Herbivory-induced signalling in plants: perception and action

Authors


I. T. Baldwin. Fax: +49 3641 571102; e-mail: baldwin@ice.mpg.de

ABSTRACT

Plants and herbivores have been interacting for millions of years. Over time, plants have evolved mechanisms to defend against herbivore attacks. Herbivore-challenged plants reconfigure their metabolism to produce compounds that are toxic, repellant or anti-digestive for the herbivores. Some compounds are volatile signals that attract the predators of herbivores. All these responses are tightly regulated by a signalling network triggered by the plant's perception machinery. Several compounds that specifically elicit herbivory-induced responses in plants have been isolated from herbivore oral secretions and oviposition fluids. Elicitor perception is rapidly followed by cell membrane depolarization, calcium influx and mitogen-activated protein kinase (MAPK) activation; plants also elevate the concentrations of reactive oxygen and nitrogen species, and modulate phytohormone levels accordingly. In addition to these reactions in the herbivore-attacked regions of a leaf, defence responses are also mounted in unattacked parts of the attacked leaf and as well in unattacked leaves. In this review, we summarize recent progress in understanding how plants recognize herbivory, the involvement of several important signalling pathways that mediate the responses to herbivore attack and the signals that transduce local into systemic responses.

INTRODUCTION

There are about 792 000 insect species, and 45% of them feed on plants (Strong, Lawton & Southwood 1984). The battle between plants and these phytophagous insects has lasted for hundreds of millions of years. Over time, plants have evolved sophisticated defence systems to counteract attacks from insects. These defences are composed of a battery of compounds that function as direct or indirect defences. Cyanogenic glucosides, glucosinolates, alkaloids, phenolics, proteinase inhibitors (PIs), etc., are direct defences that have toxic, repellant or anti-digestive effects on herbivores (Bennett & Wallsgrove 1994); direct defences also include physical barriers for herbivores, such as cuticles, trichomes and thorns. Attacked plants as well emit volatile compounds or produce extrafloral nectar that attract the predators of insect herbivores to gain an indirect means of protection (Kessler & Baldwin 2001; Kessler & Baldwin 2002; Heil 2008). According to the timing of the deployment of the defences, they can be categorized as constitutive (or ‘static’) and induced (or ‘active’) defences (Gatehouse 2002). Both can be costly to plants, and without temporal and spatial regulations, these defences can compromise plants' growth and reproduction, ultimately reducing their fitness (Baldwin & Preston 1999; Tian et al. 2003; Zavala et al. 2004). Therefore, plants use sophisticated regulatory networks to maintain a balance between growth and defence response when attacked by herbivores.

After being attacked by herbivores, plants quickly generate herbivory-specific signals, and through complicated networks, these signals are further converted to large scale biochemical and physiological changes; in addition to these changes in attacked leaves, certain signals are also conveyed to different parts of the plants where they activate ‘systemic’ defences (Green & Ryan 1972; Li et al. 2002). Cell membrane depolarization, ion flux, mitogen-activated protein kinase (MAPK) activation, phytohormone modulation, production of reactive oxygen species (ROS) and probably nitric oxide (NO) are all associated with herbivore attack and the activation of these responses (Orozco-Cardenas, Narvaez-Vasquez & Ryan 2001; Orozco-Cardenas & Ryan 2002; Maffei, Mithofer & Arimura 2004, 2006; Wu et al. 2007).

Several excellent reviews on plant–herbivore interactions have been published in recent years (Walling 2000; Gatehouse 2002; Kessler & Baldwin 2002; Felton & Tumlinson 2008; Howe & Jander 2008; Mithofer & Boland 2008). Here we complement these reviews by focusing on the perception of herbivory and the herbivory-induced early responses.

PERCEPTION OF HERBIVORE ATTACK

Although both mechanical wounding and herbivory lead to tissue damage and loss, plants respond differently to these two stresses (Baldwin 1990; Alborn et al. 1997; Halitschke et al. 2001; Reymond et al. 2004; Wu et al. 2007). In plant–pathogen interactions, many microbial (or pathogen)-associated molecular patterns (MAMPs or PAMPs) have been found and intensively studied (Bittel & Robatzek 2007; Schwessinger & Zipfel 2008). To distinguish the attack of herbivores from those from other biotic agents, plants are thought to have evolved the ability to perceive herbivory-associated molecular patterns (HAMPs) (Felton & Tumlinson 2008; Mithofer & Boland 2008) but our understanding of these is still in the very early stages. To date, HAMPs can be classified into two categories: (1) chemical elicitors derived from herbivore oral secretions (OS) and oviposition fluids; and (2) those that originate from the specific patterns of wounding.

Several elicitors that are components of herbivore OS have been identified. The effect of applying these elicitors to the specific host plants effectively mimics the effect produced by herbivory. A glucose oxidase (GOX) was isolated from Helicoverpa zea OS and it appears that GOX suppresses plants' defence responses (Eichenseer et al. 1999; Musser et al. 2005). β-Glucosidase from the OS of the white cabbage butterfly larvae, Pieris brassicae, elicits the production of volatiles from cabbage plants (Mattiacci, Dicke & Posthumus 1995). In addition to proteins, certain peptides are also perceived by plants and these activate herbivory-induced responses. After cowpea (Vigna unguiculata) plants are attacked by fall armyworm (Spodoptera frugiperda), they emit ethylene and elevate the levels of jasmonic acid (JA) and salicylic acid (SA), thereby increasing their volatile emissions and the concentrations of defensive compounds. High-performance liquid chromatography (HPLC) separation of OS revealed that a peptide named inceptin is the elicitor in fall armyworm's OS that triggers all these reactions. Intriguingly, inceptins are proteolytic products of plant chloroplastic ATP synthase γ-subunit (cATPC) (Fig. 1a): even at a very low concentration (1 fmol/leaf), inceptin induces JA, ethylene and SA accumulation in cowpea (Schmelz et al. 2006). A further study demonstrated that cATPCs are digested in the caterpillar midgut to produce inceptins, which are perceived by plants and induce defences when they are introduced into plant wounds during feeding (Schmelz et al. 2007). It will be very interesting to determine how cowpea plants perceive inceptins, and whether ion fluxes, MAPKs, etc., are involved in the signal transduction.

Figure 1.

Structures of several herbivore-derived elicitors. (a) Inceptin. (b) Fatty acid–amino acid conjugates. (c) Caeliferin. (d) Bruchin. (e) Benzyl cyanide.

Fatty acid–amino acid conjugates (FACs) are a second group of elicitors that allow plants to monitor the feeding activity of larvae. FACs are composed of two moieties (Fig. 1b): a fatty acid moiety usually consisting of either linolenic acid or linolenic acid and their derivatives (Alborn et al. 1997; Pohnert et al. 1999; Spiteller & Boland 2003; Maffei et al. 2004) and an amino acid moiety, Gln or Glu. The fatty acids and amino acids are plant and insect derived, respectively, whereas the modification of fatty acid and conjugation of fatty acid and amino acid both happen in the caterpillar midgut (Pare, Alborn & Tumlinson 1998). Volicitin, a hydroxyl FAC (N-17-hydroxylinolenoyl-L-glutamine; Fig. 1b), was the first FAC identified in Spodoptera exigua (beet armyworm) OS, which induces volatile release in maize seedlings (Alborn et al. 1997). Since then, different forms of FACs have been found in other insect species (Pohnert et al. 1999; Halitschke et al. 2001; Mori et al. 2001; Spiteller & Boland 2003). The effects of applying FACs have been intensively studied in Nicotiana attenuata. When synthetic FACs are applied to wounded N. attenuata leaves, they activate higher levels of MAPK activity and amplify and modify wound-induced changes in transcriptome, proteome and levels of defensive secondary metabolites (Halitschke et al. 2001, 2003; Giri et al. 2006; Wu et al. 2007). FACs are not only found in caterpillars: a recent study showed that they exist also in the cricket Teleogryllus taiwanemma and the fruit fly Drosophila melanogaster larvae (Yoshinaga et al. 2007), indicating their yet unidentified physiological functions for insects.

The existence of putative receptors of FACs was demonstrated in Zea mays. Truitt, Wei & Pare (2004) showed that [3H]-L-volicitin rapidly, reversibly and saturably binds to the plasma membranes isolated from maize leaves. Furthermore, after treatment with methyl jasmonate (MeJA), the amount of radio-labelled volicitin that binds to cell membranes increases, suggesting a JA-dependent transcription regulation of this putative receptor. FACs are amphiphilic compounds that have detergent-like properties. Maischak et al. (2007) found that FACs have ion channel-forming activities, and this property may account for some of their biological functions in plants (Maffei et al. 2004). Several important receptors involved in the perception of plants' phytohormone and pathogen elicitors have been identified and functionally characterized using forward genetics (reviewed in Tichtinsky et al. 2003; Chow & McCourt 2006). However, given that most of the plant species that respond to FACs are non-model organisms for genetics (Arabidopsis, for example, does not appear respond to FAC elicitation), the possibilities of using forward and reverse genetics to identify the genes that encode receptors for FACs are limited.

A new class of elicitors known as caeliferins was recently isolated from the American bird grasshopper (Schistocerca americana) (Alborn et al. 2007; Fig. 1c). They are saturated and monounsaturated sulfated α-hydroxy fatty acids in which the ω-carbon is functionalized with either a sulfated hydroxyl or a carboxyl conjugated to glycine via an amide bond. Similar to applying FACs, applying caeliferins to maize seedlings elicits the release of volatile terpenes, although the ecological function of caeliferin-induced volatiles remains unknown (Alborn et al. 2007).

Many herbivorous insects lay eggs on plants and some plants respond to oviposition by forming neoplasm and necrotic tissue, producing ovicidal substances and emitting volatile signals that attract parasitoids (Hilker & Meiners 2006). Oviposition fluid has been identified as containing the functional elicitors (Meiners & Hilker 2000; Hilker et al. 2002; Schroder et al. 2007); to date, two of them have been structurally elucidated. Bruchins are long-chain α, ω-diols that are esterified at one or both oxygens with 3-hydroxypropanoic acid (Fig. 1d). They were isolated from the oviposition fluid of the pea weevil (Bruchus pisorum L.), and applying as little as 1 fmol (0.5 pg) results in neoplastic growth on the pods of particular genotypes of pea, which effectively extrudes an egg from the pod surface and thereby functions as defence against oviposition (Doss et al. 2000). Recently benzyl cyanide (BC) isolated from Pieris brassicae oviposition fluid has been identified as the second elicitor in oviposition fluid (Fig. 1e): 1 ng of BC elicits defence responses in Brussels sprouts plants (Brassica oleracea var. gemmifera cv. Cyrus) that arrest the egg parasitoid Trichogramma brassicae (Fatouros et al. 2008).

In addition to chemical elicitors, the wounding resulting from the feeding activity of herbivores in some plant species can trigger herbivory-specific responses in plants. Different herbivores have distinct patterns of feeding involving variations in the way leaf tissue is removed, the frequency and the time period of feeding. It is possible that some plants perceive these herbivore-specific patterns of wounding and use this information to produce herbivory-specific responses. Wounding lima bean leaves with a ‘MecWorm’, a device that mimics the timing but not the mechanical properties of the feeding damage resulting from Spodoptera littoralis larvae and the snail Cepaea hortensis feeding, elicits a volatile release similar to the natural feeding of these herbivores (Mithofer, Wanner & Boland 2005), indicating that some plant species recognize the timing of wounding to tailor their defence responses. The generality of this recognition system needs further study.

Herbivores such as aphids and whiteflies (Hemiptera) suck on plants' phloem tissue with their stylets to obtain nutrition. Although the wounds elicited by this type of feeding are hard to observe with the naked eye, these insects still induce remarkable changes in both signalling and secondary metabolism (Walling 2000; Voelckel, Weisser & Baldwin 2004; Kempema et al. 2007; Zarate, Kempema & Walling 2007). Unlike the insects that remove large quantities of tissue during their feeding, Hemiptera induce many responses in plants that are similar to those induced by pathogen attack (Walling 2000; Smith & Boyko 2007). Genetic analyses indicated that plants' resistance to Hemiptera is mediated by R genes (Kaloshian 2004; Goggin 2007). An Mi-1 gene encoding a nucleotide-binding site–leucine-rich repeat, a typical R gene (NBS-LRR) protein was found to confer resistance to aphids, whiteflies, as well as nematodes (Kaloshian, Lange & Williamson 1995; Rossi et al. 1998; Nombela, Williamson & Muniz 2003). During feeding, aphids secrete numerous enzymes such as oxidases, pectinases and cellulases (Miles 1999); whether these enzymes elicit defence reactions in plants remains to be examined.

Volatile organic compounds released from damaged plants can also elicit defence responses. Clipped Artemesia tridentata plants emit MeJA into the ambient air that can induce defence or sensitizing (priming) responses in neighbouring plants (Farmer & Ryan 1990; Karban et al. 2000; Kessler & Baldwin 2004), perhaps by directly acting as JA after hydrolysis in vivo. Several studies have shown that other volatile compounds, such as terpenes and C6 volatiles induced by herbivory, also activate defence responses in adjacent plants (Arimura et al. 2000, 2002; Kost & Heil 2006; Heil & Ton 2008). How these compounds mediate the transfer of information between plants remains elusive.

HERBIVORY-INDUCED SIGNALLING IN PLANTS

Although little is known about how plants perceive herbivory, the responses elicited by this perception have been better studied. These include ion fluxes, cell membrane depolarization, MAPK activation, ROS and NO production, and modulation of phytohormone levels (Fig. 2).

Figure 2.

A model summarizing early signalling events in herbivore-attacked plants. After herbivore attack, herbivore elicitors (here FAC) bind to putative receptors on plasma membranes and activate further responses. Through an unknown mechanism, Ca2+ influx is initiated, which depolarizes cell membranes. Increased Ca2+ (likely together with a CDPK) greatly enhances NADPH oxidases located in cell membrane and leads to ROS production. MAPKs (at least SIPK and WIPK) are quickly activated; they transcriptionally regulate many genes involved in JA and ethylene biosynthesis, as well as NADPH oxidase and WRKY transcription factors (TFs). SIPK is likely also involved in NO production; both ROS and NO modify amino acids in proteins and induce transcriptional changes of various defence-related genes. A yet to-be-indentified pathway triggers JA biosynthesis. JA is further converted to JA-Ile by JAR; binding of JA-Ile to SCFCOI1 initiates the degradation of JAZ proteins that negatively regulate JA-responsive genes. Without phosphorylation, ACS is degraded through 26S proteasome pathway; after being phosphorylated by SIPK, it gains higher stability and enhances ethylene biosynthesis. Red arrows represent directly phosphorylation; blue arrows represent transcriptional regulation. AOC, allene oxide cyclase; AOS, allene oxide synthase; CDPK, calcium-dependent protein kinase; JAZ, jasmonate ZIM-domain; LOX, lipoxygenase; OPDA, 12-oxo-phytodienoic acid; OPR3, OPDA reductase 3; NO, nitric oxide; NOA, NO-associated protein; NR, nitrate reductase; ROS, reactive oxygen species; SCF, Skp, Cullin, F-box; SIPK, salicylic acid-induced protein kinase; WIPK, wound-induced protein kinase.

Cell membrane depolarization and calcium influxes

One of the earliest cellular responses to herbivory is membrane depolarization. Maffei et al. (2004) measured the membrane potentials of cells adjacent to the bite zones produced by Spodoptera littoralis larvae on lima bean leaves. Within minutes, a large membrane depolarization was observed; moreover, cells at different distances to the bite zone had distinct depolarization profiles. FAC species and their concentrations have variable effects on membrane potentials. The change of membrane potentials is a result of ion fluxes across the cell membranes. Ca2+ is a second messenger known to be involved in various signalling transduction pathways that mediate the transduction of environmental and developmental signals into physiological changes (Lecourieux, Ranjeva & Pugin 2006). With results consistent with the membrane potential data, Maffei et al. (2004) used a fluorescence dye to demonstrate that cells in close proximity (20–200 µm) to the bite zone of an herbivore undergo a rapid Ca2+ influx. The intensity of the fluorescence signal generated by herbivore feeding is higher than that induced by wounding, and calcium channel blocker verapamil considerably suppresses the signal. The authors also argue that the detergent-like FACs may directly, albeit incompletely, contribute to the Ca2+ influx; this was further confirmed by the ion channel forming capability of FACs in planar lipid bilayer membranes (Maischak et al. 2007). The interaction of Ca2+ and other early signalling components are considered in the next sections.

MAPK signalling

The MAPK signalling cascade is a conserved pathway involved in modulating various cellular responses in eukaryotes (Herskowitz 1995; Chang & Karin 2001; MAPK Group 2002). Many studies have demonstrated that MAPKs play important roles in mediating plants' responses to diverse stress stimuli (Ligterink & Hirt 2001; Tena et al. 2001; Zhang & Klessig 2001; Jonak et al. 2002), especially their responses to pathogens (Pedley & Martin 2005). MAPKs transcriptionally regulate WRKY transcription factors and may directly phosphorylate transcription factors that results in changes of their activity (Kim & Zhang 2004; Menke et al. 2005). Using a reverse genetic approach, Wu et al. (2007) showed that in N. attenuata, two MAPKs, salicylic acid-induced protein kinase (SIPK) and wound-induced protein kinase (WIPK) play critical roles in transducing OS elicitation into downstream molecular reactions (Fig. 2). As soon as 5 min after OS elicitation, plants dramatically increase their SIPK activity; this activity peaks at about 10 min and slowly wanes to basal level after 3 h. Compared with wounding alone, applying OS to wounds amplifies the kinase response and prolongs their activation. This study also revealed that the OS elicited increases in phytohormones (JA, ethylene and SA), as well as the transcript levels of many defence-related genes, are controlled by the MAPK pathway. Similar results were reported in a study of tomato plants overexpressing prosystemin, in which silencing the orthologues of SIPK and WIPK greatly compromised the plants' PI-based defence, as seen in the attenuated JA levels, which allowed the Manduca sexta larvae to gain more mass (Kandoth et al. 2007).

ROS

The involvement of ROS [or active oxygen species; these include including the superoxide anion (O2–), hydrogen peroxide (H2O2) and the hydroxyl radical (OH)] in plants' defence against pathogens has been well recognized (reviewed in Lamb & Dixon 1997). Increased levels of H2O2 were found in wounded tomato and sweet potato plants (Orozco-Cardenas & Ryan 1999; Jih, Chen & Jeng 2003). Similarly, herbivory also elicits ROS production. Feeding of H. zea larvae on soybean leaves results in the accumulation of H2O2 (Bi & Felton 1995). In Medicago truncatula, no detectable ROS is produced after wounding, but is after herbivory (Leitner, Boland & Mithofer 2005). ROS was also observed in herbivore-attacked lima bean leaves using a ROS-sensitive fluorescence dye (Maffei et al. 2006).

NADPH oxidases are responsible for the production of pathogen-elicited ROS (Simon-Plas, Elmayan & Blein 2002; Torres, Dangl & Jones 2002; Yoshioka et al. 2003; Torres & Dangl 2005). Some evidence suggests that NADPH oxidases are also the main source of wounding- and herbivory-induced ROS in plants (Orozco-Cardenas & Ryan 1999; Orozco-Cardenas et al. 2001; Sagi et al. 2004). In N. attenuata, silencing Rboh D, a NADPH oxidase, dramatically reduces OS- or FAC-induced H2O2 levels (Wu & Baldwin, unpublished data). The regulation of NADPH oxidase activity is likely associated with the ion fluxes associated with the rapid changes in membrane potential. All plant NADPH oxidases located in the membrane carry two Ca2+ binding EF hands, which suggests that their activity is modulated by Ca2+ (Keller et al. 1998). Indeed, without other cytosolic components, plant NADPH oxidases are directly activated in the presence of Ca2+ (Sagi & Fluhr 2001). In guard cells, abscisic acid (ABA) signalling involves both Ca2+ and ROS. Arabidopsis mutants lacking functional AtrbohD and AtrbohF (two NADPH oxidases) do not respond to ABA with stomata closure, ROS production, cytosolic Ca2+ elevation and activation of plasma membrane Ca2+-permeable channels; supplying H2O2 rescues both Ca2+ channel activation and stomatal closure in these mutants, indicating Ca2+ influx occurs downstream of ROS in ABA signalling (Pei et al. 2000; Kwak et al. 2003). Similarly, both genetic and pharmacological evidence revealed the complex interaction between ROS production and Ca2+ influxes in plant development and in their response to stresses: Ca2+ appears to be both upstream and downstream of ROS, reflecting a spatio-temporal Ca2+ and ROS regulation network (Levine et al. 1996; Piedras et al. 1998; Grant et al. 2000; Foreman et al. 2003). Protein kinases are involved in the regulation of ROS. In potato, two calcium-dependent protein kinases (CDPKs), StCDPK4 and StCDPK5, phosphorylate NADPH oxidase at the N terminal, which enhances its activity, resulting in turn in ROS bursts (Kobayashi et al. 2007). More recently, the transcript levels of NADPH oxidase have been shown to be dependent on SIPK and a close homologue NTF4 in N. benthamiana (Asai, Ohta & Yoshioka 2008). Little is known about how herbivory activates NADPH oxidase. However, the large similarity of pathogen- and herbivory-induced early signalling events suggests the involvement of protein kinases and Ca2+ influxes.

The function of ROS in plants' development and defence against pathogens has been intensively studied (reviewed in Lamb & Dixon 1997; Apel & Hirt 2004; Torres & Dangl 2005). In tomato, H2O2 induced by wounding is JA-dependent, and inhibiting NADPH oxidases blocks the elevation of several herbivore-resistant genes after wounding, systemin, oligosaccharides and MeJA treatments (Orozco-Cardenas & Ryan 1999). Another study in tomato also highlighted the importance of ROS in wound-induced PI expression (Sagi et al. 2004). Further studies on the regulation of herbivory-induced NADPH oxidase activity, herbivore growth assays on ROS-deficient plants and analysis of defence-related traits in these plants are needed to better understand how ROS are involved in plant–herbivore interactions.

NO

NO, one of the ‘active nitrogen species (NO and peroxynitrite)’, is another small molecule involved in plant development, stomatal closure and stress responses (Bolwell 1999; Besson-Bard, Pugin & Wendehenne 2008; Hong et al. 2008; Neill et al. 2008; Wilson, Neill & Hancock 2008). In Arabidopsis, two enzymes involved in NO production have been isolated: NO-associated protein 1 (AtNOA1) (Guo, Okamoto & Crawford 2003; Zemojtel et al. 2006) and nitrate reductase (NR) (Desikan et al. 2002). AtNOA1 was first thought to be a NO synthase (NOS); biochemical data ruled out that it is a bona fide NOS, although it somehow regulates NO levels in plants (Crawford et al. 2006; Zemojtel et al. 2006). NR, which is usually associated with a plants' nitrogen assimilation, also produces NO from nitrite (Dean & Harper 1986; Rockel et al. 2002). In guard cells, ROS is required for NO production (Bright et al. 2006). Evidence from tobacco cell cultures treated with cryptogein (a pathogen elicitor) combined with inhibitors of protein kinase, NO synthase, and calcium flux and NO donors indicated that NO production is dependent on phosphorylation events and Ca2+ influx; NO production also contributes to the release of Ca2+ from intracellular stores, and H2O2 does not trigger NO production (Lamotte et al. 2004). Consistently, SIPK is involved in the regulation of NO production after pathogen elicitor INF1 treatment (Asai et al. 2008).

The function of NO in plants' resistance to pathogens has been well documented (reviewed in Hong et al. 2008). Little is known about the wounding-induced NO in plants. Using a NO donor and a NO scavenger, Orozco-Cardenas & Ryan (2002) showed that NO negatively regulated PI transcript levels after wounding, systemin, oligosaccharides and JA treatments. Huang et al. (2004) demonstrated an increase of NO concentration in Arabidopsis epidermal cells after wounding; moreover, applying gaseous NO leads to SA accumulation and elevated transcript levels of AOS, LOX2 and OPR3, although no JA accumulation was detected; this was likely caused by the suppressive effects of elevated SA. Wounding-induced NO has also been found in marine macroalga Dasycladus vermicularis (Ross, Kupper & Jacobs 2006). There is still no report about the herbivory-induced NO in plants.

The reactive nature of NO free radicals allows for the rapid modification of proteins. Given this, it is unlikely that NO is perceived by a particular receptor but rather directly reacts with proteins to modify activity and thus functions indirectly as a signalling molecule. One of the most common protein modifications is S-nitrosylation: NO directly reacts with sulfhydryl groups of cysteines on proteins to form S-nitrosothiols, which in turn changes the activity or function of the protein (Wang et al. 2006). A critical regulator of salicylic acid-induced plants' defence against pathogens, NPR1, is also S-nitrosylated, which leads to conformational changes of NPR1 (Tada et al. 2008). In addition, outward-rectifying K+ channels, metacaspase 9 (AtMC9), a MYB2 transcription factor (AtMYB2) and ribulose-1,5-biphosphate carboxylase/oxygenase (RuBPCase) have also been shown to be modified by S-nitrosylation (Sokolovski & Blatt 2004; Belenghi et al. 2007; Serpa et al. 2007; Abat, Mattoo & Deswal 2008). Another modification is nitrosylation. NO reacts with O2•− to form peroxynitrite (ONOO-, NO + O2•− → ONOO-), which rapidly modifies many amino acid residues and results in nitrosylation of proteins. Pathogen-induced hypersensitive response in Arabidopsis plants is associated with a higher degree of protein nitrosylation (Romero-Puertas et al. 2008); this increased nitrosylation was also observed in INF1-elicited tobacco suspension cells (Saito et al. 2006). NO-induced protein modifications very likely play important roles in plants' resistance to herbivores, but much more research is required to put this assertion on strong experimental footings.

Phytohormones (ethylene, jasmonic acid and salicylic acid)

JA and ethylene have been long recognized as pivotal hormones that regulate a myriad of plant developmental and environmental responses (Reymond & Farmer 1998; Chen, Etheridge & Schaller 2005; Broekaert et al. 2006; Wasternack 2007). The biosynthesis and perception of ethylene have been intensively studied (Bleecker & Kende 2000; Wang, Li & Ecker 2002; Chow & McCourt 2006). Two steps are central to ethylene biosynthesis: conversion of S-adenosyl-L-Met to 1-aminocyclopropane-1-carboxylic acid (ACC) by ACC synthases (ACSs) and oxidation of ACC to form ethylene by ACC oxidases (ACOs). The first step is usually the rate-limiting step. Liu & Zhang (2004) elegantly showed that in Arabidopsis, MPK6 (orthologue of tobacco SIPK) directly phosphorylates ACS2 and 6, which are quickly degraded by 26S proteasome pathway without phosphorylation (Joo et al. 2008), greatly enhances their stability. This phosphorylation event leads to dramatically increased ethylene release. Furthermore, the authors found that activating MPK6 accounts for only half of pathogen elicitor-induced ethylene emission (Liu & Zhang 2004). Similarly, in N. attenuata, silencing SIPK abolished half of the ethylene produced by wounding and OS elicitation (Wu et al. 2007). It will be interesting to identify the regulatory element that accounts for the production of the other half of ethylene production after pathogen or herbivore elicitation, which is probably an unidentified calcium-dependent protein kinase (Tatsuki & Mori 2001). Besides being post-translationally regulated, both ACS and ACO genes are regulated transcriptionally after herbivory elicitation. In N. attenuata, at least, two ACOs and one ACS transcripts are amplified by OS- and FAC-elicitation (von Dahl et al. 2007), and these augmented transcript levels are kinase-dependent (Wu et al. 2007).

Ethylene is perceived in Arabidopsis through five receptors: ETR1; ETR2; ERS1; ERS2; and EIN4 (Chen et al. 2005). CTR1, the next downstream component, encodes a putative Raf-like MAPKKK and interacts with the ETR1 receptor, but its biochemical activity and molecular actions are unclear (Huang et al. 2003). Loss-of-function receptor and CTR1 mutants show constitutive ethylene responses, indicating that both are negative regulators of ethylene signalling (Kieber et al. 1993; Hua & Meyerowitz 1998). EIN2 and then EIN3 are positive regulators further downstream of CTR1 (Chao et al. 1997; Alonso et al. 1999). EIN3 and other EIN3-like proteins are transcription factors that stimulate the transcription of other transcription factors, such as ERF1 (Solano et al. 1998); these transcription factors positively or negatively mediate the transcription of various ethylene-responsive genes (Ohme-Takagi & Shinshi 1995).

In tomato, both applying an ethylene inhibitor and using plants impaired in ethylene biosynthesis demonstrated that ethylene is necessary for the elicitation of PI mRNA (O'Donnell et al. 1996). Silencing ACOs and blocking ethylene perception in N. attenuata by ectopically expressing a mutant ETR1 gene revealed the regulatory function of ethylene in wound-induced nicotine biosynthesis (von Dahl et al. 2007).

Many studies have shown that in plant–herbivore interactions, JA is the most important hormone that regulates a plant's defence levels. Its significance has been demonstrated in plants with impaired JA biosynthesis or perception whose defences against herbivores are remarkably decreased (Howe et al. 1996; McConn et al. 1997; Kessler & Baldwin 2002; Halitschke & Baldwin 2003; Paschold, Halitschke & Baldwin 2007), and in microarray analyses indicating its central role in regulating wounding- and herbivory-induced transcriptome changes (Reymond et al. 2000, 2004). Almost all the enzymes involved in JA biosynthesis have been identified in Arabidopsis and other species (reviewed in Delker et al. 2006; Wasternack 2007). α-Linolenic acid derived from chloroplast membrane is converted to 12-oxo-phytodienoic acid (OPDA) through a pathway composed of several chloroplastidial enzymes: lipoxygenase (LOX); allene oxide synthase (AOS); and allene oxide cyclase (AOC); the OPDA is further transported to peroxisomes, where it is transformed to JA by OPDA reductase 3 (OPR3) and then three steps of β-oxidation (Wasternack 2007). After wounding or herbivore attack, JA levels increase very quickly, usually faster than the upregulation of the transcript levels of JA biosynthetic genes; overexpressing AOC or AOS constitutively does not lead to higher levels of JA unless plants are wounded (Howe et al. 2000; Laudert, Schaller & Weiler 2000; Stenzel et al. 2003). All these lines of evidence suggest a rate-limiting step exists in early steps of the JA biosynthesis pathway. A recent study in Arabidopsis shows that two lipases, DGL and DAD1, are necessary and sufficient to produce JA after wounding: overexpressing DGL greatly enhances basal JA levels, whereas silencing DGL in the dad1 mutant decreases both basal and wound-induced JA to undetectable levels (Hyun, Choi & Hwang 2008). It is likely that the release of α-linolenic acid from chloroplast membranes is the rate-limiting step in the biosynthesis of JA. The rate of JA production indicates that this step is probably controlled post-translationally. Whether DGL is also involved in herbivore-induced JA accumulation and with what mechanism plants activate DGL are interesting questions that deserve more attention. In addition, MAPKs play pivotal roles in regulating the biosynthesis of JA. Silencing SIPK and WIPK in Nicotiana and tomato largely abolishes wound- and herbivory-induced JA; transcript analyses indicated that these kinases regulate the transcript levels of many genes involved in JA biosynthesis (Kandoth et al. 2007; Seo et al. 2007; Wu et al. 2007). Whether MAPKs, especially SIPK and WIPK, regulate the transcript levels of DGL and whether herbivore elicitation directly activates DGL through a unknown pathway, such as calcium binding, phosphorylation, or nitrosylation, merit further research.

In addition to JA, precursors of JA and JA derivatives (JA–amino acid conjugates) are effective compounds in eliciting anti-herbivore defences. Arabidopsis opr3 mutant is devoid of JA but the resistance of this mutant to Bradysia impatiens larvae is not compromised, and gene expression analyses using microarrays showed that wounding opr3 plants induces transcript changes in several genes similar to those genes' changes in WT plants (Stintzi et al. 2001). This study indicates that OPDA is not only an intermediate product in JA biosynthesis, but also a signalling compound involved in wounding- and herbivore-induced responses. A milestone in understanding how JA acts was the finding that JAR1 (related to adenylate-forming enzymes of the firefly luciferase family) catalyses the conjugation reaction between JA and Ile to form JA-Ile (Staswick & Tiryaki 2004). JAR-silenced plants or jar mutants lack most JA-induced responses; furthermore, with the exception of JA-Ile, applying other JA–amino acid conjugates does not restore these responses, indicating the critical importance of JA-Ile in both wounding- and herbivore-induced defences (Staswick & Tiryaki 2004; Kang et al. 2006; Wang et al. 2008). However, applying JA-Ile to N. attenuata plants impaired in JA biosynthesis (antisense LOX3 plants) does not fully restore these plants' resistance traits to herbivores; consistently, microarray analyses also indicate that JA and JA-Ile play largely overlapping but somewhat distinct roles in herbivore resistance (Wang et al. 2008).

coi1 Mutants were identified to be insensitive to coronatine, a compound produced in Pseudomonas syringae that has a similar structure to JA-Ile (Feys et al. 1994; Staswick & Tiryaki 2004; Katsir et al. 2008). coi1 mutants and plants having silenced COI1 show remarkably reduced responses to JA, wounding and herbivory (Reymond et al. 2000; Devoto et al. 2005; Paschold et al. 2007). COI1 encodes an F-box protein; together with Skp1 and cullin they assemble SCF ubiquitin ligase complex, SCFCOI1 (Xie et al. 1998; Xu et al. 2002). This suggests that COI1 negatively regulates JA-induced responses by degrading JA-response repressors through the ubiquitin/26S proteasome pathway. Although this hypothesis was proposed a decade ago, our understanding of JA perception did not emerge until recent work that identifies jasmonate ZIM-domain (JAZ) proteins as the repressors of JA-responsive genes (Chini, Fonseca & Fernandez 2007; Thines et al. 2007). After wounding or herbivory, the induced JA is quickly converted to JA-Ile by JARs; although direct measurement of the binding between JA-Ile and COI1 is still absent, a yeast two-hybrid and in vitro pull-down assays indicated that JA-Ile specifically promotes the interaction between COI1 and JAZ proteins. This interaction leads to the degradation of JAZ proteins through the ubiquitin/26S proteasome pathway (Chini et al. 2007; Thines et al. 2007). JAZ proteins directly bind to MYC2, a transcript factor that positively regulates the levels of many JA-responsive genes (Chini et al. 2007), although it is still unclear how JAZ proteins suppress the action of MYC2 and likely other JA-responsive transcription factors. Intriguingly, the perception of JA strikingly resembles that of IAA (Woodward & Bartel 2005; Badescu & Napier 2006) and gibberellins (Hirano, Ueguchi-Tanaka & Matsuoka 2008; Jiang & Fu 2007), where the repressors are degraded through ubiquitin/26 proteasome pathway after binding of IAA and gibberellins to F-box proteins TIR1 and GID1, respectively.

Two key enzymes are involved in the SA biosynthesis in Arabidopsis, phenylalanine ammonia lyase (PAL) and isochorismate synthase (ICS), using phenylalanine and chorismate as the substrates (Verberne, Muljono & Verpoorte 1999; Wildermuth et al. 2001).The function of SA in plants' defence responses to herbivory remains unclear. M. sexta only induces minor accumulation of SA in N. attenuata (Diesel & Baldwin, unpublished data), suggesting that SA may play a minor role in plants' resistance to chewing insects. In comparison, phloem-feeding insects, such as aphids and silverleaf whiteflies, induce SA-dependent responses. Transcriptome analyses indicated that feeding of these insects elicit SA-regulated transcripts; JA- and ethylene-regulated genes are induced transiently or at moderate levels (Walling 2000; Moran & Thompson 2001; Moran et al. 2002; de Vos, Kim & Jander 2007). Although the activation of SA signalling is a common event in plants attacked by phloem-feeding insects, its role in plants' defence varies among species. SA is important for tomato's resistance to potato aphids (Li et al. 2006); however, in Arabidopsis SA has been shown to have neutral and negative effect on aphid and silverleaf whitefly growth, respectively (Pegadaraju et al. 2005; Zarate et al. 2007).

Herbivory elicits various early signalling pathways, including membrane depolarization, ion fluxes, kinase activation, production of ROS and active nitrogen species, and phytohormone changes (summarized in Fig. 2). There is a great similarity between pathogen- and herbivore-induced signalling events that gives rise to the question, how does the activation of similar signalling pathways lead to different metabolism reconfigurations against different stresses? We assume that although these two biotic stresses activate many common signalling events, the magnitude of individual signalling component's action, the timing and the location of these actions may determine the output of transcriptional changes and in turn the specific reconfiguration of plants' defence metabolism.

SYSTEMIC SIGNALLING

After wounding or herbivore attack, plants accumulate defensive compounds in both damaged leaves and intact distal leaves (Green & Ryan 1972). Thus, a signal transmits ‘wounded’ or ‘herbivore attacked’ message to those leaves. The nature of this mobile signal has been long debated. Both electric and hydraulic signals were proposed to be this mobile signal (Malone, Alarcon & Palumbo 1994; Stankovic & Davies 1997); involvement of the vascular transport system was also supported experimentally (Jones et al. 1993; Orians, Pomerleau & Ricco 2000; Schittko & Baldwin 2003). Systemin was believed to the mobile signal that travels to systemic leaves in tomato (McGurl et al. 1992). However, an elegant analysis indicated that systemin is involved only in propagating local wound-induced JA signal, but not in a long-distance signal transmission; this study also revealed that JA itself or a JA-elicited signal moves to systemic leaves and induces PI activity (Li et al. 2002). In line with this, by using N. attenuata irjar4/6 plants, which have highly impaired JA and Ile conjugation activity, Wang et al. (2008) demonstrated that JAR activity is necessary for both local and systemic induction of PI. Furthermore, the authors proposed that JA-Ile is synthesized in local and systemic leaves after herbivory, but is unlikely the mobile signal; probably this signal is JA or another JA metabolite.

In addition to the defence reactions in distal intact leaves, how undamaged parts of an attacked leaf respond to herbivory remains to be fully understood. After wounding the tip of a tomato leaf, higher levels of PI mRNA were found at the base of the leaf but not at the wounded tip region (Howe et al. 1996). Using in-gel kinase assay and mRNA blotting as imaging techniques, it was shown that after applying OS to a certain wounded region in an N. attenuata leaf, a mobile short-distance signal moves rapidly (within 10 min) to certain regions of the leaf and activates both kinase signalling and JA biosynthesis (Wu et al. 2007). Treating leaves at base or tip leads to different quantitative distribution of kinase and JA levels; mRNA blotting showed that the induction of many defence-related genes in different regions of a leaf also correlates with specific spatial patterns of kinase activation. This study revealed that after herbivory, FACs bind to a certain receptor and quickly induce the short-distance signal that conveys herbivory-elicited reaction to other parts of the attacked leaf; the long-distance signal that travels to systemic leaves appears to be downstream of this short-distance signal. Identification of this mobile short-distance signal will not only greatly enrich our knowledge about kinase signalling pathway, it will shed light on how plants respond to herbivory on a spatial–temporal scale, which is relevant for an actively feeding herbivore.

CONCLUSION

The recent years have seen tremendous progress in the research on the early signalling events in herbivory-elicited plants. Many signalling pathways that are involved in the plant–pathogen interaction have also been identified as playing roles in aspects of plant–herbivore interactions. However, much remains unknown. How plants perceive herbivore attack, how exactly the kinase, JA, ethylene, ROS and NO pathways are regulated, what are the interactions among these pathways, and how different parts of plants communicate to deploy defence are all interesting and challenging questions. Emerging tools in molecular biology and genetics, as well as in imaging and analytical chemistry provide the opportunity to unravel the complicated signalling networks in plants that have evolved during the long history of plant–herbivore arms race.

ACKNOWLEDGMENTS

We thank the Max-Planck Society for supporting the research in our lab, Dr. Jinsong Wu for sharing unpublished data and Emily Wheeler for editorial assistance. Because of space limitation, we could not fully cite all the publications in this field, thus we apologize to all the researchers whose relevant work has not been mentioned here.

Ancillary