Using ‘mute’ plants to translate volatile signals


*(fax +49 3641 571 102; e-mail


When attacked by herbivores, plants release volatile organic compounds (VOCs) that attract natural enemies of the herbivores and function as indirect defenses. Whether or not neighboring plants ‘eavesdrop’ on these VOCs remains controversial because most studies use unrealistic experimental conditions and VOC exposures. In order to manipulate exposures of wild-type (WT) Nicotiana attenuata‘receiver’ plants, we elicited transformed ‘emitter’ plants, whose production of herbivore-induced C6 green leaf volatiles (GLVs) or terpenoid volatiles was genetically silenced, and placed them up-wind of WT ‘receiver’ plants in open-flow experimental chambers. We compared the transcriptional and secondary metabolite defense responses of WT receiver plants exposed to VOCs from these transgenic emitter plants with those of plants exposed to VOCs from WT emitter plants. No differences in the constitutive accumulation of defense metabolites and the signal molecule jasmonic acid (JA) were found. Additional elicitation of receiver plants revealed that exposure to WT, GLV-deficient and terpenoid-deficient volatile blends did not prime induced defenses, JA accumulation, or the expression of lipoxygenase 3 (NaLOX3), a gene involved in JA biosynthesis. However, exposure to wound- and herbivore-induced VOCs significantly altered the transcriptional patterns in receiver plants. We identified GLV-dependent genes by complementing the GLV-deficient volatile blend with a mixture of synthetic GLVs. Blends deficient in GLVs or cis-α-bergamotene regulated numerous genes in receiver plants that did not respond to the complete VOC blends of WT emitters, indicating a suppressive effect of GLVs and terpenoids. Whether these transcriptional responses translate into changes in plant fitness in nature remains to be determined.


The many different volatile organic compounds (VOCs) that plants release into the atmosphere mediate various pollination and defense mutualisms that plants have evolved with insects (Dobson and Bergstrom, 2000; Reddy and Guerrero, 2004). The ability of plants to interact with other organisms via VOCs fuels the expectation that they communicate similarly with each other (Dicke et al., 2003). Although the word ‘communication’ is loaded, signifying different things to different researchers, most would agree that, for the term to apply, information should be exchanged regardless of ‘intent’ or fitness benefit for either party. Two decades ago, researchers reported that wounding (W) or herbivore attack resulted in changes in herbivore resistance, or in the secondary metabolites mediating this resistance, not only in the attacked plants (‘emitters’) but also in adjacently growing plants (‘receivers’; Baldwin and Schultz, 1983; Rhoades, 1983). In some experiments, aerial transfer of information was the most parsimonious explanation of the results (reviewed in Baldwin et al., 2002).

Once the receiver plants are exposed to VOCs, they must be recognized before a response can take place. Research with plants has lagged behind research with animals mainly because knowing when a plant has responded is difficult. While animals respond with visually apparent changes in behavior, plant responses may be observed as changes in the transcriptome, proteome or metabolome – all of which require sophisticated analysis – that may or may not increase plant resistance to further herbivore attack (Figure 1). Recent analyses of the transcriptional responses of plants to volatile exposure have rekindled interest in the phenomenon (Baldwin et al., 2002; Dicke and Bruin, 2001). VOC-induced transcript accumulation of genes involved in oxylipin and ethylene signaling as well as terpene biosynthesis has been shown for detached lima bean (Phaseolus lunatus) leaves (Arimura et al., 2000a,b, 2001, 2002) and Arabidopsis seedlings (Bate and Rothstein, 1998). Analysis of this early stage of the total phenotypic response of the plant (Figure 1) allows the identification of signals that plants have perceived but ignored. However, how to manipulate ecologically realistic exposures to the volatile signals remains a largely unmet challenge. A brief review of previous experimental efforts highlights the problems.

Figure 1.

Overview of the experimental schedule and the hierarchy of possible responses in plants exposed to volatile organic compounds (VOCs) released from wounded or herbivore-attacked plants.
After the signal is recognized by the receiver plant, the plant may respond with changes in its transcriptome, proteome, and metabolome that may or may not result in functionally significant changes in its phenotype (e.g. herbivore performance). All experiments were conducted in open-flow chambers each containing emitter plants of different genotypes that were genetically altered in their elicited VOC emissions [wild type (WT), a green leaf volatile (GLV)-deficient line of Nicotiana attenuata silenced in expression expression of as-hp1 and a line of N. attenuata silenced in the expression of as-lox] and WT receiver plants. The release of VOCs from emitter plants was repeatedly elicited over 2 days (with the exception of experiment VI in which Manduca sexta larvae fed on emitter plants for 5 days). After the 2-day VOC exposure, all emitter and receiver plants were removed from the chambers, and receiver plants were designated either for the analysis of ‘constitutive’ or of ‘induced’ VOC-elicited changes in defense-related transcripts [by quantitative reverse transcriptase–polymerase chain reaction (qRT-PCR) and microarrays], secondary metabolites [nicotine and trypsin proteinase inhibitors (TPIs)], and the endogenous signal that elicits them [jasmonic acid (JA)]. ‘Constitutive’ changes were analyzed in receiver plants that were only exposed (= exposed + unwounded) to the different volatile blends. To determine whether VOC exposure primes the induction of these defense responses, we additionally elicited (potentiated) receiver plants directly after termination of exposure (at time 0) with standardized puncture wounds with M. sexta larvae oral secretions and regurgitants (W + R = exposed + potentiated by W + R). Leaf material for the analysis of defense responses was harvested from both undamaged and potentiated receiver plants at the times of maximal response as determined by previous experimentation. To avoid problems of pseudoreplication (see Fowler and Lawton, 1985), the number of VOC-exposure chambers determined the degrees of freedom for all statistical analyses of all data obtained in the experiments.

There is no question that plants respond to being ‘fumigated’ with high concentrations of VOCs; whether these responses are relevant to plant–plant signaling hinges on whether the exposures realistically represent what occurs in nature. Given that measuring a response is already a challenge for researchers working with sedentary organisms, it is not surprising that many researchers have chosen experimental designs that maximize the probability of detecting a response by increasing exposures. Most studies have been performed in airtight chambers in which the volume of air surrounding a receiver plant is kept small, thereby minimizing the dilution of possible signals. The restrained airspace in these chambers not only increases VOC concentrations, but also influences the physiological status of the plant. If the chambers are illuminated, normal photosynthetic processes will reduce CO2 levels in the sealed chambers, curtailing carbon fixation. Consequently, as plants reach their CO2 compensation points, they may respond by increasing the number of stomata in the open state, thereby increasing the exposure of mesophyll cells to the volatile signals. Studies that use open-flow systems that provide a continuous air supply avoid these complications. Considering the type of exposure set-up employed, in combination with the use of intact plants or detached leaves or branches as receivers, previous glasshouse-based studies can be categorized into the following three groups with decreasing ecological relevance: (i) open-flow/intact receiver plant (Pettersson et al., 1999; Preston et al., 1999; Shulaev et al., 1997); (ii) sealed chamber/intact receiver plant (Arimura et al., 2001, 2002; Baldwin and Schultz, 1983; Bate and Rothstein, 1998; Birkett et al., 2000; Engelberth et al., 2004; Farmer and Ryan, 1990; Shulaev et al., 1997; Tscharntke et al., 2001); (iii) sealed chamber/detached receiver leaves or branches (Arimura et al., 2000a,b, 2001, 2002; Farmer and Ryan, 1990). The interpretation of responses in excised receiver organs as indicative of ‘communication’ works well for the analysis of pheromone perception in ablated insect antennas (Birkett et al., 2000), but the lack of discrete tissues in plants with receptors for volatile signals and the recognition that ablation results in dramatic changes in all response variables obscure the results of such studies. Using excised leaves as emitters of volatiles (Arimura et al., 2000a,b, 2002; Birkett et al., 2000; Engelberth et al., 2004; Tscharntke et al., 2001) is also problematic, as excised leaves emit more herbivore-induced volatiles with a different blend compared with intact plants (Schmelz et al., 2001).

The availability of synthetic volatiles allows researchers to dissect the released blends into their active constituents, but few experiments conducted to date with synthetic mixtures have demonstrated that the concentrations used mimic natural exposures (Bate and Rothstein, 1998; Engelberth et al., 2004; Ruther and Kleier, 2005). Furthermore, which component of the blend contains the ‘information’ for a neighboring plant remains unknown. Herbivore-induced volatile blends are composed of products from various biosynthetic pathways (Pare and Tumlinson, 1997) such as the fatty acid/lipoxygenase pathway (e.g. green leaf volatiles; GLVs, Figure 2), the isoprenoid pathway (e.g. monoterpenes and sesquiterpenes), and the shikimic acid/tryptophan pathway (e.g. indole). ‘Crosstalk’ could result from the interaction of two or more volatile compounds and/or their effects on the receiver plant. The constituents are released with different kinetics (Turlings et al., 1998), and prominent diurnal patterns (De Moraes et al., 2001; Loughrin et al., 1994) superimposed on the temporal feeding patterns of the insect herbivores eliciting them. Until the relevant information for receiving plants is identified, these compositional and temporal dynamics represent a major challenge for studies relying on synthetic blends to simulate responses.

Figure 2.

A scheme of the oxylipin signaling cascade.
The function of the hydroperoxide lyase (HPL) and lipoxygenase (LOX3) enzymes in the production of wound- and herbivore-induced volatile organic compounds (VOCs; highlighted in gray; the numbers refer to the labeled peaks in the chromatograms of Figures 3, 4) and signals eliciting defensive secondary metabolites are illustrated. The substrates of HPL, the hydroperoxides 13-HPOT and 13-HPOD, are formed by dioxygenation of linolenic or linoleic acid by lipoxygenase (LOX). HPL cleaves HPOT and HPOD to form C6-aldehydes that are further metabolized by alcohol dehydrogenase (ADH) and isomerization factors (IF). The release of these green leaf volatiles after wounding is dramatically decreased by silencing the expression of HPL by antisense expression (as) of NaHPL in the as-hpl lines of N. attenuata (Table 2; Figure 3). 13-HPOT is also the precursor of jasmonic acid (JA), and silencing NaLOX3 by antisense expression in as-lox lines significantly decreases both the accumulation of JA after elicitation and the emission of cis-α-bergamotene after herbivory (Table 2; Figure 4), as well as the defense metabolites nicotine and TPIs.

Field studies, in contrast, meet the criteria of ecological realism concerning environmental conditions (Dolch and Tscharntke, 2000; Rhoades, 1983; Tscharntke et al., 2001), although the realism of some emitter treatments remains debatable (Karban, 2001; Karban and Maron, 2002; Karban et al., 2000; Kessler et al., 2005). However, the identification of the active components of the volatile blend is a Herculean task in field studies in which plants are exposed to a multitude of additional signals from their environment. Moreover, not all studies carried out under natural conditions support the claim that information is exchanged between infested and neighboring uninfested plants (Fowler and Lawton, 1985; Preston et al., 2001).

The use of mutants or transgenic plants (e.g. antisense or RNAi-silenced) defective in the release of defined parts of the VOC bouquet as emitters in experimental designs with receiver plants placed in open-flow chambers represents a novel solution to the challenges described above. The native diploid tobacco Nicotiana attenuata is a particularly useful system in which to study herbivore-induced responses, as the responses elicited by attack from a diverse herbivore community have been characterized in terms of the signal cascades, transcriptional responses, and down-stream direct and indirect defense responses (Baldwin and Preston, 1999; Halitschke and Baldwin, 2004; Steppuhn et al., 2004; Zavala et al., 2004a). Many of the herbivore-responsive genes of N. attenuata have been identified by cDNA differential display, subtractive hybridization, and cDNA-amplified fragment-length polymorphism display (Halitschke and Baldwin, 2003; Halitschke et al., 2001; Hermsmeier et al., 2001; Hui et al., 2003; Schittko et al., 2001; Voelckel and Baldwin, 2003). These genes have been spotted onto microarrays and their expression behavior analyzed in response to various environmental stresses (Halitschke and Baldwin, 2003; Hui et al., 2003; Izaguirre et al., 2003; Voelckel and Baldwin, 2004). When N. attenuata leaves are damaged by attack from mammalian browsers, they rapidly release large amounts of GLVs (Hatanaka, 1996), increasing their levels of nicotine (Baldwin, 1999) and the activity of trypsin proteinase inhibitor (TPI, Baldwin, 1999; Van Dam et al., 2001; Zavala et al., 2004a,b), all of which require jasmonate (JA) signaling (Halitschke and Baldwin, 2003). This wound response is reorganized when larvae of the tobacco hornworm (Manduca sexta), a specialized lepidopteran herbivore of N. attenuata, causes the damage. The reorganization of the wound response – which occurs when eight fatty acid amino acid conjugates, present in the oral secretions and regurgitants (R) of M. sexta, are introduced into plant wounds during feeding – begins with a dramatic JA burst in the attacked leaves (Schittko et al., 2000). The JA burst alters the expression of numerous genes and the accumulation and release of secondary metabolites, including terpenoid VOCs (Halitschke and Baldwin, 2003; Halitschke et al., 2000; Halitschke et al., 2001; Kahl et al., 2000; Roda et al., 2004).

Here we use a recently characterized GLV-deficient line of Nicotiana attenuata silenced in expression of (as-hplHalitschke et al., 2004) as emitter plants to study the role of GLVs in between-plant signaling. Hydroperoxide lyase (HPL) is a key enzyme of the oxylipin pathway that cleaves lipoxygenase-derived fatty acid hydroperoxides to form C6-aldehydes (Figure 2). These aldehydes, together with their corresponding C6-alcohols and -esters, constitute a substantial part of the VOCs released by freshly wounded leaves (Hatanaka, 1996). The expression of HPL in antisense orientation dramatically decreases the amounts of GLVs in the headspace of wounded as-hpl plants compared with those of wounded wild-type (WT) N. attenuata plants, but does not affect the emission of other herbivore-induced VOCs (e.g. cis-α-bergamotene; Halitschke et al., 2004; Table 2). Additionally, we used a line of N. attenuata silenced in the expression of as-lox as emitter plants to determine whether the terpenoid VOCs that are released after herbivore attack function in between-plant signaling. as-lox plants are impaired in their ability to accumulate JA in response to herbivory and release significantly fewer terpenoid VOCs after herbivore attack, whereas the release of GLVs from as-lox plants does not differ from that of WT plants (Halitschke and Baldwin, 2003; Kessler et al., 2004).

Table 2.  Volatile compounds released by Nicotiana attenuata emitter plants
GenotypeTreatmentVolatile released by emitter plants (ng)
Z3-hexenalE2-hexenalZ3-hexenolE2-hexenolZ3-hexenyl acetateZ3-hexenyl butyratecis-α-bergamotene
  1. Mean values ± standard errors are shown for the volatile compounds released.

  2. Plants of different genotypes [wild type (WT), a green leaf volatile (GLV)-deficient line of Nicotiana attenuata silenced in expression of the hydroperoxide lyase gene (as-hpl), and a line of N. attenuata silenced in the expression of the lipoxygenase 3 gene (as-lox)] were treated by mechanical wounding (W) or application of Manduca sexta R to mechanical wounds (W + R), or were left untreated (ctrl). Volatile compounds were analyzed inside open-flow chambers containing pairs of identically treated emitter plants during 48 h of repeated elicitation (four times per light period) starting immediately after the first treatment. The volatile organic compound (VOC) blend of wounded as-hpl plants was complemented with one of two synthetic GLV mixtures (MixA and MixB), both of which contained the major GLVs released by N. attenuata in different concentrations. After each wounding of as-hpl plants, 10 μl of the GLV mix was applied to a Q-tip located next to the plants. Different letters designate significantly different amounts of released VOCs (Bonferroni post-hoc test; P < 0.05).

WTctrl0.61 ± 0.61a0.00 ± 0.00a0.99 ± 0.72a0.00 ± 0.00a0.00 ± 0.00a0.71 ± 0.71a0.47 ± 0.47a
WTW501.47 ± 80.82b223.76 ± 5.34c2375.48 ± 372.30e534.97 ± 45.45e64.61 ± 24.62e135.30 ± 45.32de0.00 ± 0.00a
as-hplW11.28 ± 10.17a16.04 ± 3.36b149.24 ± 7.71bc39.71 ± 6.85b7.87 ± 0.35c10.67 ± 1.38bc6.82 ± 1.88a
as-hplW + MixA0.34 ± 0.34a97.36 ± 25.01c276.04 ± 53.43cd185.53 ± 41.52d328.19 ± 62.53f16.58 ± 4.82bc27.74 ± 21.89a
as-hplW + MixB0.57 ± 0.57a47.27 ± 2.71bc512.19 ± 60.85d114.09 ± 13.23cd51.68 ± 6.93de81.30 ± 10.70d9.41 ± 1.92a
WTW + R0.62 ± 0.32a48.57 ± 29.42bc312.86 ± 28.57cd58.14 ± 13.29bc21.51 ± 5.09cd43.83 ± 14.04cd469.69 ± 240.58b
as-loxW + R0.00 ± 0.00a43.37 ± 8.06bc84.30 ± 9.10b58.44 ± 5.63bc1.75 ± 0.41b4.58 ± 0.27ab2.28 ± 0.22a

We examined the effects of VOCs released from WT, as-lox and as-hpl N. attenuata plants on adjacently growing WT conspecifics under continuous air flow in open-flow chambers with potted plants that could only exchange information through the air. Undamaged WT receivers were exposed to wound- and herbivore-induced volatiles [induced by applying M. sexta regurgitant (R) to puncture wounds or by M. sexta feeding] released from the different genotypes of emitter plants. The exposed WT receiver plants were analyzed for changes in ‘constitutive’ trypsin proteinase inhibitor (TPI) activity, nicotine and JA accumulation, and in defense-related transcripts. The GLV-deficient emissions of as-hpl plants were complemented by applying synthetic GLVs and the responses in WT receivers were reanalyzed. In addition to the analysis of constitutive metabolite and transcript levels, we analyzed the induced defense responses of VOC-exposed receiver plants (‘potentiated’) as the perception of herbivore-induced volatiles from damaged plants may allow neighboring unattacked plants to prepare for attack and more rapidly launch defense responses once attacked (Engelberth et al., 2004). The terms ‘potentiated’ and ‘constitutive’ therefore define receiver plants with or without additional elicitation treatment after the exposure. We used transformed emitter plants, impaired in the release of specific VOCs, in replicated open-flow experimental chambers to remove specific compounds or compound groups from the VOC bouquet in order to identify the compounds that could be responsible for between-plant signaling.


VOC profiles of the elicited plants

Wounding dramatically increased the release of all analyzed GLVs from N. attenuata WT plants (Table 2; anova, F6,13 ≤ 19.576, P < 0.0001, Bonferroni-corrected P < 0.05). The release of these GLVs was significantly suppressed in wounded as-hpl plants (Figure 3, Table 2, Bonferroni-corrected P > 0.05). Adding mixtures of synthetic GLVs partially complemented the GLV-deficient VOC blend of as-hpl plants (Table 2; Figure 3: as-hpl W + GLVmix A and B, where GLVmix is a mixture of six synthetic GLVs). By adding mix A to chambers with wounded as-hpl plants, we restored the E2-hexenal and E2-hexenol components of the VOC WT bouquet, but enriched the Z3-hexenyl acetate component (Table 2). By adding mix B, we were able to mimic the WT bouquet more closely. The accumulation of the esterified GLVs was recovered to levels released by wounded WT plants (P > 0.05), whereas the release of the aldehyde and free alcohol component was still higher from wounded WT plants (Table 2; Figure 3).

Figure 3.

Representative chromatograms of green leaf volatiles (GLVs; monitored by ions 67, 69, 82, 83, 93, 104, 132 and 161) that were trapped from wounded emitter plants in flow-boxes as they were used in this study.
The chromatograms illustrate the differences in GLVs trapped from undamaged wild-type (WT) plants (WT control), wounded WT plants (WT wounded), and wounded as-hpl plants (as-hpl wounded; as-hpl is a GLV-deficient line of Nicotiana attenuata silenced in expression of the hydroperoxide lyase gene). Four fully developed leaves from each of the two emitter plants (Figure 1) were wounded with a fabric pattern wheel and the wound-induced volatiles were trapped for 10 h starting immediately after wounding. Wounded as-hpl plants released significantly fewer GLVs such as Z-3-hexenal (1), E-2-hexenal (2), Z-3-hexenyl-acetate (3), 1-hexanol (4), Z-3-hexenol (5), E-2-hexenol (6) and Z-3-hexenyl-isobutyrate (7) compared with wounded WT plants (IS, internal standard). The addition of synthetic GLVs complemented the lack of GLVs in the volatile organic compound (VOC) bouquet of wounded as-hpl plants. We applied 10 μl of the GLV mixes on a Q-tip next to the as-hpl plant after each wounding. GLVmix A contained synthetic GLVs (peaks numbered 1–6) in equal amounts, whereas GLVmix B contained the same substances in ratios adjusted to mimic the release of WT plants.

Treatment of wounds with M. sexta R elicited the release of cis-α-bergamotene in WT plants (Figure 4, Table 2; anova, F6,13 ≤ 14.807, P < 0.0001, Bonferroni-corrected P < 0.05). The elicitation of this release is known to involve the jasmonate signaling cascade (Halitschke and Baldwin, 2003), and the release of cis-α-bergamotene was not increased in as-lox plants treated with R (Figure 4, Table 2; P > 0.05). These results demonstrate that our combination of treatment-specific elicitation and genetically manipulated biosynthetic pathways altered the release of target volatile compounds (i.e. GLVs or cis-α-bergamotene) and exposed receiver plants to specifically manipulated ‘natural’ volatile blends.

Figure 4.

Representative chromatograms of terpenoid volatile organic compounds (VOCs) that were trapped from untreated wild-type (WT) plants (WT control), from wounding and Manduca sexta regurgitant (W + R)-treated WT plants and plants of a line of N. attenuata silenced in the expression of (as-lox monitored by ions 93, 161 and 132).
The release of terpenoid VOCs [β-pinene (8), sabinene (9), 3-carene (10), limonene (11), cis-α-bergamotene (12) and trans-caryophyllene (13)] was elicited by applying M. sexta regurgitant (R) to puncture wounds (W + R). These volatiles are systemically released from attacked plants in maximum quantities 24 h after elicitation and the chromatograms are of 10-h air samples from the open-flow chambers, 24 h after elicitation (IS, internal standard).

VOC exposure did not elicit defense responses

TPI activity (Figure 5a, anova, F2,17 = 0.173; P = 0.8425) and nicotine concentrations (anova, F2,17 = 0.865; P = 0.4389) in receiver plants that had been exposed to VOCs released from wounded WT or as-hpl emitter plants (Table 1: experiment IA, C, E) did not differ significantly from those in receiver plants located downwind of unwounded emitter plants. Similar results were obtained in experiments in which the effects of R-elicited VOCs released from WT and as-lox emitter plants on neighboring plants were analyzed. The exposure to the different VOC blends alone (Table 1; experiments IIA, C, G) did not significantly alter any measured direct defense metabolites (TPI activity: anova, F2,17 = 1.058, P = 0.3688, Figure 5(b); nicotine: anova, F2,17 = 0.247; P = 0.7836). Receiver plants exposed to VOCs released from M. sexta-attacked WT and as-hpl emitter plants (Table 1, experiment V) did not differ in their TPI activity [Figure S1(a), anova, F2,9 = 2.674; P = 0.1226] and nicotine accumulation [Figure S1(b), data log-transformed, anova, F2,9 = 2.335; P = 0.1525] from receiver plants that had been exposed to VOCs from untreated WT emitter plants.

Figure 5.

Trypsin proteinase inhibitor (TPI) activity (mean ± standard error of the mean for seven replicate plants from seven replicate flow chambers) measured in the +2 node leaves of volatile organic compound (VOC)-exposed (‘unwounded’) or VOC-exposed and additionally elicited (‘potentiated’) receiver plants.
Potentiation of receiver plants was achieved by wounding and treatment with Manduca sexta regurgitant (W + R).
(a) TPI activity in Nicotiana attenuata plants after exposure to VOCs released from unwounded (ctrl) wild-type (WT) or wounded (W) as-hpl (a GLV-deficient line of N. attenuata silenced in expression of the hydroperoxide lyase gene) and WT plants (Table 1: experiment I).
(b) TPI activity in receiver plants that were exposed to VOCs released from unwounded WT (ctrl) or WT and as-lox plants (a line of N. attenuata silenced in the expression of the lipoxygenase 3 gene) which were elicited by W + R (Table 1: experiment II). TPI activity in unwounded and in potentiated receiver plants was measured 3 days after termination of exposure, the time of maximum TPI accumulation in unexposed W + R-elicited N. attenuata plants.

Table 1.  Summary of open-flow chamber exposure experiments
ExperimentTreatmentReplicatesaEmitter plantsReceiver plants
Genotypeb# of plantsc ElicitationdGenotype# of plants Treatmente Analyzed traitsf Microarraygfigure
  1. aEach replicate represents an individual open-flow chamber containing a set of emitter and receiver plants.

  2. bEmitter plants were either wild-type (WT), a GLV-deficient line silenced in the expression of the hydroperoxide lyase gene (as-hpl) or a cis-α-bergamotene-dificient line of Nicotiana attenuata silenced in the expression of the lipoxygenase 3 gene (as-lox).

  3. cNumber of identically treated emitter plants contained in each replicate exposure chamber.

  4. dEmitter plants were repeatedly elicited by mechanical wounding alone (W) or by wounding and application of Manduca sexta regurgitant (W + R) or were left untreated (ctrl). Additionally, we complemented the green leaf volatile (GLV)-deficient volatile organic compound (VOC) blend of wounded as-hpl plants with a mixture of synthetic GLVs (W + MixA or B). Each treatment was repeated four times during the light period on two subsequent days. In experiments V and VI, emitter plants were elicited by M. sexta larvae (caterpillar) feeding over 2 or 5 days.

  5. eAfter exposure, receiver plants were either left undamaged or additionally elicited by wounding and M. sexta regurgitant treatment (potentiation) to amplify putative responses to VOC exposure.

  6. fWe analyzed undamaged and potentiated receiver plants for changes in transcriptional patterns (MA, microarray analysis; NaLOX3, quantitative real-time PCR of expression of the lipoxygenase 3 gene) and in defense and signaling metabolites (JA, jasmonic acid; TPI, trypsin proteinase inhibitor activity; NIC, nicotine accumulation).

  7. gNumbers of the microarray hybridizations refer to the description in Figure 8(a). as-lox, a line of N. attenuata silenced in the expression of the lipoxygenase 3 gene; WT, wild type.

IA7WT2ctrlWT2 TPI, MA 5a, 8b
B7  ctrlWT2PotentiationTPI 
D7  WWT2PotentiationTPI 
E7as-hpl2WWT2 TPI, MA2
F7  WWT2PotentiationTPI 
G7as-hpl2W + Mix AWT2 MA3
IIA7WT2ctrlWT2 JA, TPI, NIC, MA 5b, 7b, 8c
B7  ctrlWT2PotentiationJA, TPI, NIC 
D7  W + RWT2PotentiationJA, TPI, NIC 
E7as-hpl2W + RWT2 JA 
F7  W + RWT2PotentiationJA 
G7as-lox2W + RWT2 JA, TPI, NIC, MA5
H7  W + RWT2PotentiationJA, TPI, NIC 
IIIA3WT2ctrlWT2PotentiationNaLOX3 7a
C3WT2W + RWT2PotentiationNaLOX3 
E3as-hpl2W + Mix BWT2PotentiationNaLOX3 
F3as-lox2W + RWT2PotentiationNaLOX3 
IVA10WT2ctrlWT4PotentiationTPI 6
B10WT2W + RWT4PotentiationTPI 
B4WT2CaterpillarWT2 JA, TPI, NIC 
C4as-hpl2CaterpillarWT2 JA, TPI, NIC 
VIA4WT4ctrlWT1 JA S2
C4WT4CaterpillarWT1 JA 
E4as-hpl4CaterpillarWT1 JA 

VOC exposure did not prime R-induced defense responses

Herbivore- or wound-induced volatiles from nearby attacked plants might not directly elicit defense traits in undamaged neighboring plants but could prepare them for future attack by priming their defense responses. To examine the effects of VOC exposure on the induction of defense traits, we analyzed receiver plants that had been exposed to the same wound- or R-induced VOC blends as described above but, additionally, were wounded and treated with M. sexta R after exposure (potentiated). As expected, the treatment by wounding and application of R elicited increased TPI activity and nicotine accumulation, but prior exposure to wound-induced VOCs [Table 1: experiments IIB, D, F; TPI activity: anova, F2,18 = 0.012; P = 0.9877, Figure 5(a); nicotine: anova, F2,16 = 0.100; P = 0.9056] or R-induced VOCs [Table 1: experiments IIB, D, F, H; TPI activity: anova, F2,18 = 1.250; P = 0.3102, Figure 5(b), nicotine: anova, F2,18 = 1.586; P = 0.2321] had no significant effects on the elicited responses. To test if exposure to herbivore-induced VOCs affects the rate of increase of TPI activity after elicitation, rather than the absolute activity, we measured TPI activity for four consecutive days in receiver plants that had been exposed to VOCs released from herbivore-induced or untreated control emitter plants and were subsequently potentiated (Table 1: experiment IV). The elicitation of TPI activity was marginally stronger in plants exposed to herbivore-induced VOCs on days 2–4 after elicitation (Figure 6; two-way anova: exposure treatment: F1,72 = 3.822; P = 0.0545; time after potentiation: F3,72 = 115.708; P < 0.0001).

Figure 6.

The speed of trypsin proteinase inhibitor (TPI) elicitation in plants exposed to volatile organic compounds (VOCs) from undamaged wild-type (WT) control plants (ctrl) or from wounding and Manduca sexta regurgitant (W + R)-elicited WT plants.
After 48 h of exposure, all receiver plants were W + R-elicited (potentiated) at time 0, and leaves of 10 replicate plants were analyzed on four consecutive days (Table 1: experiment IV).

VOC exposure did not elicit or prime JA accumulation

Exposure to different blends of VOCs neither influenced the constitutive levels of TPI activity and nicotine nor primed the elicitation of these two defense metabolites. As these two direct defenses represent only two of many potential defenses that are elicited by JA, the possibility remains that other JA-dependent traits are elicited or primed by VOC exposure. We therefore wished to determine whether VOC exposure primes the activation of the octadecanoid pathway and the accumulation of JA, as has been shown for maize (Zea mays; Engelberth et al., 2004). To extend our analysis to the transcriptional level, we performed an additional experiment with the same set-up (Table 1: experiment III) and measured the accumulation of NaLOX3 transcripts in receiver plants that had been potentiated after the VOC exposure. NaLOX3 encodes a key enzyme of the octadecanoid pathway that plays a central role in JA biosynthesis (Halitschke and Baldwin, 2003). Previous research had shown that the amount of NaLOX3 transcripts in N. attenuata plants is highest 1–2 h after wounding and M. sexta R treatment (Halitschke and Baldwin, 2003). We therefore harvested leaf material from all receiver plants at 2 and 6 h after potentiation to determine if VOC exposure altered the rate of transcript elicitation. As expected, the amount of NaLOX3 transcripts in the receiver plants was strongly increased 2 h after elicitation, but the amplitude of this increase was not influenced by the composition of the VOC blends that receiver plants were exposed to (Figure 7a, two-way anova, data log-transformed: emitter: F5,23 = 1.617; P = 0.1954, time after potentiation: F1,23 = 58.423; P < 0.0001). Additionally, we determined if VOC exposure increased JA levels in receiver plants or primed its elicitation. Neither the constitutive nor the induced JA levels of receiver plants exposed to W + R-induced VOCs from WT (Table 1: experiments IIC, D), as-hpl (experiments IIE, F) and as-lox emitter plants (experiments IIG, H; all Bonferroni-corrected P > 0.05, Figure 7b) differed significantly from levels in receivers that had been exposed to VOCs from untreated WT emitter plants (experiments IIA, B). Similar results were obtained in receiver plants that had been exposed to M. sexta-elicited VOCs (Table 1: experiments V and VI). Two days of exposure to VOCs elicited by caterpillars feeding on WT and as-hpl emitter plants (Table 1: experiment V) did not significantly change JA levels in receivers compared with control receiver plants (data log-transformed, anova, F2,6 = 1.849; P = 0.2505, Figure S1c). Extending the exposure time from 2 to 5 days did not influence amounts of JA in either undamaged (Table 1: experiment VIA, C, E: anova, F2,13 = 1.399; P = 0.2816, Figure S2) or potentiated receiver plants (Table 1: experiments VIB, D, F: data log-transformed: anova, F2,13 = 0.464; P = 0.6389, Figure S2).

Figure 7.

Volatile organic compound (VOC) exposure does not influence the transcription of the Nicotiana attenuata lipoxygenase 3 gene (NaLOX3) or jasmonic acid (JA) accumulation after elicitation by wounding and Manduca sexta regurgitant application (W + R; potentiation).
(a) Transcription of NaLOX3, a gene essential for JA biosynthesis, is not primed by VOC exposure. Transcripts were measured by real-time quantitative reverse transcriptase–polymerase chain reaction (qRT-PCR) in receiver plants exposed for 48 h to the VOCs released from the following emitter plants: unwounded wild-type (WT) plants (WT ctrl), wounded WT plants (WT W) and as-hpl plants (as-hpl W; as-hpl is a GLV-deficient line of N. attenuata silenced in expression of the hydroperoxide lyase gene), W + R-elicited WT plants (WT W + R) and as-lox plants (as-lox W + R; as-lox is a line of N. attenuata silenced in the expression of the lipoxygenase 3 gene) that had been elicited by applying M. sexta regurgitant (R) to standardized puncture wounds and from wounded as-hpl (as-hpl W + GLV B) plants that had a mixture of synthetic GLVs added to their VOC bouquet (Table 1: experiment III). All receiver plants were potentiated by W + R treatment after exposure and leaf samples were taken from three replicate plants 2 and 6 h after elicitation, respectively. NaLOX3 transcripts were normalized to the amount of sulfite reductase (ECI) transcripts in each sample. Values are shown as the mean (± standard error) of three biological replicates in arbitrary units resulting from a calibration with a 10× dilution series of cDNAs prepared from RNA samples that contain NaLOX3 transcripts. Different emitter types had no significant influence on the expression rate of NaLOX3.
(b) Effects of VOC exposure on JA accumulations in unwounded plants, which were only exposed to elicited VOCs, and in potentiated plants, which were potentiated after being exposed to the elicited VOCs (see Figure 1; Table 1: experiment II). Receiver plants were exposed for 48 h to VOCs released from WT, as-hpl and as-lox plants that had been elicited by W + R; control receivers, in contrast, were only exposed to VOCs from unelicited WT plants (ctrl). Potentiation (W + R treatments) of two receiver plants commenced immediately after 48 h of VOC exposure, while the remaining two receiver plants remained unelicited after exposure (‘unwounded’). JA levels in the leaves of unwounded and potentiated receiver plants were measured 35 min after elicitation, the time of maximum JA accumulation in W + R-elicited N. attenuata plants. Values are the mean ± standard error of the mean for seven plants from seven replicate chambers. Neither uninduced nor potentiated JA levels in plants exposed to the VOC bouquet from unwounded WT control plants differed significantly from those of plants exposed to the VOC bouquets from elicited WT, as-hpl or as-lox plants.

VOC exposure elicited transcriptional responses

The exposure of WT N. attenuata plants to wound- or R-induced VOCs did not elicit increases in defense metabolites or in a signal that mediates their elicitation; however, plants may respond in more subtle ways that can only be discerned by examining their transcriptomes (Figure 1). Exposure to VOCs has been reported to elicit transcriptional changes of defense-related genes (Arimura et al., 2000a; Bate and Rothstein, 1998) and therefore we extended our analysis of VOC-exposed plants to a broader transcriptional analysis of genes known to be regulated after herbivore attack in N. attenuata.

With the first two microarray hybridizations, we analyzed the effects of wound-induced VOCs on unattacked neighbors (Table 1: experiment I). RNA (transcribed into cDNA) from WT plants exposed to VOCs released from WT and as-hpl plants was hybridized against cDNA prepared from control WT plants exposed to VOCs from untreated emitter plants (Figure 8a, hybridizations nos. 1, 2). Receiver plants that were exposed to VOCs released from wound-induced WT plants had 59 genes significantly regulated (31 up-regulated and 28 down-regulated). Only nine of these genes had the same regulation (one up-regulated and eight down-regulated) in plants exposed to wound-induced VOCs emitted from GLV-deficient as-hpl plants, while 37 genes can be considered to be GLV-associated as their regulation changed when GLVs were absent from the volatile blend (Figure 8b).

Figure 8.

Summary of the transcriptional changes in receiver plants 48 h after exposure to different blends of wound- or wounding and Manduca sexta regurgitant (W + R)-elicited volatiles.
Genes that are up-regulated (↑) are shown in light gray, while down-regulated genes (↓) are shown in dark gray boxes. Non-regulated (−) genes are not colored.
(a) Description of five oligonucleotide microarray hybridizations. Samples were taken from wild-type (WT) receiver plants exposed to volatile organic compounds (VOCs) from different genotypes of Nicotiana attenuata [wild type (WT), a green leaf volatile (GLV)-deficient line silenced in expression of the hydroperoxide lyase gene (as-hpl), and a line silenced in the expression of the lipoxygenase 3 gene (as-lox)] that were treated as follows: wounded (W), wounded and the wounds treated with M. sexta regurgitant (W + R), or the addition of a mixture of synthetic GLVs (+GLVmix A). Treatments of emitter plants are shown as superscripts to the genotypes. cDNA samples hybridized together on a given array are connected by an arrow. Each sample involved cDNA extracted from receiver plants from each of seven replicate open-flow chambers, which were pooled, and labeled with either Cy3 (exposed) or Cy5 fluorescent dyes (control).
(b) Exposure to wound-induced VOCs released from WT plants caused up-regulation of 28 and down-regulation of 32 genes. Only nine of these genes showed the same regulation in receiver plants that were exposed to the GLV-deficient volatiles from wounded as-hpl plants, while the majority of regulated genes changed (two) or lost (35) their regulation. Adding synthetic GLVs to the volatile bouquet of as-hpl plants restored the up- or down-regulation of 11 genes that had previously lost their regulation after exposure to GLV-deficient volatiles from as-hpl plants. This result demonstrates that GLV exposure was responsible for the differences in transcript abundances.
(c) Exposure to VOCs released from W + R-elicited WT plants resulted in the up-regulation of 46 and the down-regulation of 28 genes. Of the 74 genes that were differently regulated in response to W + R-elicited volatiles from WT plants, 23 were similarly regulated when exposed to the cis-α-bergamotene-deficient VOCs from W + R-elicited as-lox plants, while 45 genes were no longer regulated.

With an additional hybridization (Figure 8a, no. 3) we investigated if the loss of regulation caused by as-hpl VOCs could be restored by adding a mixture of synthetic GLVs (GLVmix A) to the wound-induced as-hpl volatile bouquet and hybridized cDNA prepared from plants exposed to these volatiles against cDNA from control receivers (Table 1: experiment IG). We found that 11 (three up-regulated and eight down-regulated) of the genes that did not respond to wound-induced VOCs from as-hpl plants were regulated in a GLV-dependent manner as they were elicited both in plants exposed to VOCs released from wounded WT plants and in plants exposed to VOCs from wounded as-hpl plants complemented with synthetic GLVs (Figure 8b, Table S1). Two of these 11 genes are known to be involved in stress responses (the Capsicum annuum Sn-1 gene and the Lycopersicon esculentum catalase gene); one gene is known to function in photosynthetic processes (the Nicotiana tabacum photosystem II gene); the functions of the remaining eight genes remain unknown (Table S1).

In N. attenuata, the composition of the volatile bouquet elicited by caterpillar feeding differs in its composition from the bouquet elicited by wounding alone in that terpenoid VOCs, of which cis-α-bergamotene is most prominent (Table 2, Figure 4), are systemically released from caterpillar-attacked plants within 24 h of attack (Halitschke et al., 2000). This suite of terpenoid VOCs can be elicited by adding M. sexta R to puncture wounds (Halitschke et al., 2000). To examine the effects of these additional volatiles, we performed experiments in which emitter plants were wounded and the wounds were treated with M. sexta R (Table 1: experiment II). We hybridized cDNA from receiver plants that had been exposed to R-induced volatiles released either from WT or as-lox plants against cDNA from untreated control WT receivers (Figure 8a, hybridizations nos. 4, 5). Exposure to R-induced volatiles regulated the expression of 74 genes (46 up-regulated and 28 down-regulated). Twenty-three (nine up-regulated and 14 down-regulated) of these genes responded similarly to the cis-α-bergamotene-deficient VOCs released by as-lox plants (Figure 8c); 49 cis-α-bergamotene-associated genes were differently regulated after exposure to VOCs from as-lox emitter plants. Because a supply of synthetic cis-α-bergamotene was not available, we were unable to conduct the supplementation experiment that would have determined how many of the 49 terpenoid-associated genes were expressed in a cis-α-bergamotene dependent fashion.

GLVs and terpenoids suppressed elicited transcriptional responses

Surprisingly, 166 genes (94 up-regulated and 72 down-regulated) that were not regulated after exposure to wound-induced VOCs from WT plants became regulated when plants were exposed to the GLV-deficient volatiles from wounded as-hpl plants (Table S1: yellow). Supplementing the GLV-deficient volatile bouquet from as-hpl plants with synthetic GLVs restored the suppression of 136 of these genes (Table S1: orange). The functions of most of these genes are unknown, although some are thought to play a role in primary metabolism (21 genes), some in secondary metabolism (19 genes), and some in signaling (16). The last two groups include defense-related genes such as SaPIN2a, LeOPR3, LePAL, α-DOX, WRKY1 and 6 and NtPLA2 (see Table S1).

The lack of cis-α-bergamotene in the R-induced volatile bouquet of as-lox emitters was associated with the regulation of 105 terpenoid-suppressed genes (52 up-regulated and 53 down-regulated), which were not regulated by volatiles released from elicitor-induced WT plants (Table S2: yellow). These results demonstrate that the transcriptome of a plant is responsive to the absence of particular volatiles.


‘Deaf’ and ‘mute’ plants ‘translate’ VOC signals

The majority of evidence for signaling between herbivore-infested plants and their undamaged neighbors comes from experiments in which plants were exposed to unnaturally large amounts of volatiles. Many of the responses analyzed in these studies were detected in experiments using enclosed plants or detached leaves, suggesting that they are unlikely to occur in the absence of these experimental contingencies. If intact plants are used as signal emitters in open-flow chambers, receiver plants are exposed to realistic amounts of volatiles in a realistic fashion. If mutants or transgenic plants deficient in the ability to release particular components of the WT volatile blend are used as emitters, the signaling function of the complex herbivore-induced blend can be dissected into its constituent parts in a realistic manner. Complementation studies, in which the defects of the volatile blend are supplemented by synthetic constituents to determine if the WT response is then restored, definitely demonstrate the function of particular volatiles.

The various biosynthetic pathways contributing constituents to the herbivore-induced volatile bouquet and their regulatory cascades represent genetic targets for manipulation. In this study, we used plants whose herbivore-induced terpenoid and GLV vocabulary had been muted by silencing genes either directly involved in the biosynthesis of particular volatiles (e.g. HPL) or involved in the oxylipin signal cascade (e.g. LOX3) that elicits their release after herbivore attack. In addition to the specific LOX that supplies fatty acid hydroperoxides to the jasmonate cascade [N. attenuata LOX3 (Halitschke and Baldwin, 2003); Arabidopsis LOX2 (Bell et al., 1995); tomato (Lycopersicon esculentum) LOXD (Heitz et al., 1997); potato (Solanum tuberosum) LOX-H3 (Royo et al., 1999)], mutants with defects in biosynthetic enzymes and other yet-to-be characterized regulators of the jasmonate cascade [tomato def-1 (Howe et al., 1996); Arabidopsis (Park et al., 2002)] could be used as emitters, as they are likely defective in some aspect of herbivore-induced VOCs. Key enzymes essential for the biosynthesis of constituents of the VOC bouquet, such as volatile jasmonates (JA carboxyl methyltransferase; Seo et al., 2001), salicylate (salicylic acid carboxyl methyltransferase; Chen et al., 2003), terpenoids (sesquiterpene cyclase; Shen et al., 2000) or indole (indole-3-glycerol phosphate lyase; Frey et al., 2000), represent additional potential target enzymes for genetic modification. The most frequently invoked candidate for a volatile signal that could transmit information among plants is the volatile hormone ethylene. Transgenic lines with reduced activity of 1-aminocyclopropane-1-carboxylic acid (ACC) synthase (Oeller et al., 1991) and ACC oxidase (Hamilton et al., 1990), both of which have decreased ethylene emissions, are available and could be used as emitters to examine the importance of this volatile hormone in between-plant signaling.

Studies of between-plant volatile signaling with ‘mute’ plants could be augmented by the use of ‘deaf’ plants unable to perceive particular volatile signals. The receptors for many of the volatile constituents of herbivore-induced blends remain to be discovered. However, the mutant ETR1 gene, which encodes an ethylene receptor unable to bind ethylene (Chang et al., 1993; Hua et al., 1995), has been transformed into WT plants and renders them unable to perceive ethylene emissions by co-opting the ethylene signal transduction cascade (Chang et al., 1993; Hua et al., 1995; Klee and Clark, 2002). Plants that are ‘deaf’ to the effects of particular volatile constituents could be valuable tools for studying between-plant signaling (Pierik et al., 2003, 2004), but attention must be paid to the volatile emissions of these plants, as ‘deaf’ plants such as the etr1 mutant may prove to be ‘screamers’ and constitutively release in large amounts the signal that they cannot perceive (C. von Dahl and ITB, unpublished results).

Volatile blends as elicitors of responses

A major challenge for researchers in this field is deciding which of the many potential response variables to measure in the receiver plants (Figure 1). The JA cascade is known to mediate herbivore resistance by mediating both VOC releases and the direct defense traits nicotine and TPI activity, which, in turn, strongly determine herbivore resistance in N. attenuata (Halitschke et al., 2003; Steppuhn et al., 2004; Zavala et al., 2004a). As N. attenuata germinates synchronously from long-lived seed banks after fires to form large genetically heterogeneous stands (Bahulikar et al., 2004; Preston and Baldwin, 1999), we reasoned that neighboring conspecifics that ‘eavesdropped’ on the VOCs released from herbivore-attacked plants and activate JA-mediated resistance traits would realize a fitness benefit if the herbivores moved to the ‘listening’ neighbors. We therefore measured JA and the direct defense metabolites; however, neither wound- nor herbivore-induced VOCs were found to have an effect on JA levels or any of the defense traits known to be elicited by JA (Figure 2a). These results demonstrate that in N. attenuata defense mechanisms which are elicited by the JA cascade are not affected by volatile exposure.

Rather than directly eliciting defense responses, VOC exposure may prime the defense signaling pathway, thereby allowing plants to elicit faster or larger defense responses once they are attacked. Such priming of herbivore-elicited VOC responses occurs in maize, and can be detected in the amount of JA that accumulates transiently in plants after elicitation (Engelberth et al., 2004). Experiments with Nicotianasylvestris, in which plants were elicited once, twice or three times with methyl jasmonate (MeJA), with sufficient time between treatments for complete relaxation to occur, demonstrated that plants have a memory of previous induction and more rapidly increase nicotine levels if there has been a previous induction (Baldwin and Schmelz, 1996). However, when we examined JA signaling and the downstream defense metabolites nicotine and TPIs in elicited plants that had been exposed to either wound- or herbivore-induced VOCs, we found no effect of VOC exposure on the induced levels of these compounds. These results are supported by the fact that the accumulation of transcripts of the herbivore-inducible NaLOX3 gene, which is essential for the biosynthesis of JA, was not altered by the exposure to different volatile blends.

Exposure to W + R-elicited VOCs marginally (P = 0.0545) increased the rate of TPI elicitation in comparison with plants exposed to control VOCs (Figure 6), but it is unlikely that these small (22% on day 4) increases translate into biologically significant differences in herbivore resistance. This conclusion is based on extensive previous experimentation which analyzed herbivore performance on N. attenuata plants that had been genetically transformed to manipulate the accumulation of defense compounds and the signal molecules that elicit their accumulation (Baldwin, 1998; Kessler et al., 2004; Steppuhn et al., 2004; Zavala et al., 2004a,b). For example, a biologically significant increase in herbivore resistance is typically associated with 2–8-fold increases in TPI activity (Zavala et al., 2004a,b). In the six experiments conducted in this study, the largest observed increase in TPI activity attributable to VOC exposure was a statistically non-significant increase of 45%. Whether such marginal increases would even be detectable in plants exposed to the multiple stresses typical of natural environments remains to be determined. However, a crucial step in evaluating the ecological relevance of these changes in defense compounds is the analysis of changes in herbivore performance when feeding on receiver plants (Dolch and Tscharntke, 2000; Karban et al., 2000; Pettersson et al., 1999).

The proteome of a plant is certainly not restricted to the resistance traits on which we focused in our analysis, and broad-scale proteomic analysis could perhaps facilitate the identification of other VOC-mediated responses important for the fitness of plants growing in natural habitats. To identify other traits that could be measured as response variables, we focused the analysis on an earlier step in the development of a phenotype: the transcriptome (Figure 1). With a 789-oligonucleotide microarray enriched in N. attenuata genes that are either elicited or suppressed by herbivore attack, we compared the transcriptional profiles of receiver plants after exposure to VOCs. The functional significance of the transcriptional responses found to be differently regulated by VOC exposure remains unknown, as the majority of these genes remain uncharacterized (Tables S1, S2). However, the analysis demonstrates that a plant receives and responds to VOC exposure. Wound-induced VOCs elicited the regulation of 37 genes (GLV-associated), which did not respond to exposure of the GLV-deficient VOCs from as-hpl plants (Figure 8b). The GLV dependence of 11 of these 37 genes was demonstrated when their regulation was restored by adding six synthetic GLVs to the as-hpl volatile bouquet (Figure 8b). As synthetic standards were not available for all compounds that were missing from the GLV-deficient blend (Figure 3), this analysis likely underestimates the number of GLV-dependent genes.

The volatile blend elicited by wounding is dominated by GLVs that are released directly after mechanical damage. The timing and composition of the volatile blend change dramatically when herbivores cause the damage and oral secretions are introduced into wounds during feeding (Table 2). Terpenoids (e.g. cis-α-bergamotene) are specifically released from N. attenuata plants 24 h after they are attacked by M. sexta larvae or are R-elicited (Table 2, Halitschke et al., 2000). Here we demonstrate that plants respond transcriptionally to components of the R-elicited blend (Figure 8c). More than half of the genes that were regulated in response to exposure to WT volatiles were no longer regulated when receiver plants were exposed to VOCs from as-lox emitters (Figure 8c). These results suggest that the expression of a number of genes is elicited by exposure to cis-α-bergamotene, but the results need to be verified by complementing the R-elicited VOCs released from as-lox plants with cis-α-bergamotene to determine if the WT pattern of gene regulation can be restored.

A striking result from this analysis was the large number of genes whose regulation is suppressed by exposure to VOCs. Many (136) genes were found to be regulated in response to GLV-deficient as-hpl volatiles, which were not regulated in response to volatiles from WT plants or to GLV-complemented volatiles from as-hpl plants (Table S1). Similarly, 105 genes were regulated in plants exposed to cis-α-bergamotene-deficient volatiles from R-induced as-lox plants (Table S2). The large number of genes whose regulation is suppressed by the WT volatile bouquet cautions against focusing only on the direct elicitation effects of VOCs. The responses that occur as a consequence of the absence of particular volatiles – the ‘sounds of silence’ (Simon, 1965) – may represent the information content of a particular herbivore-induced volatile bouquet.

The ‘sounds of silence’ phenomenon could also result from compounds that suppress responses in receiver plants. An established example of a volatile that suppresses defense responses helps to illustrate the phenomenon. When Nicotiana plants are wounded, they increase nicotine production and accumulation in direct proportion to the amount of wounding (Ohnmeiss et al., 1997). However, when M. sexta larvae feed, the R that is introduced into wounds elicits a dramatic ethylene release from the plants, which transcriptionally suppresses wound-induced nicotine biosynthesis (Kahl et al., 2000; Winz and Baldwin, 2001). Whether the ethylene released from an attacked plant can suppress nicotine production in a neighboring plant is unknown; however, these results suggest that researchers should pay more attention to the suppression of responses in receiver plants if they are to fully understand this type of signaling.

In summary, our results demonstrate that wound- and elicitor-induced volatiles emitted from N. attenuata plants influence gene expression in neighboring conspecifics. The effects include the elicitation of specific genes as well as the suppression of others. Plant ecologists interested in interpreting this transcriptional response face a major challenge: while the genetic basis of traits important for herbivore resistance is rapidly being worked out (Kessler and Baldwin, 2002), the traits that are important for competitive ability and tolerance of other environmental stresses that determine plant fitness in nature are simply not understood. Volatiles from damaged neighboring plants could, for example, influence resource allocation patterns or the architecture (Stowe et al., 2000) of receivers. Currently, however, we are simply not able to recognize the transcriptional signatures of such responses.


The use of ‘mute’ plants as emitters represents a valuable tool for manipulation of volatile signaling among plants growing in ecologically complex environments; such an approach returns the study of between-plant signaling to the real world. In highly replicated experiments with WT and genetically transformed native tobacco plants, we found no evidence that VOC exposure influences the well-characterized resistance traits of the plant or its defense-eliciting signal cascades. VOC exposure did elicit transcriptional responses, but interpreting these transcriptional ‘answers’ will likely require a more thorough understanding than currently exists of the traits that are important for performance in complex environments.

Materials and methods

Plant growth and collection of M. sexta regurgitant

Seeds of the 11th generation of an inbred line (generated from a seed collection in 1988 in a natural population at the DI ranch in southwestern Utah, USA) of Nicotiana attenuata Torr. Ex Watts (synonymous with Nicotiana torreyana: Solanaceae) were used as the wild type (WT) genotype in all experiments. In addition, we used transgenic N. attenuata plants, which had the N. attenuata hydroperoxide lyase NaHPL gene (as-hpl; line A-337) or the lipoxygenase NaLOX3 gene (as-lox; line A-300) silenced (Halitschke and Baldwin, 2003; Halitschke et al., 2004; Krügel et al., 2002) in the WT genetic background. Seed germination and growth were conducted as described by Krügel et al. (2002). All experiments were performed with rosette-stage plants, and plants with the most uniform growth were randomly assigned to treatment groups.

M. sexta larvae were hatched from eggs (Carolina Biological Supply, Burlington, NC, USA) and reared on fresh leaf material of N. attenuata in a growth chamber (Percival, Perry, IA, USA) under the following conditions: 30°C/16 h light, 26°C/8 h darkness. Regurgitant (R) was collected from third- to fourth-instar caterpillars and diluted 1:1 [volume/volume (v/v)] with deionized water before being added to puncture wounds.

Experimental set-up and volatile exposure

All experiments were performed in polypropylene open-flow growth chambers (18.5 l; Rubbermaid, Wooster, OH, USA) that were covered with a sheet of UV-transparent Plexiglas (Preston et al., 1999). All chambers were open at both ends and a fan (7 cm diameter) located at one of these openings produced a continuous air flow (10 l min−1) through the chambers from the emitter plants and across the receiver plants (Figure 1). During the experimental VOC exposures, all chambers were placed in a glasshouse maintained under the following conditions: 28°C/16 h light, 22°C/8 h darkness, 65% air humidity. Two or four emitter plants (see Table 1) of the different genotypes (WT, as-hpl or as-lox) were placed inside the chamber, upwind from the two or four receiver plants, which in turn were all of the WT genotype (Figure 1). Each exposure treatment was replicated at least three times, with each replicate consisting of an individual chamber (Table 1) to avoid problems of pseudoreplication (see (Fowler and Lawton, 1985).

Feeding of M. sexta larvae elicits the release of GLVs and terpenoid VOCs (e.g. cis-α-bergamotene) from WT N. attenuata plants, whereas mechanical wounding is known to elicit only the release of GLVs from WT plants and not the release of terpenes (Halitschke et al., 2000; Kessler and Baldwin, 2001). In order to manipulate the release of particular compound groups, we used different elicitation treatments in combination with mutants impaired in the production of specific components of the released VOC blend. The following treatments were used to achieve standardized elicitation of VOC release (Table 1). Mechanical damage (‘W’) was created by producing one row of puncture wounds parallel to the midrib on the four youngest fully developed leaves with a fabric pattern wheel (Dritz, Spartanburg, SC, USA). Herbivore-induced VOCs were elicited either by placing two M. sexta larvae (2nd instar) on each receiver plant for 2 days (‘caterpillar’) or by wounding and applying 5 μl of M. sexta R to each row of puncture wounds (‘W + R’). The latter treatment elicits all known transcriptional, direct and indirect responses to M. sexta attack (Halitschke et al., 2001; Roda et al., 2004) while preventing differences in elicitation resulting from different feeding behaviors of M. sexta larvae on the different genotypes (Halitschke and Baldwin, 2003; Halitschke et al., 2004). Untreated WT plants (ctrl) were used as emitters for control exposures in all experiments.

To analyze GLV-mediated responses, we compared the effects of VOCs from as-hpl emitter plants with those from WT emitters (Table 1: experiments I, II, III, V and VI). The release of wound-induced GLVs is substantially impaired in as-hpl mutants (Halitschke et al., 2004; Table 2), whereas the amount of terpenoid VOCs (e.g. cis-α-bergamotene) released after W + R treatment or caterpillar feeding is comparable to that released from WT plants (Table 2). Use of this genotype as an emitter plant in combination with different types of treatments to elicit VOC release (Table 1) allowed us to separate out the effect of GLVs. A second mutant line of N. attenuata (as-lox) was used to analyze the effects of R-induced terpenoid volatiles (Table 1: experiments II and III). The emission of terpenoid VOCs elicited by wounding and application of R is suppressed in as-lox plants while WT amounts of GLVs are emitted by these plants (Halitschke and Baldwin, 2003; Table 2).

In a third treatment, GLVs that were missing from the VOC bouquet of wounded as-hpl plants (Table 1: experiments IG and IIIE) were complemented with synthetic GLVs. For these experiments we used two different mixtures of synthetic GLVs, both containing the following compounds: hexanal (Sigma-Aldrich, Steinheim, Germany), E-2-hexenal (Sigma-Aldrich), hexanol (Sigma-Aldrich), E-2-hexenol (Dragoco, Holzminden, Germany), Z-3-hexenol (Sigma-Aldrich), and Z-3-hexenyl acetate (Dragoco). The two GLV mixtures differed only in the relative amounts of the individual compounds. Previous studies had demonstrated that a wounded N. attenuata leaf of a rosette-stage plant releases approximately 200 ng of Z-3-hexenol during the first minute after wounding (Halitschke and Baldwin, 2003). The release decreases after wounding but remains detectable for 20 min (RH and ITB, unpublished data). Therefore, we estimated that 5–15 μg of Z-3-hexenol is released from four wounded leaves and used this value as the basis for the composition of GLVmix A, which contains all compounds at 10 μg (10 μl)−1 acetone. In order to more accurately mimic the GLV release of WT plants, we used a second GLV mixture (GLVmix B) to analyze NaLOX3 expression (Table 1: experiment III) in which the concentration of each individual compound was adjusted to the relative GLV ratios released from wounded WT plants. The substances in GLVmix B had the following concentrations [μg (10 μl)−1]: hexanal 0.1; E-2-hexenal 0.4; hexanol 0.1; E-2-hexenol 0.7; Z-3-hexenol 10; and Z-3-hexenyl acetate 0.2. Ten microliters of either mixture was applied to a Q-tip located upwind of the wounded as-hpl emitter plants.

The treatments (with the exception of caterpillar feeding) were repeated every 4 h during the light period (starting at 10:00 h) on two consecutive days in order to mimic continuous wounding or herbivore feeding, respectively.

Analysis of VOC emission

To quantify the amount of VOCs that receiver plants were exposed to during the 2-day volatile exposures, we trapped volatiles from WT, as-hpl and as-lox plants (n = 3) in open-flow chambers with the same set-up and conditions as described above, but excluded the receiver plants from the flow boxes in order to analyze only the VOCs released from the emitter plants. WT and as-hpl plants were wounded as described and wound-induced VOCs were trapped continuously for 2 days. Additionally, we trapped volatiles from wounded as-hpl plants whose VOC blends were complemented by the addition of the two synthetic GLV mixtures. Herbivore-induced volatiles were elicited by treating puncture wounds with R. Volatiles were trapped on 30 mg of a polymerous adsorbent (SuperQ; Alltech, Deerfield, IL, USA) and analyzed as previously described (Halitschke et al., 2000). Traps were changed every 10 h at the end of the light or dark period. Empty chambers were trapped to determine the background concentration of volatiles. All peak areas were normalized using peak areas of an internal standard (tetralin). Absolute amounts of E-2-hexenal, Z-3-hexenol and E-2-hexenol were calculated from the regression of peak areas from a dilution series of standards (10, 1, 0.1, and 0.01 ng μl−1). Absolute amounts of released cis-α-bergamotene were calculated from the regression of peak areas from a dilution series of caryophyllene (20, 10, 2, 1, 0.2, and 0.1 ng μl−1). Table 2 presents the absolute amounts of VOCs trapped from the flow boxes over the 2 days when receiver plants were elicited four times in each light period (eight elicitations in total). Data were log-transformed and analyzed by separate analyses of variance (anovas) for each compound. Bonferroni post-hoc tests were used for multiple comparisons between the different emitters.

Treatment of receiver plants

Four hours after the final treatment, all emitter plants were removed from the chambers (Figure 1). Receiver plants were either left undamaged and used for the analysis of constitutive changes resulting from VOC exposure (Table 1: experiments IA, C, E, G; IIA, C, E, G; V; VIA, C, E), or additionally elicited immediately after VOC exposure to determine if exposure primes herbivore-induced defense responses (Table 1: experiments IB, D, F; IIB, D, F, H; III; IV; VIB, D, F). We elicited the latter group of plants with eight rows of puncture wounds on the four youngest fully developed leaves as described above, and 20 μl of R was immediately applied to the puncture wounds. The elicitation of receiver plants after VOC exposure is referred to as ‘potentiation’ (Table 1) to distinguish this treatment from the elicitation of emitter plants. Plant material for the analyses of defense traits was harvested from all receiver plants at the described time points (Figure 1), which were the same for undamaged and potentiated receiver plants.

To determine if exposure to herbivore-induced VOCs altered the elicitation of TPI activity, we performed an additional experiment (Table 1: experiment IV). We exposed four receiver plants to VOCs released from untreated control plants and from WT plants that had been wounded and R-treated for the previous 2 days, and potentiated all four receivers after termination of exposure as described above. Leaf material for the analysis of the TPI activity induction was harvested from one of the four plants on four subsequent days after elicitation.

In order to analyze the accumulation of NaLOX3 transcripts in elicited receiver plants, we performed another experiment (Table 1: experiment III) with the same set-up and emitter-treatment combinations as described above. After 2 days of exposure to the different VOC blends, the emitter plants were removed and all receiver plants were potentiated. Leaf material for the analysis of NaLOX3 expression was harvested from these receiver plants 2 and 6 h after elicitation. Total RNA was extracted from all samples, and transcripts of NaLOX3, a key enzyme of the JA synthesis pathway (Figure 2), were analyzed by quantitative real-time polymerase chain reaction (PCR).

Analysis of direct defense traits and JA accumulation

We analyzed JA accumulation, TPI activity, and nicotine accumulation in both receiver plants that had only been exposed to VOCs, as well as in potentiated receiver plants, after the times shown in Figure 1, in which time 0 is the time of potentiation of receiver plants (Figure 1). Leaf tissue (100–150 mg) for JA analysis was harvested after 35 min from receiver plants and analyzed as previously described (Krügel et al., 2002). After 3 days, we sampled 100–150 mg of leaf material to analyze TPI activity by the radial diffusion activity assay, as previously described (Van Dam et al., 2001). To determine if VOC exposure influenced the rate of TPI elicitation, we harvested additional leaf material from elicited receiver plants on four subsequent days after the elicitation of the receivers (Table 1: experiment IV).

Leaf tissue (100–150 mg) for the analysis of nicotine contents was sampled 5 days after the termination of exposure from undamaged and elicited receiver plants. The accumulation of nicotine was analyzed by high-performance liquid chromatography (HPLC) as previously described (Halitschke and Baldwin, 2003).

If necessary, nicotine, TPI activity and JA data were log-transformed to fulfill the requirements for an anova which was used to analyze all data. The respective manipulated parameters (emitter genotype, emitter treatment and receiver treatment) were used as independent factors.

Analysis of nucleic acids

Leaf tissue (0.5–2.0 g) from one plant of each replicate open-flow chamber was harvested 6 h after VOC exposure for the microarray analysis. Leaf material from the seven replicate receiver plants of each exposure treatment was pooled and ground to a fine powder in liquid nitrogen. Total RNA was extracted using the guanidinium thiocyanate-phenol-chloroform method (Winz and Baldwin, 2001). Isolation of Poly(A)+ RNA, generation of cDNAs, and fluorescent labeling were performed as previously described (Halitschke and Baldwin, 2003). The labeled cDNAs were hybridized to oligonucleotide microarrays (789 50-mer oligonucleotides of N. attenuata genes that respond to herbivore attack, spotted onto epoxy-coated glass slides), with each oligonucleotide being spotted four times (Quantifoil Microtools, Jena, Germany) as previously described (Halitschke and Baldwin, 2003; Heidel and Baldwin, 2004). An Affymetrix 428TM Array Scanner (Affymetrix, Santa Clara, CA, USA) was used to scan the hybridized microarrays with sequential scanning for Cy5- and Cy3-labeled cDNA at a maximum resolution of 10 μm/pixel with a 16-bit depth. The data analysis followed the methods previously described (Halitschke and Baldwin, 2003; Heidel and Baldwin, 2004). A transcript was considered to be differentially regulated when all of the following criteria were met: (i) the average expression ratio (ER) for the four spots exceeded the thresholds for up- or down-regulation (1.5 and 0.67, respectively); (ii) the individual expression ratios were significantly different from 1 as determined by t-test; (iii) the combined signals from both Cy3- and Cy5-labeled cDNA averaged over the four spots were greater than 1000. The five hybridizations that were performed are depicted in Figure 8(a). Genes that were found to be regulated in response to a specific blend of volatiles were classified into three groups. Genes that lost their regulation as a result of the absence of a specific class of VOCs are considered to be associated with the presence of the compounds, while genes whose regulation was restored by complementing the volatile bouquet with synthetic versions of the missing substances are considered to be dependent on these volatiles. The third category of regulated genes contains those that are regulated only when specific volatiles are absent from the normal VOC blend of the plant. These genes are designated as being suppressed by specific VOCs.

Total RNA for the analysis of NaLOX3 expression was extracted as described above and cDNA was synthesized from 150ηg RNA using MultiScribeTM reverse transcriptase (Applied Biosystems, Darmstadt, Germany). Quantitative real-time PCR (ABI PRISMTM 7000; Applied Biosystems) was conducted using the qPCRTM core reagent kit (Eurogentec, Seraing, Belgium), specific TaqMan® primers (MWG-Biotech, Ebersberg, Germany) and a NaLOX3 probe (Halitschke and Baldwin, 2003). The relative gene expression was calculated using the comparative 2ΔΔCt method (Livak and Schmittgen, 2001) with sulfite reductase as endogenous control gene (Bubner and Baldwin, 2004) which under the experimental conditions used here is not regulated (B. Bubner and ITB, unpublished data). The data were log-transformed and statistically analyzed by a two-way anova.


We thank C. Diezel for indefatigable assistance with sample preparations, W. Kröber and T. Hahn for array hybridization and data analysis, E. Wheeler for editorial assistance, two anonymous reviewers for comments on the manuscript, and the Max Planck Society for funding.