Indirect Defense Responses to Herbivory in Grasses

Insect resistance in grasses is the result of many defense mechanisms that act in parallel to limit the damage of herbivore attacks. Many of these defense mechanisms are based on plant secondary metabolites or defensive proteins that directly affect the herbivore due to their toxic or deterring properties. Less than two decades ago, a new type of defense mechanism, termed indirect defense, was first described in maize (Zea mays; Turlings et al., 1990). Central to this type of defense is the release of a volatile plant signal that attracts natural enemies of the herbivore. Indirect defense has been reported in more than 15 different plant species—both monocots and dicots—after attack by arthropod herbivores (Dicke, 1999; Kessler and Baldwin, 2002; Turlings and Wäckers, 2004). For several of these plants, the benefits of indirect defense result in the reduction of subsequent herbivory and an increase in reproductive fitness (Bernasconi et al., 1998; De Moraes et al., 1998; Thaler, 1999; van Loon et al., 2000; Kessler and Baldwin, 2001). In recent years, our knowledge of volatile-mediated defenses in plants has been rapidly expanding. In grass crops like maize and rice, indirect defenses have been the subject of much attention because they might offer new strategies for crop protection against herbivores. In an effort to review some of the most interesting developments in this field, this article will focus on volatile-mediated defense mechanisms in grasses.

HERBIVORE-INDUCED VOLATILES ARE CENTRAL TO PLANT INDIRECT DEFENSES

All grasses assayed to date respond to herbivore damage with the emission of a volatile blend consisting mostly of terpenes and products of the lipoxygenase pathway (Gouinguene et al., 2001; Degen et al., 2004; Köllner et al., 2004; Cheng et al., 2007; Yuan et al., 2008). The amount and composition of the volatile signal emitted by the plant is dependent on the type of herbivore cue. In maize, mechanical wounding of the leaves or infestation with aphids only induces a moderate response, whereas damage by larvae of the lepidopteran Spodoptera littoralis results in a far greater release (Turlings et al., 1998; Schnee et al., 2006). In maize, this increase in volatile production is due to larval elicitors like volicitin [N-(17-hydroxylinolenoyl)-l-Gln] that are formed in the larval gut and introduced into the leaf during larval feeding (Alborn et al., 1997; Spiteller et al., 2000; Tumlinson and Lait, 2005). Volicitin appears to be relatively specific for the induction of maize and other grasses because it does not induce volatiles in the dicot lima bean (Phaseolus lunatus; Koch et al., 1999). An important intermediate of the signal transduction pathways from herbivore damage to volatile production is jasmonic acid in both maize and rice (Schmelz et al., 2003; Xu et al., 2003; Cheng et al., 2007). The biosynthesis of jasmonate derivates in response biotic stress has been best studied in dicot plants where jasmonic acid, methyl jasmonate, amino acid conjugates of jasmonic acid, and pathway intermediates like cis(+)-12-oxophytodienoic acid are formed and subsequently activate different parts of a complex signaling network (Devoto and Turner, 2005; Wasternack, 2007). Some of the enzymes involved in this pathway, including several lipoxygenases, have also been characterized in maize (Nemchenko et al., 2006; Gao et al., 2008), but our knowledge is still limited compared to dicots.

In Nicotiana attenuata, an Ile conjugate of jasmonic acid is crucial for the induction of nicotine production after herbivore damage to the plant. This conjugate interacts with the factor COI, an F-box protein essential for plant signaling processes (Paschold et al., 2007). In Arabidopsis (Arabidopsis thaliana), a complex of jasmonic acid conjugate and COI protein targets a repressor of the JAZ family for degradation by the 26S proteasome and thereby activates genes of plant defense (Thines et al., 2007). Further transcription factors involved may be those of the WRKY family, which have already been implicated in the jasmonate-dependent expression of a terpene synthase in cotton (Gossypium hirsutum) responsible for the production volatile (+)-δ-cadinene (Xu et al., 2004). A recent review by Howe and Jander (2008) provides a comprehensive overview of the early signaling events after herbivory and the induction of the biosynthetic pathways involved in the production of plant volatiles.

In grasses, especially rice and maize, the biosynthesis of volatile terpenes has been studied in detail. Responsible for the high number of volatile mono- and sesquiterpenes in both species is the enzyme class of terpene synthases. In rice, three herbivore-induced terpene synthases are sufficient to produce the majority of the terpene volatiles (Yuan et al., 2008). The terpene blends of maize are formed by at least six sesquiterpene synthases (Köllner et al., 2004). Three of these enzymes, TPS1, TPS10, and TPS23, are strongly induced by herbivore damage and produce the major sesquiterpene components of herbivore-induced volatiles (Schnee et al., 2002, 2006; Köllner et al., 2008a). These terpene synthases can also be induced by jasmonic acid treatment of the plant (T.G. Köllner and J. Degenhardt, unpublished data).

Beyond their role in plant defense, the volatiles of maize plants were also shown to elicit responses in neighboring plants. This phenomenon, generally known as priming, involves increased transcription of defense-related genes and allows the plant to respond faster and more vigorously to herbivore attack (Baldwin et al., 2006). In maize, these interactions were elicited by products of the lipoxygenase pathway, the green leaf volatiles (Engelberth et al., 2004; Farag et al., 2005; Ruther and Kleier, 2005). Prolonged exposure of maize plants to these volatiles primes many herbivore-induced genes such that they will respond faster to subsequent herbivore damage (Ton et al., 2006). After herbivore damage, primed plants are better defended against herbivory and release volatiles that are more attractive to parasitic wasps. Priming of crop plants might therefore offer novel strategies for protection (Turlings and Ton, 2006).

INDIRECT DEFENSES ABOVE GROUND

One of the first indirect defenses was identified in maize fed upon by larvae of the lepidopteran Egyptian cotton leaf worm, S. littoralis. The volatiles released in response to S. littoralis attack attract the parasitic braconid wasp Cotesia marginiventris (Hymenoptera), which oviposits into the larvae (Fig. 1; Turlings et al., 1990). The larva of C. marginiventris will develop inside its lepidopteran host until it kills the host upon emergence. Because parasitized larvae consume less plant material and are not able to reproduce, parasitism can benefit the fitness of the plant. In maize, an increase in fitness based on production of a higher number of seeds has been observed (Hoballah and Turlings, 1999; Hoballah et al., 2004). The volatile blend emitted by herbivore-damaged maize plants consists of indole, products of the lipoxygenase pathway, and a large number of mono- and sesquiterpenes (Fig. 2; Turlings et al., 1990; Köllner et al., 2004). Volatile indole is formed by the maize indole-3-glycerol phosphate lyase, an enzyme activated by herbivory and the herbivore elicitor volicitin (Frey et al., 2000). A function of indole in the attraction of parasitic wasps, however, could not be clearly demonstrated (D'Alessandro et al., 2006).

Figure 1.

Scheme of terpene-mediated interactions of a maize seedling above and below ground. Damage of maize leaves by lepidopteran herbivores activates the terpene synthases TPS10 and TPS23, which produce a blend of volatile terpenes. This blend attracts several species of parasitic wasps. Damage of the roots by D. v. virgifera activates the terpene synthase TPS23. The volatile terpene produced by TPS23, (E)-β-caryophyllene, attracts entomopathogenic nematodes. Drawings by Tobias G. Köllner.

Figure 2.

Volatile emission of maize in response to herbivory by lepidopteran larvae. A, Volatiles from control leaves and leaves damaged by S. littoralis were collected and separated by gas chromatography. The major terpene compounds were identified as β-myrcene (1), hexenyl acetate (2), (E)-β-ocimene (3), linalool (4), 4,8-dimethylnona-1,3,7-triene (5), phenylmethyl ester (6), phenylethyl ester (7), indole (8), geranyl acetate (9), (E)-β-caryophyllene (10), (E)-α-bergamotene (11), (E)-β-farnesene (12), germacrene D (13), β-sesquiphellandrene (14), (E)-nerolidol (15), and 4,8,12-trimethyltrideca-1,3,7,11-tetraene (16). Depicted are traces of the total ion current detector IS, internal standard nonyl acetate. B, The sesquiterpenes (E)-β-caryophyllene, (E)-α-bergamotene, and (E)-β-farnesene are volatile plant defense compounds.

Attempts to identify the terpene compounds crucial for the attraction of parasitic wasps have been hampered by the complexity of the blends and the difficulty of obtaining individual terpenes with the correct chirality for bioassays (Turlings et al., 1991; D'Alessandro et al., 2006). Fortunately, identification of genes involved in the biosynthesis of these volatiles has provided molecular tools to demonstrate which of the compounds are attractive to the parasitic wasp. The key enzymes of terpene biosynthesis are the terpene synthases, which form the basic carbon skeletons of these compounds (Gershenzon and Kreis, 1999). A unique feature of these enzymes is their ability to form mixtures of many terpenes from only one substrate. Biochemical characterization of the terpene synthase gene family in maize resulted in the identification of the herbivore-induced terpene synthase TPS10, which produces the major sesquiterpene volatiles of maize (Schnee et al., 2006). TPS10 forms (E)-β-farnesene, (E)-α-bergamotene, and other herbivory-induced sesquiterpene hydrocarbons from the substrate farnesyl diphosphate (Figs. 1 and 2B). Overexpression of TPS10 in Arabidopsis resulted in plants emitting high quantities of TPS10 sesquiterpene products identical to those released by maize. Using these transgenic Arabidopsis plants as odor sources in olfactometric assays showed that females of the parasitoid C. marginiventris learn to exploit the TPS10 sesquiterpenes to locate their lepidopteran hosts after prior exposure to these volatiles in association with the host (Schnee et al., 2006). This behavior was based on associative learning of C. marginiventris, which can utilize both learning and innate, preformed responses to locate its host (D'Alessandro and Turlings, 2005; Hoballah and Turlings, 2005). Overexpression of the herbivore-induced (E)-β-caryophyllene synthase TPS23 in Arabidopsis showed that the volatile (E)-β-caryophyllene product can also be associated with host presence by C. marginiventris (Köllner et al., 2008).

This gene-based dissection of the herbivore-induced volatile blend demonstrates that a single gene such as tps10 or tps23 can be sufficient to mediate the indirect defense of maize against herbivore attack. Furthermore, associative learning can also adapt parasitoids to alterations of the herbivore-induced volatile blend due to genotype, plant age, and abiotic conditions (Takabayashi et al., 1994; DeMoraes et al., 1998; Schmelz et al., 2003; van den Boom et al., 2004). However, females of C. marginiventris are also attracted to the full blend of maize volatiles without prior association, indicating that the blend contains additional attractive compounds like the so-called green leaf volatiles (Z)-3 hexenal, (E)-2 hexenal, (Z)-3 hexenol, (Z)-2-hexenyl acetate, and (Z)-3-hexenyl acetate that might elicit a positive chemotactic response innate to C. marginiventris (D'Alessandro and Turlings, 2005; Hoballah and Turlings, 2005). Interestingly, the emission of herbivore-induced volatiles is not always beneficial for the maize plants, but can also attract additional herbivores like larvae of Spodoptera frugiperda (Carroll et al., 2006).

In rice, the induction of volatiles was observed after herbivory by the lepidopteran Spodoptera litura and the brown plant hopper, Nilaparvata lugens (Hemiptera). The volatiles induced by S. litura are repellent to N. lugens, although this hemipteran was not deterred by volatiles induced by its own herbivory (Xu et al., 2002). The volatiles induced by N. lugens, however, are attractive to Cyrtorhinus lividipennis, a predator of N. lugens eggs (Lou and Cheng, 2003). Also attracted is a mymarid egg parasitoid, Anagrus nilaparvatae, which is able to extract information about both the time and severity of the infestation from these volatiles (Lou et al., 2005b). This information might increase the chances of survival for the offspring of the wasps. Volatiles released after treatment of rice with jasmonic acid attracted the egg parasitoid A. nilaparvatae, indicating an indirect defense mechanism that reduces the egg load on the rice plant (Lou et al., 2005a). A major component of rice volatiles is (E)-β-caryophyllene, a sesquiterpene olefin that is the product of the sesquiterpene synthase OsTPS3 isolated from rice (Cheng et al., 2007). Overexpression of OsTPS3 in rice resulted in transgenic plants that emit high quantities of (E)-β-caryophyllene, but only in response to treatment with methyl jasmonate. Behavioral assays with transgenic and wild-type plants both treated with methyl jasmonate showed that the parasitoid A. nilaparvatae is more attracted to the transgenic plants emitting higher levels of (E)-β-caryophyllene, indicating a possible role of OsTPS3 in the indirect defense of rice (Cheng et al., 2007).

INDIRECT DEFENSES BELOW GROUND

Despite being covered by soil, roots are also subject to attack by herbivores. Although very little is known about indirect defense mechanisms below ground, it was often assumed that entomopathogenic nematodes are attracted to damaged roots via chemical cues (Boff et al., 2001; Van Tol et al., 2001; Bertin et al., 2003). Recently, such a below-ground defense against arthropods was elucidated in detail. The defense is targeted against larvae of the beetle Diabrotica virgifera virgifera (western corn rootworm), an important root pest of maize. In response to feeding by D. v. virgifera larvae, maize roots release a signal that strongly attracts the entomopathogenic nematode Heterorhabditis megidis (Fig. 1; Rasmann et al. 2005). The attractive signal was identified as (E)-β-caryophyllene (Fig. 2B). Most North American maize lines do not release (E)-β-caryophyllene from their roots, whereas many European lines and the closest wild relatives of maize, teosinte, do so in response to D. v. virgifera attack. Field experiments showed a 5-fold higher nematode infection rate of D. v. virgifera larvae on a maize variety that produces the signal than on a variety that does not (Rasmann et al., 2005). The (E)-β-caryophyllene signal is produced by the (E)-β-caryophyllene synthase TPS23, which is independently regulated in leaves and roots in response to damage by different herbivores (Köllner et al., 2008). Above and below ground, the signal is involved in the defense against herbivores with completely different sites and modes of attack. The ability of TPS23 to produce (E)-β-caryophyllene is widely distributed among the wild relatives of maize and was shown to be under positive selection pressure. However, the loss of (E)-β-caryophyllene production in most North American maize varieties is not due to inactive alleles of the tps23 gene itself, but caused by an alteration of the signal transduction network that abolishes herbivore-induced gene transcription (Köllner et al., 2008).

DO HERBIVORE-DAMAGED GRASSES SPEAK A COMMON VOLATILE LANGUAGE?

The volatile blends released by grasses in response to herbivory vary greatly in quantity and composition. In a sample of 32 maize lines, release rates from 0.7 to 54.2 μg h⁻¹ g⁻¹ leaf dry weight were observed and suggest that some maize varieties are much more capable of attracting herbivore enemies than others (Degen et al., 2004). The volatiles released by these lines consisted of the same set of 20 major compounds, but the relative proportions of these compounds varied greatly. A principal component analysis of these volatiles detected very little correlation between blends of closely related genotypes. Only the sesquiterpene (E)-β-caryophyllene was found to appear in European flint lines, but not in the Minnesota 13 complex and Early Dent lines (Degen et al., 2004). The herbivore-induced volatiles from five teosinte species (Gouinguene et al., 2001), rice (Cheng et al., 2007), and several other species of the Poaceae (T.G. Köllner and J. Degenhardt, unpublished data) consist of the same major compounds and exhibit the same degree of variation found in the maize lines.

The composition and biological activity of volatiles also changes over time after herbivore attack (Hoballah and Turlings, 2005). At the onset of herbivory, the blend is dominated by the green leaf volatiles, (E)-2-hexenal and (Z)-3-hexen-1-ol, while mono- and sesquiterpenes appear about 2 h later. In grasses, the species of attacking herbivore does not seem to affect the composition of the volatiles very much. Only slight differences between the volatile blends were reported in maize attacked by the lepidopteran larvae S. littoralis and Ostrinia nubilalis (Turlings et al., 1998), and rice infested with Pseudaletia separata and Helicoverpa armigera (Yan and Wang, 2006) possess no distinct differences in their volatile profiles. For some herbivore enemies, however, minor differences between the volatile blends that are undetectable by gas chromatography can strongly affect host-seeking behavior. For the parasitic wasp Cotesia kariyai, for example, the host-finding behavior is affected even by the developmental stage of the herbivore that induces the maize volatiles (Takabayashi et al., 1995).

The high variation between volatile blends argues against a single volatile signal that is common to grasses. Especially in maize, the variation in volatiles between different genotypes appears too large to be perceived as a specific signal. However, C. marginiventris, like most other parasitic wasps, can associate a successful oviposition experience with the volatiles encountered at that time (Turlings et al., 1990). Therefore, the parasitoid locates its hosts by both innate responses and associative learning of volatiles (Hoballah and Turlings, 2005). This allows the parasitoid to adapt and optimize its host-finding strategy toward the most rewarding plant signals. In field experiments, parasitic wasps even show cross-recognition between different grass species. Intercropping maize with the molasses grass, Melinis minutiflora, significantly increased larval parasitism of stem borers by Cotesia sesamiae and decreased levels of infestation by stem borers in the crop (Khan et al., 1997). This interaction is thought to occur since volatile components released by intact M. minutiflora are similar to those produced by herbivore-damaged maize plants (Khan et al., 1997).

Below ground, the sesquiterpene (E)-β-caryophyllene was identified as a single compound attracting entomopathogenic nematodes to maize roots by D. v. virgifera (Rasmann et al., 2005). Induction of (E)-β-caryophyllene after herbivore damage was also not only observed in several genotypes of maize, but also observed in six teosinte species. This indicates that this defense signal occurs in many grasses related to maize (Köllner et al., 2008). Entomopathogenic nematodes are not only attracted to (E)-β-caryophyllene, but recognize other plant volatiles as well (Rasmann and Turlings, 2008). However, (E)-β-caryophyllene diffuses more efficiently through soil than most other sesquiterpenes, making it a rather specific defense signal against soil herbivores (Hiltpold and Turlings, 2008). A field experiment with maize varieties producing different amounts of (E)-β-caryophyllene was able to demonstrate the specificity and efficiency of this defense against D. v. virgifera in an agricultural setting (Rasmann et al., 2005).

VOLATILES MIGHT PROVIDE NEW STRATEGIES FOR CROP PROTECTION

Biological control methods are often proposed as alternatives to synthetic insecticides in an effort to reduce the environmental impact of modern agriculture. In this context, natural enemies of herbivores show great promise to limit crop damage in an environmentally safe manner if they can be summoned in sufficient abundance during outbreaks. The complexity of the interaction between crop, herbivore, and its enemies requires very careful implementation of indirect defense strategies in agroecosystems (Degenhardt et al., 2003). The most important requirements are plants that send out a sufficiently strong and attractive volatile signal after herbivore attack. The large variation found among maize volatiles suggests that not all genotypes are capable of emitting these signals (Degen et al., 2004). Most northern American genotypes have even lost the ability to produce (E)-β-caryophyllene and thereby lost the defense mechanism against D. v. virgifera that is associated with it (Köllner et al., 2008). The identification of tps23 provides a molecular tool to restore (E)-β-caryophyllene production in nonproducing maize lines and should allow for the development of alternate strategies for D. v. virgifera control in an agricultural setting. The second requirement for the implementation of indirect defense strategies is that a suitable herbivore enemy is present in the region where the crop is grown. This enemy species must be able to control herbivore populations sufficiently to significantly decrease plant damage. While some herbivore enemies were shown to strongly reduce herbivory to plants in field experiments (Thaler, 1999; Kessler and Baldwin, 2001), an interaction between the large cabbage white butterfly and the parasitic wasp Cotesia glomerata resulted in a higher consumption of leaf material in the parasitized caterpillars (Coleman et al., 1999). Third, the effect of an indirect defense will be reduced when additional herbivores are attracted by the plant volatiles. In rice, the volatiles emitted after treatment with jasmonic acid also attracted females of the herbivore N. lugens indicate such an unfavorable interaction (Lou et al., 2005a). Last, the indirect defenses need to be compatible with other defenses of the plant. If the herbivore is subjected to toxins or feeding deterrents formed by direct defenses of the plant, it might not constitute a good food source or host for herbivore enemies. The combination of several synergistic defense strategies, however, should provide plants with high resistance against herbivores.

Integrated pest management strategies to manipulate the abundance and distribution of natural enemies were already demonstrated to be successful (Khan et al., 1997, 2006, 2008). In these push-pull strategies, herbivore enemies are attracted to crops by intercropping with plants that release high levels of volatiles and thus minimize pest problems in an environmentally safe manner.

Alternatively, the manipulation of volatile emission in crop grasses may be a valuable strategy to improve attraction of herbivore enemies. This strategy might be aided by engineering of plants that emit strong, readily detectable volatiles that match the preferences of a particular enemy species (Degenhardt et al., 2003; Turlings and Ton, 2006). The development of such plants is now feasible due to the elucidation of the pathways responsible for the biosynthesis of volatile compounds. The effectiveness of these tritrophic interactions needs to be in synergy with the direct defenses of the plant to extend the time that herbivores remain vulnerable to attack from foraging enemies. The interplay between indirect defense and Bt-toxin-mediated herbivore resistance in transgenic maize plants has already been addressed by two studies (Turlings et al., 2005; Dean and De Moraes, 2006). Although no significant changes in the composition of the herbivore-induced volatiles were apparent in these studies, the overall volatile release of the transgenic plants was strongly reduced due the altered feeding pattern of the herbivore. Further studies of the interactions between grasses, their herbivores, and the enemies of their herbivores are necessary to facilitate the application of indirect defenses in the cultivation of crop grasses.

CONCLUSION

In recent years, a growing number of volatile-mediated defense responses have been discovered in grasses. The complexity of the volatile blend and the large numbers of different herbivore enemies suggests that many more of these indirect defenses remain to be characterized. As we get a better understanding of the enzymes of volatile production and the pathways that regulate their activity, we acquire the molecular tools to engineer plants with a specifically altered volatile emission. These plants are valuable to identify the functions of particular volatiles and to develop novel defense strategies for crop plants. Application of these defense strategies in an agricultural setting might offer new, environmentally friendly approaches to increase insect resistance in grass crops.