An amino acid substitution inhibits specialist herbivore production of an antagonist effector and recovers insect-induced plant defenses

Plants respond to insect herbivory through the production of biochemicals that function as either direct defenses or indirect defenses via the attraction of natural enemies. While attack by closely related insect pests can result in distinctive levels of induced plant defenses, precise biochemical mechanisms responsible for differing responses remain largely unknown. Cowpea ( Vigna unguiculata ) responds to Fall armyworm (FAW; Spodoptera frugiperda ) herbivory through the detection of fragments of ATP synthase γ -subunit (cATPC) proteins, termed inceptin-related peptides, present in larval oral secretions (OS). In contrast to generalists like FAW, OS of the legume specializing Velvetbean caterpillar (VBC; Anticarsia gemmatalis ) does not elicit ethylene production and demonstrates significantly lower induced volatile emission in direct herbivory comparisons. Unlike all other Lepidoptera OS examined, which preferentially contain inceptin [ Vu -In; + ICDINGVCVDA - ], VBC OS contains predominantly a C-terminal truncated peptide Vu -In -A [ + ICDINGVCVD - ]. Vu -In -A is both inactive and functions as a potent naturally occurring antagonist of Vu -In induced responses. To block antagonist production, amino acid substitutions at the C-terminus were screened for differences in VBC gut proteolysis. A valine substituted peptide [ Vu -In Δ V ; + ICDINGVCVDV - ] retaining full elicitor activity was found to accumulate in VBC-OS. Compared to the native polypeptide, VBC that previously ingested 500 pmols of the valine modified cATPC precursor elicited significantly stronger plant responses in herbivory assays. We demonstrate that a specialist herbivore minimizes the activation of defenses by converting an elicitor into an antagonist effector and identify an amino acid substitution that recovers these induced plant defenses to a level observed with generalist herbivores.

Excluding microorganisms, plants and insects constitute over 75 percent of the total species on earth (Hawksworth and Kalin-Arroyo, 1995). At a numerical species level, insect herbivores significantly outweigh plants and together display exquisitely complex ecological interactions (Price, 2002). Herbivorous insects are often broadly categorized as either generalists or specialists to denote feeding preferences on either a wide variety or select subset of plant species, respectively. It is widely envisioned that host plant defenses have less impact on specialists as compared to generalist herbivores.
Comparatively, specialists often manipulate plants to their benefit and demonstrate improved tolerance of host-specific biochemicals. For example, sawfly larvae in the genus Pontania induce protective closed galls in numerous willow species (Salix) that result in dramatically reduced interior tissue defenses (Nyman and Julkunen-Tiitto, 2000). Swallowtail caterpillars, such as Papilio polyxenes, utilize plants in the family Apiaceae through efficient cytochrome P450 monooxygenase enzymes that directly detoxify dietary furanocoumarins (Cohen et al., 1992;Berenbaum, 2002). Cabbage white butterfly (Pieris rapae) larvae prevent toxic isothiocyanate formation through the action of a gut nitrile-specifier protein that redirects typical hydrolysis of glucosinolates to instead form nitriles that are excreted in the frass (Wittstock et al., 2004). Although specialists often exhibit greater tolerance to specific biochemicals than generalists, in many cases they are still negatively impacted by high levels of these plant defenses (Adler et al., 1995;Zalucki et al., 2001;Berenbaum, 2002;Agrawal and Kurashige, 2003).
A hallmark of herbivore-inducible plant defense is variability. One method to non-destructively measure this variation is through assessment of indirect defenses such as herbivore-induced volatile emissions that attract natural enemies of the offending pest. In some cases, different feeding habits such as leaf herbivory versus stem boring readily account for this variation; however, even similar patterns of leaf herbivory can result in different volatile emission patterns and parasitoid attraction (Turlings et al., 1998;De Moraes et al., 1998). Oral secretion (OS)-derived fatty acid amino acid conjugate (FAC) elicitors, such as N-(17-hydroxylinolenoyl)-L-glutamine and related analogs, promote plant recognition of Lepidopteran herbivores and trigger induced volatile emission in diverse plant species including maize (Zea mays), eggplant (Solanum melongena) and soybean (Glycine max) (Alborn et al., 1997;Schmelz et al., 2009). A combination of mechanical damage and FAC elicitors can constitute the initial stimuli leading to production of the defensive phytohormones jasmonic acid (JA) and ethylene (ET); however, elicitation can be phylogenetically idiosyncratic and even absent from closely related plants (Schmelz et al., 2009). In receptive plants, a potential for specificity comes from differences in the biochemical content of herbivore OS. For example, in Nicotiana attenuata, herbivory and OS of the specialist Manduca sexta elicits greater production of JA and ET than comparable treatments with the generalist Spodoptera exigua (Diezel et al., 2009). These phytohormone differences are coincident with markedly different levels of FAC elicitors and also salivary-derived glucose oxidase (GOX) in the herbivore species.
As a salivary secretion, GOX is one of the few herbivore-derived effectors implicated in the suppression of inducible defenses, namely reduced nicotine accumulation in tobacco and improved growth of H. zea (Musser et al., 2002). The suppressive activity of GOX in Nicotiana species is in part through the formation of H 2 0 2 , elicitation of induced salicylic acid (SA) production and associated negative cross-talk interactions that attenuate induced JA and ET signaling (Diezel et al., 2009).
Transcriptional evidence for similar SA-mediated suppression of JA signaling also exists in different feeding guilds. For example, in Arabidopsis thaliana, whitefly (Bemisia tabaci) feeding appears to repress accumulation JA-regulated defense transcripts such as PDF1.2 (Zarate et al., 2007). Although the effector remains unknown, B. tabaci suppression of JA-regulated resistance is associated with strong up-regulation of the SA signaling pathway and associated gene transcript levels (Zarate et al., 2007).
Regardless of diet breadth, herbivores cause responses in their hosts that are variable and more recently the paradigm of generalists and specialists herbivores inducing predictable plant responses has been challenged (Ali and Agrawal, 2012).
Improved utilization of host plants by any biotic attacker often involves a combination of detoxification and prevention of defense activation. The vast majority of mechanistic advances demonstrating suppression of inducible defenses have come from plant-pathogen research (Jones and Dangl, 2006;Zhou and Chai, 2008;Hogenhout et al., 2009). A well characterized component of plant basal immunity to pathogenic bacteria is recognition of microbe associated molecular patterns (MAMPs) such as the flagellin peptide fragment, termed flg22. The FLS2 receptor binds flg22 to initiate defense; however, successful pathogens such as Agrobacterium tumefacians and pathovar variants of Xanthomonas campestris pv campestris exhibit altered sequences of flg22 that neither associate with FLS2 nor activate defense (Felix et al., 1999;Zipfel et al., 2004;Sun et al., 2006). To counter elicitation, pathogenic bacteria, such as Pseudomonas syringae pv. tomato DC3000 (Pst DC3000), inject into plant cells a broad array of multifunctional effector proteins via the type III secretion system that suppress activation of innate immunity (Jones and Dangl, 2006;Zhou and Chai, 2008). Moreover, Pst DC3000 also secretes the jasmonoyl-isoleucine phytohormone mimic coronatine as a small-molecule effector that co-opts signaling by the E3 ubiquitin ligase COI1 and transcription factor JIN1/MYC2 to suppresses MAMP mediated resistance responses (Millet et al., 2010). In a conceptually useful scheme detailing the probable evolutionary interplay of resistance and disease susceptibility, plants are hypothesized to have first evolved MAMP recognition systems to trigger innate immune responses. Subsequently pathogens overcame recognition through effectors (virulence factors) that were able to suppress those immune responses (Bent and Mackey, 2007). Plant R-proteins then evolved to either directly recognize pathogen effectors (avirulence factors) or products of their activities on guarded plant proteins to promote defense activation and resistance. In contrast to pathogens, relatively little is known about the mechanisms Lepidopteran herbivores use to evade plant recognition and activation of defense responses (Bonaventure et al., 2011).
Cowpea (Vigna unguiculata) and common bean (Phaseolus vulgaris) recognize insect herbivory through the detection of trace amounts of cyclic disulfide-bridged inceptin-related peptides present in caterpillar OS, which are derived from proteolytic fragments of chloroplastic ATP synthase γ -subunit (cATPC) proteins (Schmelz et al., 2006;Schmelz et al., 2007). Inceptin-related peptides have thus far been described in the OS of Fall armyworms (FAW; Spodoptera frugiperda), which prefer grasses but are confirmed generalists known to feed on over 60 different diverse plant species including cowpea (Luginbill, 1928). To better understand the adaptations of a legume specialist, we examined the Velvetbean caterpillar (VBC; Anticarsia gemmatalis), a devastating defoliator of tropical soybean and other legumes, whose host range includes Vigna and Phaseolus species in the Americas (Buschman et al., 1977;Piubelli et al., 2005). Direct comparative assays of induced responses in cowpea to both natural herbivory and OS of both VBC and FAW revealed that VBC induced significantly lower plant defense responses. Compared to eight different Lepidoptera pest species examined, VBC demonstrate a unique preferential processing of inceptin-related peptides into a biologically inactive form, namely Vu-In -A ( + ICDINGVCVD -). Moreover, Vu-In -A functions as a naturally occurring antagonist of plant defense responses triggered by inceptin (Vu-In; + ICDINGVCVDA -). Sequential screening of Vu-In peptide libraries with substituted C-terminal amino acids revealed a single conservative amino acid change that reduces VBC proteolysis of active inceptin-related peptides and recovers induced plant defense responses to this pest.

RESULTS
Decreased Defense Activation by a Legume Specialist Herbivore. In cowpea, we previously established that FAW OS contains the elicitor Vu-In that drives the coordinated production of the defense-related phytohormones including jasmonic acid (JA), ethylene (ET) and salicylic acid (SA) as well as increases in emission and leaf pools of the predominant volatile (E)-4,8-dimethyl-1,3,7nonatriene (DMNT) (Schmelz et al., 2006;Schmelz et al., 2007;Carroll et al., 2008). Volatile DMNT is detected by numerous parasitoid wasps and attracts natural enemies of herbivorous pests (Gouinguene et al., 2005;Kappers et al., 2005). Cowpea leaves experiencing a FAW feeding bout from larvae devoid of Vu-In, fail to elicit DMNT production (Schmelz et al., 2006). In a direct comparison to FAW OS, larval secretions from the legume specialist VBC elicited 8.8-and 2.2-fold lower levels of ET and DMNT, respectively, when applied to wounded cowpea leaves ( Fig. 1A-B). DMNT leaf pools present at 4 h following VBC OS treatment were not significantly different from mechanical damage+H2O alone (Fig.   1B). Quantified changes in leaf tissue pools of DMNT are used as an estimate for differences in volatile emission at the local site of treatment and herbivore attack. To examine cowpea defense responses to actual herbivory, we performed both short-term assays, which allowed single feeding bouts, and longterm assays where plants were subjected to intensive sustained attack by either FAW or VBC. Analysis of cowpea leaf tissue surrounding short-term feeding sites which received equivalent damage (Fig. 1C) and long-term whole plant volatile emissions (Fig. 1D) both demonstrate that VBC herbivory results in significantly lower levels of DMNT compared to FAW herbivory. Thus, cowpea plants produce quantitatively different defense responses to defoliation by these generalist and specialist Lepidopteran herbivores.

Unlike Other Lepidoptera, VBC Predominantly Contain an Inactive Inceptin-related Peptide.
Large-scale purification of FAW OS previously revealed the presence of four inceptin-related peptides, consisting of predominantly Vu-In and lesser amounts of Vu-E+ In [ + EICDINGVCVDA -], Vu-GE+ In [ + GEICDINGVCVDA -], and Vu-In -A [ + ICDINGVCVD -]. Unlike the three inceptin-related peptides with additional N-terminal amino acids, the C-terminal truncated Vu-In -A lacked ET, SA and DMNT inducing activity (Schmelz et al., 2007). Consistent with our previous findings, FAW accumulate mainly Vu-In and only low amounts of Vu-In -A when fed peptide precursors containing the inceptin core sequence ( Fig. 2A). In contrast, VBC OS contains largely the inactive Vu-In -A and only low levels of active Vu-In Corn earworm (Helicoverpa zea). After feeding on cATPC precursors containing the inceptin core sequence, only VBC exhibited significantly lower percentages (< 25%) of active OS peptides (sum of Vu-In, Vu-E+ In, and Vu-GE+ In) compared to the total of all four inceptin-related peptides (Fig. 2C).
Compared to the three highest accumulators of active inceptin-related peptides (S. exigua, O. nubilalis and H. zea), T.ni larvae also exhibited a modest ability to accumulate Vu-In -A (Fig. 2C). While not a preferred host plant, in laboratory settings T. ni is able to complete multiple instars on plants responsive to inceptin such as P. vulgaris (Soo Hoo et al., 1984). Nonetheless, unlike the other 8 Lepidopteran pests tested, VBC exhibit preferential processing of inceptin into the predominate inactive OS peptide Vu-In -A .

Vu-In -A is a Naturally-Occurring Antagonist of Inceptin-induced Responses.
In plants, C-terminal amino acid deletions in synthetic peptide signals can result in the creation of competitive antogonists of binding and inhibitors elicitor-induced plant responses (Pearce et al., 1993;Meindl et al., 1998;Meindl et al., 2000). To examine if Vu-In -A functions as a naturally occurring antagonist in cowpea, leaves were

Biochemical Screen of Substituted Peptide Precursors Recovers Active Elicitors in VBC OS.
Given that amino acid sequences influence protease susceptibility (Poreba and Drag, 2010), one strategy to minimize the production of antagonists and restore plant defense elicitation following VBC attack is to modify the inceptin C-terminal. Subtle sequence changes have the potential reduce VBC production of Vu-In -A relative to active elicitors. Towards this goal, we examined if the C-terminal alanine of Vu-In can withstand amino acid changes and still retain elicitor activity. As a first step, a Vu-In ΔX substitution library was synthesized and assayed for induced ET production in cowpea leaves. With the exception of the basic amino acids, H, K, R (histidine, lysine, arginine) and P, W (proline, tryptophan), all other Vu-In C-terminal substitutions maintained significant elicitor activity (Fig. 4A). To estimate elicitor stability following larval ingestion, all active 11mer Vu-In ΔX C-terminal substituted peptides were then fed to VBC larvae and recollected from the OS. The Vu-In ΔX C-terminal substituted peptides D, N, Q, S, T, V and Y were recoverable from VBC OS at equal or greater levels relative to Vu-In -A , and thus predictably sufficient to maintain elicitation ( Fig. 3B and 4B). Lepidopteran larvae generate inceptin-related peptides through both N-and C-terminal proteolysis of larger cATPC precursors. Given this complexity, we examined larval production of 6 substituted peptide candidates from their respective model 19mer-Vu-In ΔX precursors requiring amide bond cleavage for activation. Following ingestion by VBC and subsequent OS collection, the precursor 19mer-Vu-In To examine both reproducibility and importance of dose, this experiment was repeated by feeding VBC larvae 5-fold greater levels of the 19mer peptide precursors. At 2.5 nmoles of peptide, herbivory triggered 2.1-, 4.1-, 2.6-and 3.5-fold greater levels of JA, SA, CA and DMNT, respectively, when VBC larvae previously ingested the modified 19mer-Vu-In Δ V compared to the native 19mer-Vu-In ( Fig. 6E-H). Plant responses to larvae that previously consumed cowpea leaves plus H 2 O or 19mer-Vu-In were not statistically different (Fig. 6E-H). On average, the 2.5 nmole dose of 19mer-Vu-In Δ V resulted in larvae that elicited slightly higher plant responses than those previously ingesting 0.5 nmoles ( Fig.   6A-H). Collectively, these results demonstrate that an alanine to valine substitution in the cATPC polypeptide sequence is able to recover the elicitation of cowpea defense responses following VBC herbivory.

DISCUSSION
The identity and activity of herbivore-associated elicitors of plant defenses have been intensively studied over the past decade (Howe and Jander, 2008). In some cases, insects may avoid the production of active elicitors by feeding on tissues lacking essential precursors while in other situations individual herbivore elicitors within a class can vary greatly between conspecific larvae without significantly influencing defense activation (De Moraes and Mescher, 2004;Roda et al., 2004). While the outcome of pathogen and insect attack relies significantly upon their capacity for host plant utilization, the documentation of herbivore effectors that suppress elicitation is quite limited. A few examples come from aphids that consume phloem for many hours once a sieve element feeding site has been established. In broad bean (Vicia faba), secreted salivary proteins from the aphid Megoura viciae suppress sieve element occlusion, which typically occurs through the formation of dispersed forisomes (i.e. expanded proteinaceous inclusions), triggered in response to damage and increased local Ca 2+ levels . These salivary-derived Ca 2+ -binding proteins also promote dispersed forisomes to convert back into their original compact structure thereby recovering the flow of phloem sap and enabling successful feeding. Similarly, a specific salivary protein of unknown function, designated as C002, has been demonstrated to be absolutely required for pea aphid (Acyrthosiphon pisum) feeding and survival (Mutti et al., 2008). Biochemical mechanisms lepidopteran larvae utilize to suppress plant responses during herbivory are less clear but can also involve salivary proteins. H. zea herbivory on tobacco (N. tabacum) results secretion of the enzyme GOX onto the leaf surface and suppresses induced nicotine accumulation in tobacco (Musser et al., 2002). In cowpea, we currently demonstrate a new mechanism of suppression where legume specialist larvae exhibit the targeted removal of functional OS elicitors in part by converting the active signal into an antagonist of defense elicitation (Fig. 7).
It is unclear if generalist and specialist herbivores consistently differ in their OS elicitors. The transcriptional response of N. attenuata to attack from two generalist herbivores, namely H. virescens and S. exigua, was more similar too each other than that of the specialist M. sexta. This difference was linked to OS-derived FACs (Voelckel and Baldwin, 2004). OS contents of the two generalists are virtually identical whereas M. sexta lacks volicitin and predominately harbors fatty acid-glutamic acid conjugates (Pohnert et al., 1999;Alborn et al., 2003). FACs from M. sexta also have a role in suppressing induced nicotine accumulation in tobacco, yet remain elicitors for indirect defense responses (Kahl et al., 2000). Unlike a number of Solanaceous plants, defense responses in cowpea are not elicited by FACs (Schmelz et al., 2009). Instead variation within inceptin-related peptides from herbivore OS mediates the activation or suppression of elicitor-induced defenses. The current evidence for specialist antagonism of elicitor-induced plant defenses is further supported by the lack of preferential Vu-In -A processing in all generalist species examined (Fig. 2C). A future survey of legume specialist Lepidoptera that include Vigna and Phaseolus as frequent host plants would reveal the breadth and scope of this mechanism in nature.
The theoretical existence of a naturally occurring peptide signal that functions as an antagonist to elicitor-induced plant responses dates back nearly two decades. In 1991, discovery of the 18 amino acid endogenous peptide from tomato foliage, termed systemin, marked the first bioactive peptide signal found in plants (Pearce et al., 1991). Shortly thereafter, structure-function studies with truncated peptides revealed that the successive removal of C-terminal amino acids inactivated the signal and resulted in peptides that blocked the subsequent binding and activity of native full-length systemin (Pearce et al., 1993;Meindl et al., 1998). Related studies with the bacterial peptide elicitor flg22 also found that C-terminal truncations resulted in competitive antagonist peptides and led to the 'addressmessage' concept that peptide signal N-and C-terminals are important for receptor binding and activation, respectively (Meindl et al., 2000). With both systemin and flg22, numerous C-terminal amino acid deletions in synthetic peptides are required to achieve strong competitive antagonism. In the current study, a single C-terminal amino acid loss resulting in Vu-In -A promotes antagonism of elicitation when present at 66% or more of the total inceptin-related peptides. While comparative phylogenetic studies are consistent with the existence of specific plant receptors for insect-derived elicitors, the precise mechanism of Vu-In mediated defense activation remains unknown (Schmelz et al., 2009). We identify Vu-In -A as truncated elicitor derivative that antagonizes inceptin-induced responses including ET, SA and DMNT; however, the action of Vu-In -A as a competitive antagonist at the receptor level remains hypothetical and subject to future empirical examination (Fig. 7).
In contrast with plant-pathogen interactions, receptor-ligand pairs have yet to be identified to regulate plant resistance to insect attack. For aphid resistance, receptors including the Mi-1.2 gene in tomato and Vat genes in melon protect against specific biotypes of Macrosiphum euphorbiae and Aphis gossypii, respectively (Rossi et al., 1998;Boissot et al., 2010). Currently, aphid associated ligands that bind these receptors are unknown. A common long-term threat in crop breeding is the eventual shift in predominant insect biotypes that favor those able to overcome plant resistance genes. For example, in 1997 a virulent isolate M. euphorbiae that performs well on tomato plants harboring the Mi-1.2 gene was identified in California and currently appears to predominate in the US (Goggin et al., 2001). Of relevance to insect biotypes that overcome plant resistance, we demonstrate that VBC posses a biochemical adaption that minimizes the activation of plant defenses while feeding. The infrequency of predominant Vu-In C-terminal processing in other Lepidoptera species examined is likely due to the high level of protease resistance associated with cyclized peptides (Horton et al., 2002). While the precise mode of cleavage has not been described, VBC larvae may produce a specific gut carboxypeptidase that is capable of cleaving the C-terminal alanine from the largely cyclic Vu-In. Thus far trypsin-like proteases have been predominantly described; however, the gut tissues of A. gemmatalis larvae also harbor bacteria potentially capable of further proteolysis Visotto et al., 2009).
In pathology terms, the presence of Vu-In in FAW OS is analogous to an 'avirulence factor' given its ability to trigger rapid plant defense responses in cowpea. Conversely, Vu-In -A in VBC OS functions as a 'virulence factor' by suppressing elicitation and subsequent induced defenses triggered by trace amounts of Vu-In. Unfortunately, these designations are accurate only in the context of each individual system examined given that virulence and avirulence factors often possess opposite activities in related plant species. Divergent activities of previously designated virulence and avirulence factors result in both confusion and conceptually restrictive usage. The term 'effector' is now favored and broadly encompasses biochemicals that either positively or negatively modify plant defense signaling and expression (Hogenhout et al., 2009). The discovery of elicitor antagonist activity of Vu-In -A compels us to view inceptin-related peptides not simply as a class of elicitors but more widely as herbivoreassociated effectors capable of multiple activities.
In cowpea, we demonstrate that a conservative amino acid change in the cATPC precursor for Vu-In can block production of an elicitor antagonist and recover elicitor-induced responses following VBC attack (Fig. 7). Curiously, potential mutations that convert precursors of Vu-In to Vu-In Δ V may be may be subject to additional constraints. In an earlier analysis of the coding region of chloroplastic atpC spanning Vu-In we observed perfect conservation of the C-terminal alanine in every plant species investigated (Schmelz et al., 2006). This might be due to interactions with the inhibitory ε -subunit that in part mediates chloroplastic ATP synthase activity (Hightower and McCarty, 1996). Thus in legumes that respond to inceptin-related peptides, the transgenic manipulation of plants to improve defense responses

MATERIALS AND METHODS
Plant and Insect Materials. FAW larvae were obtained from Dr. R. Meagher (USDA-ARS-CMAVE, Gainesville, FL) and reared on a pinto bean based diet (Schmelz et al., 2006). Preliminary trials with VBC larvae utilized wild caught individuals from outbreak areas on the surrounding margins of peanut fields (Williston, FL) and these results were readily replicated using commercially obtained insects. All experiments utilized VBC, T. ni, H. virescens, P. includens, A. ipsilon, S. exigua, O. nubilalis (Sun Gro Horticulture Distribution, Inc) supplemented with 14-14-14 Osmocote (Scotts Miracle-Gro). All plants were maintained in a greenhouse with a 12 h photoperiod, minimum of 300 μmol -2 s -1 of photosynthetically active radiation supplied by supplemental lighting, 70% relative humidity and temperature cycle of 24ºC/28ºC (night/day).

Cowpea Leaf Bioassays and Plant Metabolite Analysis.
All experiments used 14-18 day old plants containing 2 fully expanded pairs of trifoliate leaves. For ET induction assays, the adaxial sides of new fully expanded leaves were superficially scratched with a razor in 3 areas, removing approximately 5% of the total waxy cuticle. The damage sites (2 cm 2 each) included the central leaf tip spanning both sides of the midrib and 2 mid-basal sections on opposite sides of the mid-rib. Test solutions in 5 μl H 2 0 were immediately applied and dispersed over the damage sites. For ET quantification, leaves remained on the intact plants for 1h, were then excised, sealed in 13-ml tubes and analyzed via GC headspace sampling as previously described (Schmelz et al., 2006). To quantify leaf tissue pools of DMNT, JA SA, and CA a 4 cm 2 section of leaf surrounding the treated site was weighed (50 to 100 mg), frozen in liquid N 2 , processed and analyzed by gas chromatography isobutane-chemical ionization mass spectrometry as described elsewhere (Schmelz et al., 2004;Schmelz et al., 2006) . Collection and quantification, of leaf volatile emission from intact plants followed established protocols (Carroll et al., 2008).
Individual Insect Feeding Studies. 6 th instar VBC and FAW larvae were allowed to feed for 12 h on cowpea leaves and then individually isolated in polystyrene 12-well Cell Culture Plates (BD Falcon™) for 1 h in the absence of plant tissue. The larvae were carefully transferred onto the leaf canopies of tightly spaced cowpea plants and allowed to feed for 20 min. At this time, all larvae were removed from plants and only insect damage sites with comparable leaf area consumed (typically 50 mm 2 ), were selected for analysis (n=6). To ensure maximal uniformity of feeding time and surplus of larval wound sites with equivalent leaf area removed, each of these trials utilized groups of 48 insects on 16 plants for each insect species or treatment comparison. To control elicitor content of larval OS, VBC in 12-well plates were allowed to consume a 65 mm 2 cowpea leaf disc containing the native and substituted 19mer cATPC peptides. Within 30 min of consumption, larvae where placed on leaves as described above.

Quantification of Inceptin-related Peptides in Insect OS. Chemical verification and quantification of
inceptin-related peptides from insect OS utilized HPLC-MS as previously described (Schmelz et al., 2006). An isotopically-labeled [ 13 C and 15 N valine; + ICDING-V*-CVDA -] Vu-In analog was used as an internal standard (ISTD) for quantification. Aliquots of crude OS, typically 50-100 μl derived from a pool of 12 larvae, were sequentially spiked with 100 ng of the ISTD peptide, 5 μl HCl, vortexed, and centrifuged 12000 x g for 5 min. The aqueous phase was mixed with an equal volume of EtOH, stored at -70ºC for 30 min, and centrifuged 12000 x g for 2 min. Samples were diluted to 5% EtOH, loaded on 100 mg RP-C18 SPE columns, washed with 2 ml of H 2 O, and eluted with 9:1 CH 3 CN:H 2 O. Samples were then concentrated to dryness under vacuum, brought up in 50 μl 5:95 CH 3 CN:H 2 O containing 10mM NH 4 COOH. Using an HPLC system comprised of a P4000 pump, an AS3000 autosampler, and a UV6000LP detector (Thermo Separation Products, San Jose, CA), 10μL injections of OS samples were onto a YMC ODS-AQ RP-C18 (250 x 4.6 mm, S-5 μm, 20 nm; Waters Corp, Milford, MA) analytical column heated to 60 Cº, using flow rate of 1 ml min -1 , with mobile phases A and B containing 95:5 H2O:CH3CN and 9:1 CH3CN:H2O, respectively, as well as 10mM NH4COOH buffer. Post column eluant was split, allowing 0.1 ml min -1 to enter the ion source for positive ion mass spectrometry (MS) using a LCQ Deca XPMAX ion trap (Thermo Electron Corp, San Jose, CA) as previously described (Schmelz et al., 2006). Chemical verification was based on comparing retention times and MS 2 daughter ion spectra mass spectra of synthetic standards against natural products. Quantification was based on extracting [M+H] + ions from full scan spectra using the following common ion-analyte pairs 1125.5 (ISTD), 1048.5 (Vu-In -A ), 1119.5 (Vu-In), 1248.5 (Vu-E+ In), 1305.5 (Vu-GE+ In), 1147.5 (Vu-In  (Table S1).
Lepidoptera Species Comparison. The OS was collected and pooled from groups of 6 larvae (n=4) for inceptin quantification. Early 6 th instar larvae (n ≥ 48) were removed from diet and isolated for 1 h prior to feeding on leaf discs (65 mm 2 ) containing 4.5 nmoles of 19mer-Vu-In. OS was collected and frozen immediately after the larvae consumed at least 50% of the treated leaf disk and required between 30-120 min depending on the species. With the exception of A. epsilon, O. nubilalis, and T. ni, which utilized either maize (Zea mays) or bok choy (Brassica rapa) leaves, all other species were fed cowpea leaves.
Resulting levels of the four inceptin-related peptides Vu-In -A , Vu-In, Vu-E+ In, Vu-GE+ In were then quantified by LC/MS as described.

Determination of Vu-In -A Inceptin Antagonist Activity.
Four paired groups of cowpea leaves (n=4, 32 total) were wounded and treated with either 5 μl H 2 0 or Vu-In -A at 0.0045, 0.045, 0.45, and 4.50 pmol leaf -1 . One min later, both leaves in each paired group were treated with a secondary application of Vu-In at either 0.0045, 0.045, 0.45, and 4.50 pmol leaf -1 . To access simultaneous application of Vu-In -A and Vu-In, all peptide treated leaves (n=4) received 10 μl of H 2 0 containing 1 pmol Vu-In in the presence of either 0, 1, 2, 4, 8, or 16 pmol of Vu-In -A . A damage plus H 2 0 only control group was included. All leaves were excised after 1 h and analyzed for ET production. To examine Vu-In -A antagonism of Vu-In activity in the context of leaf metabolites at 4h, a subset of these treatments was re-examined. Cowpea leaves (n=4, 48 total) were wounded and treated with either 5 μl H 2 0 or 1 pmol Vu-In, Vu-In + Vu-In -A (1:1, 1 pmol leaf -1 for each peptide), Vu-In + Vu-In -A (1:4 ; 1 and 4 pmol, respectively), or 1 pmol Vu-In -A alone. To address the similarity of Vu-In ΔV activity to Vu-In, leaves were also treated with 1 pmol Vu-first analyzed by LC/MS for purity, normalized for concentration and tested at 10 pmole leaf -1 for Einducing activity in cowpea. Vu-In C-terminal substituted peptides with activity similar to Vu-In were then fed to 14 groups of 12 VBC larvae (20 μg leaf -1 disk) and OS was collected 6 h later and pooled for HPLC-MS analysis. To examine VBC proteolytic processing of precursors into inceptin-related peptides, 6 of the more promising Vu-In    Vu-In (i.e. C-terminal =A). Different letters (a-c) represent significant differences (All ANOVA P values were < 0.0001. Tukey test correction for multiple comparisons: P < 0.05).      Previous VBC leaf disc diet treatments (nmoles 19mer-Vu-In ΔX ) H 2 O A V 0.5 2.5