks after infestation with one neonate aphid. Mutant lines were compared with the wild-type with Student’s t-test (*P<0.05; **P<0.01).

Transcriptomic signature of wrky22-3

To study the role of WRKY22 in resistance to aphids, mRNA sequencing (RNA-seq) analysis was performed on wrky22-3 and wild-type plants that had received one of three treatments: (1) an empty clip cage for 48h, (2) a clip cage for 48h with addition of aphids in the last 6h, and (3) a clip cage with aphids for 48h. For each treatment three biological replicates were sampled, each consisting of a pool of five leaves from different plants. Samples were sequenced for single end 50-bp reads, and for each sample at least 9.5 million reads mapped to unique loci on the Arabidopsis reference genome (Supplementary data file S1). The total number of differentially expressed (DE) genes increased with the duration of infestation, from approximately 700 at 6 hpi to 1000 at 48 hpi in the wild-type. In both treatments, wrky22-3 contained twice as many up- and down-regulated genes as the wild-type (Fig. 5A). Principal component analysis showed that the duration of infestation was the major factor explaining differential expression (Fig. 5B–D). Highly abundant transcripts that were up-regulated in wrky22-3 compared with the wild-type included the JA reporter VSP1 (6 hpi) and pathogenesis-related genes such as PR2 and PR5 (48 hpi) (Fig. 6). Photosynthesis- and water-transport-related genes were down-regulated in wrky22-3 at 48 hpi (Fig. 6). Gene ontology (GO) enrichment analysis of the total set of DE genes revealed an overrepresentation of up-regulated JA-, SA- and abscisic acid (ABA)-responsive genes in wrky22-3 at both 6 and 48 hpi (Fig. 7A). The JA pathway was mainly characterized by up-regulation of genes of the ethylene response factor (ERF) branch (Vos et al., 2015), e.g. the AP2/ERF transcription factor RAP2.6, and PDF1.2 (Table 2). The majority of the DE genes associated with JA and ABA showed a peak at 6 hpi in wrky22-3, whereas most SA-responsive genes reached their highest level at 48 hpi (Fig. 7B). The induction of SA-responsive genes in the T-DNA line at 48 hpi coincided with a suppression of genes associated with auxin (AUX) responsiveness, plant growth and cell wall loosening (Fig. 7A, B).

Fig. 5.

Differentially expressed (DE) genes between treatments with and without aphids in the wild-type and wrky22-3. (A) The number of DE genes between control and infestation treatments (green bars: up-regulated, red bars: down-regulated). (B) Biplot of the two first principal components of differentially expressed genes between control and infestation treatments (DE genes ≥2-fold). (C) Overlap in up-regulated genes, and (D) down-regulated genes.

Fig. 6.

Gene transcripts of aphid-infested wild-type and wrky22-3 plants. Differentially expressed genes between wild-type and knockout plants (≥2-fold change) are shown in red. Axes depict the square-root transformation of the normalized number of transcripts (the number of fragments per kilobase of transcript per million reads mapped (FPKM)); genes ≤2500 FPKM are shown (including all DE genes in the dataset). Annotations include gene name and biological process; wounding: wound responsive; pathogens: pathogen responsive; cadmium: responsive to cadmium; abiotic: responsive to several abiotic stresses.

Fig. 7.

Enriched biological processes in wrky22-3. (A) Over-representation of biological processes in the knockout relative to the wild-type. Balloons refer to a process, or to the biosynthesis of, or responsiveness to the respective compound (SA: salicylic acid; JA: jasmonic acid; ABA: abscisic acid; AUX: auxin; ET: ethylene; CK: cytokinin; clock: circadian clock). Balloon colour indicates enrichment in the knockout (green: up-regulated; red: down-regulated; green/red: both up- and down-regulated; the total number of DE genes is depicted below the charts). (B) Relative expression patterns between treatments within each plant line. Only the dominant pattern (≥50% of the genes) of significant perturbations (≥2-fold, q-value <0.05) between treatments with and without aphids is shown. (Wall: cell wall loosening; n: number of genes associated with the biological process; e: empty clip cage; 6: 6 hpi.)

Table 2.

Differentially expressed genes (≥2-fold) of over-represented biological processes in wrky22-3 relative to the wild type

GO enrichment and gene classification are according to the BiNGO Cytoscape app (SA: salicylic acid; JA: jasmonic acid; ABA: abscisic acid; AUX: auxin; ET: ethylene; CK: cytokinin) (Cline et al., 2007; Maere et al., 2005).

Process	Treatment	Direction	Name	AGI code	Description
ABA	Empty cage, 6 hpi	Up	ANNAT4	At2g38750	Annexin, Golgi-mediated secretion
	6 hpi	Up	HAI1	At5g59220	HIGHLY ABA-INDUCED PP2C gene 1
	6 hpi	Up	HD-Zip-I	At3g61890	Homeodomain leucine zipper class I
	48 hpi	Up	ACR8	At1g12420	ACT DOMAIN REPEAT 8
	48 hpi	Up	AMY1	At4g25 000	ALPHA-AMYLASE-LIKE 1, starch mobilization
	48 hpi	Up	Dehydrins	At3g50970, At1g20440	Membrane located, freeze tolerance
	48 hpi	Up	ERF48	At2g40340	ABA responsive AP2/ERF transcription factor
	48 hpi	Up	LTI78	At5g52310	LOW-TEMPERATURE-INDUCED 78
	48 hpi	Up	WRKY63	At1g66600	ABA responsive WRKY transcription factor
AUX	6, 48 hpi	Down	CCA1	At2g46830	Negative regulator of circadian rhythm
	48 hpi	Down	AXR3	At1g04250	AUXIN RESISTANT 3
	48 hpi	Down	GH3s	At2g47750, At5g13360	GH3 auxin responsive gene family
	48 hpi	Down	SAURs	At1g20470, At1g29500, At1g29510, At3g03820, At4g22620, At4g38840, At4g38850, At4g38860, At5g18020, At5g18030, At5g18050	SAUR(-like) auxin-responsive proteins
CK	48 hpi	Down	ARRs	At1g19050, At1g74890, At3g57040, At5g62920	Arabidopsis response regulator (ARR) family
JA	Empty cage, 6, 48 hpi	Up	JAZs	At2g34600, At5g13220, At1g17380, At1g19180	JAZ7, JAZ10, JAZ5, JAZ1, Jasmonate-Zim- domain proteins
	Empty cage, 6, 48 hpi	Up	MDHAR4	At3g09940	Monodehydroascorbate reductase
	Empty cage, 6, 48 hpi	Up	MYB47	At1g18710	JA-responsive MYB transcription factor
	Empty cage, 6, 48 hpi	Up	TAT3	At2g24850	Tyrosine aminotransferase, JA responsive
	Empty cage	Up	VSP2	At5g24770	VEGETATIVE STORAGE PROTEIN 2
	Empty cage, 6, 48 hpi	Up	VSP1	At5g24780	VEGETATIVE STORAGE PROTEIN 1
	Empty cage, 6 hpi	Up	AOCs	At3g25760, At3g25780	Allene Oxide Cyclase family, JA biosynthesis
	Empty cage, 6 hpi	Up	OPR3	At2g06050	OXOPHYTODIENOATE-REDUCTASE 3, JA biosynthesis
	Empty cage	Up	EXT4	At1g76930	Extensin
	Empty cage	Up	JR1	At3g16470	JASMONAtE RESPONSIVE 1
	6, 48 hpi	Up	PDF1.2	At5g44420	PLANT DEFENSIN 1.2
	6 hpi	Up	DAD1	At2g44810	DEFECTIVE ANTHER DEHISCENCE 1, JA biosynthesis
	6 hpi	Up	JAR1	At2g46370	Jasmonate-amido synthetase
	6 hpi	Up	LOX3	At1g17420	LIPOXYGENASE 3
	6 hpi	Up	RAP2.6	At1g43160	AP2/ERF transcription factor
SA	6, 48 hpi	Up	GRX480	At1g28480	Glutaredoxin family, suppresses PDF1.2
	6, 48 hpi	Up	LURP1	At2g14560	Resistance to Hyaloperonospora parasitica
	6, 48 hpi	Up	WRKY18	At4g31800	WRKY18
	48 hpi	Up	WRKYs	At5g01900, At5g22570	WRKY38, WRKY62
	48 hpi	Up	MYB77	At3g50060	MYB77
	48 hpi	Up	WAK1	At1g21250	CELL WALL-ASSOCIAtED KINASE 1
JA, SA, ABA	6 hpi	Up	CIR1	At5g37260	MYB transcription factor
	48 hpi	Up	MYBs	At1g06180, At1g57560, At5g67300, At2g16720	MYB13, MYB50, MYB44, MYB7
	48 hpi	Up	MPK11	At1g01560	MAP KINASE 11
	48 hpi	Up	PDR12	At1g15520	ABC transporter family, MAPK cascade
Camalexin	Empty cage, 48 hpi	Up	PAD3	At3g26830	PHYTOALEXIN DEFICIENT 3, camalexin biosynthesis
	48 hpi	Up	P450	At4g39950	Cytochrome P450, indo-3-acetaldoxime (IAOx) biosynthesis
Terpenoids	Empty cage, 6, 48 hpi	Up	TSP4	At1g61120	TERPENE SYNTHASE 4
	Empty cage	Up	TPS10	At2g24210	TERPENE SYNTHASE 10
Cell wall	48 hpi	Down	Expansins	At1g20190, At1g26770, At1g69530, At2g20750, At2g40610	Expansin family, cell wall loosening and multidimensional cell growth

Enhanced negative JA–SA crosstalk in wrky22-3

Upon aphid infestation, the wrky22-3 transcriptome showed evidence of initial suppression, but eventual up-regulation of SA signalling. The expression of PR1, a robust SA-reporter gene (Pieterse et al., 2012), was 2-fold down-regulated at 6 hpi, but 3.5-fold up-regulated at 48 hpi in wrky22-3 compared with the wild-type. Transcript levels of JA-reporter genes PDF1.2 and VSP1 were consistently more abundant in wrky22-3 (Table 3), suggesting a possible role of negative JA–SA crosstalk (Spoel and Dong, 2008; Pieterse et al., 2012). A potential antagonizing candidate is NIMIN-2, encoding an SA-suppressing protein (Weigel et al., 2005), which was up-regulated in wrky22-3 6 hpi (Table 3). Apart from SA antagonism, there were also signs of JA antagonism. Several up-regulated genes in wrky22-3, i.e. GRX480, WRKY51, and WRKY62 (Table 3), have previously been implicated as potential suppressors of JA signalling (Mao et al., 2007; Ndamukong et al., 2007; Gao et al., 2011).

Table 3.

JA- and SA-signalling-related gene expression in wrky22-3 plants compared with wild-type plants with and without aphids

Differentially expressed genes with at least 2-fold absolute change are shown (ns: not significant; emp: empty clip cage; SA/JA sig: SA/JA signalling; SA/JA suppr: suppression of SA/JA signalling).

			Fold change
Gene	Name	Role	Emp	6 hpi	48 hpi	Reference
VSP1	VEGETATIVE STORAGE PROTEIN 1	JA sig	13.0	2.5	2.8	(Anderson et al., 2004; Lorenzo et al., 2004)
VSP2	VEGETATIVE STORAGE PROTEIN 2	JA sig	7.5	ns	ns	(Anderson et al., 2004; Lorenzo et al., 2004)
PDF1.2, 1.2C	PLANT DEFENSIN 1.2A, 1.2C	JA sig	ns	2.3	3.5	(Lorenzo et al., 2003; Penninckx et al., 1998)
PR1	PATHOGENESIS-RELATED GENE 1	SA sig	ns	0.4	3.5	(van Loon et al., 2006)
NIMIN-1	NIM1-INTERACTING 1	SA suppr	ns	ns	2.5	(Weigel et al., 2001; Weigel et al., 2005)
NIMIN-2	NIM1-INTERACTING 2	SA suppr	ns	2.0	2.5	(Weigel et al., 2001; Weigel et al., 2005)
GRX480	Glutaredoxin	JA suppr	ns	2.8	2.8	(Ndamukong et al., 2007)
WRKY51	WRKY DNA-BINDING PROTEIN 51	JA suppr	ns	ns	4.3	(Gao et al., 2011)
WRKY62	WRKY DNA-BINDING PROTEIN 62	JA suppr	ns	ns	2.8	(Mao et al., 2007)

JA induction by mechanostimulation in wrky22-3

Remarkably, the treatment with empty clip cages changed the expression of almost 150 genes in the knockout relative to the wild-type. Most of them were up-regulated in wrky22-3 and showed a significant overrepresentation of JA-responsive genes, including VSP2 of the touch- and wound-responsive MYC-branch of the JA signalling pathway (Lange and Lange, 2015) (Fig. 7 and Table 2). In order to see if mechanostimulation by clip cages had affected the plant’s response to aphids, RT-qPCR was conducted on aphid-infested leaves without clip cages (see Materials and methods). We found that in clean wrky22-3 plants, PDF1.2 expression was lower compared with wild-type plants (Fig. 8). After 6h of aphid infestation, PR1 was up-regulated in wrky22-3 while PDF1.2 only showed a non-significant increase (Fig. 8). These results suggest that, in the RNA-seq analysis, the over-represented JA response may have been clip cage-induced and was likely involved in the suppression of the SA response to aphids at 6 hpi.

Fig. 8.

The effect of aphid infestation without clip cage on the expression of JA and SA reporter genes. RT-qPCR measurements of expression of the JA reporters VSP2 (A) and PDF1.2 (B), and the SA reporter PR1 (C) in wild-type and wrky22-3 plants (Student’s t-test and the Mann–Whitney U-test; different letters denote significant differences). Aphids were contained on the leaves without inflicting major mechanical stimulation (see Materials and methods).

Known resistance genes

Several genes described in earlier studies as being involved in enhancing resistance to M. persicae were up-regulated in wrky22-3. Expression of the MYB-domain-containing protein MYBR1, found to affect M. persicae reproduction under the influence of harpin proteins (Liu et al., 2010), was more than 2-fold induced at 48 hpi (Supplementary data file S1). Also several other genes with affiliation to known resistance factors for M. persicae were up-regulated, such as the phloem protein PP2-A12 at 48 hpi, the myrosinase-binding protein MBP1 at 48 hpi, and several xyloglucan endotransglucosylases/hydrolases (XTHs) at 6 and 48 hpi (Mewis et al., 2005; Divol et al., 2007; Zhang et al., 2011; Louis and Shah, 2013).

Differential expression of cell wall-related genes

After 48h of aphid infestation, the wrky22-3 transcriptome was characterized by down-regulation of genes associated with cell wall loosening (Fig. 7 and Table 2). To assess all cell wall-related processes, DE genes with cell wall annotation were selected and grouped into categories based on their name and function (Fig. 9). We did not observe an up-regulation of touch-responsive xyloglucan endotransglucosylases/hydrolases (XTHs) by the empty clip cage. The empty clip cage did, however, cause a 3- to 5-fold up-regulation of the cellulose synthase-like genes CSLA1, CSLA10, and CSLA15, involved in hemicellulose biosynthesis (Liepman et al., 2005). Aphids up-regulated XTHs and down-regulated, for example, expansins, involved in cell-wall loosening, and pectin lyases, involved in pectin breakdown. While cell-wall loosening is a prerequisite for cell elongation, a process mainly regulated by CK and AUX (Taiz, 1984; Yadav et al., 2009; Albersheim et al., 2011), the transcriptomic patterns indicate an aphid-induced arrest of symplastic cell growth in wrky22-3.

Fig. 9.

Expression of cell-wall related genes in wrky22-3 and the effect of mechanostimulation by empty clip cages and aphid infestation. Genes and treatments are clustered according to the number of differentially expressed genes (≥2-fold change), using Ward’s minimum variance method (red: down-regulated; green: up-regulated in wrky22-3 compared with the wild-type).

Discussion

The effect of WRKY22 on M. persicae aphids

Plant responses to aphids are known to involve many transcriptional perturbations including multiple phytohormonal pathways (De Vos et al., 2005; De Vos et al., 2007; Smith and Boyko, 2007; Foyer et al., 2015). It is, therefore, a challenge to unravel the genetic basis of effective defence mechanisms against aphids. In this study, we explored natural variation in Arabidopsis to find genes related to impaired feeding behaviour of M. persicae. Natural variation in the occurrence of long probes, a proxy for the success rate of phloem ingestion, was associated to several genomic regions, including the WRKY22 locus. Polymorphisms in the WRKY22 promoter and in the last exon most strongly correlated with variation in aphid feeding behaviour. Even though the associations had low statistical power and heritability, knockout lines confirmed an effect of WRKY22 on aphid performance. Without a functional WRKY22 protein, it was more difficult for aphids to penetrate the epidermis and mesophyll and they arrived later at the vascular bundle. One WRKY22 overexpression line showed the opposite trend, although the impact was smaller than in the wrky22 knockouts, most likely due to the moderate extent and short time frame of the overexpression (3- to 5-fold change, induced 24h before the experiments). The effects of WRKY22 on aphid performance were marginal in the first 8h, but more substantial after an infestation period of 2 weeks as reflected by aphid population size. Overall, these assays indicate that WRKY22 promotes susceptibility to M. persicae aphids. Our RNA-seq and RT-qPCR analyses indicate that WRKY22-mediated susceptibility is associated with the suppression of SA signalling. Aphid infestation of wrky22-3 plants resulted in faster and potentially stronger up-regulation of the SA pathway than in wild-type plants. Even though it has been described that JA-induced defences are most effective against aphids (De Vos et al., 2007; Walling, 2008), SA-induced mechanisms have been shown to have a detrimental impact on aphids as well (Li et al., 2006; Moloi and Westhuizen, 2006). The down-regulation of pectin lyases and expansins (Fig. 9) suggests that there is less degradation of pectin and less loosening of the cell wall matrix in the wrky22 mutant. Fortification of the primary cell wall may have hampered the penetration of the mesophyll apoplast by aphids. This would explain why aphids required more time in probing the epidermis and mesophyll and were delayed in reaching the vascular bundle on the wrky22 mutants (Fig. 3 and Supplementary Table S3). We can, however, not exclude the involvement of other resistance factors in the mesophyll, such as the accumulation of reactive oxygen species or secondary metabolites.

Involvement of WRKY22 in biotic and abiotic stress responses

Apart from WRKY22’s responsiveness to aphids, we observed a strong activation of the JA pathway in wrky22-3 as a result of the use of clip cages. JA accumulation is known to be induced by mechanical stimuli (Ichimura et al., 2000; Chehab et al., 2012), wounding, and damage-associated molecular patterns (DAMPs) (Doares et al., 1995; Denoux et al., 2008; Vidhyasekaran, 2014). Since there were no obvious signs of plant damage, the up-regulation of the JA pathway was most likely triggered by touch-induced surface stimulation of our samples. WRKY22 has been shown to be induced in response to touch and wounding (Lee et al., 2004; Kilian et al., 2007). Our data suggest that WRKY22 acts as a suppressor of JA signalling in response to these stimuli. Many other abiotic stimuli have been described to induce WRKY22 as well, such as prolonged darkness, submergence, cold acclimation, light perception, salinity, potassium starvation, and exposure to ozone (Folta et al., 2003; Hampton et al., 2004; Monte et al., 2004; Lee et al., 2005; Tosti et al., 2006; Chawade et al., 2007; Kilian et al., 2007; Zhou et al., 2011; Göhre et al., 2012; Hsu et al., 2013; Kim et al., 2013; Sugimoto et al., 2014). Hsu et al. (2013) identified several potential downstream targets of WRKY22, including genes involved in drought resistance and phosphate starvation. The accumulating evidence for its involvement in abiotic stress responses warrants a change of view, i.e. that WRKY22 is not solely involved in PTI. A parallel can be drawn with WRKY40, previously known as a repressor of PTI but recently also recognized as a central player in ABA inhibition during abiotic stress (Chen et al., 2010; Friedel et al., 2012). Although abiotic and biotic stimuli are most likely perceived via stress-specific mechanisms and require differential plant responses, signal-transduction pathways might converge via common regulators, such as WRKY22, in order to fine-tune the interplay between phytohormones.

SA–JA signal integration

One of the major questions is whether WRKY22 is an activator or repressor of SA and JA signalling. Our transcriptome analysis of wrky22-3 revealed up-regulation of JA signalling upon mechanostimulation and up-regulation of SA signalling upon aphid infestation. This would suggest that in wild-type plants, WRKY22 is a suppressor of JA and SA signalling. From previous studies we know, however, that WRKY22 and WRKY29 confer resistance to (hemi)biotrophic and necrotrophic pathogens (Asai et al., 2002; Hsu et al., 2013), and that they are induced by PAMP-triggered MAPK cascades which result in the activation of SA, JA and ET signalling (Zipfel et al., 2004). There is no direct evidence that WRKY22 and its homologue WRKY29 induce SA, JA and ET signalling, and the possibility exists that they are involved in MAPK-triggered processes independent of SA and JA signalling. Nevertheless, their requirement for PTI makes them unlikely candidates for consistent suppression of plant defence hormones. Rather, WRKY22 could be an integrator of SA and JA signals, inhibiting or enforcing both pathways, depending on their interaction with other transcription factors and signalling pathways (Fig. 10). Similarly, WRKY70 has been proposed to be capable of inducing and inhibiting both SA and JA signalling, depending on the strength of the induction (Li et al., 2004; Ülker et al., 2007). Although many questions remain with regard to the underlying mechanism, our study shows that WRKY22 plays a role in both SA and JA signalling and is involved in transcriptional reprogramming in response to mechanostimulation and aphid infestation. To understand the function of WRKY22, its transcriptional network needs to be further unravelled under multiple biotic and abiotic stress conditions. With respect to aphids, WRKY22 increases susceptibility. Its responsiveness to aphid infestation and its potential to suppress JA and SA signalling would make WRKY22 an excellent target for aphids to manipulate JA- and SA-dependent host plant defences for their own benefit.

Fig. 10.

Hypothetical model of WRKY22’s role in plant response to abiotic (left) and biotic (right) stresses. Changes in, for example, light, temperature and touch are perceived via sensors and ion channels; plant invasion by organisms such as bacteria and aphids is mainly perceived via pattern-recognition (PR) receptors. These stimuli induce WRKY22 directly via MAPK cascades (Ichimura et al., 2000; Asai et al., 2002), or indirectly via SA accumulation (Miao et al., 2004; Miao and Zentgraf, 2007). Alternatively, aphid effectors secreted via the saliva may induce WRKY22 via PR-receptor-independent routes. WRKY22 subsequently integrates signalling of the JA and SA pathway, by inhibiting or activating specific transcription factors (TFs) and other regulatory genes.

Supplementary data

Supplementary data are available at JXB online.

Data file S1. Differentially expressed genes between wrky22-3 and wild-type rosette leaves with and without aphid infestation.

Table S1. Primers used for PCR and RT-qPCR.

Table S2. Percentage of aphids making long probes (>25min) 4.5 hour after inoculation on 344 natural Arabidopsis accessions.

Table S3. Aphid feeding behaviour, measured by 8-hour EPG recordings on wild-type (Col-0) and wrky22 T-DNA lines (wrky22-3 and wrky22-4).

Table S4. Aphid feeding behaviour, measured by 8-hour EPG recordings on wild-type (Col-0) and wrky22-inducible overexpression lines (OE.c and OE.e).