Why Does Herbivore Attack Reconfigure Primary Metabolism?

A plant's resistance to herbivore attack is thought to be principally determined by its secondary metabolism, which can be remarkably plastic and responsive to different grades and types of herbivory. Newer unbiased “omic” approaches, which characterize transcriptomic, metabolomic, and proteomic changes in herbivore-attacked plants, have laid to rest the notion that metabolism can be neatly parsed into “secondary metabolism,” which functions to meet environmental challenges, and “primary metabolism,” which supports growth. The hundreds of genes regulated during the plant-herbivore or -pathogen interaction have been analyzed with microarray studies, and almost all aspects of metabolism are represented, with a substantial fraction coming from primary metabolism (Hui et al., 2003; Both et al., 2005; Major and Constabel, 2006; Mozoruk et al., 2006; Ralph et al., 2006; Schmidt and Baldwin, 2006; Tian et al., 2006; Kant and Baldwin, 2007) Here, we consider four overlapping functional explanations for this reconfiguration:

1. Resistance traits are costly and frequently up-regulated after attack, requiring reductions in growth, reproduction or storage, and/or increases in assimilation to meet their metabolic demands (Fig. 1). These changes in resource allocation can be either acute, driven by immediate reductions in resources, or anticipatory, occurring before resource supply limits defense activation (Smith and Stitt, 2007).

2. Rather than supporting defense responses, reconfiguration could support the physiological adjustments plants must make to tolerate herbivory and reduce the negative fitness consequences of herbivore attack (Fig. 1).

3. Primary metabolites could function as signals in defense pathways (Fig. 2).

4. Induced changes in primary metabolism could themselves be defensive (Fig. 2).

Figure 1.

Dependency of resistance traits (defenses and tolerance) and primary metabolism. Primary metabolism is fueled by energy and resources, which the plant gains from its environment. Primary metabolism involves growth, storage, and reproduction. Tolerance depends on primary metabolites and energy, both of which are taken from pools for reproduction, storage, and/or growth, and later reinvested in reproduction. Defenses from secondary metabolism are based on energy and resources from primary metabolism, which can be partially resupplied to primary metabolism. Parts of primary metabolism can function as direct defense.

Figure 2.

Resistance signaling is elicited differently from simple wounding when herbivore-specific elicitors (FACs) are introduced into wounds during caterpillar feeding. Signaling depends on primary metabolites. Herbivory induces a large reorganization of primary metabolism, including altered photosynthesis and altered sink/source relations. These changes are coordinated by a signaling network that is only partially understood. The expression of defense and tolerance traits requires changes both in primary and secondary metabolism.

We consider these four hypotheses in an overview of the literature that addresses how assimilation and the partitioning of assimilates are altered by herbivory and how primary metabolites function as signals and as defenses. In conclusion, we consider the challenges that plant biologists face in attempting to falsify these hypotheses. Compared to the falsifications of hypotheses about the defensive function of secondary metabolism, tests of the above hypotheses will seriously challenge the procedures that we use to understand resistance mechanisms and perhaps even challenge the reductionist paradigm that has proved so useful for understanding gene function in much of biology in the last century.

HERBIVORE-INDUCED CHANGES IN ASSIMILATION AND PARTITIONING OF ASSIMILATES

If the resource demands of defense production compete with those of growth and reproduction, and defenses are costly to produce (Steppuhn, 2007), then changes in the partitioning of resources among growth, storage, and reproduction would be expected (Fig. 1). Changes in how resources are partitioned according to function can be avoided if the rate of resource assimilation increases. Similar predictions hold for the activation of physiological changes that allow plants to better tolerate the negative fitness consequences of herbivory. Tolerance, which measures a plant's ability to compensate for the negative fitness effects of tissue damage, is usually described as a reaction norm of the fitness of specific genotypes at various damage levels (Strauss and Agrawal, 1999; Stowe et al., 2000). Tolerance is thought to result from the activation of dormant meristems, changes in plant architecture, resource allocation, or photosynthetic capacity. In short, the herbivory-induced activation of both defense and tolerance responses is predicted to alter resource assimilation and source-sink relationships, and the literature provides general support for these predictions.

The photosynthetic apparatus frequently responds to herbivore attack, usually with decreases in CO₂ assimilation in the attacked leaf that are proportionally greater than the leaf area that is actually damaged (Zangerl et al., 2002). Otherwise, the photosynthetic response depends on the type of attacker and the age of the tissues that are measured (Welter, 1989). Defoliating herbivores can increase the photosynthetic activity of unattacked leaves, whereas stem borers and mesophyll feeders tend to decrease activity. On the other hand, transcript levels of photosynthetically related genes are commonly down-regulated (Hui et al., 2003; Ralph et al., 2006; Tang et al., 2006). It is thought that down-regulation of the photosynthetic apparatus protects it from oxidative damage (Niyogi, 2000), but decreased photosynthetic activity may also free up resources, especially nitrogen-rich compounds, making these available for use in secondary defense pathways. Decreased photosynthetic rates may be part of a global inhibition of protein synthesis, which may anticipate the need to redirect resources to defensive functions. As decreases in photosynthetic rates are more common than increases, altered photosynthesis has only rarely been correlated with tolerance (e.g. Cullen et al., 2006). Increases in photosynthetic rates could also be caused by changes in source-sink relationships resulting from the increased demand for energy and carbon (C)-based resources that the production of defensive compounds entails; separating where and for what additional C and energy are used is difficult. The activation of dormant meristems and thus new sinks has been shown to be central in tolerance in some species (Bergelson et al., 1996; Mabry and Wayne, 1997). For example, in Nicotiana attenuata, increased branching compensates for leaf damage (Schwachtje et al., 2006), and jasmonic acid (JA) signaling, which is responsible for activating several defense responses in this species, appears to suppress regrowth and contribute to apical dominance (Zavala and Baldwin, 2006).

How herbivore attack alters source-sink relationships remains unclear other than by reducing source strength when herbivores consume and damage leaves. As well as serving a variety of developmental functions, invertases are involved in the regulation of sink strength by cleaving Suc into Glc and Fru, thereby altering the osmotic gradient of Suc and leading to altered carbohydrate partitioning by turning specific tissues into metabolic sinks for carbohydrates (Roitsch and Gonzalez, 2004). Hence, invertases are often regulated after insect attack. For example, increased sink strength is elicited by JA treatment and gypsy moth feeding via the increased activity of cell wall invertases in the sink leaves of hybrid Populus deltoides × Populus nigra (Arnold and Schultz, 2002) and the wounding of leaves in Solanum lycopersicum, Solanum peruvianum, and Pisum sativum, which increases the activity of soluble (vacuolar) and cell wall invertase in damaged leaves (Zhang et al., 1996; Ohyama et al., 1998). Root wounding was shown to induce vacuolar and cell wall invertase in Beta vulgaris (Rosenkranz et al., 2001). Changes in assimilate flux after herbivore attack may occur along the transport routes, where sugar transporters are involved in Suc loading and unloading. Wounding is known to elicit a Suc transporter, AtSUC3, in sieve elements of different sink tissues (Meyer et al., 2004) and a monosaccharide transporter, STP4, in Arabidopsis (Arabidopsis thaliana; Truernit et al., 1996).

Tolerance to herbivore attack can be acquired by changing resource allocation when stored reserves are used (for example, those of root tissues). This strategy favors biennial or perennial species that normally accumulate reserves during their growing season for later growth during short-day periods (Wyka, 1999; Wise and Cummins, 2006). With nutrients stored in safe tissues, e.g. roots, plants have the possibility to regrow later in the growing season when the pressure from aboveground herbivores may have decreased. If plants are attacked by root herbivores, assimilates can be remobilized above ground. The highly tolerant Centaurea maculosa responds to root herbivory by the knapweed moth by reducing its nitrogen (N) uptake but also shifting N to aboveground tissues (Newingham et al., 2007), suggesting that N allocation can be a determinant of tolerance. This idea is supported by the finding that after its leaves were clipped, the dwarf shrub Indigofera spinosa increased its root N uptake (Coughenour et al., 1990) and that Quercus serrata accumulates higher N levels in leaves (Takashima et al., 2004). Moreover, N allocation to roots has been observed after methyl-JA treatment of Medicago sativa (Meuriot et al., 2004).

Carbon is allocated to roots in response to leaf damage or herbivory in several species, for example, after grasshopper damage to Zea mays (Holland et al., 1996) and Panicum coloratum (Dyer et al., 1991), after the defoliation of Lolium perenne (Bazot et al., 2005) and of two C₄ perennial grasses (Briske et al., 1996), and after methyl-JA treatment of Populus tremuloides (Babst et al., 2005). Recently, a SnRK kinase has been found to regulate the reallocation of photoassimilates in response to herbivory, facilitating a tolerance response (Fig. 2; Schwachtje et al., 2006). SnRK kinases are involved in regulating isoprenoid, amino acid, and especially carbohydrate metabolism (Halford and Paul, 2003). The β-subunit of the kinase complex is rapidly down-regulated in the source leaves of N. attenuata after simulated attack by the tobacco hornworm, leading to 10% more photoassimilate being partitioned to roots. The same effect was seen in JA-deficient asLOX plants, which are silenced for the JA-biosynthetic enzyme lipoxygenase, making this response demonstrably independent of JA signaling. This rapid bunkering of C into root tissues is elicited when wounds are treated with fatty acid-amino acid conjugates (FACs), which are the insect-specific elicitors that activate most defense responses via the jasmonate cascade (Halitschke et al., 2003). At the end of its growing season, N. attenuata gains a measure of tolerance by reusing its additional root resources to prolong flowering, leading to increased capsule production late in the season. Recently, wild-type Arabidopsis plants overexpressing JA were observed to have reduced fitness but the same tolerance of defoliation, which is consistent with the idea that there is a JA-independent mechanism of tolerance (Cipollini, 2007).

The increased flux of C to the roots in response to herbivory would be expected to increase the rate of root growth, but in young seedlings of N. attenuata, for example, sometimes just the opposite occurs (Hummel et al., 2007). Unlike the FAC-elicited C flux, this rapid inhibition of root growth requires an intact JA-signaling cascade (G.M. Hummel, U. Schurr, I.T. Baldwin, and A. Walter, unpublished data) and may be one of the plant's anticipatory responses. Changes in growth that are anticipated in advance of resource limitations (and therefore differ from acute responses) have acquired growing importance as physiologists have shifted their focus to understanding the relationships between C balance and growth (Smith and Stitt, 2007).

By studying the growth dynamics of plants unable to synthesize starch due to a mutation in plastidial phosphoglucomutase in combination with experimental conditions in which the dark cycle was extended, researchers have discovered that plants anticipate the length of the dark period and adjust their synthesis and catabolism of starch to exactly meet energy demands during the dark period (Gibon et al., 2004; Smith and Stitt, 2007). For reasons that are not completely clear, starch accumulation at the end of the dark period is inversely correlated with growth rate (Cross et al., 2006). It will be interesting to see how these anticipatory changes in allocation and resource partitioning that are likely coordinated by a plant's circadian clock are modified when plants are elicited by insect-specific elicitors.

PRIMARY METABOLITES AS SIGNALS AND DEFENSES

A plant's resistance response to insect feeding is coordinated by different signaling pathways that depend on primary metabolites; in addition, the integration of the different signals induced by wounding and insect-specific elicitors results in a complex rearrangement of primary and secondary metabolism (Fig. 2). JA is a crucial player in defense signaling (Devoto and Turner, 2005) and requires kinases, such as WIPK and SIPK (Wu et al., 2007), and transcription factors such as WRKYs (Hui et al., 2003). After elicitation, Ile production is amplified by Thr deaminase (TD). Two hours after elicitation, the mRNA levels of N. attenuata's TD are increased by as much as 30 times (Kang et al., 2006). The Ile that is produced at the attack site is rapidly conjugated to JA, forming JA-Ile, a key activator of defense signaling (Chini et al., 2007; Thines et al., 2007). In addition to the JA-dependent signaling, several JA-independent responses to wounding and FACs have been documented (Leon et al., 1998; Rojo et al., 1999; LeBrasseur et al., 2002; Gross et al., 2004; Schwachtje et al., 2006), but knowledge about the underlying mechanisms is limited. Recently, the signaling role of sugars has received increased attention because several sugar-induced resistance genes have been found. For example, Suc, Glc, and Fru act as specific regulatory signals on the wound-inducible expression of an extensin gene (SbHRGP3) in Glycine max (Ahn et al., 1996; Ahn and Lee, 2003); additionally, a putatively defensive vegetative storage protein is Suc as well as JA induced (Berger et al., 1995). Moreover, transcripts of a hexokinase, which can function as a sugar sensor or photosynthesis repressor (Rolland et al., 2006), are induced by wounding and are sensitive to trehalose-6-P (Claeyssen and Rivoal, 2007), which itself is involved in the feedback regulation of photosynthesis and developmental transitions (Paul, 2007; Ramon and Rolland, 2007). Trehalose and SnRK protein kinases have been shown to interact (Schluepmann et al., 2004), as have sugars and lectins, which also can be induced by JA, suggesting lectins play a role in signal transduction (Chen et al., 2002; Van Damme et al., 2003; Gabius et al., 2004; Lannoo et al., 2006). Furthermore, an antagonistic interaction between Glc and ethylene, which is involved in defense signaling (von Dahl and Baldwin, 2007), has been reported (Zhou et al., 1998).

Several metabolites that play well-studied roles in primary metabolism have been found to possess defensive functions. Their dual function has been discovered because very high levels of them accumulate in plants, or because their induction patterns after herbivore attack are similar to those of defensive secondary metabolites. In the case of TD, for example, the function of the enzyme, degrading Thr, led to the hypothesis that it functioned in the insect's gut to degrade this essential amino acid. TD's regulatory domain was found to be removed by insect proteases, suppressing its negative feedback regulation by Ile (Chen et al., 2007). TD then continuously degrades Thr in the gut lumen, leading to amino acid starvation. Two TD isoforms are known in S. lycopersicum, one of which is stable in insect guts (Chen et al., 2007); in N. attenuata, in contrast, one TD serves both primary and secondary functions (Kang et al., 2006).

High levels of calcium oxalate (CaOx), a primary metabolite, accumulate in plants (up to 80% of dry mass), and in some plants CaOx synthesis is induced by herbivory (Molano-Flores, 2001; Ruiz et al., 2002). CaOx regulates bulk levels of the Ca that is involved in cell signaling and in several biochemical processes. The morphologically diverse CaOx crystals are either stored in the vacuoles of specialized cells, the crystal idioblasts, or are associated with the cell wall (Franceschi and Nakata, 2005). Crystals can be located around tissues, e.g. vascular bundles, to provide a physical barrier against chewing insects by an abrasive effect that blunts insects' mandibles (Korth et al., 2006). Moreover, CaOx is thought to act as an antinutritive defense by decreasing the efficiency with which ingested food is converted (Korth et al., 2006).

Vegetative storage proteins and lectins play dual roles in primary metabolism and resistance, and some of them are JA induced. For a detailed description of defensive proteins, see Zhu-Salzman et al. (2008). Carbohydrates can also directly function as defenses. Galactose markedly reduced larval growth of western spruce budworm when added to artificial diet in quantities of 6%, but Glc and Fru increased growth (Zou and Cates, 1994). In Acacia, mutualistic ants are attracted by nectar sugars (Heil et al., 2005). However, only after Suc is cleaved by an invertase, which is present in extrafloral nectar, are nonsymbiotic ants repelled and mutualistic ants attracted.

TESTING HYPOTHESES ABOUT DEFENSIVE FUNCTION

Whether a given secondary metabolite plays a role in herbivore protection is best determined by planting isogenic plants that both do and do not produce the metabolite into the plants' native environment where they can be confronted with native herbivore communities. With the development of transformation systems and the identification of genes that control the biosynthesis and flux into secondary metabolism, it is now possible to create these isogenic plants and to test their function. For example, nicotine-, TPI-, and JA-signaling-deficient N. attenuata plants have provided strong proof for the defensive function of individual secondary metabolites, for defensive synergies among different secondary metabolites, and for the role of JA signaling in activating metabolic changes (Kessler et al., 2004; Steppuhn et al., 2004; Zavala et al., 2004a; Steppuhn and Baldwin, 2007).

The defensive metabolites of a plant have long been thought to act synergistically; the combination of different effects is assumed to be more than their parts. A defensive synergism between nicotine and trypsin protease inhibitors (TPIs) was discovered in N. attenuata when the production of nicotine or TPI or both was silenced, and when plants were attacked by the second most common lepidopteran herbivore in this tobacco's natural habitat, Spodoptera exigua. The compensatory feeding response of this herbivore to TPI-induced amino acid starvation was inhibited by the larvae's limited ability to tolerate nicotine and the leaf area consumed was reduced when both secondary metabolites were present (Steppuhn and Baldwin, 2007). Similarly, artificial diet experiments showed that plant-derived phytic acid, a primary metabolite, reduces the detoxification of xanthotoxin, a defensive furanocoumarin, by the parsnip webworm, likely by inhibiting insect CYP450 monooxygenases (Green et al., 2001). These results demonstrate that in planta tests are essential for pinpointing defensive function because the chemical milieu in which a metabolite is expressed can profoundly influence how an herbivore responds.

These secondary-metabolite-deficient plants also provided strong support for the hypothesis that secondary metabolites, previously thought to be directed solely at agents outside the plant, play a physiological role inside the plant. Silencing a TPI gene not only increased plants' susceptibility to herbivores but also increased growth and seed production. This increase in plant fitness, apparent not only when TPI production was silenced but also when TPI production was restored in an ecotype naturally deficient in TPI production (Zavala et al., 2004b), is not likely explained by the liberation of resources that are no longer invested in TPI production. While the mechanisms remain to be worked out, it is likely that TPIs down-regulate growth through their modulation of a yet-to-be described signaling cascade.

Molecular biology made it possible to uncover the defensive functions of secondary metabolites because it was possible to silence the accumulation of a secondary metabolite with simple RNAi constructs driven by constitutive promoters without simultaneously affecting plant growth. Tests of the defensive function of primary metabolites will require subtle silencing tools that allow silencing to be both tissue specific and controlled at very precise times; the goal is to minimize the growth and developmental effects of gene silencing while determining herbivore performance and resistance under native conditions. Integrative approaches that compare the effects of gene silencing at different levels in the signaling hierarchy will be necessary to determine how resources are allocated and source-sink relations adjusted. Asking the proteome, metabolome, and transcriptome for answers using unbiased analytical tools will, we hope, identify those regulatory nodes that are altered by herbivore attack. While (ultra) high-throughput analytical and data handling platforms will be important for handling the torrent of data produced, student training that emphasizes real-world familiarity with the plant and its natural history will also be important. Students will need to be trained to use an experimental approach that inverts the normal sequence of events in the biological discovery process. Instead of proceeding step-by-step from gene, to transcript, protein, metabolite, glasshouse phenotype, and, only when the plants are fully characterized, to a field test, field tests will need to be carried out earlier in the analysis. In this way, biological intuition will again become a valued trait among plant biologists.

The discovery of the function of one of N. attenuata's RNA-direct RNA polymerases (RdR1) in mediating resistance responses to herbivore attack illustrates the procedure (Pandey and Baldwin, 2007). RdRs are essential in siRNA biogenesis, but their organismic-level function was unknown. When plants silenced in endogenous RdR1 expression were planted into native populations, they were found to be highly susceptible to attack from native herbivores, which was associated with reduced nicotine levels, which in turn are known to require JA signaling. Further analysis revealed that RdR1 was involved in coordinating the signaling of phytohormones, namely, ethylene and JA, and, after sequencing the small RNA transcriptome, a number of small RNAs that matched sequences for phytohormone biosynthetic genes were found (Pandey and Baldwin, 2008). This example demonstrates the value of field experiments in rapidly unraveling whole-organismic functions that are overlooked when plants are grown in protected glasshouse environments.

Heterotrophy was well established long before the evolution of photosynthesis. Plants have always had to cope with the ravages of consumers that want access to the resources that plants control. Their sophisticated means of defending themselves likely use all aspects of their metabolism. Their prodigious anabolic potential allows plants to throw just about everything at consumers to protect themselves. Our challenge will be to figure out what parts of metabolism are currently being maintained by natural selection as defenses. As Rick Karban predicted in his “Moving Target Hypothesis” (Karban et al., 1997), plants are very plastic and this plasticity itself may well be part of its defensive repertoire.

Although nature provides the best laboratory for testing gene function, we have trained a generation of plant biologists who are unfamiliar with field work. When unbiased transcriptional responses are used to “ask the plant” which genes are regulated in response to herbivore attack, the plant provides testable hypotheses about which genes are important in tolerance or defense. Gene annotations classify genes and specify their putative biochemical function based on sequence similarity. These annotations are extremely valuable, but they should be viewed with caution as they do not exclude other biochemical functions or functions at other levels. An example of the TD gene from N. attenuata illustrates the point. Silencing TD generated plants with stunted growth because TD is involved in Ile biosynthesis. However, other TD-silenced plants grew normally but were found to be highly susceptible to herbivores (Kang et al., 2006). Further analysis of these plants revealed a strong reduction of defense responses to herbivory, due to reduced levels of JA-Ile; defense signaling was hampered, but the effects on primary metabolism were negligible. This illustrates that the defense function of a primary gene can be studied with transformants that exhibit “mild” phenotypes and are not impaired in development.

Clearly, there will be much to be learned by “asking the plant” and using the “omic” tools for deciphering the plant's answer in the genes, proteins, and metabolites that it regulates differently when attacked by herbivores. If we are sufficiently forward thinking to ignore the gene annotations, to silence these regulated responses in ways that do not dramatically influence growth, and then to ask the community of herbivores that naturally attack plants whether the plant is more resistant to or tolerant of herbivore attack, we will undoubtedly learn much that is new about how plants survive in the real world. Although at present technical and regulatory issues impede the adoption of this procedure, the blinders that come with specialized scientific training will be as difficult to remove as the other challenges.