An Elicitor of Plant Volatiles from Beet Armyworm Oral Secretion

See allHide authors and affiliations

Science  09 May 1997:
Vol. 276, Issue 5314, pp. 945-949
DOI: 10.1126/science.276.5314.945


The compoundN-(17-hydroxylinolenoyl)-l-glutamine (named here volicitin) was isolated from oral secretions of beet armyworm caterpillars. When applied to damaged leaves of corn seedlings, volicitin induces the seedlings to emit volatile compounds that attract parasitic wasps, natural enemies of the caterpillars. Mechanical damage of the leaves, without application of this compound, did not trigger release of the same blend of volatiles. Volicitin is a key component in a chain of chemical signals and biochemical processes that regulate tritrophic interactions among plants, insect herbivores, and natural enemies of the herbivores.

The intriguing defensive reaction of plants, whereby plant volatiles induced by insect herbivore injury attract natural enemies of the herbivores, is triggered when a substance in the oral secretion of the insect herbivores contacts damaged plant tissue. We have isolated, identified, and synthesized N-(17-hydroxylinolenoyl)-l-glutamine (Fig. 1), which we have named volicitin, from the oral secretion of beet armyworm (BAW) (Spodoptera exiguaHübner) caterpillars. Synthesized and natural volicitin induce corn (Zea mays L.) seedlings to release the same blend of volatile terpenoids and indole released when they are damaged by caterpillar feeding (1). This blend of volatile compounds attracts females of the parasitic wasp Cotesia marginiventris, natural enemies of BAW caterpillars, to the damaged corn plants (2).

Figure 1

Structure of volicitin. (A) The biosynthetic pathway leading to jasmonic acid in plants and (B) the biosynthetic pathways leading to prostaglandins and leukotrienes in animals.

The similarity of the structure of the insect-produced elicitor with the structures of the precursors of eicosanoids and prostaglandins involved in signaling in insects and other animals (3) and the components of the octadecanoid signaling pathway in plants (Fig. 1) (4) indicates a link between these two systems. The octadecanoid pathway is involved in induction of biosynthesis and release of volatiles in response to insect herbivore feeding (5, 6).

We collected oral secretions (about 5 μl per caterpillar) by squeezing third to fifth instar BAW caterpillars that had been fed on corn seedlings, causing them to regurgitate (1). Biological activity was determined by collection and capillary gas chromatographic (GC) analysis (7) of volatiles from corn seedlings treated with oral secretion or subsequent fractions thereof (8).

The crude oral secretion was acidified, centrifuged, and filtered to remove proteins and solids (9). When the filtered oral secretion was fractionated on a reversed phase solid-phase extraction cartridge (9), the total activity of the original crude secretion eluted from the reversed phase cartridge with 50% CH3CN in H2O, indicating a molecule of medium polarity.

We further purified the active material from solid-phase extraction by a series of reversed-phase high-performance liquid chromatography (rpHPLC) fractionations using three sets of conditions (10). Only one active component, detected by monitoring ultraviolet (UV) absorption at 200 nm and with no absorption above 220 nm, eluted from the final column, and its biological activity was equivalent to that of the original crude oral secretion. The active material was extracted into CH2Cl2 from an acidified (pH 3) aqueous solution but not from an aqueous solution at pH 8 (11). All biological activity could be extracted from the organic phase back into pH 8 buffer, indicating lipid character and an acidic functional group. A CH2Cl2 solution of the active material was also fractionated on a normal-phase diol column (12). The active component eluted from this column with MeOH (Me designates methyl). Rechromatography of the active component on the final rpHPLC column indicated it was greater than 99% pure.

Purified active compound was applied to artificially damaged leaves of intact corn seedlings (2). The volatiles released were the same as those induced by treatment with BAW oral secretion. Application of only buffer resulted in the release of significantly smaller quantities and different proportions of volatiles (Fig.2).

Figure 2

(left). Average amount (nanograms per 2 hours) (n = 4) of volatiles collected from three intact Ioana corn seedlings that had been artificially damaged and treated with (A) 15 μl of BAW oral secretion per seedling on the damage sites, (B) 15 μl of oral secretion equivalents of pure natural volicitin, (C) 15 μl of buffer (8), or (D) undamaged control plants. At 9:00 p.m. a 1-cm2 area of the second leaf of three-leaf seedlings was scratched with a clean razor blade and the test solution immediately rubbed over the damaged site. The next morning at 9:00 a.m. the seedlings were cut off above the root, and volatiles were collected and analyzed as described (7, 8). Bars with the same retention time in each graph represent the following compounds: 1, hexenyl acetate; 2, linalool; 3, (E)-4,8-dimethyl-1,3,7-nonatriene; 4, indole; 5, α-trans-bergamotene; 6, (E)-β-farnesene; 7, (E)-nerolidol; 8, (3E,7E)-4,8,12-trimethyl-1,3,7,11-tridecatetraene.

The active compound was identified by mass and infrared spectroscopy and by chemical transformations. Fast atom bombardment mass spectrometry (FAB-MS) analysis (13) indicated the presence of only one compound with diagnostic peaks at mass-to-charge ratio (m/z) 423.280 (M + H)+ in the positive ion mode and at m/z 421.2733 (M − H) in the negative ion mode. The addition of sodium chloride to the FAB matrix resulted in reduced intensity of the m/z 423.280 and the appearance of m/z 445.2628 (MH + Na)+. Thus, the active compound is a weak acid with a molecular weight of 422.274 daltons for the neutral molecule in acid form, and its elemental composition is C23H38N2O5 (422.278 daltons).

Daughter ions of the sodium salt, m/z 445, obtained by FAB-MS/MS (13), included a dominant ion atm/z 427 (445 − 18), whereas daughter ions of m/z423 gave a strong peak at m/z 405 (423 − 18), both indicating a loss of H2O. The lower mass region showed a characteristic pattern of peaks at m/z 147, 130, 129, 101, 84, 67, and 56, which is most consistent with the electron impact (EI) mass spectrum of glutamine (14). Subtraction of glutamine, linked by an ester or amide bond, gave C18H30O3 as the elemental composition of the second part of the molecule, which is consistent with a hydroxy C18 acid with three double bonds.

When the active sample was lyophilized and treated with MeOH and acetic anhydride (15), GC/MS analysis (16) of the product revealed two prominent peaks. Chemical ionization (CI)/MS analysis of the first of these peaks with a retention time of 21.05 min revealed a prominent (M + 1)+ion at m/z 144, and EI/MS analysis revealed a molecular ion at m/z 143 and diagnostic ions at m/z 84 (base peak), 56, and 41, identifying it as the methyl ester of pyroglutamate, which confirmed the presence of glutamine. The CI mass spectrum of the second peak (retention time of 27.55 min) contained a very weakm/z 309 (M + 1)+ ion and a predominant ion atm/z 291 due to loss of H2O (M + 1 − 18)+. Loss of MeOH gave an ion at m/z 277 (M + 1 − 32)+, and the loss of both H2O and MeOH gave an ion at m/z 259 (M + 1 − 18 − 32)+. The EI spectrum of the same peak showed no molecular ion but a strong peak atm/z 290 due to the loss of H2O (M − 18)+, and a fragmentation pattern of ions characteristic of a straight-chain unsaturated hydrocarbon. These results were consistent with the methyl ester of a hydroxy acid. A smaller peak in the chromatogram had retention characteristics (retention time of 26.09 min) and a mass spectrum consistent with the acetate of the same hydroxy acid methyl ester (17).

Fourier transform infrared analysis (18) of the hydroxy acid methyl ester peak from GC produced a spectrum with a weak absorption at 3646 cm−1, indicating an alcohol, and absorption bands at 3019, 2935, and 2865 cm−1, typical of an unbranched, nonconjugated unsaturated hydrocarbon chain. The absence of absorption bands in the 2000 to 2500 cm−1 and the 960 to 980 cm−1 regions eliminated the possibility of an acetylene or trans double bond, respectively. The intensity of the 3019 cm−1 peak indicated three cis double bonds. Absorption at 1758 cm−1 confirmed the presence of a methyl ester.

Partial reduction (19) of the methyl ester of the C18 hydroxy acid resulted in a mixture of mono- and di-unsaturated products as identified by GC/MS. Subsequent ozonolysis (20) of this mixture and GC/MS analysis produced three diagnostic GC peaks with (M + 1)+ ions at m/z187, 229, and 271, corresponding to H(CO)(CH2)n(CO)OCH3 withn = 7, 10, and 13, respectively. Methyl linolenate treated in the same way gave identical products. Thus, the olefinic bonds in the chain are located on carbons 9, 12, and 15, and the alcohol group is on either the 17th or 18th carbon.

The methyl ester was saturated by treatment with PdO/H2 overnight (21). GC/MS (EI) analysis of the product showed an m/z 299 (M − 15)+ ion andm/z 270/271 (M − 44/M − 43)+ ions, indicative of β-cleavage of a secondary alcohol on C17. EI mass spectra of a pyrrolidide derivative (21) of the saturated product produced diagnostic ions at m/z 309 (M − 44)+and m/z 338 (M − 15)+, confirming the C17 location of the hydroxyl group.

Only an amide bond between glutamine and the acid moiety of the hydroxy acid would be consistent with the results of all analyses. Thus, the active compound isolated from the oral secretion of BAW larvae is N-(17-hydroxylinolenoyl) glutamine, which we name volicitin.

Racemic 17-hydroxylinolenic acid was synthesized (Fig.3) (22) and coupled with d- andl-glutamine (23). The crude synthetic conjugates purified on HPLC showed retention characteristics identical to the natural product. FAB-MS/MS and GC/MS analyses showed the synthetic and natural products to be identical.

Figure 3

(right). Synthesis scheme for volicitin.

Volicitin synthesized with d- or l-glutamine contained about 15% of the opposing form (24), indicating racemization during the step in which the fatty acid was coupled to glutamine. However, the active natural compound consisted exclusively of the l-glutamine form. Enantiomerically pured- and l-glutamine forms of synthetic volicitin were collected from a chiral column (24) for bioassay.

The concentration of natural volicitin was estimated by HPLC detector response to be about 20 pmol per microliter of BAW oral secretion. Bioactivity of the l-glutamine form of the synthetic volicitin was equivalent to that of oral secretion (Fig.4). The d-glutamine form was not active (Fig. 4). Racemic 17-hydroxylinolenic acid, d-glutamine, and l- glutamine at 300 and 900 pmol per excised corn seedling showed no activity. Thus, the biologically active compound isN-(17-hydroxylinolenoyl)-l-glutamine. At this time the configuration about the asymmetric 17th carbon in the fatty acid chain remains unknown.

Figure 4

Relative release of volatiles collected for 2 hours from three LG11 corn seedlings that had been treated with 10, 30, 100, 300, or 1000 pmol per plant of thed-glutamine (d-Synthetic) or thel-glutamine (l-Synthetic) forms of volicitin in 500 μl of buffer (8), or with buffer only. The combined amount in nanograms of caryophyllene, α-trans-bergamotene, (E)-β-farnesene, (E)-nerolidol, and (3E,7E)-4,8,12-trimethyl-1,3,7,11-tridecatetraene was used to calculate the release relative to that of seedlings treated with 15 μl of BAW oral secretion, which is equivalent to ∼300 pmol of natural volicitin. The error bars represent standard error (n = 5). Data points topped by the same letter do not differ significantly [Tukey test, P ≤ 0.05 (33)].

In plants, the synthesis and release of volatile compounds appear to be induced by jasmonic acid, which is produced from linolenic acid by the octadecanoid signaling pathway (6). Jasmonates also stimulate other physiological and defensive processes in plants (4, 6, 25), and the amino acid conjugates of jasmonic acid are involved in physiological and developmental processes in many plants (6, 26). Therefore, the presence of an elicitor that is an octadecatrienoate conjugated to an amino acid suggests that the elicitor molecule is involved with the octadecanoid pathway in the herbivore-damaged plants.

This elicitor activity is not diet-related and thus does not originate from the plants (1), although the fatty acid moiety is probably derived from linolenic acid obtained from the diet. The oral secretion of insects fed an artificial diet or filter paper is as active as that from insects fed on plants (1). Volicitin is also related in structure to eicosapentaenoic and arachidonic acids from the fungus Phytophthora infestansthat elicit the production of fungitoxic sesquiterpenes in potato (27).

The octadecanoid signaling pathway in plants is similar in many ways to the eicosanoid pathways in animals that produce prostaglandins and leukotrienes (4, 25). In insects eicosanoids may mediate cellular responses to bacterial infections as well as regulate other physiological functions (3).

Both corn and cotton respond to BAW damage and to the oral secretions of BAW applied to damaged leaves by producing and releasing terpenoids and indole (1, 2, 28-30). Although some compounds, such as indole, ocimene, and farnesene, are released by both plants, others are unique to each plant. Both plants respond systemically to BAW oral secretion by releasing induced volatiles from undamaged leaves of injured plants (29, 31). In cotton the induced volatile compounds are known to be synthesized de novo (30).

Volicitin accounts for the total activity of BAW oral secretion, and boiling the secretion for 30 min did not diminish its activity; thus, there is no evidence for enzymatic activity in eliciting volatiles in this case. In contrast, a β-glucosidase in the saliva ofPieris brassicae caterpillars elicits the release of volatile compounds from cabbage leaves (32). The sequence of events between the introduction of β-glucosidase and the emission of volatiles is unknown (5), but the simple release of a terpenoid or other volatile compound from a glycoside by the direct action of such an enzyme from the attacking herbivore obviously occurs by a different mechanism than the induction of de novo biosynthesis by a small molecule like volicitin. Also, it cannot account for the delayed release of herbivore-induced volatiles (28) or the systemic release of induced volatiles (29,31). Thus, different plant species may use different mechanisms to produce and release volatiles and may respond to different elicitors. Whether closely related insect species, particularly those that feed on the same types of host plants, produce volatile elicitors with the same or very similar structures remains unknown.

  • * Present address: University of Neuchatel, Institute of Zoology, CH-2007 Neuchatel, Switzerland.

  • Present address: Community Research Service, Kentucky State University, Frankfort, KY 40601, USA.

  • To whom correspondence should be addressed. E-mail: tumlnsn{at}


Stay Connected to Science

Navigate This Article