Insect eggs suppress plant defence against chewing herbivores

Authors


For correspondence (fax +41 21 692 41 95; e-mail philippe.reymond@unil.ch).

Summary

Plants activate direct and indirect defences in response to insect egg deposition. However, whether eggs can manipulate plant defence is unknown. In Arabidopsis thaliana, oviposition by the butterfly Pieris brassicae triggers cellular and molecular changes that are similar to the changes caused by biotrophic pathogens. In the present study, we found that the plant defence signal salicylic acid (SA) accumulates at the site of oviposition. This is unexpected, as the SA pathway controls defence against fungal and bacterial pathogens and negatively interacts with the jasmonic acid (JA) pathway, which is crucial for the defence against herbivores. Application of P. brassicae or Spodoptera littoralis egg extract onto leaves reduced the induction of insect-responsive genes after challenge with caterpillars, suggesting that egg-derived elicitors suppress plant defence. Consequently, larval growth of the generalist herbivore S. littoralis, but not of the specialist P. brassicae, was significantly higher on plants treated with egg extract than on control plants. In contrast, suppression of gene induction and enhanced S. littoralis performance were not seen in the SA-deficient mutant sid2-1, indicating that it is SA that mediates this phenomenon. These data reveal an intriguing facet of the cross-talk between SA and JA signalling pathways, and suggest that insects have evolved a way to suppress the induction of defence genes by laying eggs that release elicitors. We show here that egg-induced SA accumulation negatively interferes with the JA pathway, and provides an advantage for generalist herbivores.

Introduction

Plants attacked by herbivorous insects develop various defence strategies (Howe and Jander, 2008). In addition to pre-existing physical and chemical barriers consisting of trichomes, thick secondary cell walls and toxic metabolites, plants deploy defences that are induced upon recognition of the enemy. In a sophisticated indirect mechanism, they emit volatiles to recruit predators of the attacker. This is superimposed on the production of anti-nutritive compounds, poisons and defensive proteins that directly affect the performance of the herbivore (Walling, 2000; Howe and Jander, 2008). At the molecular level, herbivores trigger massive transcriptional changes that are mainly controlled by the JA pathway (Reymond et al., 2004; De Vos et al., 2005; Kempema et al., 2007). Accordingly, Arabidopsis mutants impaired in the JA response are more susceptible to insect feeding (Bodenhausen and Reymond, 2007; McConn et al., 1997; Reymond et al., 2004, Stotz et al., 2002).

Insect eggs appear to be a passive stage of herbivores, but nevertheless represent a real threat for the plant as they give rise to feeding larvae. Direct and indirect responses to oviposition have been reported (Hilker and Meiners, 2006). Direct defences include the growth of undifferentiated cells on pea pods, which elevate weevil eggs and thus increase the risk of desiccation and predation of the larvae (Doss et al., 2000), production of the ovicidal substance benzyl benzoate in rice in response to eggs of the white-backed plant hopper (Seino et al., 1996), or development of a necrotic zone at the site of egg deposition in Brassica nigra and potato, which results in egg desiccation and mortality (Shapiro and Devay, 1987; Balbyshev and Lorenzen, 1997). Furthermore, the production of oviposition deterrents prevents female butterflies from laying more eggs on the host (De Vos et al., 2008). Indirect defences include the emission of volatiles (mono- and sesquiterpenes) or modification of the plant surface chemistry in response to oviposition, resulting in attraction of egg parasitoids. These tritrophic interactions have been observed in several species and with various predators (Meiners and Hilker, 2000; Hilker et al., 2002; Fatouros et al., 2005).

There is thus evidence that plants can perceive egg deposition, but knowledge on the nature of the elicitors and the cellular and molecular responses to oviposition is still limited. The only characterized elicitors of a direct defence are those responsible for the formation of tumour-like structures on pea pods. These elicitors are long-chain fatty acid-derived molecules called bruchins that are found in the eggs and adults of bruchid beetles (Doss et al., 2000). Elicitors triggering the release of volatiles have not yet been identified, but some are present in oviduct secretions coating the egg (Hilker et al., 2005). The anti-aphrodisiac male pheromone benzyl cyanide, which is found in the accessory gland secretion released by mated female butterflies, is the elicitor responsible for leaf surface changes that attract egg parasitoids (Fatouros et al., 2008). Some egg predators can exploit this sexual signal to attach themselves to the mated female butterflies and locate the host plant (Huigens et al., 2009).

In response to oviposition, plants reduce their photosynthetic activity (Schroder et al., 2005) and the emission of ethylene (Schroder et al., 2007). Recently, we analysed the expression profile of Arabidopsis leaves after oviposition by two species of pierid butterflies. Eggs laid by the large cabbage white butterfly Pieris brassicae modified the expression of hundreds of genes, including defence and stress-related genes that are induced in plants undergoing programmed cell death. Furthermore, callose deposition, hydrogen peroxide production and cell death occurred at the site of egg deposition, and we showed that egg-derived elicitors triggered these events (Little et al., 2007). Moreover, necrotic zones were observed at the oviposition site in plants related to Arabidopsis (Bruessow and Reymond, 2007).

The genes induced by oviposition include several known targets of SA. This signalling molecule is a potent inducer of pathogenesis-related genes, and is involved in the resistance against biotrophic pathogens (Glazebrook, 2005). Interestingly, it is known that SA and JA signalling pathways interact antagonistically (Beckers and Spoel, 2006; Koornneef and Pieterse, 2008). For example, there is evidence that SA can suppress JA-dependent defence responses (Doares et al., 1995; Gupta et al., 2000; Spoel et al., 2003, 2007; Cipollini et al., 2004; Koornneef et al., 2008). Activation of the SA pathway weakens the plant response to attackers that are resisted through the JA pathway, as demonstrated using a necrotrophic fungus (Spoel et al., 2007) and a chewing herbivore (Cipollini et al., 2004). This raises the question of whether eggs trigger an accumulation of SA, and whether this could in turn inhibit the defence responses against insect herbivores. We show here that SA levels increase at the site of oviposition, and that induction of herbivore-responsive genes is compromised after a further attack by chewing insects. Moreover, larvae of a generalist herbivore grew better on plants treated with egg extract, suggesting that eggs can manipulate plant defence signalling by interfering with the response to herbivores. These data suggest an intriguing phenomenon whereby insects may indirectly protect their offspring by releasing as yet unknown egg elicitors that target the defensive mechanism of the plant.

Results

Elicitors are present in eggs of various insects

We recently reported that P. brassicae eggs activate the expression of hundreds of genes in Arabidopsis (Little et al., 2007). Using an Arabidopsis transgenic line containing the promoter of the defence gene PR1 coupled to the β-glucuronidase (GUS) reporter gene, we showed that this activation was localized to the site of oviposition and could be mimicked by application of soluble P. brassicae egg extracts (Little et al., 2007). To test whether this phenomenon also occurred with eggs of other insects, we monitored the response of Arabidopsis GUS reporter lines to application of various egg extracts. In addition to PR1, we used promoters of three other genes that are strongly induced by oviposition: a gene encoding a trypsin inhibitor (TI), a chitinase gene (CHIT) and a senescence-associated gene (SAG13). Strong and localized GUS staining was observed for all marker genes in response to egg extracts from the specialist P. brassicae, the generalist Spodoptera littoralis, and the non-herbivore Drosophila melanogaster (Figure 1a). On the contrary, we found that extracts of bacteria recovered from P. brassicae eggshells and extracts from the yeast Saccharomyces cerevisiae did not activate the marker genes (F.B., unpublished results). These results indicate that Arabidopsis recognizes one or several generic elicitors present in the eggs of distantly related insects, and that these molecules are probably insect-specific.

Figure 1.

 Egg-derived elicitors activate the expression of reporter genes.
(a) Application of 2 μl of egg extracts from various insect species onto leaves of Arabidopsis promoter:GUS lines. GUS expression was analysed by histochemical staining 72 h after treatment. PR1, TI, CHIT and SAG13 are egg-induced genes. The top left panel is a photograph of an entire leaf, and the other panels are close-up views of the spotted area. P.b., Pieris brassicae; S.l., Spodoptera littoralis; D.m., Drosophila melanogaster.
(b) Activation of the PR1 promoter in response to application of eggshell preparation, egg extract treated with proteinase K, egg extract filtered through a 3 kDa filter, or lipids extracted from the eggs. GUS staining was performed 72 h after each treatment.

We then performed an initial characterization of the egg-derived elicitor(s). Application of empty eggshells of freshly hatched P. brassicae larvae to PR1::GUS plants for 72 h did not cause GUS staining, suggesting that the gene-induction activity resides within the embryo (Figure 1b). GUS staining was enhanced by proteinase K treatment of the egg extract, and was conserved after filtration of the soluble fraction through a 3 kDa filter, indicating that the elicitor is a small molecule without enzymatic activity (Figure 1b). Finally, we found that the eliciting activity is enriched in the fraction containing total egg lipids (Figure 1b).

SA accumulates at the oviposition site

The SA pathway activates numerous genes that are induced in response to oviposition (Little et al., 2007). Furthermore, ICS1, which encodes the key enzyme required for SA biosynthesis (Wildermuth et al., 2001), is strongly up-regulated by oviposition (Little et al., 2007). We thus quantified SA in leaf discs underneath the eggs and in distal Arabidopsis leaves in response to oviposition by P. brassicae butterflies. The SA levels increased gradually at the oviposition site during the 4 days after oviposition, but remained almost constant in distal leaves. After 4 days, SA levels were more than 10-fold higher in leaves containing eggs, than in control leaves, reaching approximately 21 μg g−1 FW (Figure 2a). Similarly, leaf discs from plants treated with P. brassicae egg extract accumulated high amounts of SA compared to untreated leaves, indicating that the treatment mimics oviposition (Figure 2b). As SA-responsive genes are induced in tissue adjacent to the oviposition sites (Little et al., 2007), and considerable amounts of SA are found underneath the eggs, we postulate that the entire leaf onto which eggs are laid, contains significant levels of SA.

Figure 2.

 SA accumulates in response to oviposition by P. brassicae.
(a) SA was quantified in Arabidopsis leaves during 4 days after oviposition. Black bars indicate leaf discs containing eggs and white bars indicate leaf discs from distal leaves. For each biological replicate, approximately 15 leaf discs of 10 mm diameter were collected underneath the eggs or on distal leaves from 10 plants. Each egg-containing leaf carried a batch of 30–35 eggs. Values are the means ± SE of three biological replicates.
(b) SA was quantified in Arabidopsis leaves during 4 days after application of 2 μl of P. brassicae egg extract. For each biological replicate, approximately 15 leaf discs of 10 mm diameter were collected. Black bars indicate treated leaves and white bars indicate non-treated leaves. Values are the means ± SE of three biological replicates.

SA has been detected in the eggs of lepidopteran insects (Tooker and De Moraes, 2007), and this could be the source of the levels found in the leaf tissue underneath the eggs. We quantified SA in P. brassicae eggs laid on Arabidopsis leaves. SA levels did not change during the 4 days after oviposition, averaging approximately 0.5 μg g−1 FW, which is about 25% of the basal level of untreated Arabidopsis leaves (Figure S1). As the SA levels in the leaves in response to oviposition are much higher than those found in eggs, simple transfer of SA from the egg to the plant cannot account for the observed accumulation. An alternative explanation is that a small amount of SA from the egg induces the synthesis of SA in the plant. This is unlikely as SA does not regulate the expression of ICS1 (Wildermuth et al., 2001). Thus, SA accumulation in response to oviposition probably results from de novo synthesis in the leaf.

Egg extract suppresses herbivore-induced gene expression

As SA accumulates in the oviposition site and as the SA pathway is known to interfere with JA-dependent defence against insects, we tested the effect of treatment with egg extract on the expression of insect-responsive genes. For this purpose, we selected ten genes (Table S1) that are strongly induced in response to chewing herbivores and controlled by the JA pathway (Reymond et al., 2004; Little et al., 2007), and measured their expression by quantitative real-time PCR. Plants were treated for 5 days with P. brassicae egg extract, and were then infested for 2 days with newly hatched larvae of the specialist P. brassicae or the generalist S. littoralis. As expected, all the selected genes were strongly induced after challenge with either P. brassicae or S. littoralis larvae in control plants that had not been treated with egg extract. However, this induction was markedly reduced in leaves that had been pre-treated with egg extract (Figure 3a,b). For example, the known anti-herbivore gene VSP2 showed an approximately 60-fold higher induction in untreated plants than in plants treated with egg extract. Overall, a similar suppression of gene expression was observed with both P. brassicae and S. littoralis larvae. This suppression was only observed in treated leaves. In distal leaves, the induction of insect-responsive genes was similar whether the plants were treated with egg extract or not (Figure S2). Furthermore, treatment with S. littoralis egg extract also suppressed defence gene expression in response to subsequent challenge with S. littoralis larvae, indicating that this phenomenon is conserved between two distantly related insects (Figure S3).

Figure 3.

 Treatment with egg extract suppresses insect-induced genes.
(a,b) Col-0 Arabidopsis plants were treated for 5 days with P. brassicae egg extract, and subsequently challenged for 48 h with P. brassicae larvae (a) or with S. littoralis larvae (b).
(c) Arabidopsis sid2-1 plants treated with P. brassicae egg extract and challenged with S. littoralis larvae.
Expression of insect-responsive genes was measured by real-time PCR. Values are normalized to the reference gene and expressed relative to unchallenged control samples. White bars indicate untreated plants. Black bars indicate plants treated with egg extract. Values are the means ± SE of three biological replicates.

Interestingly, the suppression of insect-responsive genes was almost completely abolished in the SA-deficient mutant sid2-1. The majority of the genes were equally induced by infestation with S. littoralis larvae in treated and untreated plants (Figure 3c). Thus, our data indicate that the suppression of defence gene expression in response to egg elicitors occurs locally and is largely controlled by SA.

To assess whether the SA-mediated suppression of insect-induced genes was targeting the JA pathway only, we monitored the expression of five JA-independent genes (Reymond et al., 2004) in plants challenged with S. littoralis. These genes were induced both by treatment with egg extract and by S. littoralis feeding. However, treatment with egg extract followed by a subsequent attack by insect larvae did not suppress their induction (Figure S4). These data suggest that egg-induced SA accumulation leads to preferential inhibition of JA-dependent defence gene expression.

Plants treated with egg extract are more susceptible to S. littoralis larvae

To test whether the suppression of insect-responsive genes had an effect on insect performance, we measured the weight of larvae feeding on plants treated with P. brassicae egg extract. Plants were treated for 5 days with P. brassicae egg extract, and were subsequently challenged for 8 days with newly hatched P. brassicae larvae. Surprisingly, the weight of larvae fed on treated or untreated plants did not differ significantly (Figure 4a). In contrast, larvae of the generalist herbivore S. littoralis gained significantly more weight on egg extract-treated plants than on untreated plants (Figure 4b). The enhanced susceptibility of egg extract-treated Col-0 plants was abolished in the sid2-1 mutant. Indeed, S. littoralis larvae reached the same weight after 8 days of feeding on treated or untreated sid2-1 plants (Figure 4c). We thus show that treatment with egg extract triggers a potent SA-mediated suppression of defence against a generalist herbivore.

Figure 4.

 Treatment with egg extract increases plant susceptibility to a generalist herbivore.
(a,b) Freshly hatched larvae of the specialist P. brassicae (a) and the generalist S. littoralis (b,c) were placed on Arabidopsis plants, and larval weight was measured after 8 days of feeding.
(c) sid2-1 plants challenged with S. littoralis larvae.
White bars indicate untreated control plants challenged with insects. Black bars indicate plants treated with P. brassicae egg extract 5 days before insect challenge. The number of larvae used in each experiment is shown within the bars. Error bars indicate SE. Asterisks indicate statistically significant differences compared to the control (Student’s t test, α = 0.001). n.s., difference not significant. Similar results were observed in several independent replicate experiments.

Suppression of plant defences by egg extract does not depend on NPR1

Studies with pathogens have shown that the regulatory protein NPR1 plays a crucial role in the inhibitory action of SA on JA-dependent defences (Spoel et al., 2003, 2007; Leon-Reyes et al., 2009). Therefore, we analysed whether NPR1 is also involved in the inhibition of insect-induced responses by egg-derived elicitors. When npr1-1 mutants were challenged with S. littoralis larvae, insect-responsive genes were strongly up-regulated. However, pre-treatment with egg extract inhibited this induction significantly, similarly to Col-0 plants (Figure 5a). Accordingly, npr1-1 plants pre-treated with egg extract were more susceptible to feeding by S. littoralis than untreated plants (Figure 5b). Collectively, our data indicate that the suppression of insect defences by egg extract requires the accumulation of SA but is independent of NPR1.

Figure 5.

 Suppression of defences by egg extract is NPR1-independent.
(a) npr1-1 plants were treated for 5 days with P. brassicae egg extract, and subsequently challenged for 48 h with S. littoralis larvae. Expression of insect-responsive genes was measured by real-time PCR. Values are normalized to the reference gene and expressed relative to unchallenged control samples. White bars indicate untreated plants. Black bars indicate plants treated with egg extract. Values are the means ± SE of three biological replicates. Control experiments with Col-0 are shown in Figure 3(b).
(b) Freshly hatched S. littoralis larvae were placed on npr1-1 plants, and larval weight was measured after 8 days of feeding. White bars indicate untreated control plants challenged with insects. Black bars indicate plants treated with P. brassicae egg extract 5 days before insect challenge. The number of larvae used in each experiment is shown within the bars. Error bars indicate SE. Asterisks indicate statistically significant differences compared to the control (Student’s = test, α = 0.001). Similar results were observed in several independent replicate experiments. The control experiment with Col-0 is shown in Figure 4(b).

Discussion

Our data provide strong evidence that egg-derived elicitors trigger the suppression of defences against chewing herbivores in Arabidopsis. This process is mediated by SA, as evidenced by the lack of gene suppression and the absence of enhanced susceptibility in sid2-1 mutants. Several observations indicate that application of egg extract mimics actual oviposition by P. brassicae: (i) PR1::GUS is activated similarly by oviposition and by application of egg extract (Figure 1 and Little et al., 2007), (ii) oviposition and egg extract treatment lead to a local accumulation of SA, with similar kinetics (Figure 2), and (iii) whole-genome analysis of gene expression in response to oviposition and egg extract treatment yields overlapping transcript profiles (data not shown).

The observation that egg extracts from distantly related insect species activate the same marker genes, together with the finding that egg extracts from both a specialist and a generalist insect are effective in suppressing defence gene expression, is an indication that some generic molecules in the eggs are recognized by the plant. This is analogous to the detection of microbe-associated molecular patterns (MAMPs) by the innate immune system of the plant (Boller and Felix, 2009), and potentially broadens the list of exogenous cues that are recognized by this surveillance mechanism. Further purification and characterization of the P. brassicae elicitor(s) should shed light on the identity of these molecules.

It is remarkable that this egg elicitor recognition system is used by the herbivore for its own benefit. Generally, plants have evolved mechanisms to detect MAMPs as a defence strategy, and as yet there is no example in which the attacker obtains a benefit. If egg detection is similar to MAMP-triggered immunity, we postulate that egg recognition by a MAMP/receptor pair evolved first as a defence mechanism for the plant and was subsequently hijacked by the herbivore for its own advantage.

We found that plants treated with egg extract were more susceptible to chewing larvae of the generalist herbivore S. littoralis but not to larvae of the specialist P. brassicae. The most straightforward explanation is that P. brassicae, a specialist of Brassicaceae plants species, is adapted to the defences of these plants. For example, the larvae of several lepidopteran species including Pieris rapae and P. brassicae contain a nitrile-specifier gut protein that detoxifies the breakdown products of glucosinolates, which are the major insect deterrents in Arabidopsis (Wittstock et al., 2004; Wheat et al., 2007). Moreover, we recently showed that larvae of the specialist P. brassicae performed equally well on wild-type Arabidopsis plants and on mutants that either lacked glucosinolates or that were impaired in the JA pathway, whereas larvae of the generalist S. littoralis were much larger when feeding on mutant plants (Schlaeppi et al., 2008). Alternatively, P. brassicae oral secretions might contain factors that suppress plant defences. Thus, the oviposition-induced and oral secretion-induced suppression of JA signalling may be redundant, and may explain why P. brassicae larvae perform equally well on treated and untreated plants.

It is intriguing that P. brassicae eggs have retained the ability to suppress herbivore defences although this does not result in increased performance of their own progeny. One reason could be that this trait preceded the acquisition of tolerance to induced defences, and therefore would have provided an advantage to the larvae early in evolution, before specialization and adaptation to its current host plants. Alternatively, the suppression of defences could still be beneficial for the larvae under less favourable conditions, for instance when the detoxification of defence compounds is less efficient due to the high yields of toxins produced in some Brassicaceae species. Finally, in addition to its role in inhibition of the JA pathway, the egg-triggered accumulation of SA might facilitate the development of the larvae. As the SA pathway is involved in defence against microbial and fungal pathogens, it could be advantageous for the hatching larvae to stimulate SA production around the oviposition site and therefore to protect the tissue from infection and to keep the full nutritive value of the early feeding site. Interestingly, the SA levels measured at the oviposition site are in the same range as those found in leaves infected with bacterial, fungal, and oomycete biotrophs (Summermatter et al., 1995; Roetschi et al., 2001; Wildermuth et al., 2001). The combined accumulation of SA (this paper) and the up-regulation of many pathogenesis-related genes at the site of oviposition (Little et al., 2007) support this hypothesis.

Our data show that eggs hijack the SA pathway for the benefit of the hatching larvae, but raise the question of whether SA accumulation is of direct benefit for the plant? Several SA-responsive genes might be involved in the programmed cell death that is observed at the site of oviposition (Little et al., 2007). In some species, this response is more severe, and results in death of the eggs or in the eggs falling from the leaf surface (Shapiro and Devay, 1987; Balbyshev and Lorenzen, 1997). Other SA-induced genes might play a direct but as yet unknown defensive role against the eggs. In addition, SA might be converted to the volatile methyl salicylate (MeSA), which has been shown to inhibit oviposition of the cabbage moth Mamestra brassicae (Ulland et al., 2008). It would be interesting to test whether Arabidopsis plants emit MeSA in response to oviposition, and whether this subsequently reduces further oviposition by P. brassicae butterflies.

Using sid2-1 plants, we found that SA accumulation is required for the suppression of insect-induced defences, but we have previously shown that the majority of egg-induced genes are induced in sid2-1 plants 3 days after oviposition (Little et al., 2007). This apparent discrepancy may be explained by the fact that the target genes are not the same. On one hand, oviposition directly controls the expression of hundreds of genes, including SA-related genes and genes involved in programmed cell death. Oviposition directly controls the expression of hundreds of genes, including SA-related genes and genes involved in programmed cell death. We have recently found that another unknown signal can partially replace SA in the late induction of these genes (data not shown). In addition, the oviposition-triggered accumulation of SA interferes with the induction of JA-dependent genes that are not induced by eggs but by chewing larvae.

We provide here strong evidence that egg-induced accumulation of endogenous SA suppresses JA-mediated defences against a generalist herbivore. Cross-talk between JA and SA signalling has been described in response to various attackers (Beckers and Spoel, 2006; Stout et al., 2006; Koornneef and Pieterse, 2008). Plants challenged with microbial pathogens or treated with SA or inducers of the SA pathway were more susceptible to herbivorous insects or necrotrophic pathogens (Thaler et al., 1999; Cipollini et al., 2004; Cui et al., 2005; Spoel et al., 2007; Leon-Reyes et al., 2009). Wild-type tobacco and Arabidopsis mutants with elevated SA levels were more susceptible to feeding by Spodoptera exigua and Trichoplusia ni larvae, respectively (Cui et al., 2002; Rayapuram and Baldwin, 2007). Infection with tobacco mosaic virus stimulated SA signalling in tobacco, and consequently increased feeding by the hornworm Manduca sexta (Preston et al., 1999). At the molecular level, exogenous treatments with SA and JA have shown that SA inhibits the JA-induced expression of several defence genes (Doares et al., 1995; Niki et al., 1998), including the well-known JA markers VSP2, PDF1.2 and LOX2 in Arabidopsis (Spoel et al., 2003; Koornneef et al., 2008).

The role of the antagonistic effect of SA on the JA pathway is not fully understood, but is thought to help the plant to optimize its response to pathogens with different lifestyles (Spoel and Dong, 2008). Indeed, by inhibiting an inappropriate defence pathway, the plant could allocate more energy for an effective defence against a particular attacker. Our finding that egg-derived elicitors suppress plant defence against insects by activating SA/JA cross-talk illustrates another modulation of the defence signalling network, but this time for the benefit of the attacker. Indeed, the recognition of eggs leads the plant to indirectly lower the defences against feeding larvae. A similar phenomenon has been reported for the phloem-feeding silverleaf whitefly (Bemisia tabaci). Nymph feeding on Arabidopsis induced SA-responsive genes and repressed JA-responsive genes. In addition, mutants that activated the SA pathway or were impaired in JA signalling allowed faster nymph development (Zarate et al., 2007). It was proposed that nymphs suppressed JA defences via SA/JA cross-talk but endogenous SA levels were not measured. In another example, attack of sorghum plants by greenbug aphids activated SA-regulated genes, although it was shown that the JA-regulated defences were effective against aphids. Again SA/JA cross-talk was suggested to explain these intriguing results, albeit without experimental evidence (Zhu-Salzman et al., 2004). Other studies have indicated that components in insect oral secretions can suppress plant defences (Kahl et al., 2000; Musser et al., 2002; Lawrence et al., 2007). The expression and activity of JA-dependent genes was higher in Arabidopsis plants that were infested with S. exigua caterpillars with impaired salivary secretions compared to intact caterpillars (Weech et al., 2008). Interestingly, this suppression of defence responses was abolished in mutants that are unable to mount a systemic acquired resistance response, and might indicate that SA is involved in this phenomenon. In support of this hypothesis, treatment with S. exigua oral secretions stimulated SA accumulation in Nicotiana attenuata (Diezel et al., 2009). There is thus emerging evidence that insects or insect-derived cues (eggs or oral secretions) are able to manipulate the SA/JA cross-talk to their advantage.

The molecular mechanisms of cross-talk between the SA and JA pathways are not fully understood. The regulatory protein NPR1 has been identified as a key component in the inhibitory action of SA on JA-dependent defences (Spoel et al., 2003). NPR1 is active in the nucleus, regulating the expression of pathogenesis-related genes (Spoel et al., 2009), but has an additional role in the cytosol, where it is proposed to negatively interact with several components of the JA pathway (Spoel et al., 2003; Beckers and Spoel, 2006). Moreover, ethylene signalling was found to override the function of NPR1 in SA/JA cross-talk (Leon-Reyes et al., 2009, 2010). In addition, the transcriptional activator WRKY70 (Li et al., 2004), glutaredoxin (Ndamukong et al., 2007) and the fatty acid desaturase SSI2 (Kachroo et al., 2001) are important modulators of SA/JA cross-talk in Arabidopsis. In this study, we found that egg-induced defence suppression against a generalist herbivore is independent of NPR1. This finding confirms earlier studies with Arabidopsis showing that inhibition of JA-induced resistance to chewing insects still occurred in npr1-1 plants treated with SA or with virulent bacterial pathogens (Cui et al., 2002; Cipollini et al., 2004). In addition, SA inhibited the induction of leaf trichomes, another component of resistance to herbivores, in an NPR1-independent way (Traw and Bergelson, 2003). There is thus emerging evidence that the negative SA/JA cross-talk is controlled differently depending on the type of attacker triggering the JA pathway. It will be interesting to elucidate the precise molecular mechanism of the suppression of insect-induced responses. We have found that inhibition of JA-induced genes by egg extract was still observed after treatment with MeJA, suggesting that the negative SA/JA cross-talk operates downstream of JA biosynthesis (data not shown).

In summary, we present here an intriguing facet of the arms race between plants and insect herbivores, in which an apparent inert stage of the insect, the egg, hijacks the SA signalling pathway for the benefit of its progeny. It will be fascinating to determine whether this phenomenon is widely distributed in nature as a potent way of circumventing the defences of plants.

Experimental Procedures

Plants and insects

Arabidopsis thaliana (Col-0), PR1::GUS (a gift from Dr Allan Shapiro, Biological Sciences Department, Florida Gulf Coast University, FL) and the mutants sid2-1 (a gift from Dr Christiane Nawrath, Department of Plant Molecular Biology, University of Lausanne, Switzerland) and npr1-1 (Nottingham Arabidopsis Stock Center, http://arabidopsis.info) in the Col-0 background were grown as described previously (Reymond et al., 2004). For GUS reporter lines, the promoter regions of TI (At1g73260), CHIT (At2g43570) and SAG13 (At2g29350) were amplified by PCR (see Table S1 for primer sequences), cloned into the Gateway donor vector pDONR/Zeo (Invitrogen, http://www.invitrogen.com), recombined into the destination vector pMDC162 cassette A (Curtis and Grossniklaus, 2003), and the plasmids transferred to Agrobacterium tumefasciens strain pGV3101 (pMP90). Transgenic plants were obtained by floral dipping and selection of the seeds on 100 μg ml−1 hygromycin.

Rearing of Pieris brassicae (large white butterfly) was performed on cabbage (Brassica oleracea) (Reymond et al., 2004). Spodoptera littoralis (Egyptian cotton worm) eggs were obtained from Dr Roland Reist (Syngenta Crop Protection, Stein, Switzerland), Drosophila melanogaster eggs were obtained from Dr Tadeusz Kawecki (Department of Ecology and Evolution, University of Lausanne, Switzerland).

SA measurements

SA was quantified in excised leaf discs by HPLC as described previously (Garcion et al., 2008). Analyses were performed on triplicate samples with good reproducibility, and the results given are mean values. For each biological replicate, approximately 15 leaf discs of 10 mm diameter were collected at the oviposition/treatment site or on distal leaves from ten plants. Fifteen batches of eggs (30–35 eggs per batch) from the same plants were analysed similarly to plant extracts.

Treatments with egg extracts

Pieris brassicae eggs laid on cabbage leaves were crushed with a pestle in Eppendorf tubes. After centrifugation (15 000 g for 3 min), the supernatant (‘egg extract’) was stored at −20°C. Egg extracts from S. littoralis and D. melanogaster were prepared similarly. For determination of GUS activity, 2 μl of egg extract, corresponding to one egg batch of approximately 20–30 eggs, was spotted onto each leaf for 72 h. The extract was then gently removed with a paint brush, and GUS staining was performed as described previously (Little et al., 2007). For protease treatment, 50 units of proteinase K immobilized on Eupergit C beads (Fluka, http://www.sigmaaldrich.com) were added to 1 ml of egg extract. The sample was incubated for 20 min at 30°C with agitation at 10 g in a thermomixer (Eppendorf, http://www.eppendorf.com). The beads were then spun down at 100 g, and the promoter:GUS plants were treated with 2 μl of the supernatant. For filtration, 100 μl of egg extract was placed in a 3 kDa Microcon filter (Millipore, http://www.millipore.com), and centrifuged for 60 min at 14 000 g at 20°C. Two microlitres of flow-through were applied to the promoter:GUS plants. For total lipid extraction, 20 ml of chloroform/ethanol (1/1) were added dropwise to 1 ml of egg extract in an Erlenmeyer flask, and incubated for 1 h at room temperature on a shaker. The clear supernatant was transferred to a beaker and the solvent evaporated to dryness on a steam bath. The extract was dissolved in pure chloroform and filtered through a funnel packed with cotton. Chloroform was evaporated under nitrogen, and the lipid extract was applied to promoter:GUS plants. Remaining eggshells of freshly hatched P. brassicae larvae were collected and mixed with distilled water before application onto promoter:GUS plants using a spatula.

For expression analyses and bioassays, two spots of 2 μl of egg extract were applied onto each leaf. In total, two leaves each from 20 plants were treated. After 5 days, the egg extract was gently removed with a paint brush, and one freshly hatched P. brassicae or two freshly hatched S. littoralis larvae were placed onto each plant. Plants were placed in transparent plastic boxes in a growth room as previously described (Bodenhausen and Reymond, 2007). After 2 days of feeding, local and distal leaves from four plants were harvested for RNA analysis, and the 16 remaining plants were left with the larvae for another 6 days. At the end of the experiment, larvae were collected and weighed. Controls consisted of untreated and/or uninfested plants.

Quantitative real-time PCR

Total RNA was extracted using an RNeasy® plant mini kit (Qiagen, http://www.qiagen.com). For cDNA synthesis, 1 μg of total RNA was reverse-transcribed using M-MLV reverse transcriptase (Invitrogen) in a final volume of 25 μl. Each cDNA sample was generated in triplicate and diluted fourfold with water. Gene-specific primers (Table S1) were designed to produce amplicons of between 70 and 200 bp from the 3′ end of the cDNA strand. Primer efficiencies were assessed by a five-step dilution regression. Quantitative real-time PCR analysis was performed in a final volume of 25 μl containing 2.5 μl of cDNA, 0.1 μm of each primer, 0.03 μm of reference dye, and 2 × Brilliant II Fast SYBR® Green QPCR Master Mix (Agilent, http://www.agilent.com). Reactions were performed using an Mx3000P™ real-time PCR machine (Agilent) with the following program: 95°C for 10 min, then 40 cycles of 10 sec at 95°C, 20 sec at 55°C and 17 sec at 60°C. Relative mRNA abundance was normalized to the reference gene At2g28390 (Arabidopsis SAND family protein), which has been shown to be a stable reference gene for quantitative RT-PCR (Czechowski et al., 2005), and expressed relative to the control sample. The values in Figure 3 are the mean of three independent biological replicates.

Acknowledgements

We thank Roland Reist (Syngenta Crop Protection, Stein, Switzerland) for providing S. littoralis eggs, Dr Tadeusz Kawecki (Department of Ecology and Evolution, University of Lausanne, Lausanne, Switzerland) for providing D. melanogaster eggs, and Blaise Tissot for help in growing plants. The Swiss National Science Foundation (Schweizerische Nationalfonds) supported this work (grant number 3100A0_118421 to P.R.).

Ancillary