Oviposition by pierid butterflies triggers defense responses in Arabidopsis

Insect eggs represent a threat for the plant as hatching larvae rapidly start with their feeding activity. Using a whole-genome microarray, we studied the expression profile of Arabidopsis thaliana leaves after oviposition by two pierid butterflies. For Pieris brassicae, the deposition of egg batches changed the expression of hundreds of genes over a period of three days after oviposition. The transcript signature was similar to that observed during a hypersensitive response or in lesion-mimic mutants, including the induction of defense and stress-related genes and the repression of genes involved in growth and photosynthesis. Deposition of single eggs by P. rapae caused a similar although much weaker transcriptional response. Analysis of the jasmonic acid and salicylic acid mutants coi1-1 and sid2-1 indicated that the response to egg deposition is mostly independent of these signaling pathways. Histochemical analyses showed that egg deposition is causing a localized cell death, accompanied by the accumulation of callose, and the production of reactive oxygen species. In addition, activation of the PR1::GUS reporter gene correlated precisely with the site of egg deposition and was also triggered by crude egg extract. This study provides molecular evidence for the detection of egg deposition by Arabidopsis plants and suggests that oviposition causes a localized response with strong similarity to a hypersensitive response. There is thus evidence that plants are capable of detecting the presence of insect eggs and that they respond by activating direct and indirect defenses. However, there is no information on the molecular changes that take place in the host following oviposition. In this study, we investigated the response of Arabidopsis to oviposition by pierid butterflies and analyzed whole-genome transcript profiles in egg-laden leaves for a period of three days. The combined results of transcriptome analysis and visualization of defense-associated markers reveal that oviposition triggers a local response similar to a programmed cell death (PCD) accompanied by the activation of many defense-related genes.


INTRODUCTION
Plants under attack from herbivorous insects develop an array of defenses that are aimed at slowing the growth or development of the aggressor and specifically tailored to protect themselves from further attack. Numerous studies have revealed that, in addition to constitutive defenses represented by trichomes, thick secondary cell walls, or poisonous compounds, plants are equipped with inducible defenses that can be grouped into indirect (production of volatile odor blends to attract natural enemies of the attacker encountered) and direct responses (production of antidigestive proteins, toxic secondary compounds, and enzymes that affect insect growth and development) (Karban and Baldwin, 1997;Walling, 2000;Dicke et al., 2003). In Arabidopsis thaliana, challenge with chewing herbivores has been shown to provoke substantial transcriptional changes largely controlled by the plant hormone jasmonic acid (Reymond et al., 2004;De Vos et al., 2005).
One important aspect of plant/insect interaction that has received less attention is the potential recognition of insect egg deposition. Eggs represent a future threat for the plant and the anticipation of damage by a preactivation of defenses could provide an advantage to the host. Both direct and indirect responses to oviposition have been observed in plants (Hilker and Meiners, 2006). In pea pods, growth of undifferentiated cells is triggered upon oviposition by the pea weevil, which results in elevating the egg from the surface, increasing the risk of desiccation, predation, or falling off the pod (Doss et al., 2000). Upon oviposition by the white-backed planthopper, rice plants produce benzyl benzoate, an ovicidal substance that causes high egg mortality (Seino et al., 1996). The development of a necrotic zone at the site of egg deposition was observed in Brassica nigra resulting in egg dessication and mortality (Shapiro and Devay, 1987). In one potato hybrid clone, a necrotic zone around the oviposition site provoked the detachment and the fall of the eggs on the soil (Balbyshev and Lorenzen, 1997).
Indirect defenses induced by egg deposition comprise the release of volatiles or the modification of the surface chemistry by egg-laden leaves resulting in the attraction of egg parasitoids. Twigs of Pinus sylvestris emit volatiles after oviposition by the pine sawfly Diprion pini, attracting the egg parasitoid Chrysonotomyia ruforum. This response was shown to be both local and systemic and could be mimicked by jasmonic acid treatment (Hilker et al., 2002). Similar tri-trophic interactions have been observed in elm (Meiners and Hilker,

Expression changes in response to Pieris brassicae eggs.
After mating, female Pieris brassicae butterflies lay eggs on the underside of leaves of cabbage family plants. Approximately 10-40 eggs are gently deposited close to each other without visible physical damage. Between four to six days after oviposition, eggs start to hatch, young larvae eat the egg chorion, and then start to eat the leaf around the oviposition site. We wanted to know whether molecular changes take place in Arabidopsis during the first days after oviposition, before egg hatching and thus before larvae start to feed on the plant.
Using near full-genome microarrays, we investigated transcriptional changes 24 h, 48 h, and 72 h after oviposition by P. brassicae. On average, butterflies laid one to two egg batches per plant and we collected five leaf disks of approximately 5 mm in diameter at the appropriate times. RNA was extracted, amplified, labeled, and hybridized to CATMA microarrays containing 22,473 gene specific tags (Allemeersch et al., 2005). As controls, an equal number of leaf disks from a distal egg-free leaf were sampled at each time point, and the RNA extracted. This ensured that any observed changes were due to the presence of eggs and not to signals or contacts generated by flying butterflies. Each experiment was repeated three times independently, and each sample was labeled either with Cy3-dCTP or Cy5-dCTP (dye-swap design), giving a total of 6 measurements per time point. Average expression values were computed and genes whose expression ratio was ≥ 2 or ≤ 0.5 and P value <0.05 (both criteria had to be met in at least one time point) were considered as differentially regulated by egg deposition.
Eggs laid by P. brassicae triggered large changes in gene expression. 303 genes were induced 24 h after oviposition, 416 after 48 h, and 671 after 72 h (Supplemental Table I). In addition, oviposition caused a down-regulation of 53, 123, and 426 genes after 24 h, 48 h, and 72 h, respectively (Supplemental Table I). This represents a maximal induction of ca. 4% and a repression of ca. 2% of the transcriptome. Analysis of the time-course showed that the 56% of the genes that are induced at 72 h are already up-regulated at 24 h or 48 h and that the expression of most genes gradually increased or stayed steady over the three days of measurements, indicating that the response is more likely due to the continuous presence of the eggs and not to a touch response to egg deposition. The majority of down-regulated genes (68%) was only repressed 72 h after oviposition (Supplemental Fig. 1 to 3). It is possible that we have missed some early changes in gene expression but we observe that the response to oviposition corresponds to consistent and long-lasting transcriptional changes. In addition, we might have underestimated the number of differentially regulated genes after oviposition by using distal leaves as control tissue. As mentioned above, the rationale was to use leaves that had been in contact with butterflies as a control to eliminate the possibility of identifying genes responding to touch or butterfly-derived chemicals. The limitation of this experimental design is that genes induced after oviposition at similar levels in oviposited leaves and in distal leaves will not be detected. However, one hybridization using leaf samples from plants that did not contain eggs indicated that the number of genes induced systemically is low (Supplemental Fig. 4).
We examined the potential function of differentially expressed genes and classified them according to gene ontology (GO) terms using the tools for GO annotations at TAIR (http://www.arabidopsis.org) (Berardini et al., 2004). The distribution of gene annotations in different functional classes revealed a relatively high abundance of defense-and stress-related genes (Table I). Since eggs represent a future threat for the plant as they generate chewing herbivores that will feed on the leaves, we were curious to compare the transcript profile after oviposition to that of plants experiencing herbivory. We performed transcriptional analyses of Arabidopsis plants challenged with third-instar larvae of the specialist Pieris rapae using CATMA microarrays. We have previously shown using a dedicated microarray containing defense-related genes that both P. brassicae and P. rapae larvae trigger highly similar expression changes (Reymond et al., 2000). Surprisingly, there was very little overlap between oviposition-and herbivory-induced genes; only 10% (75/766) of the genes induced by P. brassicae eggs were also induced by P. rapae herbivory. These genes included members of the phenylpropanoid pathway (prephenate dehydratase, At3g44720; PAL1, At2g37040; C4H, At2g30490; TAT3; At2g24850; flavonol synthase, At5g05600) implicated in the synthesis of defense compounds; several genes involved in the synthesis of tryptophan (ASA1, At5g05730; TRP1, At5g17990; TSB1, At5g54810; TSB2, At4g27070); genes involved in redox balance or oxidative stress response (thioredoxin; At1g45145; glutaredoxin, At1g28480; two peroxidases, At5g64120 and At5g05340); and stress-responsive transcription factors (HSF4, At4g36990; ZAT10, At1g27730; ZAT12, At5g59820). They also included genes potentially involved in the synthesis and response to jasmonic acid (JA), salicylic (SA), and ethylene (ET) defense signals (LOX4, At1g72520; OPR3, At2g06050; ACX1, At4g16760; JR1, At3g16470; EDS5, At4g39030; ACC oxidase, At1g05010; ATERF1, At4g17500; ERF2, At5g47220) (Supplemental Table I).
Looking into the list of induced genes in more details we identified several markers of hypersensitive response (HR), which is a particular case of PCD. HR is an induced response triggered by the specific recognition of bacterial pathogens, viruses, fungi, and nematodes and is characterized by a localised cell death at the site of infection that prevents the progression of the disease (Kombrink and Somssich, 1995). We thus compared the expression profile after oviposition by P. brassicae with the profile observed after an HR caused by the bacterial  Table I).
Interestingly, we observed that most of the 75 genes induced both by oviposition and insect herbivory ( Fig. 1A) were also induced in acd2-2 mutant (62/75) and in P. syringae infected plants (64/75), indicating that these genes are likely markers of a general stress response.
Similarly, the majority of the 10 genes repressed both by oviposition and insect herbivory were also down-regulated in acd2-2 mutants (8/10) and by P. syringae treatment (8/10) (Supplemental Table I).
Egg deposition by P. brassicae caused the upregulation of many defense-related genes (Table II). Classical marker genes that accumulate during HR were induced, including pathogenesis-related genes PR2, PR3, PR4, and PR5, and regulators of innate immunity, like EDS1, PAD4,and SAG101. Some avirulence-or HR-responsive genes of unknown function were also up-regulated. Genes related to PCD included the anti-apoptosis BAX- INHIBITOR-1 (Matsumura et al., 2003); two genes that repress cell death and promote cell growth, BONZAI1 (BON1) and BON1-associated protein (BAP1) (Hua et al., 2001;Yang et al., 2006); two proteases belonging to the group of metacaspases that regulate plant PCD (Sanmartin et al., 2005); and a zinc finger (MYND type) family protein predicted to be associated with PCD. Cells that undergo HR synthesize callose and phenolic compounds. A gene encoding a callose synthase (CALS1) and several members of the phenylpropanoid pathways were induced by oviposition. Cellular protection against an oxidative burst as well as defenses responses are also activated in adjacent cells surrounding the site of HR. Several genes involved in the regulation of oxidative stress, like a superoxide dismutase, an Lascorbate oxidase, and a glutathione-S transferase were induced. Up-regulated genes encoding defense proteins included chitinases, lectins, and proteinase inhibitors. The tryptophan pathway provides precursors for several defense compounds like camalexin and indole glucosinolates in Arabidopsis (Wittstock and Halkier, 2002;Glawischnig et al., 2004).
Several genes involved in the biosynthesis of indole glucosinolates were induced by oviposition. Among other interesting genes potentially involved in defense or signaling, we identified 17 WRKY transcription factors that are typically involved in the regulation of defense-or senescence-induced programs (Eulgem et al., 2000), several MYB, NAC, ERF/AP2 and zinc-finger transcription factors, 29 protein kinase genes, 13 genes associated with calcium signaling, 21 genes encoding various transporters, and 7 genes involved in redox metabolism (Supplemental Table I).
In addition, we observed that 41 receptor-like kinases (RLKs) were induced by eggs from 24 h to 72 h after oviposition. RLKs are transmembrane proteins containing an intracellular kinase domain and a variable extracellular domain thought to interact with different extracellulars ligands. There are more than 600 RLKs in Arabidopsis (Shiu and Bleecker, 2001) and the role of only a few of them has been characterized, notably in development, growth, symbiosis or defense (Zipfel and Felix, 2005). RLKs are known to be involved in the perception of pathogen-derived elicitors that activate defense responses.
Recent studies with bacterial elicitors have shown that they induce more than 100 RLKs (Zipfel et al., 2004;Zipfel et al., 2006). By comparing our data with those of Zipfel et al.
(2006) we found that 22 of the oviposition-induced RLKs were also induced after treatement with the bacterial elicitors flg22 or elf26 (Table III).
Genes down-regulated by P. brassicae eggs consisted of genes involved in cell wall metabolism, cuticle biosynthesis, and fatty acid synthesis. This was accompanied by the repression of many genes that participate in photosynthetic activities (Table II).

PR1 is induced by oviposition
The up-regulation of two representative genes by P. brassicae oviposition was verified by quantitative real-time PCR after 72 h. A chitinase gene (At2g43570) was highly induced at and near the oviposition site, but only very weakly in the distal leaf (Fig. 2). The expression of a trypsin inhibitor gene (At1g73260) showed a highly localized response to P. brassicae eggs, as the gene was more than 1000-fold more expressed at the egg site than in the near, distal, or control samples. We also analyzed the expression of PATHOGENESIS-RELATED 1 (PR1) since it is a known defense marker gene involved in innate immunity. A probe for this gene was lacking in the current CATMA microarray. We found that PR1 is highly induced (60 to 70 fold compared to control plants) 72 h after oviposition, both at the site of oviposition and in a region surrounding the egg batch ( Fig. 2). There was also a small induction in the distal leaf.

Role of JA and SA pathways in egg responses.
We next determined the contribution of known signaling pathways involved in defense similar. We found almost no genes whose induction depended strictly on the JA or the SA pathway, i.e. genes that were induced in wild-type plants and not in the mutants after oviposition (Fig. 3). On the contrary, some of the induced genes were slightly more induced or more repressed in coi1-1 than in wild type, illustrating a potential negative regulation of these genes by JA in wild-type plants; 29% of the genes induced by P. brassicae oviposition were significantly more induced in coi1-1 than in wild-type plants (Fig. 3A, Supplemental Table II). This biais was less pronounced with sid2-1 plants where only 16% of egg-regulated genes showed a statistical difference between mutant and wild-type plants, somes genes being more induced in sid2-1 and some more in wild-type plants (Fig. 3B, Supplemental Table II).
We did not identify enrichment of particular functional categories among the genes that were differentially regulated by oviposition between coi1-1 and wild-type plants or between sid2-1 and wild type plants. Moreover, most of the genes differentially regulated between mutant and wild-type plants were not the same in coi1-1 and in sid2-1 plants (Supplemental Table II). An interesting exception consisted of two genes encoding NPR1-interacting proteins (NIMIN-1, At1g02450 and NIMIN-2, At3g25882) that were significantly more induced by oviposition in coi1-1 and less induced in sid2-1. cell wall remodeling exhibited also a similar expression biais, including a pectin methyl esterase (At3g47380), two endoxyloglucan transferases (At2g14620 and At4g18990), and a cellulase (At1g65610). Proteases and protease inhibitors were also enriched in this category, including two asparrtyl proteases (At3g12700 and At5g37540), a cysteine proteinase (At5g43060), a protease inhibitor (At3g22620), and a serpin (At2g26390) (Supplemental Table II).
Two known PR genes (PR2, At3g57260 and HEL, At3g04720) and an avirulence responsive gene (AIG2, At3g28930) were induced two-fold more by oviposition in sid2-1 than in wild-type plants. On the opposite, five WRKY transcription factors were not or less induced by oviposition in sid2-1 plants (Supplemental Table II). Accordingly, WRKY transcriptions factors have been implicated in the control of SA-dependent defense gene induction (Ulker and Somssich, 2004). In summary, although we do not see a major contribution of SA and JA in the control of oviposition-induced gene expression changes, more work will be necessary to assess the relative importance of these two pathways in the fine modulation of plant response to egg deposition.

Expression changes after oviposition by Pieris rapae.
To gain more insight in the specificity of egg-induced expression changes, we extended our analysis to a species with a different oviposition behavior. In contrast to P. brassicae, P.
rapae butterflies lay only one egg per site. We analyzed transcript profiles of Arabidopsis leaves 24 h to 72 h after oviposition by P. rapae. Each Arabidopsis leaf contained approximately one to two single eggs and we collected leaf disks of ca 5 mm at the site of oviposition. Samples from three independent experiments for each time point were hybridized to the microarrays. We observed that single egg deposition by P. rapae triggered a much weaker response than that by P. brassicae; only 23, 45, and 59 genes were up-regulated after 24 h, 48 h, and 72 h, respectively. Of those, 25 genes were induced in at least two time-points, indicating that although the effect was small these changes were not due to random experimental variability. Furthermore, 15 of these 25 genes were also induced after oviposition by P. brassicae (Supplemental Table III). In general, P. brassicae eggs caused a larger induction than P. rapae eggs but we also identified some genes specifically induced by P. rapae eggs, including a pathogenesis-related Bet v1 allergen protein (At1g70830), and others showing a quantitatively similar response to each treatment, including the hevein-like gene (HEL/PR4) (Fig. 4). Only 4 genes were down-regulated 72 h after oviposition. However, three of these genes, including one germin-like protein (At5g20630) and two glycine-rich proteins (At1g04800, At4g29020,) were also down-regulated after oviposition by P. brassicae (Supplemental Table III).

Oviposition induces cellular changes associated with cell death
A careful examination of egg-laden Arabidopsis plants did not reveal any physical damage. Furthermore, we could not see obvious phenotypical changes, nor any sign of infection at the site of oviposition during the days between oviposition and hatching.
Scanning electron microscope images of leaves containing P. brassicae and P. rapae eggs showed that eggs are fixed to the plant surface with a glue forming a visible meniscus surrounding the base of the egg, without apparent modification of the leaf epidermis ( Fig. 5A to 5C). Since the analyses of expression changes after oviposition identified many genes associated with PCD, we carried-out histochemical experiments to analyze known cellular changes associated with these responses. First, trypan blue was used to stain dead cells in leaves 72 h after oviposition by P. brassicae. We observed a strong staining corresponding closely to the site where the egg batch was deposited ( Fig. 5D and 5E). As trypan blue is a dye that enters permeable dead or damaged cells, a control experiment was done where leaves were stained just after oviposition. In this case, we only saw a faint circular staining corresponding to the contact point between each egg and the leaf surface (data not shown), indicating that careful detachment of the eggs before the staining procedure was not damaging the plant surface and hence was not causing the strong cell death response observed after three days.
Second, we examined callose deposition in egg-laden leaves. The accumulation of autofluorescent phenolic compounds and deposition of callose are markers associated with lesion formation in response to pathogen invasion (Koga et al., 1988) and lesions found in lesion-mimic mutants (Dietrich et al., 1994). Leaves were stained with aniline blue 72 h after P. brassicae oviposition and viewed under a fluorescent microscope. We observed a strong accumulation of callose closely associated with the site of egg batch deposition ( Fig. 5F to 5H).
Third, PCD is often preceded by an early production of reactive oxygen species (ROS) (Levine et al., 1994). To visualize the production of hydrogen peroxide (H 2 O 2 ) after oviposition, leaves were submerged in a 3,3'-diaminobenzidine (DAB) solution, a stain that polymerizes with H 2 O 2 and that produces a reddish-brown precipitate. In accordance with the detection of dead cells and callose deposition, the leaves stained with DAB 72 h after oviposition showed an accumulation of H 2 O 2 at the site of egg deposition (Fig. 5I). Fourth, having observed that the defense gene PR1 was highly induced by P.
brassicae eggs ( Parafilm egg-like structures to leaves for 72 h did not activate PR1::GUS reporter gene either (data not shown). In another test, eggs were extracted with MeOH or CHCl 3 , the extract was dried and resuspended in 0.05% Tween 20. The solution was applied to leaves for 72 h but no PR1::GUS activity could be detected (data not shown). Finally, we observed PR1::GUS activation after oviposition of single eggs by P. rapae ( Figure 6E and 6F), although staining was only seen in less than 50% of the cases.
To carry out an initial study on the nature of the elicitor(s) responsible for activating PR1::GUS expression, we collected P. brassicae eggs deposited on Arabidopsis leaves, gently crushed them in an Eppendorf tube and painted a few microliters of crude extract on a reporter plant. After 72 h, plants were stained for GUS activity and clearly showed activation of the PR1 promoter at the site of application ( Fig. 6G and 6H, left side). When the extract was boiled for 3 min before application, no activity could be detected ( Fig. 6G and 6H, right side). We then found that the supernatant of centrifuged crude egg extract (10'000 g, 1 min) kept a strong activity after storage at -20°C (Fig. 6I). This provides a useful information for the future isolation of active fractions. To confirm that eggs, or compounds associated with eggs, are responsible for the induction of PR1, we dissected female butterflies and removed eggs from the ovaries. A portion of the eggs (equivalent to a normal batch or about 30 eggs) was applied to a leaf. After 72 h, a strong GUS staining was observed at the site of application

Egg-induced responses
In this study, we show that Arabidopsis plants are able to detect the presence of eggs of the pierid butterflies P. brassicae and P. rapae, and trigger a response that has strong similarities with an HR. In the case of pathogenesis, HR is elicited after the recognition of specific pathogen-derived molecules by plant specific resistance genes and triggers a localized response that is characterized by a restricted necrosis at the site of infection, the production of ROS, the accumulation of secondary metabolites, and the induction of pathogenesis-related genes. Cells directly below the oviposition site stained strongly with trypan blue indicating that these cells were negatively affected by the presence of the eggs and were undergoing cell death, though not to the extreme result of developing a necrotic zone around the eggs. Indeed, there was no physical difference between the leaf surface where eggs were recently removed and other parts of the leaf. However, the plant was responding strongly as was further evidenced by the very specific accumulation of callose at the oviposition site, along with the induction of a callose synthase gene. Callose has been implicated in plant defense, for example in the wound response where it acts as a plug to seal the wound site. It is also believed that callose is rapidly deposited as a physical barrier to impede microbial and fungal attack (Stone and Clarke, 1992). However, it has also been shown that an Arabidopsis mutant lacking a functional callose synthase is more resistant to powdery mildew (Nishimura et al., 2003) indicating that the role of callose in resistance has to be carefully evaluated in each biological interaction.
Another supporting factor of a defensive response by the plant was the production of hydrogen peroxide at the oviposition site. Hydrogen peroxide is often produced as a result of external biotic and abiotic stimuli and has been shown to play a role in the control of HR (Levine et al., 1994) and in the induction of PR proteins in tobacco (Chen et al., 1993;Chamnongpol et al., 1998). Moreover, oviposition provoked gene expression changes that were very similar to those observed during bacteria-induced HR or in lesion mimic mutants.
Well-known PCD and defense-associated genes were induced by oviposition. One of the strongly induced genes was SAG13, a short-chain alcohol dehydrogenase that has been identified as a marker of PCD in the lesion mimic mutant acd11 (Brodersen et al., 2006). The PR1::GUS reporter line showed that the induction of the defense marker gene PR1 was specifically associated with the oviposition locations of both P. brassicae and P. rapae eggs.
Interestingly, egg deposition induced EDS1, PAD4, and SAG101, three lipase-related proteins that are thought to play a fundamental role in transducing redox signals upon both biotic and Arabidopsis ecotypes might reveal some genetically-based variation in the response to egg deposition.
In addition, we observed a down-regulation of several expansins, a cellulose synthase, pectin-modifying enzymes, and genes involved in cuticle biosynthesis by P. brassicae eggs, indicating that there might be cell wall remodeling at the site of oviposition. Many photosynthesis-related genes were also repressed, corroborating the previous observation of a net decrease of photosynthetic activity in Scots pine after egg deposition by Diprion pini (Schroder et al., 2005). Since defense production has been shown to be costly (Baldwin, 1998), plants have likely to reallocate ressources for this metabolic process at the expense of growth and photosynthesis.

Nature of the elicitor
We showed that a compound present in the supernatant of crude egg extracts activated PR1::GUS expression, indicating that the response to oviposition was not triggered by a physical damage or by simple contact of the butterfly during egg deposition. Extracts from accessory glands that produce the egg cement had however a weaker eliciting activity. of pine twigs to oviposition (Hilker et al., 2002). It has been hypothesized that JA could act as an egg-derived signal although it is not known whether it is transferred from the chorion to the plant surface. We measured JA levels in P. brassicae eggs and found 25 ± 6 ng JA/g egg tissue (n=3), a value comparable to resting levels of unchallenged Arabidopsis leaves (Reymond et al., 2004). This amount is similar to levels measured in Manduca sexta, a species where egg JA-content was considered to be low and substantially lower than the 400 µg/g measured in Spodoptera exigua eggs (Tooker and De Moraes, 2005). Given the fact that JA levels in P. brassicae eggs are low and that the induction of most egg-responsive genes are independent on a functional JA pathway, it is unlikely that JA plays a role as an egg-derived elicitor.
Eggs may contain symbionts that are transmitted by the female and are also exposed to pathogens (Kellner, 2002). The response to oviposition could thus be due to microorganisms associated with the eggs. However, in both reports of HR-like responses to egg deposition, the authors could not isolate plant pathogens associated with the eggs and concluded that the induction of the necrotic response by bacteria was unlikely (Shapiro and Devay, 1987;Balbyshev and Lorenzen, 1997). In our experiments, we could not detect obvious signs of bacterial disease while we observed the induction of a transcript pattern similar to that caused by the infiltration of high doses of a plant pathogen. Finally, we showed that two species of insects with different oviposition behavior provoked a similar response. It seems unlikely that eggs from different species carry the same plant pathogens. There is thus little evidence that microorganisms are the cause of egg-derived elicitors but further experiments will be necessary to identify the chemical nature and precise origin of these elicitors.   Zipfel et al., 2006). Since about half of the egg-induced RLKs were also found to respond to flagellin and EF-Tu (Zipfel et al., 2006), we cannot exclude that some of them were induced by egg-associated microorganisms. Further studies on the nature of the egg elicitors will help in elucidating the molecular basis of RLK activation.

Role of egg-induced genes in plant resistance
Since the egg shell is composed of proteins and wax, our finding that proteases and lipases are induced by egg deposition could indicate that these enzymes act as a direct defense against the egg. In addition, several chitinase genes were also induced by P. brassicae eggs.
Chitin is a major component of insects where it plays a scaffolding role supporting the cuticle as well as peritrophic matrices in the gut epithelium, but it does not seem to be present in the eggshell (Merzendorfer and Zimoch, 2003). Chitinases might thus rather play a defensive role against newly hatched larvae and their early induction might anticipate the feeding damage.
Similarly we also observed the induction of lectins and proteases inhibitors that are known to have some anti-insect properties. Moreover, we observed that two terpene cyclases were upregulated by egg deposition. Oviposition might thus also trigger the release of volatiles to attract parasitoid wasps, as it has been documented in other plant/insect interactions. Although this phenomenon has not yet been observed in Arabidopsis, these two genes are potential candidates for the synthesis of indirect defense compounds. In summary, our data suggest that the transcriptional response to oviposition constitutes a direct defense against the egg and that it can also contribute to defense in the early stages after egg hatching. At present we do not know whether the strong and localized HR-like response is really effective against egg deposition by interfering with the egg development and/or hatching rate, or whether it slows down the initial larval development. Interestingly, we noticed that hatching P. brassicae larvae always start to feed at the oviposition site, supporting a role for a localized plant defense activation. Further studies using Arabidopsis mutants impaired in the induction of egg-responsive genes might help to answer this question.
In conclusion, we provide here a molecular evidence for the detection of egg deposition by Arabidopsis plants and show that oviposition causes a strong and localized defense response that resembles an HR. It remains to be shown whether this response constitutes a direct defense against the eggs or whether plants have evolved a mechanism to anticipate the threat posed by future feeding larvae.

Plant and Insect Growth Conditions
Arabidopsis (Arabidopsis thaliana) ecotype Col-0 and mutants were grown as described previously (Reymond et al., 2000). Mutant seeds of acd2-2 were obtained from the Nottingham Arabidopsis Stock Center. Seeds of coi1-1 (non-glabrous) were a gift from Dr. Qiagen). RNA was then amplified using the MessageAMPII aRNA Kit (Ambion).
Fluorescent probes for hybridization were prepared using 5 µg of aRNA combined with 2 µg of 9N primers in a total volume of 9 µl. The mix was heated to 70°C for 10 min, chilled on ice, and then reverse-transcribed for 2 h at 42°C using 300 U SuperscriptII ( and Cy5 channels a normalization factor was computed using the Loess method. Signal values <500 (two to three times the average background standard deviation) were raised to 500 to avoid extreme expression ratios.
Normalized signal intensities were used to calculate expression ratios. We previously observed that a threshold of 2-fold and a P value <0.05 strongly increase the chance of identifying differentially expressed genes (Reymond et al., 2004) and therefore used both criteria to identify genes differentially expressed by oviposition. We performed a Student's ttest with log 2 -transformed expression ratios (one sample hypothesis, H 0 : log 2 ratio = 0) and

RT-PCR Analysis
Leaf disks for quantitative real-time PCR reactions were harvested 72 h after oviposition. All samples were harvested using a scalpel to enable cutting the precise area of the egg batch within 1 mm of the eggs (egg sample). The near sample was the 1-3 mm section of leaf surrounding previously cut egg disk. The distal leaf disk sample was cut to a similar size as the egg batch from a leaf without eggs but on the same plant as the eggs. A control (CTL) sample was also harvested consisting of leaf disks of similar size to the egg leaf disks but on a plant that was not exposed to the butterflies. The eggs were immediately and carefully removed, avoiding damage to the surface of the leaf disk, and the leaf sample was frozen in liquid nitrogen for later RNA extraction.
RNA samples were first reverse-transcribed using Superscript II (Invitrogen) based on the manufacturer instructions with 5 µg total RNA (purified with an RNeasy Qiagen kit followed by a DNaseI treatment) in a final volume of 100 µl. The resulting cDNA samples were stored at -80°C. RT-PCR was performed in triplicate using FullVelocity SYBR green kit

Histochemical Staining
For visualization of cell death, eggs were very carefully removed from leaves 72 h after oviposition avoiding leaf damage and leaves were submerged in trypan blue solution (5 mL lactic acid, 10 mL 50% glycerol, 1 mg trypan blue, and 5 mL phenol) at 30°C for 2-3 hours.
Leaves were then destained in 70% chloral hydrate followed by washes in 70 to 95% ethanol.
Callose deposition was detected 72 h after oviposition. Eggs were removed, leaving a few on the border of the egg batch for localization purposes. Ethanol-destained leaves were placed in 0.15 M phosphate buffer pH 8.5 containing 0.01% aniline blue for 1.5-3 h. Leaves were examined under a Leica MZ16FA epifluorescence microscope (excitation 340-380 nm, emission 420 nm filter) and images were acquired using a Leica DC300F camera. Hydrogen peroxide accumulation was measured 72 h after oviposition. After the eggs were removed, the leaves were submerged in a 1 mg/ml solution of diaminobenzidene (DAB) and placed in a phytotron (100 µmol m -2 sec -1 , 20°C and 65% relative humidity) for 8 h. The leaves were then destained by boiling in 95% ethanol for 10 min, and stored at room temperature in 95% ethanol until photographs could be taken.
GUS staining for PR1 expression analyses was performed on leaves 24 h, 48 h, and 72 h after oviposition. Eggs were removed and leaves were treated with 90% acetone at room temperature for 30-60 min, rinsed in a 50 mM phosphate buffered mixture containing 0.5 mM potassium ferricyanide, 0.5 mM potassium ferrocyanide, and 0.05% Triton X-100. Leaves were then submerged in the same solution with the addition of 0.5 µg/µl X-Gluc, vacuuminfiltrated for 20 min, and left at 37°C overnight. Chlorophyll was further destained in 70% ethanol and photographs were taken.
For PR1::GUS activity after application of P. brassicae egg extract, egg batches were collected on Arabidopsis plants 2 to 3 h after oviposition and crushed with a conical pestle in an Eppendorf tube. A few microliters of crude extract were directly applied on an Arabidopsis leaf with a pipette tip and the plant was left for 72 h before GUS staining. An aliquot of the extract was also boiled for 3 min and painted on a leaf. In another experiment, the crude egg extract was centrifuged (10,000 g, 1 min) and the supernatant was stored at -20°C before the application of 1 or 2 µL to a leaf, equivalent to 1/3 or 2/3 of an egg batch, respectively. For the treatment with female's ovaries, the P. brassicae reproductive system was removed with a scalpel, homogenized, and applied to a leaf. Isolated accessory glands from P. brassicae female butterflies were obtained from J.J.A. van Loon (Wageningen University). Six glands (stored at -80°C for 3 days) were crushed in an Eppendorf tube, and centrifuged (10,000 g, 1 min). Leaves were spotted with 2 µL of supernatant (equivalent to one accessory gland) and the plant was left for 72 h before staining. Aliquots of the solid pellet equivalent to one accessory gland were also spotted on the leaves. Expression ratios of oviposited vs egg-free leaves are plotted against expression ratios of oviposited vs control leaves from non oviposited plants.
Supplemental Table I. List of genes regulated by Pieris brassicae eggs, and comparison with infection with Pseudomonas syringae AvrRPM1, and lesion formation in acd2-2.
Supplemental Table II. List of genes regulated by Pieris brassicae eggs in wild-type and in the mutant plants coi1-1, and sid2-1.
Supplemental Table III. List of genes regulated after oviposition by P. rapae.
Supplemental Table IV. List of genes regulated after feeding by P. rapae.
Supplemental Table V Table   I.