Integration of two herbivore‐induced plant volatiles results in synergistic effects on plant defence and resistance

Abstract Plants can use induced volatiles to detect herbivore‐ and pathogen‐attacked neighbors and prime their defenses. Several individual volatile priming cues have been identified, but whether plants are able to integrate multiple cues from stress‐related volatile blends remains poorly understood. Here, we investigated how maize plants respond to two herbivore‐induced volatile priming cues with complementary information content, the green leaf volatile (Z)‐3‐hexenyl acetate (HAC) and the aromatic volatile indole. In the absence of herbivory, HAC directly induced defence gene expression, whereas indole had no effect. Upon induction by simulated herbivory, both volatiles increased jasmonate signalling, defence gene expression, and defensive secondary metabolite production and increased plant resistance. Plant resistance to caterpillars was more strongly induced in dual volatile‐exposed plants than plants exposed to single volatiles.. Induced defence levels in dual volatile‐exposed plants were significantly higher than predicted from the added effects of the individual volatiles, with the exception of induced plant volatile production, which showed no increase upon dual‐exposure relative to single exposure. Thus, plants can integrate different volatile cues into strong and specific responses that promote herbivore defence induction and resistance. Integrating multiple volatiles may be beneficial, as volatile blends are more reliable indicators of future stress than single cues.

Although plant perception of individual environmental cues is relatively well understood, less is known about the capacity of plants to integrate multiple environmental cues (Finch-Savage & Leubner-Metzger, 2006). Integrating multiple cues may enable plants to obtain more reliable information of a given environmental condition than individual cues. Many volatiles that are released from leaves upon herbivore attack are also released constitutively by other sources, including flowers, bacteria, and fungi (Piechulla, Lemfack, & Kai, 2017;Tholl, Sohrabi, Huh, & Lee, 2011), and thus do not provide reliable information about the presence of an herbivore on a neighbouring plant (Baldwin et al., 2006). By contrast, the overall composition of herbivore-induced volatile blends is often highly species and stress-specific and may thus indicate the presence of herbivores more reliably (Junker et al., 2017;McCormick, Unsicker, & Gershenzon, 2012). Whether plants can integrate multiple volatile cues into defence responses is not well understood (Erb, 2018;Ruther & Kleier, 2005).
The perception of herbivore-induced plant volatiles has been studied in detail in maize (Zea mays). Maize plants that are exposed to volatile blends from herbivore-attacked plants respond more rapidly and more strongly to subsequent herbivore attack (Engelberth, Alborn, Schmelz, & Tumlinson, 2004;Ton et al., 2007). This form of priming includes higher amounts of jasmonates, higher expression of defencerelated genes, and higher emission of terpene volatiles (Engelberth et al., 2004, Ton et al., 2007. Furthermore, caterpillar growth is reduced and herbivore natural enemies are more strongly attracted to herbivoreattacked maize plants that are exposed to herbivore-induced volatiles (Ton et al., 2007). So far, two components of the herbivore-induced volatile blend of maize have been identified to trigger defence priming.
GLVs are specific for plants, but are released in response to many stresses including drought, mechanical wounding, herbivore attack, and pathogen infection (Ebel, Mattheis, & Buchanan, 1995;Scala, Allmann, Mirabella, Haring, & Schuurink, 2013). By contrast, indole is produced by many different organisms and plant tissues (Bailly et al., 2014;Stamm, Lottspeich, & Plaga, 2005), but its release from plant leaves seems to be specific to herbivore attack, as herbivore-derived elicitors, but not wounding alone induce strong indole emissions (Frey et al., 2000), and the indole biosynthesis gene ZmIGL is induced by herbivore attack, but not by other stresses such as salt stress or fungal infection (Erb et al., 2009). Thus, GLVs and indole complement each other in terms of the information they convey, and the simultaneous presence of GLVs and indole may be a better predictor of the presence of a herbivore-attacked plant than each cue alone. As both GLVs and indole prime jasmonate defenses, it is conceivable that they may have additive effects on defence priming.
Based on these considerations, we investigated how simultaneous exposure of maize plants to HAC and indole affects maize defenses.
We first quantified the impact of HAC and indole individually on phytohormone production, defence gene expression, and defence metabolite accumulation in plants that were induced by simulated herbivory and measured the influence of these volatiles on plant resistance to herbivores. We then compared the effects of individual volatile exposure with the effects of simultaneous exposure to HAC and indole.
We tested for synergistic effects of HAC and indole exposure by comparing the effects elicited by simultaneous exposure with the calculated additive effects of the individual exposures (Machado, Arce, McClure, Baldwin, & Erb, 2018). Our experiments reveal that maize plants integrate two different herbivore-induced volatiles into strong and specific defence signatures.

| Plants and herbivores
The maize (Z. mays) genotype B73 was used in this study. Maize seedlings were grown as previously described . Fourteenday-old plants were used for all experiments. Spodoptera littoralis eggs were provided by the University of Neuchâtel and reared on artificial diet as previously described (Maag et al., 2014). Herbivore oral secretions were collected from third instar S. littoralis larvae, which had been feeding on maize leaves for 48 hr. Briefly, the S. littoralis larvae were held with a pair of lightweight forceps, and regurgitation was induced by gently pinching their heads with another pair of forceps.
Oral secretions were collected using a micropipette and collected in Eppendorf tubes on ice. Oral secretions were stored at −80°C and diluted 1:1 in autoclaved Milli-Q water prior to use.
Control dispensers were prepared the same way using empty glass vials. Dispensers were prepared 24 hr before the start of experiments.

| Plant volatile exposure
To expose maize plants to synthetic indole and/or HAC, different sets of dispensers were individually introduced into 2-L glass vessels containing maize seedlings. The glass vessels were connected to a multiple air-delivery system via PTFE tubing. Purified air entered the glass vessels at a flow rate 0.3 L min −1 and was released through additional openings. This set-up ensured sufficient ventilation to avoid the buildup of unnatural volatile concentrations while effectively isolating the headspaces of the different plants. The volatile exposure system was placed into a greenhouse cabin (26 ± 2°C; 14: 10 hr, light [8 a. m.-10 p.m.]: dark; 55% relative humidity). Dispensers were added into the glass vessels in the evening (8 p.m.) before herbivore induction.
The following treatment combinations were used in all experiments: Control (empty dispenser), HAC (HAC dispenser), indole (indole dispenser); HAC + indole (HAC dispenser and indole dispenser). Although HAC is released 1 hr earlier than indole upon simulated herbivory , both volatiles are released continuously and simultaneously from maize leaves that are attacked by real caterpillars . We therefore exposed maize plants to HAC and indole using the same timing. After 16 hr of exposure (at 10 a.m.), the plants were carefully removed from the glass vessels, placed on a table in the same greenhouse cabin, and induced as described in the next section.

| Plant induction by simulated herbivory
To test how indole and HAC influence herbivore-induced plant responses, the pre-exposed maize plants were induced by wounding two leaves over an area of~0.5 cm −2 on both sides of the central vein with a razor blade, followed by the application of 10 μl of S. littoralis oral secretions. This treatment results in plant defence responses similar to real S. littoralis attack (Erb et al., 2009) and is referred to as "simulated herbivory" or "induction" throughout the rest of the manuscript.
In three different experiments, leaves were either harvested at 0 min (no herbivore induction), 45 min, or 5 hr after simulated herbivory and then flash frozen and used to quantify phytohormones, expression of defence-related genes, benzoxazinoids, and volatiles. Whole maize leaves, excluding the damaged area, were harvested. All analyses within time points were performed on the same leaf samples.

| Gene expression analysis
The influence of volatile exposure on the herbivore-induced expression of signalling and defence genes was determined by quantitative  (Erb et al., 2009). The relative gene expression levels of the target genes were calculated using the 2 −ΔΔCt method (Wong & Medrano, 2005). The primers of all tested genes are provided in Table S1.

| Volatile analyses
To assess the impact of volatile exposure to herbivore-induced volatile production, maize leaves were analysed 5 hr upon simulated herbivory. At this time point, volatile priming significantly increases terpene release in maize (Engelberth et al., 2004;Erb et al., 2015). Frozen leaf powder was analysed with solid-phase microextraction-gas chromatography-mass spectrometry (SPME-GC-MS; n = 5). This approach allows for the measurement of leaf volatile contents, which are highly correlated with volatile release rates in maize during daytime (Seidl-Adams et al., 2015). Fifty milligrams of leaf powder were placed in a 10-ml glass vial. An SPME fibre (100-μm polydimethylsiloxane coating; Supelco, USA) was then inserted into the vial and incubated at 60°C for 35 min. The incubated fibre was immediately analysed by GC-MS (Agilent 7820A GC interfaced with an Agilent 5977E MSD, USA) following previously established protocols (Huang et al., 2016). Major volatile compounds were identified by comparing mass spectra with the NIST Mass Spectral Library (USA) as well as authentic standards, and the abundance of each compound was determined by integrating individual peak areas.

| Herbivore resistance assays
To quantify the impact of volatile exposure on herbivore growth and plant resistance, individual preweighed second instar S. littoralis larvae were introduced into cylindrical mesh cages (1-cm height and 5-cm diameter) and then clipped onto the leaves of individual maize plants that were previously exposed to different volatile combinations (n = 10). The position of the cages was moved every day to provide sufficient food supply for the larvae.
Larval weight was recorded 4 days after the start of the experiment. For damage quantification, the remaining leaves were scanned, and the removed leaf area was quantified with Digimizer 4.6.1 (Digimizer).

| Statistical analyses
Gene expression, phytohormone, benzoxazinoid, volatile, larval growth, and leaf damage data were analysed by analysis of variance in plants that were pre-exposed to HAC, indole, or both volatiles simultaneously (HAC + Indole) and induced by simulated herbivory (+SE, n = 5). (f) Average transcript levels of ZmLOX10, ZmAOS, ZmPR1, and ZmPR5 (+SE, n = 5). FW, fresh weight. n.s., not significant. Treat., treatment. Gene expression is shown relative to the expression level of the control treatment. P values of one-way analyses of variance (ANOVAs) are shown (*P < 0.05, **P < 0.01, ***P < 0.001). Dashed lines indicate calculated additive effects of single volatile exposures. Letters indicate significant differences between different volatile exposure treatments (P < 0.05, oneway ANOVA followed by multiple comparisons through FDR-corrected LSMeans). Stars indicate a significant difference between the double exposure treatment and the calculated additive effect of both single treatments (*P < 0.05, Student's t tests) discovery rate (FDR) method (Benjamini & Hochberg, 1995). Normality was verified by inspecting residuals, and homogeneity of variance was tested through Shapiro-Wilk's tests using the "plotresid" function of the R package "RVAideMemoire" (Herve, 2015). Datasets that did not fit assumptions were log e -transformed to meet the requirements of equal variance and normality. Potential synergism was evaluated using a previously described approach (Machado et al., 2018 (Chapman, Schenk, Kazan, & Manners, 2002). Raw data were scaled with the "scale" function in R, and PCAs were then performed using the "MVA" function of the "RVAideMemoire" package and the "rda" function of the "vegan" package (Herve, 2015;Oksanen et al., 2013). Permutational ANOVAs were then conducted using the "adonis" function of the "vegan" package with 999 permutations. All statistical analyses were conducted with R 3.2.2 (R Foundation for Statistical Computing, Vienna, Austria) using the packages "car," "lsmeans," "vegan," and "RVAideMemoire"

| Accession numbers and data availability
The sequence data of maize genes can be found in the GenBank/ EMBL database under the following accession numbers: ZmActin

| Pre-exposure to HAC and indole specifically and synergistically increases the expression of defence genes in induced plants
To further explore the interactions of HAC and indole in regulating plant defence responses, we measured the expression levels of four defensive marker genes in volatile pre-exposed plants 5 hr after induction by simulated herbivory: the putative proteinase inhibitors ZmMPI (Farag et al., 2005;Tamayo, Rufat, Bravo, & San Segundo, 2000), ZmSerPIN and ZmCyst Ton et al., 2007), and the insecticidal ribosome-inactivating protein ZmRIP2 (Chuang et al., 2014). Exposure to HAC and indole individually increased the expression of ZmMPI, ZmSerPIN, and ZmRIP2 (Figure 2a-c). ZmCyst expression was increased by HAC, but not by indole (Figure 2d).

| Pre-exposure to HAC and indole synergistically regulates BX biosynthesis in induced plants
Benzoxazinoids (

| Pre-exposure to HAC and indole does not synergistically regulate volatile production in induced plants
Exposure of plants to both HAC and indole individually can prime herbivore-induced terpene emissions (Engelberth et al., 2004;Erb et al., 2015). As terpene biosynthesis in maize are regulated by jasmonates (Schmelz, Alborn, Banchio, & Tumlinson, 2003;, we expected additive or synergistic effects of simultaneous HAC and indole exposure on volatile production similar to the defence marker genes and BXs. Exposure of maize plants to HAC and indole individually followed by simulated herbivory increased the production of linalool, (3E)-4,8-dimethyl-1,3,7nonatriene (DMNT), (E)-α-bergamotene, (E)-α-farnesene and indole 5 hr after induction (Figure 4a-e). Simultaneous exposure to HAC and indole did not further increase volatile production. For indole, we even detected significantly lower amounts in plants exposed to both volatiles than would be expected in an additive scenario. Transcript levels of genes involved in terpene synthesis, including ZmCYP92C5, ZmTPS2, ZmTPS3, ZmTPS10, and ZmIGL (Frey et al., 2000;Richter et al., 2016;Schnee et al., 2006), showed a similar pattern ( Figure 4f). 3.5 | Pre-exposure to HAC and indole increases herbivore resistance of maize in an additive manner To investigate how HAC and indole pre-exposure influences herbivore performance and plant resistance, we measured S. littoralis growth and damage on volatile-exposed plants. Pre-exposure to HAC or indole individually reduced S. littoralis growth and plant damage ( Figure 5). Simultaneous pre-exposure to HAC and indole further increased this effect, with reductions of larval growth and damage attaining 40% ( Figure 5). Thus, HAC and indole enhance plant resistance against herbivores in an additive manner.
3.6 | Pre-exposure of HAC, but not indole, directly induces defence gene expression To investigate whether the observed synergistic effects on plant defenses are due to priming or direct induction by volatile exposure, we measured the expression of the different defence marker genes upon HAC and indole exposure without further induction. HAC preexposure significantly increased the expression of the tested jasmonate, volatile, and benzoxazinoid biosynthesis genes as well as other defence genes ( Figure 6). By contrast, indole pre-exposure did not directly induce any defence marker genes ( Figure 6). Expression of the SAresponsive genes ZmPR1 and ZmPR5 was not changed by HAC or indole exposure ( Figure 6). Simultaneous exposure to HAC and indole resulted in similar gene expression patterns as HAC alone, with the exception of the DMNT biosynthesis gene ZmCYP92C5, whose expression was synergistically enhanced by double exposure (Figure 6). Thus, HAC, but not indole, directly induces a broad spectrum of defence genes. Furthermore, most of the synergistic effects observed upon double exposure after induction by simulated herbivory (Figures 1-4) are likely due to priming rather than direct induction by HAC and indole. Average growth rate of Spodotera littoralis caterpillars feeding on plants that were pre-exposed to HAC, indole, or both volatiles simultaneously (HAC + Indole, +SE, n = 10). (b) Average consumed leaf area (+SE, n = 10). n.s., not significant. Treat., treatment. The results of one-way analyses of variance (ANOVAs) are shown (**P < 0.01, ***P < 0.001). Dashed lines indicate calculated additive effects of single volatile exposures. Letters indicate significant differences between different volatile exposure treatments (P < 0.05, one-way ANOVA followed by multiple comparisons through FDR-corrected LSMeans) marker genes, and the production of defensive secondary metabolites in plants. Dual exposure also markedly suppressed herbivore growth and plant damage. These patterns are unlikely due to direct induction, as HAC, but not indole, directly increased defence gene expression.

|
Instead, HAC and indole primed maize plants together to respond more strongly upon induction. A likely mechanism to explain this pattern is convergence of HAC and indole in early defence signalling.
Both HAC and indole act upstream of the jasmonate signalling pathway, possibly by priming the activity of MAP kinases (Ye, Glauser, Lou, Erb, & Hu, 2018) and/or WRKY transcription factors (Engelberth, Contreras, Dalvi, Li, & Engelberth, 2013;Mirabella et al., 2015). As most of the measured downstream defenses are under the control of jasmonates (Dafoe et al., 2011;Moraes et al., 2008;Schmelz, Alborn, Banchio, et al., 2003;Stotz et al., 2002;Ton et al., 2007), the synergistic effects of HAC and indole on jasmonate signalling likely explain the enhanced defence responses observed in this study. We  Pre-exposure to (Z)-3-hexenyl acetate (HAC), but not indole, directly induces defence gene expression in maize plants. (a) Average transcript levels of genes involved in JA biosynthesis, SA signalling and benzoxazinoid biosynthesis in plants that were pre-exposed to HAC, indole, or both volatiles simultaneously (HAC + Indole) without subsequent induction (+SE, n = 5). (b) Average transcript levels of putative proteinase inhibitors and a ribosomeinactivating gene ZmRIP2 (+SE, n = 5). (c) Average transcript levels of genes involved in terpene and indole biosynthesis (+SE, n = 5). n. s., not significant. Treat., treatment. Gene expression is shown relative to the expression level of the control treatment. P values of one-way analyses of variance (ANOVAs) are shown (*P < 0.05, **P < 0.01, ***P < 0.001).
Dashed lines indicate calculated additive effects of single volatile exposures. Letters indicate significant differences between different volatile exposure treatments (P < 0.05, one-way ANOVA followed by multiple comparisons through FDR-corrected LSMeans). Stars indicate a significant difference between the double exposure treatment and the calculated additive effect of both single treatments (*P < 0.05, Student's t tests) From an ecological point of view, the integration of HAC and indole into stronger defence priming may allow plants to adjust their defence investment according to the reliability of the perceived cues.

ZmMPI
As GLVs can be emitted in response to many stresses, including for instance mechanical injury in the absence of herbivory (Ebel et al., 1995;Scala et al., 2013), they cannot be used as reliable cues by plants to anticipate herbivory. The same is true for indole alone, which can emanate from various environmental sources (Bailly et al., 2014;Stamm et al., 2005) but is emitted from leaves in much greater quantities upon contact with herbivore-elicitors than wounding alone (Frey et al., 2000;Zhuang et al., 2012). The simultaneous presence of indole and GLVs on the other hand may be a relatively robust predictor of herbivore attack due to the complementary nature of their information contents. Given that priming can be costly (van Hulten, Pelser, van Loon, Pieterse, & Ton, 2006), adjusting the magnitude of priming according to the reliability of the perceived cues may be beneficial.
Especially when the reliability of individual volatile cues is low, the ability to integrate multiple volatile cues may confer important advantages to plants. However, it is important to point out that the integration of multiple volatile cues is not always necessary to obtain reliable information from the environment. Insect pheromones, for instance, can be fairly specific and may be sufficient to reliably indicate the presence of a herbivore. In line with this argument, Solidago altissima plants respond similarly to the exposure to a single pheromone component of the goldenrod gall fly as to the full volatile blend of the herbivore (Helms et al., 2017).
Double exposure to HAC and indole enhanced direct defenses but had no clear effect on the emission of induced volatiles, which are often viewed as indirect defenses that attract natural enemies (Turlings & Erb, 2018). Recent work in tomato furthermore demonstrates that changes in light quality leading to phytochrome B inactivation shifts tomato defenses from direct to volatile-mediated indirect defenses (Cortés, Weldegergis, Boccalandro, Dicke, & Ballaré, 2016). Thus, plants seem to be able to integrate various environmental cues to regulate their relative investment into direct and indirect defenses. Regarding the results of the present study however, we would like to remain cautious with our interpretation, as the effects of the observed patterns on indirect defenses have not been quantified, and the ecological interpretation of defence responses of a domesticated plant warrants caution due to possible pleiotropic effects of domestication. Nevertheless, exploring if and how the composition of volatile cues influences the relative investment of plants into direct and indirect defenses is an exciting prospect of this work.

| CONCLUSIONS
Plants perceive a variety of volatiles from the environment. Our work lends support to the concept that plants are also able to integrate multiple volatile cues into specific, and possibly adaptive, defence responses. Understanding the mechanisms and ecological factors that shape the evolution of signal integration will be important to improve our understanding of plant responses to complex volatile blends. Plants were pre-exposed to HAC, indole, or both volatiles simultaneously (HAC + Indole, n = 5) prior to induction by simulated herbivory. PCAs include data on defence gene expression at 0 min, phytohormones and signalling related gene expression at 45 min, and defence gene expression and secondary metabolite production at the 5 hr time point. Data points represent individual replicate samples. Vectors of individual defence markers are shown as grey arrows. P values of permutational analyses of variance ("Adonis test") between treatments are shown ACKNOWLEDGMENTS We thank Monika Hilker and Thomas Schmülling for the invitation to contribute to this special issue. We further thank Christelle A.M.
Robert for the help with volatile analyses and for insightful discussions and Klaus Schlaeppi for the help with statistics. The comments of two anonymous reviewers helped to improve the paper.