•Plants under herbivore attack emit mixtures of volatiles (herbivore-induced plant volatiles, HIPVs) that can attract predators of the herbivores. Although the composition of HIPVs should be critical for the attraction, most studies of transgenic plant-emitted volatiles have simply addressed the effect of trans-volatiles without embedding in other endogenous plant volatiles.
•We investigated the abilities of transgenic wishbone flower plants (Torenia hybrida and Torenia fournieri) infested with spider mites, emitting a trans-volatile ((E)-β-ocimene) in the presence or absence of endogenous volatiles (natural HIPVs and/or floral volatiles), to attract predatory mites (Phytoseiulus persimilis).
•In both olfactory- and glasshouse-based assays, P. persimilis females were attracted to natural HIPVs from infested wildtype (wt) plants of T. hybrida but not to those of T. fournieri. The trans-volatile enhanced the ability to attract P. persimilis only when added to an active HIPV blend from the infested transgenic T. hybrida plants, in comparison with the attraction by infested wt plants. Intriguingly, floral volatiles abolished the enhanced attractive ability of T. hybrida transformants, although floral volatiles themselves did not elicit any attraction or avoidance behavior.
•Predator responses to trans-volatiles were found to depend on various background volatiles (e.g. natural HIPVs and floral volatiles) endogenously emitted by the transgenic plants.
Behavioral responses of carnivores toward synthetic volatile compounds and various HIPVs produced by herbivore- or chemical-induced plants or transgenic plants have been tested widely in laboratory- and field-based studies (Arimura et al., 2005; Sabelis et al., 2007). The use of transgenic plants, in particular, represents a novel solution to the challenges of studying the ecological relevance of HIPVs (e.g. plants’ indirect defense) and the potential contribution of HIPVs to biological pest control (Aharoni et al., 2005). For instance, it has been reported that transgenic Arabidopsis plants emitting (3S)-(E )-nerolidol and (E )-4,8-dimethyl-1,3,7-nonatriene ((E )-DMNT), or only (3S )-(E )-nerolidol, attract carnivorous predatory mites (Phytoseiulus persimilis) (Kappers et al., 2005). Other transgenic plants emitting six maize sesquiterpene synthase (TPS10) products were reported to cause significant attraction of parasitoid Cotesia marginiventris females (Schnee et al., 2006). Moreover, Degenhardt et al. (2009) successfully demonstrated the full potential of this strategy by restoring the emission of (E )-β-caryophyllene, which attracts entomopathogenic nematodes that infect and kill a voracious root pest. However, it should be noted that most studies, including the examples noted here, have simply been conducted to assess whether one or multiple trans-volatiles can attract carnivores without considering the role of endogenous plant volatiles, and therefore these findings still leave us far from elucidating the complicated nature of bodyguard attractions in cases mediated by the volatiles blended in natural HIPVs and other endogenous plant volatiles such as floral and fruit volatiles (Dicke & Baldwin, 2010). Indeed, this issue is critical for assessing the actual impact of genetic engineering of HIPVs in horticultural pest control; for example, natural enemies of pest herbivores may cease to respond to novel HIPV blends from transgenic, ornamental plants at flowering stages, so as to protect themselves against intraguild predation by flower-visiting generalist predators. While considerable evidence about the responses of pollinators to floral volatiles has been accumulated, little is known about carnivore response to floral volatiles or floral volatiles blended in HIPVs (Dicke & Baldwin, 2010).
The two-spotted spider mite Tetranychus urticae is a serious pest of agricultural, vegetable, fruit and ornamental plants (Helle & Sabelis, 1985). The mite-induced plant volatiles enhance the prey-searching efficacy of predatory mites (P. persimilis), and the attraction of P. persimilis results in the extermination of T. urticae from the plants (Takabayashi & Dicke, 1996). In the current study, we show a potential function of HIPVs emitted from torenia (wishbone flower), which is grown as a garden plant and may suffer severely from spider mite attack during its cultivation. In addition, the genetic variability of plants often causes different behaviors of predators as a result of variations in the individual HIPV profiles of these accessions (Degen et al., 2004; Huang et al., 2010; Kappers et al., 2010; Snoeren et al., 2010). Therefore, we tested two species –Torenia hybrida cv Summer Wave Blue (SWB) and Torenia fournieri cv Crown White and Violet (CrW and CrV, hereafter called Cr for the two cultivars together) – to assess transgenic torenia plants emitting monoterpene HIPVs. We especially focused on (E)-β-ocimene, an acyclic monoterpene hydrocarbon compound, because, first, it is one of the most common volatile chemicals released from plants in response to herbivory, including herbivorous mite attack (Dicke et al., 1990; Paré & Tumlinson, 1999; Pichersky & Gershenzon, 2002; Arimura et al., 2009); secondly, it has been reported that the entire blend of HIPVs, including (E)-β-ocimene, rather than this monoterpene alone, plays an important role in the attraction of predatory mites (Ishiwari et al., 2007; Zhang et al., 2009; Kappers et al., 2011); and thirdly, it is, however, still unclear what combination of HIPVs, including (E)-β-ocimene and other plant volatiles (e.g. floral volatiles), functions for the attraction. To answer this question, this paper presents the results of a set of diverse methodological approaches (i.e. laboratory and glasshouse assays) that, by focusing on (E)-β-ocimene, examined what effect metabolic engineering, adding (E)-β-ocimene to the natural composition of HIPV bouquets, has on attraction. We also evaluated whether floral volatiles blended with transgenic plant volatiles measurably affect the attraction of predators, because torenia is a horticultural plant that can flower profusely throughout the year in glasshouses. The results revealed composite effects of the volatile blends emitted from the transgenic plants on the attraction of predatory mites.
Materials and Methods
Plants, mites and treatments
Torenia plants (SWB, CrW and CrV) were maintained under sterile conditions in a plant box containing 1/2 Murashige and Skoog medium supplemented with 0.2% gellan gum in a growth chamber at 25°C (160 μE m−2 s−1 with a 16 h photoperiod). To prepare potted plants, each plant was propagated by transferring herbaceous cuttings to soil in a plastic pot. Individual plants were grown in plastic pots in a glasshouse (23 ± 3°C, 60 ± 10% relative humidity (RH), 16 h photoperiod) for 3–5 wk until use. T. urticae was reared on kidney bean plants (Phaseolus vulgaris cv Nagauzuramame) in a climate-controlled room (25 ± 1°C, 60 ± 10% RH, 16 h photoperiod). P. persimilis, obtained from a commercial source (Koppert Biological Systems, Berkel en Rodenrijs, The Netherlands), was reared on T. urticae-infested bean plants in a climate-controlled room (25 ± 1°C, 60 ± 10% RH, 16 h photoperiod). Fertilized adult female predators 3–5 d after the final molting were used for the bioassays. To make starved predators, the predators were individually placed in sealed plastic tubes (1.5 ml), each containing a drop of water (3 μl) only, for 24 h in a laboratory (25 ± 1°C, 16 h photoperiod).
To prepare infested plants for bioassays and chemical analyses, a potted torenia plant growing in a plastic pot was infested with T. urticae (100 adult females) for 3 d in a climate-controlled room (25 ± 1°C, 60 ± 10% RH, 16 h photoperiod at a light intensity of 80 μE m−2 s−1); the numbers of T. urticae females that remained on SWB, CrW and CrV 3 d after the inoculation were compared by a generalized linear model (GLM) with a logit link, and binomial distribution of sampling error. Furthermore, to prepare intact plants treated with methyl jasmonate (MeJA) in chemical analyses, a solution containing MeJA (0.5 mM, pH 5.8–6.0; Sigma-Aldrich, St. Louis, MO, USA) in 20 ml of water was evenly sprayed onto a potted torenia plant 1 d before the volatile collection.
Generation of transgenic torenia plants
The full-length coding region of lima bean β-ocimene synthase (PlOS; EU194553 (Arimura et al., 2008)) was inserted into the GUS reporter gene site downstream of the constitutive cauliflower mosaic viral (CaMV) 35S promoter of the binary vector pBI121 (Clontech, Mountain View, CA, USA). The resulting plasmid, pBI121-PlOS, was transformed into Agrobacterium tumefaciens strain EHA105 by electroporation. Torenia transformation was performed as described previously (Aida & Shibata, 2001), and pBI121 plasmid was similarly transformed as a control. Briefly, leaf fragments of in vitro plants were cocultured with Agrobacterium for 7 d at 22°C under dark conditions. Resistant shoots were regenerated by subsequent culturing on medium containing 300 mg l−1 kanamaycin as a selection reagent at 25°C under a 16 h photoperiod regime. Since torenia is a heterozygous plant developed by a combination of extensive hybridization and mutation breeding, it is difficult to obtain homozygous lines (Aida, 2008). Accordingly, the resulting transgenic plants (T0) were maintained in vitro by subculturing every month until use.
We collected the volatiles from a potted, nonflowering plant (c. 0.5–1 g fresh weight (FW)) following different treatments (i.e. T. urticae damage or MeJA spray) or five cut flowers wrapped in moist cotton wool that were kept in a glass container. In addition, volatiles from an infested, nonflowering plant (c. 0.5–1 g FW) together with three cut flowers were collected. Headspace volatiles from potted plants and flowers were collected using 100 mg of Tenax-TA resin (20/35 mesh; GL Science, Tokyo, Japan) packed in a glass tube (3.0 mm i.d., 160 mm length) for 2 and 1 h, respectively, at a flow rate of 100 ml min−1 with clean air (25 ± 1°C, light intensity of 80 μE m−2 s−1). n-Tridecane (0.1 μg) was also added to the glass container. The volatile compounds collected were analyzed by gas chromatography-mass spectrometry (GC-MS) according to the method described by Arimura et al. (2004). The identification of the volatiles was made by comparison of their retention times with those reported in the literature, and by comparison of the data with authentic standards, with the exception of α-zingiberene, γ-curcumene, and 3-ethyl-4-methylpentanol, which were identified only by comparing their mass spectra because of the lack of their authentic compounds. (E)-β-Ocimene was quantified using its authentic compound (SAFC, Sigma-Aldrich, St. Louis, MO, USA).
Reverse transcription (RT) and real-time PCR
Total RNA was isolated from leaf tissues using a Qiagen RNeasy Plant RNA kit and an RNase-Free DNase Set (Qiagen, Hilden, Germany) following the manufacturer’s protocol. First-strand cDNA was synthesized using a PrimeScript RT reagent Kit (Takara, Otsu, Japan), and 0.5 μg of total RNA at 37°C for 15 min. Real-time PCR was done on an ABI Prism® 7000 Sequence Detection System (Applied Biosystems, Tokyo, Japan) using FastStart Universal SYBR Green Master (ROX) (Roche Applied Science, Indianapolis, IN, USA), cDNA (1 μl from 10 μl of each RT product pool), and 300 nM primers. The following protocol was used: initial polymerase activation: 10 min at 95°C; then 40 cycles of 15 s at 95°C and 60 s at 60°C. PCR conditions were chosen by comparing threshold values in a dilution series of the RT product, followed by a nonRT template control and nontemplate control for each primer pair. Relative RNA levels were calibrated and normalized with the level of ACT3 (AB330989) mRNA. The primers used were as follows: PlOS (5′-CAACAATGCATGGGTCTCAG-3′ and 5′-TGCTGCTTCCCCTCTCTCTA-3′) and ACT3 (5′-CAACTGCAGAGCGTGAAATC-3′ and 5′-ATCATCGATGGCTGGAAAG-3′).
Each olfactory bioassay was performed in a Y-tube olfactometer in a laboratory (25 ± 1°C, light intensity of 80 μE m−2 s−1), according to the method described by Shimoda (2010). The odor sources used were mostly divided into the following five types: uninfested plants, infested plants, flowers, infested plants plus flowers, and clean air. As described earlier, a potted, nonflowering plant was infested with T. urticae (100 adult females) for 3 d. For all the assays except that in Fig. S5, three or four infested plants (c. 3 gFW, sum) were wrapped in moist cotton wool and aluminum foil. To prepare uninfested plants, potted, nonflowering plants were incubated for 3 d in the same conditions as used for infested plants. To prepare flowers as an odor source, six to eight SWB flowers or 10–12 Cr flowers (c. 2 g FW, sum) were cut and then wrapped in moist cotton wool and aluminum foil. The moist cotton wool and aluminum foil were used for assays with clean air. Note that transient detachment of plants during assays should not markedly influence our interpretation since the emission rates of HIPVs from the detached plants after T. urticae damage did not differ from those from attached plants (Supporting Information, Fig. S1).
Predators were individually introduced at the start point in the olfactometer, and the numbers of predators choosing either sample or control odor sources were recorded. Predators that did not choose within 5 min (‘no choice’ subjects) were excluded from the statistical analysis. Assays using 20 predators were carried out as a single replicate in a day. Three or four replications (i.e. 40–80 predators in all) were carried out on different days. The results from three or four replications of each experiment were subjected to a replicated G-test; the pooled G-value (Gp, df = 1 in each) was used to test the null hypothesis that the predators exhibited a 50 : 50 distribution over the sample and control odor sources in each experiment (Sokal & Rohlf, 1995). We also confirmed that there was no significant heterogeneity among replications in each experiment (replicated G-test, df = 2 or df = 3, P >0.05 for each Gh), suggesting good reproducibility of the two-choice test.
Glasshouse experiments were performed in Tsukuba, Japan, during April–June and September 2011. The temperature in the glasshouse was maintained between 25 and 31°C with a photoperiod of 16 h (natural + supplemental light at a light intensity of 80–110 μE m−2 s−1), by an air conditioner and shading with nonwoven curtains (Love Sheet, Unitika Ltd, Osaka, Japan). The odor sources used were mostly divided into the following three types: a potted, uninfested, nonflowering plant; a potted, nonflowering plant infested with T. urticae (100 adult females) for 3 d; and a potted, flowering plant infested for 3 d. The biomasses of a nonflowering plant and a flowering plant (having two to three flowers) were c. 3 and 3.7 g FW, respectively. The potted plants and P. persimilis placed in the sealed plastic tubes were adapted to glasshouse conditions for 2–3 h before use in the assays. Volatiles from the potted plants that had been preconditioned in the glasshouse were determined under controlled conditions in the laboratory (Fig. S2).
During bioassays, 80 females of starved P. persimilis were released on a moist paper filter at the center of a white board of polyvinyl chloride (90 × 90 cm, Takiron Plate, ES9700A 1.0, Takiron Co., Ltd., Osaka, Japan) on which four plant pots consisting of two pairs of plants (i.e. four plants in all) were placed at the corners of the board at the same distance (30 cm) from the predator-release point (Fig. 8a). Predators that reached each pot were immediately collected using a fine paintbrush and recorded for 2 h after the onset of predator release. Predators that did not visit within 2 h (‘no choice’ subjects) were excluded from the statistical analysis. Assays using 80 predators were carried out as a single replicate in a day. Three replications (i.e. 240 predators in total) were carried out on different days; the results were subjected to a replicated G-test.
Volatile profiles of torenia wildtype (wt) plants and their flowers
Torenia is a genus of flowering plants that have many hybrid species. In this study, we first investigated the emission of HIPVs from two wt species, T. hybrida with blue floral color (SWB) and T. fournieri with white (CrW) or violet (CrV) floral color, in response to infestation by spider mites. Nonflowering SWB plants infested with T. urticae released a blend of HIPVs comprising a monoterpene (linalool), a homoterpene ((E)-DMNT) and sesquiterpenes (α-zingiberene, α-bergamotene, γ-curcumene and unidentified sesquiterpene) (Fig. 1). By contrast, nonflowering CrW and CrV plants scarcely released HIPVs, except for (Z)-3-hexen-1-yl acetate in the case of infested CrV, in response to feeding spider mites. 1-Octen-3-ol and (E)-DMNT were constitutively released from T. fournieri plants (compounds 2 and 10 from both CrW and CrV, Fig. 1). These differences were not likely caused by differences in the degree of T. urticae performance on different host species, for the following reasons. First, T. urticae females survived similarly on SWB, CrW and CrV plants for 3 d after the inoculation (means ± SE) of percent survival: SWB, 53.4 ± 28.1; CrW, 52.8 ± 25.1; CrV, 53.2 ± 23.9; GLM, df = 2, χ2 = 0, P =1). Moreover, both CrW and CrV scarcely released volatiles after treatment with MeJA (a powerful HIPV inducer), in contrast to the observation that the MeJA-treated SWB emitted the similar blend of mite-induced HIPVs (Fig. S3). Also, our preliminary analysis using Spodoptera litura larvae as herbivores did not result in significant emission of HIPVs from infested T. fournieri plants, although the larvae fed well on the plants (data not shown). Altogether, it is highly likely that T. fournieri species genetically lack the potential for HIPV emission.
An aromatic, 3-ethyl-4-methylpentanol (peak 6, Fig. 1), was detected in floral vapors of all the studied species. SWB flowers contained many additional volatiles, such as monoterpenes ((+)-2-carene, α-terpinene, p-cymene and limonene (peaks 3, 5, 7 and 8, Fig. 1)), but none of the HIPVs induced by T. urticae were detected.
Response of P. persimilis to volatiles released from T. urticae-infested plants and flower tissues
Using a Y-tube olfactometer, we investigated the influence of HIPVs on the predator’s foraging behavior. P. persimilis females were not attracted to volatiles from uninfested SWB, CrW or CrV compared with control clean air (SWB: Gp = 0.067, P =0.796; CrW: Gp = 2.416, P =0.12; CrV: Gp = 0.424, P =0.515, df = 1 each, replicated G-test; Fig. S4). By contrast to the control experiments, the predator significantly preferred volatiles emitted from infested SWB plants to those from uninfested SWB plants (Gp = 6.8, P <0.01; Fig. 2a), showing the predator-attractiveness of HIPVs from infested SWB plants. Infestation of T. urticae, however, did not make CrW or CrV attractive to predatory mites (CrW: Gp = 0.276, P = 0.599; CrV: Gp = 0.601, P =0.438), probably as a result of the lack of sufficiently induced emission of HIPVs (see Fig. 1). P. persimilis females were not attracted to floral volatiles of SWB, CrW or CrV compared with control air (SWB: Gp = 0.153, P =0.696; CrW: Gp = 0.067, P =0.796; CrV: Gp = 0.601, P =0.438; Fig. 2b).
Generation of transgenic torenia plants emitting (E)-β-ocimene
To clarify the impact of exogenously manipulated HIPV compound(s) in the natural HIPV blend, we generated transgenic plants of the respective torenia species that constitutively emitted (E)-β-ocimene. Five representative transgenic plants exhibiting substantial (E)-β-ocimene emission are shown in Fig. 3; the emission ranged from 0.16 μg g–1 FW h−1 (uninfested plants of line CrV41) to 1.30 μg g–1 FW h−1 (uninfested plants of line CrW9) (Fig. 3f), whereas the respective wt species and pBI121-transformed control plants showed no emission (see Fig. 1). Furthermore, even in response to T. urticae infestation, the emission levels were not altered in transgenic line SWB12 (Fig. 3d–f). A trace of (Z)-isomer of β-ocimene was also present in all the headspace volatiles (2% of the total β-ocimene products, Fig. 3b), at a concentration that corresponded to the composition of β-ocimene isomers generated from the recombinant PlOS protein expressed in Escherichia coli using geranyl diphosphate as a substrate (Arimura et al., 2008).
Curiously, a trace of (E)-β-ocimene was emitted by flower tissues of PlOS-transgenic plants (4–33 ng g–1 FW h−1, Fig. 3c,e,g). Given the fact that SWB flowers are capable of producing a certain amount of monoterpenes (Fig. 1) but only traces of PlOS transcript were detected in floral tissues (Fig. 4), the low emission of (E)-β-ocimene was most likely the result of limited de novo PlOS activity rather than a lack of isoprenoid precursors such as isopentenyl diphosphate and dimethylallyl diphosphate. A similar result was also observed in transgenic torenia plants that expressed lower levels of GUS marker protein in both petals and flower buds than leaf tissues under the control of the CaMV 35S promoter (Sasaki et al., 2011).
It should be noted that none of the transgenic lines exhibited any differences in their morphology or the emission rates of other volatiles (i.e. HIPVs or floral volatiles) compared with wt plants (compare, for example, the GC profiles between SWB wt plants and SWB12 plants; Figs 1, 3 and S1).
Response of P. persimilis to volatiles released from the shoots and flowers of torenia transformants
We investigated the influence of (E)-β-ocimene on the predator’s foraging behavior, using the nonflowering stage of four transgenic lines, SWB3, SWB12, CrW15 and CrV43, which emitted similar concentrations of (E)-β-ocimene (660–670 ng g–1 FW h−1; Fig. 3f). Contrary to our expectations, P. persimilis did not discriminate volatiles of uninfested transformants from those of uninfested, wt plants (SWB3: Gp = 0.45, P =0.502; SWB12: Gp = 0, P =1; CrW15: Gp = 2.463, P =0.117; CrV43: Gp = 0.267, P =0.605; Fig. 5a,d). Increasing the amount of plant material used in the assay (from 3 to 20 g FW) did not affect the behavioral patterns (SWB3: Gp = 2.136, P =0.144; SWB12: Gp = 0.317, P =0.574; CrW15: Gp = 0.601, P =0.438; CrV43: Gp = 1.07, P =0.301; Fig. S5). Regarding Cr transformants, neither type of infested CrW15 or CrV43 was attractive when compared with uninfested wt (CrW(wt) or CrV(wt)) plants (CrW15: Gp = 0.267, P =0.605; CrV43: Gp = 0.424, P =0.515; Fig. 5b) or when compared with infested wt plants (CrW15: Gp = 2.416, P =0.12; CrV43: Gp = 1.07, P =0.301; Fig. 5c). The ineffectiveness of adding (E)-β-ocimene was presumably the result of the lack of predator-attractiveness of the original HIPVs released from infested CrW15 and CrV43, similarly to the ineffectiveness found for the respective wt plants (Fig. 1).
By contrast to these results, (E)-β-ocimene enhanced the attractivity of P. persimilis when the SWB-derived HIPVs, the active infochemicals, were blended. P. persimilis preferred volatiles from infested SWB3/12 plants to those from uninfested SWB (wt) plants (HIPVs + (E)-β-ocimene vs basal volatiles; SWB3: Gp = 17.984, P <0.001; SWB12: Gp = 5.211, P <0.05; Fig. 5e) or to those from infested SWB (wt) plants (HIPVs + (E)-β-ocimene vs HIPVs; SWB3: Gp = 11.53, P <0.001; SWB12: Gp = 8.258, P <0.01; Fig.5f).
We further tested the effects of floral volatiles on the predator’s foraging behavior mediated by HIPVs and (E)-β-ocimene. Similar to the lack of effect of wt floral volatiles (Fig. 2b), P. persimilis showed no preference for floral volatiles of SWB3/12 over clean air (SWB3: Gp = 0, P =1; SWB12: Gp = 0.424, P =0.515; Fig. 6a), which confirms that transgenic floral volatiles did not affect any attraction or avoidance behaviors. Curiously, the enhanced attractiveness of volatiles from infested SWB3/12 plants compared with those from infested SWB(wt) plants (HIPVs + (E)-β-ocimene vs HIPVs; see Fig. 5f) was abolished when infested SWB3/12 plants were supplemented with their respective flowers (HIPVs + (E)-β-ocimene + floral volatiles vs HIPVs; SWB3: Gp = 0.114, P =0.736; SWB12: Gp = 0.013, P =0.91; Fig. 6c). However, volatiles from infested SWB3/12 plants were still attractive despite the presence of floral volatiles, when compared with volatiles from uninfested SWB(wt) plants (HIPVs + (E)-β-ocimene + floral volatiles vs basal volatiles; SWB3: Gp = 13.165, P <0.001; SWB12: Gp = 8.398, P < 0.01; Fig. 6b). Interestingly, infested SWB(wt) plants supplemented with their flowers were equally attractive when compared with infested SWB(wt) plants (HIPVs + floral volatiles vs HIPVs; Gp = 1.674, P =0.196; Fig. 6c), or more attractive when compared with uninfested SWB(wt) plants (HIPVs + floral volatiles vs basal volatiles; Gp = 9.208, P <0.01; Fig. 6b). These results indicate that floral volatiles can suppress the supporting effect of (E)-β-ocimene embedded into natural HIPVs but not HIPV-mediated attraction of P. persimilis.
We also examined several other combinations of plants and flowers in the assays and confirmed these results. For example, P. persimilis preferred volatiles from infested SWB3/12 plants supplemented with their respective flowers, when compared with those from nonflowering, uninfested SWB3/12 plants (HIPVs + (E)-β-ocimene + floral volatiles vs (E)-β-ocimene; SWB3: Gp = 5.484, P <0.05; SWB12: Gp = 11.649, P <0.001; Fig. 7a), whereas the reverse was true in the case of comparison to volatiles from nonflowering, infested SWB3/12 plants (HIPVs + (E)-β-ocimene + floral volatiles vs HIPVs + (E)-β-ocimene; SWB3: Gp = 4.968, P <0.05; SWB12: Gp = 7.641, P <0.01; Fig. 7b). These results suggest that floral volatiles specifically suppress the supporting effect of (E)-β-ocimene, but not HIPVs’ attractivity. Moreover, compared with infested SWB(wt) plants supplemented with their flowers, the predatory mites did not prefer infested SWB3/12 plants supplemented with their respective flowers (HIPVs + (E)-β-ocimene + floral volatiles vs HIPVs + floral volatiles; SWB3: Gp = 1.07, P =0.301; SWB12: Gp = 0.017, P =0.896; Fig. S6) or with SWB(wt) flowers (HIPVs + (E)-β-ocimene + floral volatiles vs HIPVs + floral volatiles; SWB3: Gp = 1.07, P =0.301; SWB12: Gp = 0.267, P =0.605). Taken together, all the results shown here confirmed that floral volatiles can suppress the P. persimilis attraction caused by HIPVs + (E)-β-ocimene but not that caused by HIPVs or by (E)-β-ocimene alone.
Response of P. persimilis to nonflowering or flowering stages of torenia plants in glasshouse
To confirm the results from the olfactometer assays and assess whether the findings noted in the previous section are confirmed at the glasshouse level, we next conducted choice tests with glasshouse-based studies. As shown in Fig. 8(a), starved P. persimilis females were released at the center of an assay board, and the predatory mites which were attracted to plants on each corner of the board were counted in the glasshouse. We first conducted assays of T. urticae-infested SWB(wt) plants vs uninfested SWB(wt) plants, and the results revealed a significant preference (Gp = 7.721, P <0.01; Fig. 8b). As seen in olfactometer assays, there was no discrimination between infested CrW(wt) or CrV(wt) plants and uninfested plants (CrW(wt): Gp = 0.766, P =0.381; CrV(wt): Gp = 0.018, P =0.892). Since these results were absolutely in line with those from olfactometer assays, it was confirmed that our glasshouse-based system is certainly useful for evaluating P. persimilis attraction.
Accordingly, using this system, we assessed transgenic torenia with different treatments. At the nonflowering stage, the predatory mites preferred infested SWB3/12 plants over infested SWB(wt) plants (HIPVs + (E)-β-ocimene vs HIPVs; SWB3: Gp = 11.803, P <0.001; SWB12: Gp = 12.381, P <0.001; Fig. 9a), while they did not discriminate between uninfested SWB3/12 plants and uninfested SWB(wt) plants ((E)-β-ocimene vs basal volatiles; SWB3: Gp = 0.046, P = 0.83; SWB12: Gp = 0.043, P =0.835; Fig. S7). Furthermore, as expected, neither type of infested Cr transformant was attractive when compared with uninfested CrW(wt) or CrV(wt) plants ((E)-β-ocimene vs basal volatiles; CrW15: Gp = 0.935, P =0.334; CrV43: Gp = 1.781, P =0.182; Fig. S8).
Flowering, infested SWB3/12 plants (intact plants with flowers) were not more attractive than infested SWB(wt) plants (HIPVs + (E)-β-ocimene + floral volatiles vs HIPVs; SWB3: Gp = 0.044, P =0.833; SWB12: Gp = 0.327, P =0.568; Fig. 9c), which indicated that floral volatiles suppress the supporting effect of (E)-β-ocimene, using flowering plants in a glasshouse. The flowering, infested SWB3/12 plants were, however, more attractive than the nonflowering, uninfested SWB(wt) plants (HIPVs + (E)-β-ocimene + floral volatiles vs basal volatiles; SWB3: Gp = 32.302, P <0.001; SWB12: Gp = 14.938, P <0.001; Fig. 9b). The predatory mites preferred flowering, infested SWB(wt) plants over nonflowering, uninfested SWB(wt) plants (HIPVs + floral volatiles vs basal volatiles; Gp = 16.558, P <0.001; Fig. 9b), while they did not discriminate between flowering, infested SWB(wt) plants and nonflowering, infested SWB(wt) plants (HIPVs + floral volatiles vs HIPVs; Gp = 0.078, P = 0.78; Fig. 9c). These results indicate that floral volatiles do not affect the attractivity of HIPVs. P. persimilis preferred flowering, infested SWB3/12 plants, over nonflowering, uninfested SWB3/12 plants (HIPVs + (E)-β-ocimene + floral volatiles vs (E)-β-ocimene; SWB3: Gp = 7.264, P <0.01; SWB12: Gp = 9.178, P <0.01; Fig. 10a), whereas the reverse was true in the comparison to nonflowering, infested SWB3/12 plants (HIPVs + (E)-β-ocimene + floral volatiles vs HIPVs + (E)-β-ocimene; SWB3: Gp = 8.338, P <0.01; SWB12: Gp = 5.091, P <0.05; Fig. 10b). Altogether, these results were in complete agreement with the data obtained from olfactometer assays.
Hitherto, most studies of the ecological relevance of transgenic plants emitting HIPVs simply addressed whether one or multiple trans-products attract natural enemies (Kappers et al., 2005; Schnee et al., 2006). However, we are now aware of the importance of the entire blend of HIPVs, rather than individual compounds included in the HIPV blend, in the attraction of predatory mites (de Boer & Dicke, 2004; van Wijk et al., 2008) and parasitoid wasps (Mumm et al., 2008). Notably, the present study using transgenic plants in laboratory and glasshouse environments demonstrated the subtle nature of special mixtures in which a single transproduct was blended with various endogenous plant volatiles (i.e. natural HIPVs and floral volatiles). Our experiments revealed the following three features.
First, we found that metabolic engineering to add a single terpene compound was not always effective in attracting predatory mites to transgenic plants, from which the endogenous HIPVs were poorly released. We tested 3 and 20 g FW of uninfested transgenic plants, which emitted (E)-β-ocimene (33 and 220 ng min−1 emissions, respectively; Fig. 3f) with trace amounts of other volatiles (Fig. 3b), and found that they were not preferred by the predatory mites, irrespective of the different (E)-β-ocimene concentrations (Figs 5a,d, S5 and S7). In addition, T. urticae-infested CrW15 or CrV43 plants also scarcely attracted predatory mites (Figs 5b, S8), and this might have been because of the lack of sufficient emission of HIPVs (Fig. 1). P. persimilis is a specialist predator of Tetranychus mites (Takabayashi & Dicke, 1996), whereas (E)-β-ocimene is frequently detected at high concentrations not only within T. urticae-induced volatiles but also within a suite of caterpillar-induced volatiles (Loughrin et al., 1994; Arimura et al., 2008) and floral, needle and fruit volatiles (Dudareva et al., 2003). Thus, (E)-β-ocimene is not likely to serve the predators as a specific cue to locate their specific prey (van Wijk et al., 2008). It should be noted, however, that (E)-β-ocimene alone or blended in HIPV blends may cause specific attraction of predatory mites depending on their previous odor experiences (Ishiwari et al., 2007) and specific physiological conditions (de Boer & Dicke, 2004). It has been reported, for example, that previous experience of P. persimilis with specific HIPVs, including (E)-β-ocimene, methyl salicylate or (E,E )-4,8,12-trimethyl-1,3,7,11-tridecatetraene, increased the ability of the mites to search for their foraging locations, using these compounds as a host-searching cue (de Boer & Dicke, 2004). This was not in line with our cases, however, since we used P. persimilis females that had not had experience with (E)-β-ocimene during the rearing with T. urticae on kidney bean leaves, which do not release (E)-β-ocimene (Maeda et al., 2006). Moreover, the attractiveness of (E)-β-ocimene and (E)-DMNT for P. persimilis may decrease with an increase in the severity of starvation, when the synthetic compounds are individually tested (Dicke et al., 1990; de Boer & Dicke, 2004; van Wijk et al., 2008). The use of starved and β-ocimene-inexperienced predatory mites in the present study might have resulted in their weak ability to respond to (E)-β-ocimene.
Secondly, our study using transgenic plants revealed that the trans-product acted as a supporting infochemical when added to an active, natural HIPV blend. The addition of the trans-product in SWB3 or SWB12 plants resulted in more potent attraction of P. persimilis than the attraction by the natural HIPV blend from infested SWB(wt) plants (Figs 5f, 9a), suggesting that (E )-β-ocimene synergizes with other compounds in an HIPV blend. Hence, our results confirmed previous studies showing that predatory mites use a prey-associated volatile mixture including (E )-β-ocimene, rather than this compound alone, to forage (Ishiwari et al., 2007; van Wijk et al., 2008); e.g. Zhang et al. (2009) reported that a reduction in (E)-β-ocimene emission in an HIPV blend from lima bean plants infested with both T. urticae and the whitefly Bemisia tacaci caused reduced attraction of P. persimilis, suggesting that a high ratio of (E)-β-ocimene in a HIPV blend is significant for attracting predatory mites. Positive relationships between the proportion of (E)-β-ocimene in a HIPV blend and the attractivity to P. persimilis were also found in cucumber (Kappers et al., 2010, 2011) and Gerbera jamesonii (Krips et al., 2001). Nonetheless, we have previously reported that the dose intensity of HIPVs including (E)-β-ocimene and other compounds, rather than their ratio, emitted from lima bean plants was positively linked to the density of feeding T. urticae and the attractivity of P. persimilis (Horiuchi et al., 2003). A large amount of (E)-β-ocimene blended in active HIPVs may function as an indicator of a high density of appropriate prey, as abundant prey are required for sustaining even a single P. persimilis mite on host plants (Sabelis et al., 2007).
Thirdly, the most intriguing finding here is the fact that the supporting effect of the trans-product embedded in endogenous HIPVs could be abolished by other endogenous plant volatiles, that is, floral volatiles. When flowers were added to infested transgenic SWB3 or SWB12 plants, the predators ceased to discriminate those plants from infested SWB(wt) plants (Figs 6c, 9c). A similar phenomenon was reported by Shiojiri & Takabayashi (2005), namely that Cotesia vestalis, a specialist parasitoid wasp of diamondback moth larvae (Plutella xylostella), ceased to prefer odors from host-infested Rorippa indica plants after the plants flowered: the parasitoids might avoid floral volatiles so as to protect themselves against intraguild predation by flower-visiting ants. However, this is not likely to have been the mechanism in our case, because the floral volatiles themselves did not elicit any positive or negative response of the predators (Figs 2b, 6a), and the cocktailed volatiles from infested transgenic plants plus flowers were still more attractive than the uninfested plant volatiles (Figs 6b, 9b). Rather, our results obtained from both laboratory and glasshouse experiments indicated the antagonism of floral volatiles against the supporting effect of the trans-product but not the attractiveness of natural HIPVs. Recently, Turner & Ray (2009) reported that 1-hexanol and 2,3-butanedione, odorants present in ripening fruits and other foods, directly inhibit CO2-sensitive neurons by acting on the Gr21a/Gr63a CO2 receptor in the antenna of the fruitfly (Drosophila melanogaster), which consequently suppress their avoidance behavior toward CO2, as a major volatile component produced by stressed flies. An inhibitory system like this one detected in insects may also act in the masking effect of torenia floral volatiles on predatory mites, even though the ecological relevance in the fruitfly of the balance between CO2 (repellent cue informing unfavorable conditions) and fruit volatiles (CO2-masking food cue) is somewhat different from the possible relevance in the predatory mites of the balance between (E)-β-ocimene (food cue informing potent prey density) and floral volatiles (β-ocimene-masking cue informing a potent decline in quality or ephemerality of prey-inhabiting leaves as a result of plant flowering stage rather than nonflowering stage). In order to understand the ecological significance of the inhibition, molecular and neuronal approaches to assess β-ocimene-detection machinery and its inhibitory system in P. persimilis should be conducted in addition to the behavioral assays performed in the current study.
Torenia is known not only as a horticultural plant but also as an experimental plant with several useful characteristics, that is, ease of genetic transformation, ability to differentiate advantageous structures, production of an embryo sac, and relatively small genome size (1.71 × 108 bp) (Aida, 2008). The two torenia species (hybrids) used in the current study showed distinct indirect defense responses to spider mite attack. SWB is probably well suited to make indirect defense responses as a result of its ability to undergo the induction of HIPV emission and the consequent bodyguard attraction, whereas CrW and CrV are not. As described earlier, similar genetic variability of the induced HIPV emissions was previously examined in maize, cucumber and Arabidopsis, and the results revealed a broad range of genetic diversity (Degen et al., 2004; Huang et al., 2010; Kappers et al., 2010; Snoeren et al., 2010). Therefore, a better understanding of the genetic control of induced volatile emissions may help in the development of plant varieties particularly attractive to carnivores and other biological control agents and perhaps more repellent against herbivores (Degen et al., 2004).
The present PlOS-based manipulation in transgenic torenia provided an intriguing tritrophic system, as predator attraction was enhanced only when nonflowering plants were infested. In brief, this study contributes to our understanding of how predatory mites may use volatile signals to locate potential prey. The data specifically show that the attractiveness of a specific volatile compound largely depends on the background odors, including both HIPVs and floral volatiles, with which it is perceived by the predators. Indeed, this raises a question about the effect of other trans-volatiles, individually and in combination with endogenous volatiles, on attracting predators. Recent laboratory studies using synthetic compounds suggested that methyl salicylate alone attracted P. persimilis (van Wijk et al., 2008) and another predatory mite, Neoseiulus californicus (Shimoda, 2010). Yet, in another case, a blend of HIPVs included repellent or inhibitory cues (e.g. oximes) that caused masking of the attractiveness of methyl salicylate to P. persimilis (Kappers et al., 2011). Because of such complexity, it is necessary to construct model system(s) using genetic ‘gain’ of particular plant volatiles in order to clarify the ecological and agricultural significance of HIPVs using a wide array of methodological approaches (e.g. laboratory, glasshouse and field assays).
We gratefully acknowledge Drs Yoshikazu Tanaka (Suntory Ltd., Japan) and Ryutaro Aida (National Institute of Floricultural Science, Japan) for providing the torenia materials; Dr Richard Karban (University of California, Davis, USA) for comments on the manuscript; Ms Kikumi Katami for volatile analysis; and Mrs Kimiko Kanbe and Mrs Yumiko Togashi for rearing mites and plants. This work was financially supported in part by the Global COE Program A06 of Kyoto University; a research grant for Exploratory Research on Sustainable Humanosphere Science from the Research Institute for Sustainable Humanosphere (RISH), Kyoto University to G.A.; a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science, and Technology of Japan (no. 21770042) to G.A. and (no. 22380038, no. 23510271) to T.S.; and the Bio-oriented Technology Research Advancement Institution.