Comprehensive Analysis of Gene Expression Profiles of the Beet Armyworm Spodoptera exigua Larvae Challenged with Bacillus thuringiensis Vip3Aa Toxin

Host-pathogen interactions result in complex relationship, many aspects of which are not completely understood. Vip proteins, which are Bacillus thuringensis (Bt) insecticidal toxins produced during the vegetative stage, are selectively effective against specific insect pests. This new group of Bt proteins represents an interesting alternative to the classical Bt Cry toxins because current data suggests that they do not share the same mode of action. We have designed and developed a genome-wide microarray for the beet armyworm Spodoptera exigua, a serious lepidopteran pest of many agricultural crops, and used it to better understand how lepidopteran larvae respond to the treatment with the insecticidal protein Vip3Aa. With this approach, the goal of our study was to evaluate the changes in gene expression levels caused by treatment with sublethal doses of Vip3Aa (causing 99% growth inhibition) at 8 and 24 h after feeding. Results indicated that the toxin provoked a wide transcriptional response, with 19% of the microarray unigenes responding significantly to treatment. The number of up- and down-regulated unigenes was very similar. The number of genes whose expression was regulated at 8 h was similar to the number of genes whose expression was regulated after 24 h of treatment. The up-regulated sequences were enriched for genes involved in innate immune response and in pathogen response such as antimicrobial peptides (AMPs) and repat genes. The down-regulated sequences were mainly unigenes with homology to genes involved in metabolism. Genes related to the mode of action of Bt Cry proteins were found, in general, to be slightly overexpressed. The present study is the first genome-wide analysis of the response of lepidopteran insects to Vip3Aa intoxication. An insight into the molecular mechanisms and components related to Vip intoxication will allow designing of more effective management strategies for pest control.


Introduction
The beet armyworm Spodoptera exigua (Hübner; Lepidoptera: Noctuidae) is a highly dispersive, polyphagous species that is a serious pest of cotton, alfalfa, tomatoes, sugar beets, and many other agricultural and flower crops worldwide [1]. The concern in today's society about the ecological damage caused by the abuse of chemical insecticides as well as the ability of insects to develop resistance to them [1] has led more attention on biological insecticides as alternatives for controlling pests. Bacillus thuringiensis (Bt) is the most widely used commercial microbial pathogen. Its spores and crystalline insecticidal proteins (Cry proteins) have been used to control insects since 1938 [2]. In 1996, a novel class of insecticidal proteins isolated from Bt and expressed during the vegetative growth phase (hence the name Vegetative Insecticidal Proteins or Vip proteins), was first reported [3]. These toxic Bt proteins represent an interesting complement to Cry toxins because similarly to Cry proteins, they show high and specific activity against a wide range of agriculturally important lepidopteran larvae [4]. Since both Bt toxins (Cry and Vip) provide excellent control of target pests such as S. exigua [5,6] with minimal environmental impact, transgenic crops expressing Cry toxin alone, and recently both Cry and Vip toxins, have been developed and are being increasingly used worldwide [7,8]. This combination of Bt proteins in transgenic plants offers the possibility for targeting a wider range of insects and minimizes the risk of resistance outbreaks in the field.
The mode of action of the insecticidal Cry toxins has been extensively studied for more than 20 years. However, some aspects remain unclear. It is commonly accepted that these crystal proteins need to be solubilized in the insect gut to be processed to the active form, which binds to specific receptors in the brush border epithelial midgut cells. This binding leads to cell lysis, and eventually insect death. The molecular mechanism by which this occurs is not completely understood and differs depending on the binding model [2,9,10]. The available information mainly supports the notion that these toxins act by forming pores [11]. Although little is known about the mode of action of Vip proteins like Vip3Aa, it has also been described to act through the formation of pores in the midgut epithelial cells [12,13]. As with Cry proteins, Vip proteins are ingested either as a protoxin or in the processed toxin form [6,14], and produce similar effects in the insect midguts, causing eventually the lysis of midgut cells [15]. However, Vip proteins have different properties than their Cry counterparts in several key steps of their mode of action, including the binding to midgut receptors [4,12,13,15].
The insect midgut is where activated Cry and Vip toxins bind to and initiate cytotoxicity. Freitak and coworkers [16] suggested that in addition to being an organ of digestion and resource assimilation, the midgut epithelial tissue is also an immune response-sensing organ, as they observed that nonpathogenic bacterial feeding could trigger an immune response cascade in the Lepidoptera Trichoplusia ni.
In insects, immunity consists of the combination of cellular responses (phagocytosis, encapsulation and melanization of invading microorganisms) and humoral responses (e.g. antimicrobial peptides secreted to the hemolymph) [17,18]. Also, the immune response in insects can be classified as either systemic or local, where the specific tissue (e.g. the gut in the instance of oral intoxication processes) responds locally to the damage [19,20]. Insects respond to microbial ligands mainly through the activation of the stress pathways Toll, Imd, JAK/STAT, JNK, and MAPK p38 [19,21]. Eukaryotic nonimmune cells (such as epithelial cells) have evolved various defense responses to cope specifically with pore forming toxins, such as through the activation of the MAPK p38 and JNK pathways [22]. In fact, the MAPK p38 pathway is activated to protect the nematode Caenorhabditis elegans or HeLa cells against the Bt Cry5 toxin [22,23,24], and it is activated in Lepidoptera and Diptera after Cry-toxin intoxication [25]. Some studies indicate that S. exigua also responds to bacterial infection through the general mechanisms involved in immune response in insects, producing antimicrobial peptides (AMPs) such as cecropins [26], gloverins [27], or attacins [28], as well as producing "REsponse to PAThogen" (REPAT) proteins in the midgut after Cry1Ca intoxication [29].
Recent studies have attempted to characterize the defense response of insects to Bt or Cry intoxication by proteomic analysis and transcriptional profiling approaches [30][31][32][33][34][35][36][37]. Moreover, these types of analyses have also been used to gain insight into the mode of action of Bt Cry toxins by comparing the transcriptional profiles of resistant and susceptible insects [38][39][40][41]. In this study, microarray technology was employed to characterize the defense response of S. exigua to Vip3Aa intoxication by monitoring gene expression levels after treatment with a sublethal dose of the insecticidal protein. A custom microarray containing more than 29,000 unigenes from a S. exigua transcriptome [42] was used for comparison of Vip3Aa-treated and non-treated larvae at two different times post-treatment. The results were validated by quantitative qRT-PCR of selected genes that we identified as having different expression patterns. The transcriptional profiling could allow for a better understanding of Vip protein action in the midguts of intoxicated larvae, and could provide clues about the larval midgut response mechanisms associated with oral Vip intoxication, useful information for future biocontrol strategies.

Materials and Methods
Insects, bacteria and toxin S. exigua larvae from the FRA colony kindly provided by M. López-Ferber (INRA, St.-Christol les Alés, France) [33], were used in the experiments. The colony was reared at 25°C, with a relative humidity of 70%, and a photoperiod of 16 h:8 h (light: dark), on an artificial diet [43].
The gene encoding the Vip3Aa protein (NCBI accession AAC37036) cloned into the pMaab9 plasmid, was kindly supplied by Bayer BioScience N. V. (Ghent, Belgium). The Escherichia coli WK6 strain was used as the expression host strain. For protein production, E. coli cultures were induced with 1 mM IPTG. Cells were pelleted by centrifugation at 8,800 g at 4°C for 30 min, frozen at -20°C, and subsequently lysed by a 30 min incubation at 37°C in 20 mM phosphate buffer (pH 7.4) containing 0.5 M NaCl, 3 mg/ml lysozyme, 10 µg/ml DNAse, and 0.1 mM PMSF. The lysate was sonicated on ice and centrifuged at 27000 g at 4°C for 30 min. The supernatant (containing the Vip3Aa1 protein), was filtered through 45 µm filters and stored at -20°C until use in the feeding experiments. The concentration of Vip3Aa toxin in the supernatant was determined by densitometry after SDS-PAGE electrophoresis, using the 1D Manager Software (TDI, Madrid, Spain). The E. coli control strain was cultured and processed in the same manner as performed for the Vip3Aa-producing E. coli strain.

Treatment of S. exigua larvae with Vip3Aa
To synchronize the insects, late third instar S. exigua larvae (L3) were selected the day before the feeding experiments.
The following day, approximately 16 h after the selection, the newly moulted L4 larvae were separated and exposed, individually, to a dose of Vip3Aa of 111 ng/cm 2 , which produced a 99% growth inhibition ( Figure S1). As a control, the filtered supernatant obtained from the E. coli control culture, diluted to the same degree as was required to dilute the Vip3Aa-containing supernatant, was used to feed the larvae.
Three independent biological replicates of the Vip3Aa feeding experiments were performed. In each, sixteen larvae were exposed to the supplemented food for 8 h and 24 h. After these times, only larvae that had fed (as determined by observing the food bites) were selected for midgut dissection. At least seven larvae were used for each time point. Midguts of the larvae from each treatment (8 h or 24 h) were pooled for further processing.

Microarray design
A 44K Agilent oligonucleotide chip was designed using the eArray application from Agilent (Agilent Technologies, Wilmington, DE, USA) and included 29,102 unigenes from S. exigua (GEO Acc. No. GPL17775). The sequences of S. exigua were obtained from an S. exigua transcriptome sequencing project, described elsewhere, specifically designed to be enriched in pathogen-induced genes [42]. Most of the unique assembled sequences (unigenes) in the microarray were represented by two 60-mer oligonucleotide probes, designed to target different sections of each unigene.

RNA purification, labeling, and hybridization
The RNA from S. exigua midguts was purified using RNAzol reagent from Molecular Research Center, Inc. (Cincinnati, OH, USA), and purified using the RNAeasy Kit (Qiagen GmbH, Hilden, Germany) following the protocols provided by the manufacturers. The quality of RNA was assessed with an Agilent 2100 Bioanalyzer using the Eukaryote Total RNA Nano protocol.
Agilent One-Color Spike-in Mix was added to the purified RNA and 600 ng of total RNA was used for complimentary RNA (cRNA) synthesis. From the resulting cRNA, 165 µg were fluorescently labeled with cyanine-3-CTP 1, fragmented, and hybridized to the S. exigua microarray chip following the One-Color Microarray-Based Gene Expression Analysis (Quick-Amp labeling) protocol from Agilent, as described in Jakubowska et al. [44]. RNA labeling and hybridization, as well as array scanning and data extraction, were performed by the Microarray Analysis Service of the Principe Felipe Research Centre (CIPF), Valencia, Spain. Microarray results are available at NCBI, GEO Acc. No. GSE51195.

Microarray data analysis
S. exigua microarray chips were scanned using a G2505B Agilent scanner and data were extracted using Agilent Feature Extraction 9.5.1 software. Before data analysis, hybridization quality control reports were verified for being correct. Data analysis was performed using free Babelomics 4.3 software (available online: http://babelomics.bioinfo.cipf.es/) [45]. First, all arrays were normalized using spike-ins and quantile normalization methods. Normalized arrays for the samples treated with Vip3Aa were compared to the normalized arrays for the controls at the two time points (8 h and 24 h after treatment), and expressed as fold-change in the expression. Fold-change is defined here as a difference in log 2 values between treated and control sample, and later reported as linear ratios. The thresholds of fold-change > 2 and p-value < 0.05 were applied to consider a gene as regulated compared with control. Previous studies showed that fold change values together with a nonstringent statistical p-value cutoff provided increased consistency in the analysis of Gene Ontology terms and pathways affected [46] and generate more reproducible results [47][48][49][50]. Therefore, the false discovery rate (FDR) [51] has not been used. The FDR values in this study ranged from 0.002 (24 h of Vip3Aa treatment) to 0.126 (8 h of Vip3Aa treatment). It is worth noting that the numbers of regulated unigenes before applying statistical t-test were very similar to the numbers of unigenes when t statistics was included, which suggested a high repeatability of the biological replicates.
Annotations of the unigenes were performed using Blast2GO [52]. Functional clustering of regulated genes, while maintaining the applied thresholds, was performed using DAVID version 6.7 software [53]. The 1,470 regulated unigenes with homology to Bombyx mori genes that were admitted by DAVID, were then analyzed using the B. mori gene list as a background for functional enrichment analysis. Resulting clusters were ranked according to the Enrichment Score, which is the overall score for the whole group of terms and is calculated based on the EASE enrichment scores of all members of the group. The EASE enrichment score was calculated using the Fischer Exact test with the p-value threshold set to 0.05.

Microarray data validation by quantitative real-time polymerase-chain reaction
To confirm the microarray data, 19 regulated genes were validated by quantitative PCR (qRT-PCR). Primers for the analysis (Table S1) were designed using Primer Express software (V 2.0.0, Applied Biosystems, Foster City, CA, USA) and verified in silico using the GenosysOligoMail ver. 2.0 program (Genosys, Sigma-Aldrich, TX, USA). The suitability of the primers was further assessed in the qRT-PCR working conditions. The ATP synthase subunit C house-keeping gene was used as an internal control for normalization of the samples [29,33,44,54]. The cDNA was synthesized from 1 µg of RNA treated with DNase I (Invitrogen, Life Technologies Corporation, CA, USA) by reverse transcription using oligo-d(T) primer and SuperScript II Reverse Transcriptase (Invitrogen) according to manufacturer protocols. The qRT-PCRs were carried out in an ABI PRISM® 7000 Sequence Detection System (Applied Biosystems). Reactions were performed using Power SYBR Green PCR Master Mix (Applied Biosystems) with 5 µl cDNA template (12.5 ng) in a final reaction volume of 25 µl. Forward and reverse primers were added to a final concentration of 300 nM. Expression ratios were calculated using the formula ΔΔCt = (Ct gene of interest, treated -Ct reference gene, treated )-(Ct gene of interest, control -Ct reference gene, control ). The final and absolute gene regulation values (or fold-change values) were obtained as 2 -ΔΔCt , and were expressed as 2 -ΔΔCt for upregulated unigenes and as 1/(2 -ΔΔCt ) for down-regulated unigenes, thus allowing a better understanding of downregulation intensities.

Microarray data analysis
The transcript profiles of Vip3Aa-treated larvae as compared to control larvae were assessed using a custom S. exigua microarray containing 38,174 probes representing more than 29,000 S. exigua unigenes. Microarray probes were designed based on the unigenes from a S. exigua transcriptome designed to be enriched for pathogen-related genes [42]. A sublethal concentration of Vip3Aa protein (causing 99% growth inhibition) was used to elicit changes in gene expression after 8 and 24 h of treatment.
The data obtained showed that 5526 unigenes were transcriptionally regulated, representing 19% of all unigenes present in the array. This is high when compared with other analyses of transcriptional changes after microbial infection or Cry intoxication in lepidopterans (from 1 to 11%) [55][56][57], coleopterans (about 1%) [34], or dipterans (around 7%) [20]. This wide transcriptional change suggests a strong response of the organism to Vip intoxication, even taking into account the characteristics of the transcriptome represented in the microarray, aimed to be enriched in pathogen-induced genes [42].
The heat map generated from the microarray data ( Figure  1A) showed that the biological replicates grouped together, indicating the robustness of the results. In addition, the genes regulated at 8 and at 24 h grouped together and were separate from the controls. Analysis of the expression profiles over time, grouped the regulated genes into nine clusters. The clusters that included more than 500 genes are represented in Figure  1B and indicate that most of the regulated unigenes showed the same type of regulation at both time points, either upregulated (3,157 and 589 unigenes, Clusters 1 and 4 respectively) or down-regulated (1,715 and 1,304 unigenes, Clusters 2 and 3, respectively).The full list of unigenes belonging to each Cluster is provided in Table S2.
Of the 5,526 regulated unigenes, there were almost the same number of regulated genes after 8 h than after 24 h of intoxication (4,121 vs. 4,123, respectively; Figure 2). An overview of previous reports on time course transcriptional responses after intoxication with insecticidal proteins or bacterial feeding, shows that the response depends on the insect and on the toxic agent [30,31,34,[55][56][57][58]. When the upand down-regulated unigenes were considered separately, the number of regulated genes that resulted was similar at both time points (2,243 and 2,323 up, and 1,878 and 1,800 down, after 8 h and 24 h respectively; Figure 2). These results are in contrast with those described in experiments of intoxication with Bt Cry toxins where the proportion of down-regulated genes was greater than that of up-regulated genes [30,34]. However, our results resembled the ones observed in insects fed with whole bacteria or virus [20,56,57]. It should be noted that the levels of down-regulation achieved by the most repressed genes (around 660-fold) were greater than the levels of overexpression achieved by the most up-regulated genes (around 160-fold).
The distribution of up-and down-regulated unigenes according to length of treatment is summarized in Figure 3.

Global effect of Vip3Aa treatment on S. exigua midgut
Around 40% of the up-regulated and about 60% of the downregulated unigenes showed homology to known genes from public sequence databases as assessed by Blast2GO annotation (Figure 2). The list of unigenes with homologies in databases, regulated at both 8 h and 24 h, or regulated only at 8 h or only at 24 h following Vip3Aa treatment, is available in Table S3.
The microarray data were validated by qRT-PCR. From the most strongly regulated, we chose eleven up-regulated and eight down-regulated unigenes (underlined unigenes in Table  1) to confirm the microarray results. The validation was also performed for repat1 (unigene Se_U19481), since it has been reported to be involved in the response of S. exigua to Bt Cry intoxication [29] and, in the present study, was found specifically regulated after 8 h of Vip3Aa feeding exhibiting 41fold up-regulation (Table S3). The up-regulated genes selected for the validation included homologous to genes involved in response to pathogens or defense, transport of proteins and metabolism. The down-regulated genes selected for validation included the three unigenes most strongly repressed (coding for unknown proteins), and unigenes homologous to genes coding for peritrophic matrix proteins and metabolism related enzymes. The qRT-PCR validation results are summarized in Figure 4. The expression values obtained by qRT-PCR confirmed the microarray results at 8 h and 24 h, thus confirming their respective profiles of expression over time. A comparison of the mean values of expression obtained from the microarray and from qRT-PCR is shown in Figure S2.
To determine the type of biological processes and pathways that were affected by Vip3Aa intoxication, the functional clustering toolbox DAVID v.6.7 was used. The analysis resulted in the identification of 21 functional clusters. Table 2 summarizes the clusters with enrichment scores higher than 0.50 (the remaining clusters are reported in Table S4). The cluster with highest scoring (Cluster 1) included genes encoding hormone-binding, odorant binding and juvenile hormone-binding proteins. Homologous genes had been also found regulated in studies on Spodoptera frugiperda [31], Choristoneura fumiferana [30] and B.mori [56] intoxicated with Cry1Ca, Cry1Ab or Bacillus bombyseptieus, respectively. Recently, it was found that an odorant binding protein (related to the immune system response) in the coleopteran Tribolium castaneum, was specifically overexpressed after exposure to toxic Cry proteins [37].
The second cluster consisted of C-type lectins and lectin-like proteins, which are sugar binding proteins involved in biological recognition pathways that are involved in the immune system [21,59,60]. The third cluster was comprised of lipocalin-related proteins, which are transporters of small hydrophobic proteins involved in many biological processes like the immune response and pheromone transport. The remaining clusters included proteins involved in pattern recognition, such as immunoglobulin-like proteins (Cluster 4); and proteins involved in biosynthesis, transport, and metabolism, such as carboxylesterases, ion binding proteins, cytochrome P450 and redox proteins (Clusters 6, 9, 10, 13, 14, 15, and 7, that grouped the redox reaction enzymes and that is the cluster with We also determined the Gene Ontology (GO) term assignments for the up-regulated and the down-regulated unigenes in the Biological Process and the Molecular Function domains, at level 3 ( Figure 5). To simplify, only classes represented by more than 1% of the total amount of regulated sequences were included. At the Biological Process domain ( Figures 5A and 5B), one of the most noteworthy differences between the up-and down-regulated unigenes was observed in the "metabolic process" category, the most abundant in the S. exigua transcriptome [42], which had about twice as many representatives in the down-regulated genes (335) than in the up-regulated genes (170). Consistently, at the Molecular Function domain, the most represented category was "hydrolase activity" in both up-and down-regulated groups, as had also been detected in the transcriptome analysis of lepidopteran immune-activated larvae [42,61], but with the particularity that we observed more assigned unigenes in the down-regulated sequences ( Figure 5C and 5D). The same observation was found for the "serine-type endopeptidase activity" or "monooxygenase activity" categories. The downregulation of these processes could be a consequence of the deceleration of metabolism after Vip feeding, and is likely linked to the high growth inhibition effect (99%) and feeding cessation. This is in agreement with genome-wide gene expression analysis performed on insect midguts after Cry toxin or bacterial feeding, which point to a general downregulation of digestive proteins, such as lipases or proteases [20,30,31,34,37].
Interestingly, at the Biological Processes domain, a large difference was observed between the distribution of up-and down-regulated genes in the "immune and defense response" category (157 genes versus 38 genes, Figures 5A and 5B). These results agree with previous gene expression studies, which showed the up-regulation of genes involved in detoxification, stress response, immune system and epithelial renewal, after bacterial infection or toxin challenge [20,30,34]. Consistent with these GO analyses, an overview of the regulated unigenes (listed in Table S3) shows a distribution of roles for the up-and down-regulated genes. Table 1 summarizes the 20 most induced and the 20 most repressed unigenes at both treatment times. The strongest up-regulated unigenes included genes encoding immune-related and response to abiotic-factors (mainly from the repat family), hormone modulation (e.g. JH binding protein), and detoxification (e.g. glutathione S-transferase) proteins. The strongly repressed unigenes were mainly genes encoding digestive enzymes (e.g. serine proteases) and proteins involved in oxidoreductase reactions (e.g. cytochrome P450). Interestingly, among the most down-regulated genes we found chitin deacetylase, an enzyme that increases the permeability of the gut peritrophic membrane (PM) [62]. This enzyme was also down-regulated in Helicoverpa armigera and in S. exigua after baculovirus infection, and this regulation was explained as a mechanism of the insect to reduce its susceptibility to the ingested pathogen by decreasing PM permeability [62]. A similar mechanism could decrease the amount of Vip3A toxin passing through the PM and binding to the midgut epithelial cells. However, up-regulation of this type of enzymes was described in the coleopteran Tenebrio molitor intoxicated with Cry3A protoxin [34]. This could be due to differences between lepidopterans and coleopterans, since the latter can survive Cry intoxication for weeks without obvious signs of paralysis [34], or to the biological role of each chitin deacetylase protein.
In addition to the 40 most regulated unigenes, Table 1 also includes three unigenes of unknown function that exhibit the strongest values of down-regulation found in this study (from 332-fold to 657-fold, at 24 h). These three unigenes were further investigated and manually assembled into a single contig of 672 nucleotides (GeneBank Acc. No. KF601929), which showed homology to a B. mori gene and a H. armigera EST. Alignment of the putative encoded proteins (named REVIP in S. exigua because was detected in REsponse-to-Vip intoxication) is shown in Figure S3. No homologues of this protein were found in other insect orders.

Immune-related genes regulated after Vip3Aa ingestion
The S. exigua larvae transcriptome represented in the microarray was specifically aimed to be enriched for pathogenrelated genes and, therefore, offers the potential to detect variations in expression level of many immune-related genes and pathways with greater accuracy. When describing the larval transcriptome, Pascual et al. [42] divided the immunerelated genes into three categories: (a) genes involved in pathogen recognition, (b) genes coding for components of the main immune-related signaling pathways (Toll, IMD, JAK/ STAT, and p38 MAPK) and melanization processes, and (c) genes of antimicrobial effectors induced by these immuneactivated pathways. In the present study, the regulation of such genes, and of other immune-related genes (such as the AMP spodoptericin or unigenes homologous to the lepidopteran immune-related genes Hdd11 and Hdd23), after Vip3Aa intoxication has been screened. The results are summarized in Table 3.
As mentioned above, we have observed a general upregulation of immune-related genes. Among the different types of genes belonging to this group, we have found two notable features: (a) genes involved in pathogen recognition, melanization, and antimicrobial effectors were regulated after Vip3Aa intoxication; and (b) genes encoding components of the main immune related signaling pathways, such as Toll, IMD, JAK/STAT, MAPK p38, and JNK MAPK pathways, were not regulated.
We have not observed a clear pattern of regulation for the pathogen recognition proteins PGRP, ßGRP, lectins and hemolin in midguts. In lepidopteran hemolymph these proteins are constitutively present and trigger pathogen responses as phagocytosis, nodule formation, encapsulation, melanization and synthesis of AMPs [59,60]. In this study, after Vip3Aa feeding some unigenes homologous to PGRP precursor, ßGRP 1, 2a, and 2, and scavenger receptor and C-type lectins (carbohydrate recognition proteins that form Cluster 2 in Table  2) were only slightly overexpressed. However, unigenes with homology to lectins 1 and 5, PGRP-C and -B (to a lesser extent), and ßGRP3 were down-regulated, especially after 8 h treatment. Neither hemolin, nor Gram negative recognition proteins (GNRP) homologous genes, both present in the microarray, were found regulated in midguts after Vip3Aa intoxication.
Melanization occurs regularly in the midgut and hindgut of lepidopterans such as B. mori, to regulate fecal microbiota [63], and is permanently induced in the hemolymph of Bt-tolerant strains of Ephestia kuehniella [64] and in Cry1Ac resistant strains of H. armigera [65]. In this work, we also observed the regulation (generally overexpression) of several genes encoding members of the melanization cascade after Vip3A intoxication. PPO activating enzyme, a necessary component for the initiation of the melanization pathway [60], was upregulated. Serpins, proteinase inhibitors that regulate the serin protease cascade for melanization, as well as other immune proteinase pathways in insects [21], were also regulated, but in this case with different behavior according to their family. Serpins 2, 4, 14, 31 and, to a greater extent, 6 and 8 were upregulated. This was similar to results observed for serpin 2 in C. fumiferana after sublethal Cry1Ab intoxication [30], and for serpin 4 in Aedes aegypti intoxicated with Cry11Aa [36], and in S. frugiperda intoxicated with Cry1Ca [31]. In contrast, serpins 5 and 7 were down-regulated. The up-and down-regulation of serpins was also found in B. mori intoxicated with B. bombyseptieus [56]. The function of the different serpins and their correlation with melanization processes, protein agglutination, and regulation of other immune related signaling pathways remains unclear [21].    *. Target Name represents the mean from two probes, unless "sense" or "anti" is indicated in which case only one probe is represented.
**. Sequences with no protein homologous in databases but highly regulated or related with lepidopteran immune reaction.   The clearest pattern of regulation of immune-related genes was observed for those genes coding for antimicrobial effectors, which include the AMPs and lysozymes. After Vip3Aa feeding, unigenes representing all the families of AMPs reported by Pascual et al. [42], were detected up-regulated: cecropins, gloverins (specifically related to the Toll pathway in S. exigua [27]), diapausins, moricins, cobatoxins, attacins and lebocins (the last ones reported as not transcriptionally stimulated in B. mori midguts after Gram-positive or Gramnegative feeding [66]). In addition, spodoptericins (called defensins in other Lepidoptera) were found up-regulated. Only gallerimycin (an AMP found up-regulated after bacterial challenge in H. armigera [67]), although present in the microarray, was not significantly regulated. Moreover, several lysozymes were also up-regulated after Vip3Aa intoxication.
Up-regulation of AMPs induced by bacterial infection, bacterial feeding (including Bacillus sp), or by Cry toxins, has been previously reported in S. exigua [26][27][28] and in other lepidopteran [16,30,32,48,49,[66][67][68]. In general, the regulated AMPs exhibited an increase in transcription with time following the same pattern previously described for S. exigua attacin [28]. The AMP exhibiting the greatest induction was diapausin (up-regulated from 17-to 45-fold depending on the unigene and the time of exposure to Vip3Aa), followed by lebocin and gloverin. The latter has been described as being related to the Bt response in S. exigua [27]. Cecropin was the first AMP described in S. exigua, and its transcription in the fat body of fifth instar larvae was enhanced after injection of dead bacteria [26]. The different cecropins are grouped into six subfamilies in this Lepidoptera [42] and, although all of them were represented in the microarray, only members of subfamilies B and D were found to be regulated by Vip3Aa feeding.     Interestingly, we did not observe regulation of the unigenes representing the main immune-related signaling pathways (Toll, IMD, JAK/STAT,p38 MAPK, and JNK MAPK), in an apparent disagreement with the up-regulation observed for AMPs since some of these pathways are responsible for inducing their synthesis. However, the initial activation of these pathways relies mainly on their post-translational modification [19,69]. It would be interesting, in further studies to determine the effect of long-term exposure to Vip3A toxins on the transcriptional regulation of these genes. In fact, Cancino-Rodezno et al. found that MAPK p38 was phosphorylated in Manduca sexta larvae after one hour of Cry1Ab exposure, but they only begin to observe an increase in the expression level of MAPKp38 gene after 24 h of the exposure to the toxin [25].
Vip3Aa feeding also revealed the overexpression of other associated immune-related genes encoding proteins from the Hdd family, such as Hdd11 and Hdd23. For example, hdd11 was found to be highly induced in midguts of a Cry1Abresistant strain of Diatraea saccharalis [39], and Hdd23 was up-regulated in Lepidoptera infected by virus or bacteria [70,71]. Another gene of this family, Hdd1, was also upregulated in midguts of T. ni immune-induced by bacterial feeding [16]. The functions of these Hdd proteins remain unclear.

Bacillus thuringiensis-related genes
The most known insecticidal proteins of Bt are Cry toxins. They have been used, either alone or in combination with Bt, as insecticides for more than 70 years. Despite this long use as biological insecticides, their mode of action has not yet been fully elucidated. To date, it has been demonstrated that Cry proteins are pore forming toxins, and that certain insect midgut proteins are required for them to exert their toxicity [2,10]. The Bt Vip proteins are also pore-forming toxins [12], of which the mode of action is yet unknown. To gain insight into the possible similarities in the intoxication response mechanisms of these two types of Bt toxins, genes encoding proteins linked to Bt or Cry tolerance or resistance, as well as genes involved in the mode of action of Cry and Vip, were sought between the unigenes regulated in response to Vip3Aa feeding.
To date, two proteins have been implicated in the mode of action of Vip3A in Lepidoptera: an X-tox-like protein [72], and the ribosomal protein S2 [73]. Nine and two unigenes *. Target name represents the mean from two probes, or represents only one probe if "sense" or "anti" is indicated.
**. Unigenes revealed by more than two targets. In these cases, the Target Name in the respectively with homology to these protein genes were present in the array but their expression was not altered after Vip3Aa feeding. The screened genes related to the Cry toxins mode of action in Lepidoptera included midgut membrane-associated proteins (such as aminopeptidases-N, alkaline phosphatases, cadherin, ABC transporter), intracellular G protein, adenylate cyclase, and protein kinase A [9,74]. Other genes encoding proteins implicated in resistance to Bt or to Cry proteins in Lepidoptera, were also investigated. For example, arylphorin (a storage protein related with immune response elicited by bacterial feeding in T. ni, and related with Bt resistance in S. exigua [16,33]), members of the REPAT family (correlated with Bt resistance in S. exigua [33]), and hexamerin or lipophorin (proteins involved in resistance to Cry1Ac and in tolerance to Cry1Ac and Cry2Ab in H. armigera [65,75]). Midgut proteases, which are not only involved in the activation of Bt toxins, but also in many other processes (e.g. digestion), were not included in this screening. Table 4 summarizes the Bt-related unigenes that were found to be regulated.
After Vip3Aa intoxication, only slight differences in expression (in general overexpression) of some of the genes involved in Bt mode of action were found. This resembles what has been found in studies on whole transcriptional profiles of insect midguts after sublethal Cry intoxication in insects as C. fumiferana [30], S. frugiperda [31], and the coleopteran T. molitor [34], that did not show clear regulation of the putative Cry receptors. Indeed, in the case of C. fumiferana some APNs were found down-regulated during early times post-intoxication (5 h) [30].
Following Vip3Aa intoxication, unigenes homologous to the Cry protein receptor cadherin, were near the cutoff threshold of 2-fold regulation at both time points. The homologues of the ABC transporter (also described as a Cry receptor) showed slight up-and down-regulations, as was described for B. mori infected with B. bombyseptieus [56]. Unigenes with homology to another family of Cry receptors, the GPI-anchored proteins (APNs and ALPs), were slightly up-regulated. Interestingly, the APN unigenes that were found regulated did not show homology with the five main classes of S. exigua midgut APNs described as related with Bt [76]. Regarding the genes not directly related to the binding of the insecticidal protein, there were slight changes in the expression of unigenes homologous to G-protein and adenylatecyclase after Vip3Aa feeding. The exceptions were some down-regulated unigenes homologous to adenylate cyclase. The lack of significant changes in the transcription levels of all the Bt-mode of action related genes may suggest that they are not involved in the Vip mode of action. Alternatively, it may also indicate that, if the mode of action of Cry and Vip toxins share biochemical processes, the mechanisms of defense against Vip toxins would not rely on transcriptional regulation of the members involved in them.
Apolipophorin was found slightly down-regulated during the entire course of Vip3Aa intoxication. This contrasts with studies performed in immune-induced lepidopterans (such as T. ni fed with bacteria [16]), or with coleopterans (such as T. molitor fed with Cry3Aa [34] or T. castaneum fed with Cry3Aa or Cry23Aa/ Cry37Aa [37]), where apolipophorin III was up-regulated.
Indeed, Cry1Ab-resistant D. saccharalis [39] showed constitutive up-regulation of an apolipophorin precursor. Typically, lipophorins I and II are considered insect hemolymph proteins involved in lipid transport, but lipophorin III has been also implied in defense mechanisms by clotting [21,61]. In H. armigera, it has been shown that, in general, lipophorins can bind to Cry1Ac and Cry2Ab monomers and sequester the toxins [75]. The nature of the down-regulation of apolipophorins detected after Vip3Aa feeding is likely due to their role as lipid transporters, and their low levels during feeding could be a consequence of the general reduction of metabolic processes caused by feeding cessation.
Arylphorin is a hexamerin related to the immune response because of its mitogen activity, which is associated with cell proliferation and the replacement of damaged cells [77]. Arylphorin was found to be up-regulated in a Bt-resistant S. exigua colony [33], in S. exigua after Bt intoxication [33], and in a Cry1Ab-resistant D. saccharalis strain [39]. In our experiments, arylphorin was down-regulated in Vip3Aa-treated larvae. The different mode of action of Cry and Vip toxins could be the reason for these different observations, or, alternatively, it could be that the regulation observed in arylphorin was independent of Bt, Cry or Vip intoxication.
REPATs are midgut infection-response glycoproteins that were first discovered in S. exigua, up-regulated after treatment with different Bt toxins or with baculovirus [29], and that were overexpressed in an insect colony resistant to Bt formulations [33]. Recently, up to 46 members of this family have been reported in S. exigua, and homologous sequences have been found in other species [54]. After Vip3Aa feeding, a broad response of repat genes was detected. Unigenes with homologies to 29 different repat genes were found regulated, which pointed to a strong involvement of these genes in the midgut response to Vip3Aa (Table 5). In general, repat unigenes were overexpressed, exhibiting about the same level of up-regulation at 8 and 24 hours. A clearly different behavior was exhibited by repat42, repat43, repat46 and another repat unigene with low homology to repat14 (e-value = 0.01, which indicates it is likely a new member of the family), which were down-regulated.
Almost all repat members that were significantly regulated belonged to REPAT class α (groups I and II), and only five of them belonged to REPAT class ß (to groups III, IV, and VI) [54]. The general up-regulation of the repat unigenes from class α together with repat45, and the down-regulation of class ß repat members 42 and 43, coincide with the transcriptional profiles reported in S. exigua treated with Cry1C by Navarro-Cerrillo et al. [54]. Although the role of REPAT proteins remains unclear, the large number of repat members regulated, their homology to transcriptional activators in other species of Lepidoptera, and their ability to form heterodimers and translocate into the nucleus [78] seems to point to a possible role in the transcriptional activation of several sets of genes in response to physiological changes in the midgut produced by Vip3A or Cry1C intoxication.
In summary, in this work, the overall transcriptional response of the midgut of a lepidopteran such as S. exigua exposed to the toxic action of Vip3Aa has been described for the first time.  A comprehensive response characterized by the overexpression of immune-related unigenes and repat family unigenes, was detected together with other singular regulations (e.g. up-regulation of hormone-binding protein unigenes or down-regulation of serin proteases and chitin deacetylases). The data reported here may contribute to a better understanding of the interaction of the insect midgut with the Vip3Aa toxin, helping to unravel the processes underlying Vip toxicity. This information may allow for the design of more effective pest-management strategies using this toxin (alone or in combination with other insecticidal agents), e.g. pointing to larval genes involved in resistance mechanisms, or to targets for RNAi mediated gene disruption. In short, it may provide new tools for crop protection. Figure S1. Growth inhibition dose-response curve of S.exigua newly moulted L4 larvae challenged with Vip3Aa.

Supporting Information
Growth inhibition values were calculated following Herrero et al. [79]. Four biological replicates of the experiment (using 8 larvae per dose) were performed. (TIFF)