Upon herbivore attack, plants activate an indirect defense, that is, the release of a complex mixture of volatiles that attract natural enemies of the herbivore. When plants are simultaneously exposed to two herbivore species belonging to different feeding guilds, one herbivore may interfere with the indirect plant defense induced by the other herbivore. However, little is understood about the mechanisms underlying such interference.
Here, we address the effect of herbivory by the phloem-feeding whitefly Bemisia tabaci on the induced indirect defense of Arabidopsis thaliana plants to Plutella xylostella caterpillars, that is, the attraction of the parasitoid wasp Diadegma semiclausum.
Assays with various Arabidopsis mutants reveal that B. tabaci infestation interferes with indirect plant defense induced by P. xylostella, and that intact jasmonic acid and ethylene signaling are required for such interference caused by B. tabaci. Chemical analysis of plant volatiles showed that the composition of the blend emitted in response to the caterpillars was significantly altered by co-infestation with whiteflies. Moreover, whitefly infestation also had a considerable effect on the transcriptomic response of the plant to the caterpillars.
Understanding the mechanisms underlying a plant's responses to multiple attackers will be important for the development of crop protection strategies in a multi-attacker context.
Plants defend themselves against herbivores through direct and indirect defense mechanisms. An important indirect defense employed by plants against insect herbivores is to emit complex blends of volatile compounds that attract carnivorous enemies of the herbivores (Dicke et al., 1990; Turlings et al., 1990; Dicke, 2009). Attraction of parasitoids or predators by herbivore-induced plant volatiles (HIPV) has been recorded for > 23 plant species from 13 families (Mumm & Dicke, 2010). The signal-transduction underlying herbivore-induced indirect plant defense is mainly mediated by the octadecanoid pathway, with the central phytohormone jasmonic acid (JA); the shikimate pathway, with the central phytohormone salicylic acid (SA); and the ethylene (ET) pathway (Ozawa et al., 2000; Kessler & Baldwin, 2002; Van Poecke & Dicke, 2002; Dicke et al., 2003). The JA signaling pathway is the most important signal-transduction pathway underlying the induction of HIPV (Ament et al., 2004; Kessler et al., 2004; Wei et al., 2011).
The majority of studies on induced plant defenses address the effect of a single herbivore species (Dicke et al., 2009). In nature, however, plants are commonly exposed to attack by multiple herbivores. Chewing herbivores predominantly activate JA-mediated defense responses, whereas phloem-feeding insects, such as whiteflies and aphids, mainly activate SA-mediated responses (Moran & Thompson, 2001; Kempema et al., 2007; Zarate et al., 2007), although exceptions do exist (Heidel & Baldwin, 2004; De Vos et al., 2005). Considering the well-demonstrated cross-talk between the JA and SA signaling pathways (Zarate et al., 2007; Koornneef & Pieterse, 2008), it is hypothesized that SA-inducing herbivores will attenuate the attraction of carnivorous enemies of attackers that induce the JA signaling pathway because the induction of the SA pathway will interfere with the emission of JA-related HIPV, and vice versa (Dicke et al., 2009). Recent studies seem to support this hypothesis. For instance, the amount of HIPV induced by caterpillars alone was significantly higher when the plants were co-infested with silverleaf whiteflies or pea aphids (Rodriguez-Saona et al., 2003; Schwartzberg et al., 2011). Furthermore, Lima bean plants simultaneously infested by spider mites and whiteflies emitted lower amounts of the induced predator-attracting terpene (E)-β-ocimene, and, consequently, the attraction of predatory mites to plants infested with both spider mites and whiteflies was attenuated (Zhang et al., 2009). To understand the mechanisms underlying interference, detailed analyses of the signal-transduction pathways mediating the volatile emission as well as a transcriptomic analysis are needed.
In the present study, we employed a model system consisting of Arabidopsis thaliana, the lepidopteran herbivore Plutella xylostella and its specialist parasitoid Diadegma semiclausum to investigate the effects of additional whitefly (Bemisia tabaci) feeding on P. xylostella-infested plants on the attraction of the parasitoid D. semiclausum and the underlying molecular mechanisms. The whitefly B. tabaci is a phloem-feeding insect that is known to induce SA-mediated defenses in Arabidopsis (Zarate et al., 2007). Plutella xylostella is a chewing insect that mainly induces JA-mediated defenses in Arabidopsis (Ehlting et al., 2008). First, we compared the response of D. semiclausum to volatiles from wildtype (WT) Arabidopsis and a number of transgenic mutants, including dde2-2 (a JA-deficient mutant; Von Malek et al., 2002), ein2-1 (an ET-insensitive mutant; Guzman & Ecker, 1990), NahG (an SA-deficient mutant; Delaney et al., 1994), and cpr-6 (constitutively high SA concentration; Clarke et al., 1998) infested with P. xylostella, B. tabaci, or both. Second, we compared the volatile emissions of WT Arabidopsis infested with P. xylostella, B. tabaci, or both. Third and finally, we addressed the global transcriptomic response of plants to infestation with P. xylostella, B. tabaci or both by microarray analysis. Our results demonstrate that intact JA and ET signaling are required for the interference by B. tabaci with the P. xylostella-induced parasitoid attraction in Arabidopsis. Whereas normal concentrations of SA are not needed for the interference, elevated concentrations of SA eliminated the interference, likely due to cross-talk with the JA pathway. Microarray analyses showed that B. tabaci strongly repressed the P. xylostella-induced transcriptional response.
Materials and Methods
Arabidopsis thaliana Heynh ecotype Columbia (Col-0) and the transgenic lines dde2-2, NahG, ein2-1 and cpr-6, were grown from seed in a climate chamber (23 ± 1°C, RH 50–60%, 8 h: 16 h (light : dark)). The transgenic lines are all in the Col-0 background. Two weeks later, the seedlings were transferred to plastic pots (5 cm diameter), containing a soil mixture that was autoclaved at 80°C for 2 h. In all experiments, 6–8-wk-old plants were used.
The lepidopteran herbivore Plutella xylostella L. (diamondback moth; Yponomeutidae) was reared on Brassica oleracea plants (var. gemmifera cv Cyrus) in a climate room (21 ± 1°C, RH 50–70%, 16 h: 8 h (light : dark)). The parasitoid wasp, Diadegma semiclausum Hellen, was reared on larvae of P. xylostella under glasshouse conditions (22 ± 2°C, RH 40–80%, 16 h: 8 h (light : dark)). Whiteflies (Bemisia tabaci (Gennadius); Hemiptera: Aleyrodidae) were maintained on tomato Solanum lycopersicum cv Moneymaker in a separate glasshouse compartment (25 ± 5°C, RH 50–70%, 16 h : 8 h (light : dark)). For bioassays, D. semiclausum cocoons were collected and transferred to a gauze cage in a climate room (21 ± 1°C, RH 50–70%, 16 h : 8 h (light : dark)). Emerging wasps were provided with water and honey, and were referred to as ‘naïve’ wasps as they did not have contact with plant material or caterpillars before the experiments; 3–7-d-old mated D. semiclausum females were used in the olfactometer tests.
Plants were subjected to the following treatments:
Jasmonic acid (JA) treatment: each plant was sprayed with 1 ml of 1.0 mM JA (Sigma-Aldrich) solution (containing 0.1% Tween 20). Treated plants were used for bioassays 48 h after JA application.
Caterpillar treatment: five first-instar larvae were transferred to the plant, and allowed to feed on the plant for 48 h.
Whitefly treatment: 50 adult whiteflies were introduced onto the plant that was kept in a cage (21.0 cm high, 13.5 cm diameter). The whiteflies were allowed to feed on the plant for 48 h.
Caterpillar plus whitefly treatment: five first-instar larvae were transferred to the plant that was placed in a cage (21.0 cm high, 13.5 cm diameter). Immediately after introducing the caterpillars, 50 adult whiteflies were introduced into the cage. The caterpillars and whiteflies were allowed to feed on the plant for 48 h.
Undamaged plant treatment: intact plants were sprayed with 1 ml of water (containing 0.1% Tween 20), and used as controls for comparison with JA-treated plants. Intact plants, that received no treatment, were used as controls for comparison with caterpillar- or whitefly-damaged plants.
During the treatments with chemical elicitor or herbivory, the undamaged plants were kept in a separate room to avoid volatile-mediated interference by herbivory- or JA-treated plants. For the subsequent bioassay, the caterpillars and whiteflies remained on the plant. For the microarray analysis, all rosette leaves were harvested from four plants as one biological replicate and flash frozen in liquid nitrogen.
The behavioral responses of parasitoids to plant volatiles were investigated in a Y-tube olfactometer (Takabayashi & Dicke, 1992). Two streams of purified air (filtered through activated charcoal) were each led through a 2 l glass container into the olfactometer arms at 4 l min−1. The base of the olfactometer was connected to a vacuum line generating a flow of 8 l min−1. The experiment started with the release of individual wasps at the base of the Y-tube. Each wasp was observed for a maximum of 10 min, and a choice for one of the two odor sources was recorded when the wasp reached the end of either arm and stayed in that arm for at least 15 s. When the wasp did not make a choice within 10 min, a ‘no choice’ was recorded. Each wasp was used only once. After testing five wasps, odor sources were interchanged to avoid any influence of unforeseen asymmetries in the setup. Four plants of each treatment were used as an odor source. In this experiment we examined the behavioral responses of parasitoids to the following odor comparisons: undamaged plants vs plants treated with single herbivore or JA, and plants infested with P. xylostella vs plants infested with P. xylostella and B. tabaci. This experiment was repeated on 3–4 d with 15 wasps per day for each odor comparison.
Volatiles emitted by Arabidopsis plants were collected using a dynamic headspace collection system in a climate room (22 ± 2°C, RH 60–70%). Two hours before volatile trapping, plants were removed from the pots, and the soil with the roots was carefully wrapped in aluminum foil. The four plants of the same treatment were placed together in a 2.5 l glass jar. The jars were closed with a Viton-lined inert glass lid having an inlet and an outlet. Air was sucked out using a vacuum pump at 100 ml min−1, and the incoming air was purified through a stainless steel cartridge filled with 200 mg Tenax-TA (20/35-mesh; Grace-Alltech, Deerfield, USA). A similar cartridge was used to trap emitted plant volatiles at the outlet. After 3.5 h of trapping under continuous light, the four plants were weighed (FW). Volatile trappings of plants of each treatment were carried out in parallel on each experimental day. For each treatment, volatile collection was repeated 3–5 times.
Chemical analysis of volatiles
Headspace samples were analysed with a Thermo TraceGC Ultra connected to a Thermo TraceDSQ quadrupole mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA). Before thermodesorption, traps were flushed with helium at 30 ml min−1 for 15 min to remove moisture and oxygen. After flushing, the collected volatiles were desorbed from the Tenax traps at 220°C (Ultra; Markes, Llantrisant, UK) for 5 min with a helium flow of 30 ml min−1. The released compounds were focused on an electrically cooled sorbent trap (Unity, Markes, Llantrisant, UK) at a temperature of 0°C. Volatiles were injected on the analytical column (30 m′ 0.25 mm ID′ 1.0-μm film thickness; ZB-5MSi, Zebron, Phenomenex) in a splitless mode by ballistic heating of the cold trap to 250°C for 7 min. The temperature program started at 40°C (7-min hold) and rose at 10°C min−1 to 280°C (2-min hold). The column effluent was ionised by electron impact (EI) ionisation at 70 eV. Mass scanning was performed from 33 to 300 m/z with a scan rate of 3.9 scans s−1. The eluted compounds were identified using Xcalibur software (Thermo, Waltham, USA) by comparing the mass spectra with those of authentic reference standards or with NIST 05 library spectra. Quantification of identified compounds was based on comparison with a set of authentic reference compounds injected in different concentrations ranging from 2.5 to 100 ng μl−1 hexane. One microliter of the reference compounds was injected on a tube filled with Tenax TA and the cartridge was then flushed with nitrogen 5.0 to remove the hexane.
Microarray experiments and data analysis
Total RNA extraction and purification was done using the RNeasy Plant Mini kit (Qiagen). One microgram of total RNA was linearly amplified using the Amino Allyl MessageAmp II aRNA Amplification kit (Ambion). Samples were labelled with Cy3 and Cy5 monoreactive dye (Amersham Bioscience) and dye incorporation was monitored by measuring the Cy3 and Cy5 fluorescence emissions using a nanodrop ND-1000 UV-Vis Spectrophotometer (Bio-Rad). Microarrays containing 70-mer oligonucleotides based on the genome of A. thaliana were obtained from the group of David Galbraith from the University of Arizona, Tucson, AZ, USA (http://www.ag.arizona.edu/microarray). These microarrays contain 29 110 probes from the Operon Arabidopsis Genome Oligo Set v3.0 (Operon). This oligo set represents 26 173 protein-coding genes, 28 964 protein-coding gene transcripts and 87 miRNAs. The majority of genes are represented by one 70-mer probe on the microarray. Immobilization of the array elements was performed according to the manufacturer's website. The arrays that we used all originated from the same printing batch, thus eliminating batch-to-batch variation. The hybridization mixture contained 100 pmol of the Cy3-labeled sample, 50 pmol of the Cy5-labeled sample, 2× SSC, 0.08% SDS, and 4.8 μl Liquid Block (Amersham) in a final volume of 80 μl. The solution was incubated at 65°C for 5 min before application to the microarray covered with a lifterslip (Gerhard Menzel, Braunschweig, Germany). The microarray was placed in a hybridization chamber (Genetix, New Milton, UK) and incubated at 50°C. After 12 h the microarray was washed for 5 min in 2× SSC, 0.5% SDS at 50°C, followed by a 5 min wash in 0.5× SSC at room temperature, and a final 5 min wash in 0.05× SSC at room temperature. The microarray was immediately dried by centrifugation for 4 min at 200 rpm. Hybridized microarrays were scanned with a ScanArray Express HT Scanner (PerkinElmer, Waltham, MA, USA). The hybridizations were done using a reference design using dye-swap pairs in such a way that all treatments were hybridized three times with Cy3 and three times with Cy5.
Mean fluorescent intensities for Cy3 and Cy5 were determined using the ScanArray Express software (Perkin-Elmer, Boston, MA). Each image was overlaid with a grid to assess the signal intensities for both dyes from each spot. Background fluorescence was subtracted and spots with adjusted intensities lower than half the background were manually raised to half the background to avoid extreme expression values. Spots were excluded from the analysis when they showed aberrant shape or were located in a smear of fluorescence. Average values per dye were calculated for each slide and expression values of each spot were corrected for the difference between this average and the average of Cy3 and Cy5 together over all slides. This type of normalization was done to remove any systematic dye effects. Normalized expression ratios for each individual spot and the mean of the four replicates were calculated and subsequently log2-transformed.
Binomial tests were performed to analyse the Y-tube olfactometer experiments. Parasitoids that did not make a choice were excluded from the analysis. The gene expression data obtained from the microarray experiments were statistically analysed by a one-way ANOVA followed by between subjects t tests (TIGR MeV). Changes in volatile emission and in the expression patterns of genes that were significantly different between treatments were analysed using projection to latent structures-discriminant analysis (PLS-DA) using the software program SIMCA-P v10.5 (Umetrics AB, Umea, Sweden).
PLS-DA is a projection method that separates groups of observations by rotating the PCAs such that a maximum separation among classes – here plant treatments – is obtained (Eriksson et al., 2006). The results of the analysis are visualized in score plots, which reveal the sample structure according to the model components, and loading plots, which display the contribution of the variables to these components, as well as the relationships among the variables. The program's cross-validation procedure evaluates the significance of each additional component (starting with none) by comparing the goodness of fit (R2) and the predictive value (Q2) of the extended model with that of the reduced model (Eriksson et al., 2006). Data were log-transformed, mean-centered and scaled to unit variance before they were subjected to the analysis. One-way ANOVA analyses were performed on the scores of the first two principle components with treatment as main factor. Post-hoc Tukey's HSD multiple comparison tests were performed to reveal significant differences among the means.
Choices of parasitic wasps in Y-tube olfactometer
In dual-choice tests, female Diadegma semiclausum wasps preferred the volatiles from Plutella xylostella-infested WT Arabidopsis over those from undamaged WT plants (binomial test, P < 0.001; Fig. 1a). Similarly, the wasps preferred the volatiles from JA-treated WT Arabidopsis over untreated WT plants (binomial test, P =0.02; Fig. 1a). To further assess the roles of JA, ET and SA signaling pathways in mediating the attraction of D. semiclausum, the response of this wasp to the volatiles from Arabidopsis mutants, including dde2-2 (JA-deficient), ein2-1 (ET-insensitive), NahG (SA-deficient) and cpr-6 (constitutively high SA concentration), was tested. The wasps did not discriminate between the volatiles from P. xylostella-infested dde2-2 plants and undamaged dde2-2 plants; by contrast, the wasps still showed a significant attraction to the volatiles from P. xylostella-infested ein2-1, NahG, and cpr-6 plants, over undamaged plants of the respective mutants (binomial test, P =0.02, P =0.05, P < 0.001, respectively; Fig. 1a). These results show that the JA, but not SA or ET, signaling pathway plays an important role in mediating the attraction of D. semiclausum to P. xylostella-infested plants.
Female D. semiclausum were not attracted to the volatiles from WT Arabidopsis infested with Bemisia tabaci, over those from undamaged WT plants (Fig. 1a).
Female D. semiclausum showed a significant preference for the volatile blend from P. xylostella-infested WT Arabidopsis over that from WT plants infested with P. xylostella plus B. tabaci (binomial test, P =0.01, Fig. 1b). This shows that the additional B. tabaci infestation interferes with the attraction of D. semiclausum to P. xylostella-infested WT Arabidopsis. In the test with JA-deficient dde2-2 plants, the wasps were not significantly more attracted to P. xylostella-infested plants than to plants infested with P. xylostella plus B. tabaci (Fig. 1b). Similarly, in the tests with either ein2-1 or cpr-6 plants, the wasps were not significantly more strongly attracted to P. xylostella-infested plants, when offered against the corresponding plants infested with P. xylostella plus B. tabaci (Fig. 1b). By contrast, in the test with NahG plants that are SA deficient, the wasps were more strongly attracted to P. xylostella-infested plants than to plants infested by P. xylostella plus B. tabaci (P =0.004, Fig. 1b). These data show that intact JA and ET signaling are required for B. tabaci's interference with P. xylostella-induced indirect defense, but not SA signaling. With regard to SA signaling, the elevated SA concentrations in cpr6 abolished the interference whereas an elimination of SA (in NahG plants) leaves the interference by B. tabaci intact. Thus, the enhanced presence of SA eliminates the interference with P. xylostella-induced indirect defense.
Reciprocal effects on feeding by caterpillars and whiteflies
It is important to keep in mind that B. tabaci infestation might influence the feeding behavior of P. xylostella on dual-infested plant, and vice versa. Such an influence could have resulted in a reduction in the damage inflicted by P. xylostella or B. tabaci on dual-infested plants, which might lead to a lower plant volatile emission, thus changing the attractiveness of the plant to parasitoids. However, we have observed that the amount of leaf damage inflicted by P. xylostella caterpillars was not affected by plant genotype, whitefly co-infestation or their interaction, neither when comparing Col-0 vs dde2-2 plants nor when comparing Col-0 vs ein2-1 plants (Fig. S1). Similarly, neither plant genotype, caterpillar co-infestation nor their interaction affected the oviposition rate of B. tabaci, neither when comparing Col-0 vs dde2-2 plants nor when comparing Col-0 vs ein2-1 plants (Fig. S2). These data indicate that the two herbivores did not interfere with each other's feeding behavior on the plant genotypes used (Notes S1). Thus, we can exclude the possibility that the interference with parasitoid attraction by whitefly feeding was due to a reduction in feeding intensity inflicted by P. xylostella or B. tabaci on dual-infested plants.
Volatile blends from Arabidopsis plants
Fifteen major compounds were detected in the volatile blends of the four plant treatments (Table S1). Plants infested by P. xylostella plus B. tabaci emitted the largest amounts of volatiles, followed by plants infested by B. tabaci, whereas control plants emitted the lowest amounts of volatiles. PLS-DA resulted in a model with five significant principal components (PCs; model statistics: R2X = 0.95, R2Y = 0.85 and Q2 = 0.62; ANOVA on scores of first PC: F3,12 = 13.8, P <0.001 and second PC: F3,12 = 19.5, P <0.001) of which the first two explained 73% of the variance and clearly separated the four treatments (Fig. 2a). The first PC significantly separated treatments with B. tabaci from treatments without B. tabaci regardless of whether P. xylostella was present (Tukey's HSD multiple comparison tests, P <0.05), whereas the second PC significantly separated treatments with P. xylostella from treatments without P. xylostella regardless of whether B. tabaci was present (Tukey's HSD multiple comparison tests, P <0.05; Fig. 2a). Undamaged control plants and plants treated with P. xylostella alone emitted volatiles in lower amounts compared to plants that were damaged by B. tabaci. Plants damaged by B. tabaci emitted relatively high amounts of α-farnesene, 1-nonanol, β-myrcene and various aldehydes (Fig. 2b). Plants damaged by P. xylostella emitted relatively high amounts of methyl salicylate and an unidentified compound and relatively little dodecane (Fig. 2b).
Gene expression changes in response to feeding by P. xylostella, B. tabaci, or both
Feeding by P. xylostella resulted in a significant transcriptional response of 85 genes of which expression was induced and five genes of which expression was repressed (Table S2; Fig. 3). Although B. tabaci feeding resulted in a weak transcriptional response of the plant, that is, 12 differentially expressed genes, this insect strongly repressed the P. xylostella-induced response when feeding simultaneously on the plant with the caterpillars (Fig. 3). The results also showed that four JA-responsive genes (At1g54040, At5g13330, At5g24780 and At5g43860) that were significantly induced upon P. xylostella infestation were not induced when B. tabaci was also feeding on the plant. The weak transcriptional response to B. tabaci feeding was also reflected in the results from a PLS-DA using all the genes that were significantly different between treatments. PLS-DA resulted in a model with six significant PCs (models statistics: R2X = 0.86, R2Y = 0.97 and Q2 = 0.75; ANOVA on scores of first PC: F3,12 = 36.3, P <0.001 and second PC: F3,12 = 47.8 P <0.001) of which the first two explained 63.5% of the variance and clearly separated the four treatments (Fig. 4a). The first PC significantly separated gene expression in the P. xylostella-treated plants from the other three treatments (control and treatments with B. tabaci; Tukey's HSD multiple comparison tests, P <0.05), whereas the second PC significantly separated treatments with P. xylostella plus B. tabaci from the single herbivory treatments (Tukey's HSD multiple comparison tests P < 0.05) and the controls from all herbivory treatments (Tukey's HSD multiple comparison tests, P <0.05; Fig. 4a). The loading plot (Fig. 4b) shows that infestation with P. xylostella affects the expression of many more genes compared to treatments with B. tabaci regardless of the presence of P. xylostella. When comparing the transcriptional response of P. xylostella-infested plants to that of plants simultaneously infested by P. xylostella and B. tabaci, 72 and 17 genes were repressed or activated, respectively, due to the presence of B. tabaci on P. xylostella-infested plants (Table S4). Genes repressed by additional whitefly feeding code mainly for enzymes involved in protein metabolism or for transcription factors (Table S4). Remarkably, three genes (At1g22410, At1g25220 and At4g39950) repressed by B. tabaci are involved in tryptophan biosynthesis, which is a pathway that can lead to the production of indole glucosinolates (www.genome.jp/kegg/pathway/map/map00400.html). By contrast, genes that were activated by additional whitefly feeding code for proteins involved in protein metabolism, transport, or transcription (Table S4).
In order to validate the microarray data, we selected four genes that showed significantly different expression levels between plants infested by P. xylostella and plants infested by P. xylostella and B. tabaci, to be analysed with qRT-PCR (see Table S3 for their respective primers). The qRT-PCR and microarray analyses showed similar expression patterns among the four treatments (Fig. S3; Notes S1), indicating the reliability of the microarray data.
In the present study, we demonstrate that additional whitefly infestation of Arabidopsis interferes with the attraction of the parasitic wasp Diadegma semiclausum to volatiles emitted from Plutella xylostella-infested Arabidopsis plants. This result is consistent with a previous finding that Bemisia tabaci infestation also interferes with indirect plant defense induced by spider mites that induce JA-signaling in Lima bean plants (Zhang et al., 2009). The present study further shows that whitefly infestation has strong effects on the composition of the blend of plant volatiles induced by P. xylostella, as well as gene transcription. These results indicate that whitefly infestation strongly modifies plant responses induced by P. xylostella and, consequently, attenuates the attraction of a parasitic wasp that attacks P. xylostella caterpillars. Headspace analyses support the behavioral data on parasitoid attraction. Infestation of plants with B. tabaci, P. xylostella or both herbivore species resulted in the emission of different volatile blends, which affected the attraction of D. semiclausum wasps. It is well-known that the composition of the blend of volatiles emitted by a plant affects the degree of attraction of carnivorous arthropods (De Boer et al., 2004; Rasmann & Turlings, 2007; Zhang et al., 2009).
A few studies have demonstrated that phloem-feeding insects affected indirect defense induced by other herbivores (Moayeri et al., 2007; Dicke et al., 2009; Zhang et al., 2009; Soler et al., 2012). The effects of phloem-feeding insects on indirect plant defense are mostly explained at the level of behavioral responses to HIPV (Rodriguez-Saona et al., 2003; Zhang et al., 2009). However, little is understood about the molecular mechanisms underlying the impact of phloem-feeding insects on indirect defense of plants in response to simultaneous attack by a herbivore species from a different feeding guild. In Lima bean plants, B. tabaci infestation interferes with the spider-mite-induced predator attraction through reducing the emission of (E)-β-ocimene, which is caused by interference with the JA signal-transduction pathway (Zhang et al., 2009). From studies on cross-talk between signal-transduction pathways involved in induced direct defenses against pathogens, it has become clear that the SA pathway and JA pathway influence each other negatively (Koornneef & Pieterse, 2008). In tomato, aphid feeding interfered with the induction of genes by caterpillar feeding, but no effects were recorded at the biochemical level and in terms of caterpillar performance (Rodriguez-Saona et al., 2010). The present study exploited the potential of the molecular genetic model plant Arabidopsis to investigate the involvement of signal-transduction pathways in the interference with indirect plant defense by whitefly feeding. This showed that JA signaling is important for P. xylostella-induced attraction of the parasitoid. This is in agreement with previous findings that the JA signaling pathway is the most important signal-transduction pathway mediating indirect plant defense (Thaler et al., 2002; Ament et al., 2004; Kessler et al., 2004) and that P. xylostella induces the JA signal-transduction pathway (Ehlting et al., 2008). Upon disruption of the JA signal-transduction pathway the parasitoids no longer discriminated between odors from plants with caterpillars plus whiteflies and odors from plants with caterpillars only. The latter is likely due to the elimination of JA-dependent induction of parasitoid attraction in caterpillar-infested plants. Yet, JA-mediated induction by P. xylostella caterpillars does not seem to be the only factor underlying induced parasitoid attraction. Although disrupting ET and SA signaling does not affect parasitoid attraction to plants infested by P. xylostella only, intact ET signaling is necessary for interference with parasitoid attraction by whitefly feeding. Moreover, enhanced constitutive SA production also eliminates whitefly interference. This may be mediated by JA–SA crosstalk, although such crosstalk did not interfere with the induction of parasitoid attraction by P. xylostella infestation that was found to be unchanged in the cpr-6 mutant plants. The JA signal-transduction pathway is composed of different subpathways that are differentially affected by other phytohormones (Pieterse et al., 2009). It will be interesting to investigate how different JA subpathways are involved in the induction of HIPV by P. xylostella caterpillars, on the one hand, and in interference by B. tabaci feeding, on the other. It has become clear that interactions among signal-transduction pathways are not only dependent on direct JA–SA crosstalk (Pieterse et al., 2009; Robert-Seilaniantz et al., 2011). Phytohormonal signal signatures influence transcriptomic patterns and additional phytohormones may modulate such interactions (De Vos et al., 2005; Robert-Seilaniantz et al., 2011). For instance, SA-dependent defense gene induction is primed in an ET-dependent manner in a context of JA induction (De Vos et al., 2006).
The data show that ethylene perception is necessary for the interference with parasitoid attraction by whitefly feeding to occur in P. xylostella-infested Arabidopsis plants. In Lima bean plants, ethylene enhances the JA-mediated production of predator-attracting HIPV in prey-infested plants (Horiuchi et al., 2001). JA- and ET-mediated defenses are also known to positively interact in other induced defense responses. For instance, ET acts in concert with JA in activating genes encoding defensive proteins, such as plant defensins in Arabidopsis (Penninckx et al., 1998) and proteinase inhibitors in tomato (O'Donnell et al., 1996). Furthermore, when exogenously applied to plant tissue, ET and JA appear to function synergistically to induce osmotin and PR1b in tobacco (Xu et al., 1994).
The interference by whitefly feeding with P. xylostella-induced plant responses is also clear from the transcriptomic analyses. Although 48 h of whitefly feeding did not alter the transcriptional pattern of Arabidopsis extensively (Fig. 4b), whitefly feeding in combination with caterpillar feeding considerably reduced the number of genes that were differentially regulated in response to P. xylostella. This may be caused by priming effects, similar to the effect of priming by ET-inducing caterpillars on the transcription of an SA-responsive resistance gene (De Vos et al., 2006). Seventy-one percent of the 90 genes that were differentially regulated by 48 h of feeding by P. xylostella were also differentially regulated after feeding for maximally 24 h by P. xylostella on another Arabidopsis accession, Ler (Ehlting et al., 2008). It is noteworthy that among the genes of which the expression was affected by P. xylostella feeding and additional whitefly feeding there are six transcription factors, indicating that the effect is more extensive.
In conclusion, this study demonstrates that intact JA and ET signal-transduction pathways are required for the whitefly-induced interference with an indirect plant defense mechanism that is directed against chewing herbivores in Arabidopsis. Elevated constitutive concentrations of SA eliminate this interference. Crosstalk between defense signaling pathways has been considered a powerful strategy employed by plants to fine-tune their defense responses towards specific herbivores (Koornneef & Pieterse, 2008; Koornneef et al., 2008; Thaler et al., 2012). However, emerging evidence shows that insect herbivores and phytopathogens can manipulate plants for their own benefit by suppressing plant defense (Shiojiri et al., 2002; Pieterse & Dicke, 2007; Hartley & Gange, 2009; Zhang et al., 2009). Moreover, our data show that interactions among signal-transduction pathways may act differently in situations of a single organism attacking the plant vs multiple attack situations.
In this study we have addressed the mechanisms underlying interference with P. xylostella-induced volatile emission by co-infestation with the whitefly B. tabaci. To do so we have exploited the valuable tools that Arabidopsis provides. Arabidopsis has been recognized as a good model for understanding mechanisms involved in indirect defense in other brassicaceous plants (Van Poecke & Dicke, 2004). Plutella xylostella and B. tabaci are specialist and generalist herbivores of Brassica species, respectively. These two species are often found on the same Brassica plant in the field. In general, phloem feeders and leaf chewers may be found co-infesting the same plant under natural conditions in a range of crops (Rodriguez-Saona et al., 2003; Dicke et al., 2009; Zhang et al., 2009). Yet, studies on the ecological effects of multiple infestation have only recently been initiated. Understanding the intricate mechanisms underlying a plant's responses to multiple attackers will be important for the development of crop protection strategies based on plant responses to different attackers in a multi-attacker context.
The authors thank Léon Westerd and André Gidding for culturing the insects and parasitoid wasps and Chris Maliepaard for statistical advice. This work was financially supported by the Earth and Life Sciences Foundation VICI Grant no. 865.03.002 (to M.D.) from the Netherlands Organization for Scientific Research, and Zhejiang Provincial Natural Science Foundation of China under Grant no. R3100692 (to P.J.Z.).