Dynamics of Membrane Potential Variation and Gene Expression Induced by Spodoptera littoralis, Myzus persicae, and Pseudomonas syringae in Arabidopsis

Background Biotic stress induced by various herbivores and pathogens invokes plant responses involving different defense mechanisms. However, we do not know whether different biotic stresses share a common response or which signaling pathways are involved in responses to different biotic stresses. We investigated the common and specific responses of Arabidopsis thaliana to three biotic stress agents: Spodoptera littoralis, Myzus persicae, and the pathogen Pseudomonas syringae. Methodology/Principal Findings We used electrophysiology to determine the plasma membrane potential (Vm) and we performed a gene microarray transcriptome analysis on Arabidopsis upon either herbivory or bacterial infection. Vm depolarization was induced by insect attack; however, the response was much more rapid to S. littoralis (30 min −2 h) than to M. persicae (4–6 h). M. persicae differentially regulated almost 10-fold more genes than by S. littoralis with an opposite regulation. M. persicae modulated genes involved in flavonoid, fatty acid, hormone, drug transport and chitin metabolism. S. littoralis regulated responses to heat, transcription and ion transport. The latest Vm depolarization (16 h) was found for P. syringae. The pathogen regulated responses to salicylate, jasmonate and to microorganisms. Despite this late response, the number of genes differentially regulated by P. syringae was closer to those regulated by S. littoralis than by M. persicae. Conclusions/Significance Arabidopsis plasma membranes respond with a Vm depolarization at times depending on the nature of biotic attack which allow setting a time point for comparative genome-wide analysis. A clear relationship between Vm depolarization and gene expression was found. At Vm depolarization timing, M. persicae regulates a wider array of Arabidopsis genes with a clear and distinct regulation than S. littoralis. An almost completely opposite regulation was observed between the aphid and the pathogen, with the former suppressing and the latter activating Arabidopsis defense responses.


Introduction
Plants are attacked by a multitude of organisms, like insects, microbes and fungi, which all infer a biotic stress. As plants are sessile organisms and cannot escape their predators they have evolved diverse mechanisms to react specifically to each attacking biotroph. Chewing herbivores, like Spodoptera littoralis, consume leaves by continuously clipping off and ingesting small pieces of tissue reducing both photosynthetic capacity and biomass of fed plants [1][2][3][4][5]. Aphids, like Myzus persicae, are sap-sucking insects that remove plant nutrients, elicit plant responses that are deleterious to plant productivity and alter the mass flow of phloem contents, resulting in changes in source-sink relationships [6][7][8][9][10]. Phytopathogenic bacteria of the genus Pseudomonas colonize the leaf surfaces of plants without causing disease [11]. Pseudomonas syringae multiplies in the plant cell apoplastic intercel-lular spaces and remains extracellular triggering plant defenses aimed to restrict bacterial growth [12].
Upon all of these biotic interactions with plants, it is crucial to understand how plants dissect and convert these different stress signals into appropriate physiological reactions.
The earliest event that is detectable as a consequence of leaf damage is depolarization of the plasma transmembrane potential (V m ), followed by a cascade of biochemical and molecular events including protein phosphorylation, activation of signaling cascades and, eventually, gene expression and translation [12][13][14][15][16][17]. Both S. littoralis direct herbivory and the insect's oral secretions have been demonstrated to induce a fast mesophyll cell V m depolarization of Arabidopsis [18] and other plant species [1,[19][20][21], whereas a significant V m depolarization is observed at almost every stylet puncture of the plant plasmalemma during M. persicae phloem feeding [22]. In plant-pathogen interactions, V m depolarization is a reliable early indicator of leaf hypersensitive response (HR) [23] A variety of experimental methods have been employed to study the complex interactions of Arabidopsis and aphid herbivores, including measurements of the transcriptional responses [8,[24][25][26][27], whereas microarray-based genome-wide transcriptomic analyses have been performed in several plant species, including Arabidopsis thaliana, upon herbivore attack by Spodoptera spp., [28][29][30][31]. Although the exact nature of the systemic acquired resistance (SAR) signal in Arabidopsis after localized infection by avirulent P. syringae remains complex and has been a matter of debate [32,33], the transcriptional changes associated with basal defense to live bacteria and the contribution of specific elicitors/effectors to the regulation of the basal defense transcriptome and other host physiological processes have been thoroughly studied [12,[34][35][36].
Although many commonly induced or suppressed defenserelated genes have been identified in plants infested with chewing or phloem-feeding insects, and bacterial pathogens, there is considerable difference in the transcriptomic response of infested plants to different insects or bacteria. In the dazzling diversity of possible differential plant responses, the most difficult aspect is to assess whether a common response exists and to which extent each pathogen or herbivore differentially expresses and regulates defense response genes. Timing appears important in the interplay among the multiple plant responses to herbivores [37] and pathogenic microorganisms [38]. The aim of this work was to use a common physiological response to the herbivores M. persicae and S. littoralis and the pathogen P. syringae and, i.e. the leaf V m depolarization, as a time point for a comparative genome-wide analysis of gene expression and regulation in Arabidopsis, when attacked by different biotic agents. The obtained results should complement other studies and provide a useful resource for future study of plant multitrophic interactions.

Results
S. littoralis, M. persicae and P. syringae induce the same strong V m depolarization in A. thaliana leaves but at different times Time-course measurements of V m in Arabidopsis showed that after S. littoralis herbivory a strong and rapid V m depolarization (with respect to mechanical damage) occurs after a few minutes from the herbivore wound, with recovery of the V m between 5 and 6 h ( Figure 1). When Arabidopsis was fed by M. persicae, almost the same extent of V m depolarization was observed (P.0.05), but the timing of V m depolarization peaked between 4 and 6 h, returning near to the basal V m value between 16 and 24 h (Figure 1). A remarkable delay in V m depolarization was observed when Arabidopsis leaves were infected by the avirulent strain of P. syringae. Even in this case V m depolarization was not significantly different (P.0.05) from the values observed after S. littoralis and M. persicae herbivory; however, the maximal V m depolarization occurred between 16 and 18 h from inoculation ( Figure 1). These results indicate that Arabidopsis responds to different biotic stress with a strong and transient V m depolarization and that the timing of this event depends on the kind of biotic stress.
Comparative gene expression at the time of V m depolarization in Arabidopsis leaves infested by S. littoralis, M. persicae and P. syringae Taking into account that different biotrophs, owing to their peculiar feeding or infesting behavior, can cause different amounts of damage or trigger plant responses at different time, samples for microarray analysis were taken at time points corresponding to the maximum V m depolarization for every biotic stress.
To identify transcriptional changes in Arabidopsis, total RNA was extracted from leaves wounded for 2 h by the herbivore S. littoralis, for 5 h by the phloem feeder M. persicae and after 16 h from inoculation of the pathogen P. syringae, these timings corresponding to maximal V m depolarization. As control for S. littoralis, leaves were mechanically damaged with a pattern wheel, whereas for M. persicae leaves were wounded with the tip of an electrophysiology microcapillary. Mechanical damage was done at the same extension as observed after herbivory. Control for P. syringae consisted of MgCl 2 leaf infiltration.
In order to evaluate robustly regulated sequences that are useful to evaluate the plant response to biotic stress, four biological replicates of each biotic stress (each consisting of several stressed leaves) were used for the gene microarray analysis. By using the stringent criteria described in material and methods (fold change $2, P#0.05), out of 38,463 sequences on the Agilent spotted slide, 190 genes fulfilled these stringent criteria for S. littoralis, 1840 genes for M. persicae and 416 for P. syringae. Among these genes, 23 were commonly regulated, whereas the comparative analysis between the biotrophs revealed the presence of 35, 38 and 152 co-regulated genes in the interactions M. persicae/S. littoralis, S. littoralis/P. syringae and M. persicae/P. syringae, respectively ( Figure 2). A few Arabidopsis commonly expressed genes are regulated upon M. persicae and S. littoralis herbivory and P. syringae infection The cluster analysis of commonly regulated genes shows that at the time of V m depolarization a common feature of Arabidopsis plants under biotic stress is a cluster of down-regulated genes including four UDP-glycosyltransferases (UGT73B2, UGT73B3, UGT73C1, UGT74E2) and a cluster of up-regulated genes made of Toll-Interleukin-Resistance (TIR-NBS-LRR class) transmembrane receptor (At5g45000), NADH pyrophosphatase (NUDX6) and heavy metal transport/detoxification superfamily protein (At5g26690) ( Figure 3 and Table 1). All other commonly regulated genes were differentially expressed and the cluster analysis showed a close linkage between P. syringae and S. littoralis, indicating a common up-regulation of the remaining genes, whereas the same genes were always down-regulated by M. persicae (Table 1 and Figure 3).
Commonly expressed Arabidopsis genes upon S. littoralis and M. persicae herbivory show different regulation patterns Most of the genes that were commonly regulated by M. persicae and S. littoralis herbivory showed opposite regulation, with many genes showing down-regulation upon M. persicae feeding and upregulation after S. littoralis herbivory ( Figure 4 and Table 2).
Genes annotated for response to stimulus showed the same expression trends between the two herbivores for OPR2 and PRN1, which were commonly down-regulated, whereas up-regulation by S. littoralis and down-regulation by M. persicae was found for ethylene response factors (ERF1, ERF2), MYB113, GAD1 and a metal-phytosiderophore (YSL2). S. littoralis down-regulated alcohol dehydrogenase (ADH1), which was up-regulated by M. persicae.
An opposite gene regulation, with up-regulation caused by S. littoralis, was also found for several genes belonging to other GO categories like transferases (UF3GT, At3g50280), hydrolases (PME2, HAD, At4g29700), transcription factors (TFs) (ANAC090) and some genes responding to biotic stress (Table 2). On the other hand, an opposite regulation, with up-regulation caused by M. persicae, was found for an invertase/pectin methylesterase inhibitor (At2g37870), a GPI-anchored arabinogalactan peptide (AGP12), ADH1, and some other genes responding to biotic stress (Table 2). A common up-regulation was found for an essential component of the TAC1-mediated telomerase activation pathway (BT2), a PQloop repeat family protein/transmembrane family protein (At4g36850) and methionine sulfoxide reductase (ATMSRA2). Both insects down-regulated protease inhibitor/seed storage/lipid transfer protein (At2g37870), late embryogenesis abundant protein (LEA) (At5g53820) and armadillo/beta-catenin repeat family protein (At2g25130).
Commonly expressed Arabidopsis genes upon S. littoralis herbivory and P. syringae infection show the same regulation pattern The GO analysis of genes commonly expressed in Arabidopsis upon infection of the pathogen P. syringae and the chewing herbivore S. littoralis classified 35 of the 38 annotated genes. Most of them were related to response to stimulus, leading to a common response to heat ( Figure 5).
Commonly expressed Arabidopsis genes upon M. persicae herbivory and P. syringae infection reveal a remarkable opposite regulation The GO analysis of genes commonly expressed in Arabidopsis upon infection of the pathogen P. syringae and the phloem feeding M. persicae classified 135 genes among 152 annotated genes. Among these annotated genes, several were typical of responses to bacterium, response to other organisms, and response to stimulus ( Figure 6 and Table S1) Up-regulation after both M. persicae and P. syringae biotic stress was found for genes related to pathogenesis (NIMIN-1, PCC1, ACD6, PR5, PR13), a receptor like protein (RLP38), an S-locus lectin protein kinase family protein (At1g11330), glutaredoxin (GRXS13), a gene up-regulated in response to Hyaloperonospora parasitica (LURP1), genes involved in SA and JA metabolism (ANK,  A common down-regulation was found for a glucosyl transferase (UGT73B5), an ABC transporter (ABCB4) and a nodulin MtN21 family protein (At1g70260).
Differential gene expression in Arabidopsis upon S. littoralis herbivory reveals up-regulation of genes involved in transcription regulation and down-regulation of heat-shock proteins Among the 94 genes specifically regulated by S. littoralis herbivory, 78 were annotated and GO analysis isolated four gene categories: genes involved in the regulation of transcription, response to heat, ion transport and response to metal ion (Figure 7) In the regulation of transcription category, most of the genes were up-regulated by S. littoralis herbivory, including several genes involved in the circadian clock, whereas two basic helix-loop-helix (bHLH) (At4g20970, ATBHLH096) and a MYB like DNA-binding protein (AtGT-3a) were down-regulated (Table 4).
In the response to heat category, all expressed sHSPs and HSPs were down-regulated, whereas in response to metal ion, a copper chaperone CCH protein (At3g56240) was up-regulated and a ferritin (FER4) was down-regulated (Table 4). In the ion transport category, two Ca 2+2 ATPase (ACA12, ACA13) and a zinc transporter (ZIP7) were down-regulated, whereas a copper (At3g56240) and a sulfate (At1g77990) transporters were upregulated. Table S2 reports values of all other differentially expressed genes upon S. littoralis herbivory.
Differential gene expression in Arabidopsis leaves upon M. persicae herbivory reveals down-regulation of secondary metabolism and up-regulation of cell wall components Among the 1630 Arabidopsis differentially expressed genes upon M. persicae herbivory, 1423 were annotated and GO analysis classified 10 gene categories ( Figure 8).
Most of genes involved in fatty acid biosynthetic process were up-regulated (particularly a D-9 desaturase-like 5 protein, At1g06360), whereas a strong down-regulation was found for a P450-dependent fatty acid omega-hydroxylase (At3g48520).
Genes involved in cell wall modification and loosening included 11 expansin genes, 10 of which were up-regulated by M. persicae feeding (particularly EXP8), whereas an expansin-related gene (EXLB1) was down-regulated (Table S3).
M. persicae also strongly regulated genes involved in action of three important hormone classes: auxins, gibberellins and JA.
The full list of all other genes regulated by M. persicae feeding can be found in Table S4.
Differential gene expression in Arabidopsis leaves upon P. syringae infection shows up-regulation of defense genes including responses to JA and SA Among the 203 genes specifically regulated by P. syringae herbivory, 186 were annotated and GO analysis identified three gene categories: defense genes in response to bacterium, to fungus and incompatible interaction, and response genes to JA and SA ( Figure 9) Almost all differentially expressed genes in the defense response to bacterium were up-regulated with the only exception of a plasma membrane polypeptide (ATPCAP1) ( Table 5). Up-regulation was also found for all genes responsive to fungus (At2g14560, At4g12490, At2g39200) and in incompatible interaction (PDF1.2A, AOC2), the latter two genes being also involved in responses to SA and JA (Table 5). Genes responsive to SA included two kinases (CRK9, WAK1) and a phytoalexin-deficient 4 protein (PAD4). In response to JA, up-regulation was found for vegetative storage protein (VSP1), JA-ZIM-domain protein (JAZ10), MYB50 and arginine decarboxylase (ADC2).
The full list of all other Arabidopsis differentially expressed genes upon P. syringae infection is reported in Table S5.

Discussion
Dynamic V m depolarization responses in Arabidopsis induced by S. littoralis, M. persicae and P. syringae are different The primary candidate for intercellular signaling in higher plants is the stimulus-induced change in transmembrane potential (V m ) [39]. Here we showed that a common event upon biotic stress is plant V m depolarization, which occurs at different times according to the biotroph damage. No matter the nature of the biotroph, this event occurs at the same intensity, which in Arabidopsis corresponds to a V m depolarization of about 40 mV. We used the timing of maximal Vm depolarization as a time point to compare the genome-wide response of Arabidopsis thaliana to three known pests: two herbivores, M. persicae and S. littoralis and a pathogen, P. syringae.
Two questions arise from our results: 1) why V m depolarization is a common event? 2) Why V m depolarization occurrs at different times? To our knowledge this is the first report on time-course V m variations upon either M. persicae or P. syringae, whereas more data are available on S. littoralis herbivory-induced V m depolarization. Previous studies have demonstrated that S. littoralis herbivory triggers a calcium-induced potassium-dependent V m depolarization in wounded tissues [40] followed by a isotropic depolarizing wave that crosses all plant tissues, from shoot to roots [18,19,[41][42][43][44][45][46]. This effect was found to be strictly dependent on insect oral secretion, since both single and repeated mechanical wounding were not able to induce such response [1]. The lowering of the plant Vm is seen as a insect's strategy to reduced plant cell responses [47], thus it is conceivable to argue that the same strategy might also be used by both M. persicae and P. syringae. Elicitors are involved in Vm depolarization [48], and both M. persicae and P. syringae produce elicitors and effectors during biotic attack [16]. Furthermore, the same V m depolarization value recorded upon all three biotic attacks suggests the occurrence of a V m threshold to be reached for successful herbivory/infection.
The second question, why Vm polarization occurs at different times, appears to be associated with the mode of biotic damage. The fast clipping and consistent plant tissue removal by chewing herbivores evidently induces a ''quantitative'' response that is proportional to tissue damage; on the other hand, the stylet probing and the phloem feeding operated by the aphid induces a reduced amount of damage, that requires more time for a plant response. Finally, bacterial growth and tissue damage takes time, which appears to be proportional to V m depolarization.

Different patterns of gene regulation responses are induced by the two insects
Our study shows that the molecular mechanism of gene regulation response to S. littoralis and M. persicae are principally different. The evidence is that M. persicae regulated almost 10-fold higher number of genes than S. littoralis did. Moreover, genes responsive to both insects were regulated mostly in opposite direction.
A remarkable and specific response to S. littoralis herbivory at the time of V m depolarization was the down-regulation of several small heat-shock proteins (sHSP). sHSPs have a high capacity to bind non-native proteins, probably through hydrophobic interaction, and to stabilize and prevent non-native aggregation [49,50]. Increasing data suggest a strong correlation between sHSP accumulation and plant tolerance to a wide range of stresses [51]. To our knowledge, this is the first report on sHSP downregulation upon herbivory. Also, down-regulation of HSPs is quite rare although gene expression of sHSPs was down-regulated by SA in Arabidopsis and in tomato [52], whereas treatment by mitochondrial inhibitors and uncouplers down-regulated HSP, suggesting that mitochondrial functions are essential for sHSP synthesis [53].
With regards regulation of transcription, S. littoralis herbivory down-regulated a few genes with a regulatory role in mediating crosstalk between signaling pathways for biotic stress responses like DREB TFs [54] and two bHLH TFs. Most of the remaining TFs were up-regulated, including genes involved in the Arabidopsis circadian clock (REVEILLE 1, LHY1, CCA1 and LHY-CCA1-LIKE5) [55][56][57][58] and AtRL2, which is expressed in the funiculus of ovules and in embryos and is involved in the control of floral asymmetry [59]. Two zinc finger TF were also up-regulated and one of them, BBX32, has a native role in mediating gene repression to maintain dark adaptation [60]. An HD-Zip I TF, ATHB23, which is under the control of GA and other activators and is involved in establishing polarity during leaf development [61], was also up-regulated by S. littoralis herbivory. Altogether, these data indicate an effect of herbivory on the regulation of the circadian clock and the plant development. Two members of calcium-transporting ATPase (ACA12 and ACA13), that are dramatically induced upon exposure to pathogens [62], were down-regulated by S. littoralis herbivory. Ferritin (FER4) was also down-regulated and in Arabidopsis the absence of ferritin induces higher levels of reactive oxygen species, and increased activity of enzymes involved in their detoxification [63], a condition that is commonly found upon S. littoralis herbivory [64].
The response of Arabidopsis to M. persicae phloem feeding was clearly distinct from responses to the the chewing herbivore S. littoralis. After a few hours from feeding, and at the time of maximal V m depolarization, the aphid was able to suppress a consistent number of genes involved in flavonoid metabolism. UVexposed plants are damaged to a lesser extent by insect herbivores than non-irradiated plants [65], and UV radiation induces the accumulation of flavonoids in Arabidopsis [66]. In general higher concentration of flavonoids constrain aphid reproduction [67]; however, the generalist M. persicae was found to perform much better than the specialist Brevicoryne brassicae, when both aphids were feeding on Brassica oleracea exposed to UV-radiation [68]. The strong down-regulation of the genes coding for anthocyanidin synthases (also known as leucoanthocyanidin dioxygenases, At4g22870 and ANS) and dihydroflavonol reductase (DRF), all involved in flavonoid synthesis [69], was correlated with the downregulation of two MYB TFs (MYB75 and MYB123). MYB123 controls the biosynthesis of flavonoids in the seed coat of Arabidopsis and MYB5 was recently proposed to be partially redundant with MYB123 in this respect ( [70] and references therein). Moreover, the up-regulation of flavonoid transferase (UF3GT) induced by S. littoralis was in contrast with the downregulation of this gene after M. persicae feeding. These data suggest that M. persicae has the ability to suppress plant defenses based on flavonoids.
Fatty acid desaturases (FADs) introduce double bonds into the aliphatic tails of fatty acids and influence plant susceptibility to a wide variety of biotic stresses due in part to their critical role in the biosynthesis of the defense hormone JA [71]. Conversely, SA accumulation is enhanced by transient suppression of FAD [72]. Since aphid resistance has been observed in Arabidopsis and tomato mutants deficient in some FADs [73], the up-regulation of FAD-related genes upon M. persicae feeding indicates a possible aphid strategy aimed at lowering plant resistance.
Aphids have not been reported to induce detectable levels of JA, probably because of the induction of SA, which can interact antagonistically with JA signaling [26]. However, recent evidence suggests that SA accumulation may not be required for the repression of JA by whiteflies [74]. Our data show that M. persicae induction of several FADs and other genes related to fatty acid biosynthetic process was associated with a JA down-regulation. In fact, most of the genes related to JA biosynthesis (OPR2, OPR3, AOC3, CYP94B3, LOX3, LOX4) and signaling (several JAZs and JRs) were down-regulated, as were 8 out of the 12 members of the JAZ family of plant-specific proteins in Arabidopsis. These data are in agreement with a down-regulation of JA-related genes by phloem feeders [73,74]. JAZ proteins act as indirect transcriptional repressors of JA-dependent responses and their target to be repressed is MYC2, a key transcriptional regulator of JA signaling [75,76]. MYC2 is proposed to regulate the transcription of defense-related genes directly or indirectly through secondary TFs that mediate transcription of defense-related genes and accumulation of defense metabolites [76]. The fact that MYC2 positively regulates flavonoid biosynthesis and the down-regulation of MYC2 and several MYC2-dependent secondary TFs (e.g., CHI, ERF1) by M. persicae suggest that the aphid is able suppress plant secondary metabolism at several levels.
M. persicae feeding induced up-regulation of three families of early auxin-responsive genes, auxin/indole acetic acid (Aux/IAA), GH3, and small auxin-up RNA (SAUR), which are usually specifically induced by auxin within minutes [77]. Interestingly, two genes involved in main auxin biosynthetic pathway in Arabidopsis, YUCCA 2(At4g13260) and TAR2 (At4g24670) were up-regulated after M. persicae attack. Therefore, we can assume that up-regulation of auxin responsive genes under M. persicae attack might be linked to increased auxin concentration due to   activation of auxin biosynthesis through YUCCA-TAA1 pathway [78]. M. persicae also induced several classes of expansins. Expansin activity is mainly determined by transcription, which in turn is finely regulated by phytohormones [79,80] and increases after cell wall acidification [81]. M. persicae induction of expansins was correlated also with the up-regulation of GA metabolism. Two GA-regulated proteins (GASA5 and GASA6) showed a strong and opposite regulation. GASA5 was strongly down-regulated by M. persicae and the expression of this gene was also found to be inhibited by heat stress but unaffected by the application of exogenous SA, whereas the expression of NPR1, a key component of the SA-signaling pathway, was down-regulated by GASA5 overexpression [82]. GASA6, which was up-regulated by M. persicae, is thought to be a secreted peptide hormone precursor [83] and has also been characterized as an in vivo sugar marker gene [84]. Recently it was shown that GASA6 is severely repressed by a zinc-finger protein (AtTZF1) over-expression, a regulator connecting sugar, ABA, GA and peptide hormone responses [85]. The finding that several zinc-finger TFs were down-regulated by M. persicae correlates with the up-regulation of GASA6. The downregulation of gene coding for a-amilase (AMY1) and DELLA (RGL3) was in line with the up-regulation of genes involved in GA synthesis.
Two groups of transporters showed opposite regulation. Many secondary metabolites are transported into cells and across the plasma membrane via endogenous membrane transporters [86]. At the time of V m depolarization during M. persicae feeding, almost all peptide transporters were down-regulated. The peptide transporter family consists of electrochemical potential-driven transporters that catalyze uptake of their solutes by a cation-solute symport mechanism [87]. M. persicae regulated the gene expression of both yellow stripe-like (YSL) proteins and the OPTs. While YSL transporters are involved in metal homeostasis through the translocation of metal-chelates, OPT proteins play a role in plant growth and development being involved in long-distance metal distribution, nitrogen mobilization, heavy metal sequestration, and glutathione transport ( [88] and references therein). MATE efflux proteins were also down-regulated by M. persicae. In Nicotiana tabacum a tonoplast jasmonate-inducible MATE transporter was found to play an important role in the nicotine translocation by acting as a secondary transporter responsible for unloading of alkaloids in the aerial parts and deposition in the vacuoles [89]. Thus, the down-regulation of secondary plant products and the suppression of JA-induced genes might be another strategy of plant defense suppression by M. persicae. Finally, a large number of TFs expressed in response to chitin were down-regulated by M. persicae, including several WRKY, MYB and ERF.
Chitin, a polysaccharide composed of b-1-4-linked N-acetyl-Dglucosamine found in the exoskeleton of insects, has been shown or implicated as a signal in plant defense by inducing regulation of different ERF, MYB and WRKY TFs [90][91][92][93]. Among these TFs, WRKY53, MYB44 and RRTF1 were down-regulated by M. persicae. WRKT53 has been shown to be required to fully silence SA biosynthesis [94] and in rice its over-expression resulted in enhanced resistance to pathogens [95]. MYB44 is required for the induction of resistance to M. persicae in Arabidopsis and also affects the repression of aphid feeding activities [96], whereas several ERFs, including RRTF1, ERF1 and ERF13 were upregulated upon chitooctaose treatment in Arabidopsis [90] or after feeding of the aphids Macrosiphum euphorbiae in tomato and Aphis gossypii in melon [97].

Common and specific response of Arabidopsis infected by P. syringae
Our results indicates that common responses between S. littoralis and P. syringae exist. We found that over 190 genes regulated by S littoralis, 32% (61 genes) were also regulated by P. syringae. Moreover, this common responsive genes were regulated in the same direction.
It is interesting to note that many genes coding for sHSPs regulated by S. littoralis herbivory were also down-regulated upon P. syringae infection but not by M. persicae. Since some HSP, like Hsp70, are essential for mediating virulence effect of virulence effector of pathogenic P. syringae (e.g., Hopl1) and play roles in basal resistance to nonpathogenic strains of P. syringae [98,99], we might argue that down-regulation of HSP could be a common strategy for P. syringae and S. littoralis to suppress plant defense. Another common feature of S. littoralis and P. syringae was the upregulation of EF-hand-containing proteins, which are likely to be the key transducers mediating Ca 2+ action [100]. Calcium signaling has been demonstrated to play an important role in both plant-pathogen [101] and plant-insect interactions [102].
As expected, the specific response of Arabidopsis to P. syringae infection was the up-regulation of genes responding to bacterium, including several SA-related genes. GDG1, a member of the GH3like gene family, was shown to be an important component of SAmediated defense against P. syringae [103], whereas the transcription factor WRKY70 positively regulates basal resistance to P. syringae and is involved in SA-signaling pathway [104]. CRK9 was   found to be involved in plant responses to biotic stress, like many other Arabidopsis CRKs [105][106][107], whereas Phytoalexin Deficient4 (PAD4) stimulates production of SA and other processes to limit pathogen growth and has a distinct function in the plant innate immune response [108]. Wall-Associated Kinase (WAK1) is a candidate receptor of oligogalacturonides [109] that triggers defense responses effective against fungal and bacterial pathogens and is induced by SA [110]. Moreover, SA is primarily synthesized from chorismate via isochorismate through the action of isochorismate synthase 1 (ICS1) [111], a gene required for plant defense that was also up-regulated upon P. syringae infection. Up-regulation was also found for a plasmodesmal protein (PDLP5) that was recently discovered to be essential for conferring enhanced innate immunity against bacterial pathogens in a SA-dependent manner [112]. P. syringae also differentially up-regulated genes responding to JA: JAZ10, which is a negative regulator of both JA signaling and disease symptom development [113]; vegetative storage protein1 (VSP1), which is a methyl-JA-inducible gene involved in ABAdependent stomatal closure in Arabidopsis guard cells [114]; MYB50, that is regulated by JA [115] and ADC2, which induction has been reported for various stresses and various growth regulators including JA [116,117]. Allene oxide cyclase (AOC2), that catalyzes an essential step in JA biosynthesis [118] and is induced by JA [119] was also up-regulated. The ethylene-and jasmonate-responsive plant defensin PDF1.2A, a known JAresponsive gene [120], was also up-regulated.

A few genes are commonly regulated by all three biotic stresses
Surprisingly only a few genes were commonly expressed at the time of V m depolarization. In particular, four UGTs were downregulated in all treatments. UGTs convey the transfer of glycosyl residues from activated nucleotide sugars to a wide range of acceptor molecules such as secondary metabolites, including SA and phytoalexins [121]. Down-regulated UGTs belonged mostly to group D, whose members are associated with disease resistance against some pathogens, and a member of Group L that plays an important role in local and systemic resistance in Arabidopsis [122]. A common up-regulation was found for AtNUDX6, the gene encoding ADP-ribose (Rib)/NADH pyrophosphohydrolase. AtNUDX6 is a modulator of NADH rather than ADP-Rib metabolism and significantly impacts the plant immune response as a positive regulator of NPR1-dependent SA signaling pathways [123]. The TIR domain is found in one of the two large families of homologues of plant disease resistance proteins (R proteins) [124]. The largest class of known resistance proteins includes those that contain a nucleotide binding site and leucine-rich repeat domains (NBS-LRR proteins) [125]. The commonly up-regulated gene At5g45000 is a TIR-NBS-LRR class domain-containing disease resistance protein, transmembrane receptor involved in signal transduction, defense response and innate immune response [125,126]. Another commonly up-regulated gene was a heavymetal-associated domain-containing protein (At5g26690) which was found in the category of SA-induced genes positively regulated by a n-butanol-sensitive pathway, implying a cross-talk between SA and phospholipase D pathways [127].
Thus, at the time of V m depolarization a common plant responses to the two herbivores and the pathogen is activation of SA-dependent responses and the triggering of transmembrane receptors, whereas the action of all three biotrophs is the suppression of plant UGTs involved in direct defense.
An unexpected relationship was found for some Arabidopsis genes commonly up-regulated by the herbivore S. littoralis and the pathogen P. syringae and down-regulated by M. persicae. The aphid was found to suppress hydrolase and kinase plant activities, which were activated by the pathogen and the chewing herbivore. Moreover, several oxidative and calcium-related encoding genes as well as some cytochrome P450s (including CYP81F4, involved in the formation of the methoxy group in the indole ring to yield 1methoxy-indole-3-yl-methyl glucosinolate in Arabidopsis [128]) were only suppressed by the aphid. A remarkable up-regulation was found upon S. littoralis herbivory or P. syringae infection, and the opposite regulation upon M. persicae feeding, for a 2oxoglutarate (2OG) and Fe(II)-dependent oxygenase (At4g10500). In plants, these enzymes catalyze the molecular oxygen reduction at the Fe(II) ion binding residue, where it reacts with 2oxoglutarate and a specific substrate through the incorporation of one atom of oxygen in each compound and have been involved in plant defense [129].

Comparison with previous transcriptome analyses
A direct comparison of our data with previous transcriptome analyses indicate similar trend of gene expression in Arabibidopsis upon herbivory by S. littoralis, with particular reference to hydrolases (beta-galactosidase, beta-1,3-glucanase, pectinesterase, xyloglucan endo-1,4-beta-D-glucanase, xyloglucan endotransglycosylase), oxido-reductases (monodehydroascorbate reductase, oxidoreductase), transcription factors (ERF/AP2, MYB, WRKY) and other genes induced by biotic stress (NAM, TIR-NBS-LRR class, RPP1, calmodulin, cysteine-rich receptor-like protein kinase). An opposite trend was found for oxophytodienoic acid reductases (OPR1), alcohol dehydrogenase (ADH), arabinogalactan-protein (AGP12), glutathione S-transferase, heat shock protein 70, and a GTP binding protein [28,130]. However it has to be considered that timing of gene extraction was longer (3-5 h) and falling outside the timing of membrane depolarization. Feeding Arabidopsis with S. exigua for longer periods (from 8 to 48 h) showed the same gene expression only for NADP-dependent oxidoreductase and the opposite expression for lipase, 12oxophytodienoic acid reductase and ERF1, with respect to our results [31,131].

Conclusions
The finding that plant plasma membranes respond with a similar V m depolarization at times depending on the nature of biotic attack allowed us to set a time point for comparative genome-wide analysis among different pests. Our results showed that the aphid M. persicae regulates a wider array of Arabidopsis genes with a clear and distinct regulation than the chewing herbivore S. littoralis. Despite the different timing of V m depolarization, S. littoralis and P. syringae share different commonly regulated genes, implying a relationship between V m depolarization and gene expression. Although several commonly expressed genes between the aphid and the pathogen were present, an almost completely opposite regulation was observed, with the aphid suppressing and the pathogen activating plant defense responses.

Plant, animal and microbial material
Arabidopsis thaliana L. (Columbia 0) plants were grown from seed in a plastic pot with sterilized potting soil at 23uC and 60% humidity using daylight fluorescent tubes (120 mE m 22 s 21 ) with a photoperiod of 16 h. Every liter of soil contained 15 g of vermiculite and 335 g of soil. Moreover, two fertilizers, Osmoco-teH and TriabonH , were added at a concentration of 1 g l 21 soil. All experiments were carried out using 20-22 daysold plants (phase 3) whose leaves turned out to be the most responsive to external stimuli. At least three leaves per plant were used for infestation.
The avirulent strain (Avr) of Pseudomonas syringae pv tomato DC3000 (Pst DC3000 AvrRPM1) was cultured at 28uC in NYGB medium (bactopeptone 5 g l 21 , yeast extract 3 g l 21 , glycerol 20 g l 21 , bactoagar 15 g l 21 ) supplemented with rifampicin (100 mg ml 21 ) and kanamycin (25 mg ml 21 ) (all supplied by Sigma, Milan, Italy). Bacterial cultures were centrifuged at 2500 g for 15 min to recover bacteria, which were resuspended in sterile water to a final OD 600 of 0.2 (equivalent to 1610 8 bacteria/ml). Dilution plating was used to confirm the number of bacteria present in the inoculum.
Virus-free Myzus persicae aphids were reared on A. thaliana plants. They were allowed to grow inside Plexiglas boxes at the temperature of 23uC with a 16 h photoperiod. Only apterous aphids were used in the experiments. Prior to each experiment, at least 20 nymphs per plant were selected for the infestation.
Larvae of Spodoptera littoralis (Boisd. 1833) (Lepidoptera, Noctuidae) (supplied as egg clutches by Syngenta, Switzerland), were reared in Petri dishes at 22-24uC with a 14-16 h photoperiod and fed artificially with a diet as detailed elsewhere [1]. Only third instar larvae (at least 5 per plant) were used for herbivory and leaves were collected when 30% fed by larvae.
Plant defense responses induced by herbivores used as controls mechanical damage, which was inferred with either a pattern wheel (to mimic S. littoralis herbivore damage) or a microforged glass micropipettes (to mimic aphid stinging). P. syringae inoculation used as a control inoculation with MgCl 2 .

Membrane potentials
Membrane potentials were determined in leaf segments in time course experiments. The transmembrane potential (V m ) was determined with glass micropipettes as previously described [1,19,137]. Based on topographical and temporal determination of V m performed previously [18] the electrode was inserted between 0.5 and 2 mm from the wounded zone, where a significant V m depolarization was found to occur. The results of V m are shown as the average number of at least 50 V m measurements.

RNA extraction from Arabidopsis leaves upon insect attack and bacterial infection and cDNA synthesis
Having assessed the timing of V m depolarization, after each experiment, leaves were collected and immediately frozen in liquid nitrogen. Samples for the evaluation of S. littoralis herbivory were sampled after 2 h from feeding, whereas samples for M. persicae phloem feeding were sampled after 5 h from feeding. Samples for the evaluation of Arabidopsis responses to P. syringae were sampled after 16 h from infiltration. Four biological replicates were run for each tested biotic attack.
One hundred mg of frozen leaves were ground in liquid nitrogen with mortar and pestle. Total RNA was isolated using the Agilent Plant RNA Isolation Mini Kit (Agilent Technologies, Santa Clara, CA, US) and RNase-Free DNase set (Qiagen, Hilden, Germany). Sample quality and quantity was checked by using the RNA 6000 Nano kit and the Agilent 2100 Bioanalyzer (Agilent Technologies) according to manufacturer's instructions. Quantification of RNA was also confirmed spectrophotometrically by using a NanoDrop ND-1000 (Thermo Fisher Scientific, Waltham, MA, US).

Gene microarray analyses (including MIAME)
Five hundred nanograms of total RNA from each treated samples, were separately reverse-transcribed into double-strand cDNAs by the Moloney murine leukemia virus reverse transcriptase (MMLV-RT) and amplified for 2 h at 40uC using the Agilent Quick Amp Labelling Kit, two-color (Agilent Technologies). Subsequently, cDNAs were transcribed into antisense cRNA and labeled with either Cy3-CTP or Cy5-CTP fluorescent dyes for 2 h at 40uC following the manufacturer's protocol. Cyanine-labeled cRNAs were purified using RNeasy Minikit (Qiagen, Hilden, Germany). Purity and dye incorporation were assessed with the NanoDrop ND-1000 UV-VIS Spectrophotometer (Thermo Fisher Scientific, Waltham, MA, US) and the Agilent 2100 Bioanalyzer (Agilent Technologies). Then, 825 ng of control Cy3-RNAs and 825 ng of treated Cy5-RNAs were pooled together and hybridized using the Gene Expression Hybridization Kit (Agilent Technologies) onto 4644 K Arabidopsis (v3) Oligo Microarray (Agilent Technologies), satisfying Minimum Information About a Microarray Experiment (MIAME) requirements [138].
After a 17 h incubation at 65uC and 10 rpm, microarrays were first washed with Gene Expression Wash buffer 1 for 1 min, then with Gene Expression Wash buffer 2 for 1 min, then with 100% acetonitrile for 30 s, and finally washed in the Stabilization and Drying Solution for 30 s.
Microarrays were scanned with the Agilent Microarray G2505B Scanner (with the extended dynamic range (XDR) scan mode to scan the same slide at two different levels and data were extracted and normalized from the resulting images using Agilent Feature Extraction (FE) software (v.9.5.1) Agilent Arabidopsis (v3) Gene Expression Microarray (44 K) was used for expression profiling treatments. The microarray experiment followed a direct 262 factorial two-color design. For each of the treatment combinations, RNA from three independent lines was extracted and used for hybridization. For each RNA samples, four biological replicates were used. This resulted in 12 two-color arrays.
The microarray data is MIAME compliant and has been submitted to GEO database with Accession number pending.

Validation of microarrays
Validation of the microarray analysis was performed by quantitative real time PCR. First strand cDNA synthesis was accomplished with 1.5 mg total RNA and random primers using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, US), according to the manufacturer's instructions. Briefly, the reactions were prepared by adding 10 ml total RNA (1.5 mg), 2 ml of 106 RT Buffer, 0.8 ml of 256 dNTPs mix (100 mM), 2 ml 106 RT random primer, 1 ml of Multiscribe TM Reverse Transcriptase and nuclease-free sterile water up to 20 ml. Then the reaction mixtures were subjected to thermal incubation according to the following conditions; 25uC for 10 minutes, 37uC for 2 hours, and 85uC for 5 seconds.
All qPCR experiments were performed on a Stratagene Mx3000P Real-Time System (La Jolla, CA, USA) using SYBR green I with ROX as an internal loading standard. The reaction was performed with 25 ml of mixture consisting of 12.5 ml of 26 Maxima TM SYBR Green/ROX qPCR Master Mix (Fermentas International, Inc, Burlington, ON, Canada), 0.5 ml of cDNA and 100 nM primers (Integrated DNA Technologies, Coralville, IA, US). Controls included non-RT controls (using total RNA without reverse transcription to monitor for genomic DNA contamination) and non-template controls (water template). Specifically, PCR conditions were the following: plant cadmium resistance 1 (PCR1, At1g14880), HSP20-like chaperones superfamily (At1g52560): initial polymerase activation of 10 min at 95uC, and 40 cycles of 30 s at 95uC, 30 sec at 56uC, and 45 sec at 72uC; methionine reductase B8 (MSRB8, At4g21840), UDP-glycosyltransferase 73B4 (UGT73B4, At2g15490): initial polymerase activation of 10 min at 95uC, and 40 cycles of 15 s at 95uC, 30 sec at 59uC, and 30 sec at 72uC; germin-like protein (GR3, At5g20630), flowering promoting factor 1 (At5g10625), cytoplasmic glyceraldehyde-3-phosphate dehydrogenase, (GAPC2, At1g13440), ubiquitin specific protease 6 (UBP6, At1g51710), b-adaptin (At4g11380), elongation factor 1B alpha-subunit 2 (eEF1Balpha2, At5g19510): initial polymerase activation of 10 min at 95uC, and 40 cycles of 15 s at 95uC, 20 sec at 57uC, and 30 sec at 72uC. Fluorescence was read following each annealing and extension phase. All runs were followed by a melting curve analysis from 55 to 95uC. The linear range of template concentration to threshold cycle value (Ct value) was determined by performing a dilution series using cDNA from three independent RNA extractions analyzed in three technical replicates. All primers were designed using Primer 3 software [139]. Primer efficiencies for all primers pairs were calculated using the standard curve method [140]. Four different reference genes (cytoplasmic glyceraldehyde-3-phosphate dehydrogenase, (GAPC2, At1g13440), ubiquitin specific protease 6 (UBP6, At1g51710), b-adaptin (At4g11380) and the elongation factor 1B alpha-subunit 2 (eEF1Balpha2, At5g19510) were used to normalize the results of the real time PCR. The best of the four genes was selected using the Normfinder software [141]; the most stable gene was the elongation factor 1B alpha-subunit 2.
All amplification plots were analyzed with the MX3000P TM software to obtain Ct values. Relative RNA levels were calibrated and normalized with the level of the elongation factor 1B alphasubunit 2 mRNA.

Statistical analyses
For V m measurements, the obtained data were treated by using the stem-and-leaf function of Systat 10 in order to calculate the lower and upper hinge from the Gaussian distribution of values. The data were then filtered and the mean value was calculated along with the SE. Paired t test and Bonferroni adjusted probability were used to assess the difference between treatments and the control.
Processing and statistical analysis of the microarray data were done in R using Bioconductor package limma [142]. The raw microarray data are subjected to background subtraction and loess normalized. Agilent control probes were filtered out. The linear models implemented in limma were used for finding differentially expressed genes. Comparisons were made for each of the treatment and genotype combinations. To reduce the complexity of the analysis, technical replicates were treated as biological replicates. Benjamini and Hochberg (BH) multiple testing correction was applied. We consider genes with both BH adjusted pvalue,0.05 and fold changes .2 as differentially expressed genes.