Antisense LOX expression increases herbivore performance by decreasing defense responses and inhibiting growth-related transcriptional reorganization in Nicotiana attenuata


For correspondence (fax +49 3641 571102; e-mail


Inhibition of jasmonic acid (JA) signaling has been shown to decrease herbivore resistance, but the responsible mechanisms are largely unknown because insect resistance is poorly understood in most model plant systems. We characterize three members of the lipoxygenase (LOX) gene family in the native tobacco plant Nicotiana attenuata and manipulate, by antisense expression, a specific, wound- and herbivory-induced isoform (LOX3) involved in JA biosynthesis. In three independent lines, antisense expression reduced wound-induced JA accumulation but not the release of green leaf volatiles (GLVs). The impaired JA signaling reduced two herbivore-induced direct defenses, nicotine and trypsin protease inhibitors (TPI), as well as the potent indirect defense, the release of volatile terpenes that attract generalist predators to feeding herbivores. All these defenses could be fully restored by methyl-JA (MeJA) treatment, with the exception of the increase in TPI activity, which was partially restored, suggesting the involvement of additional signals. The impaired ability to produce chemical defenses resulted in lower resistance to Manduca sexta attack, which could also be restored by MeJA treatment. Expression analysis using a cDNA microarray, specifically designed to analyze M. sexta-induced gene expression in N. attenuata, revealed a pivotal role for LOX3-produced oxylipins in upregulating defense genes (protease inhibitor, PI; xyloglucan endotransglucosylase/hydrolase, XTH; threonine deaminase, TD; hydroperoxide lyase, HPL), suppressing both downregulated growth genes (RUBISCO and photosystem II, PSII) and upregulated oxylipin genes (α-dioxygenase, α-DOX). By genetically manipulating signaling in a plant with a well-characterized ecology, we demonstrate that the complex phenotypic changes that mediate herbivore resistance are controlled by a specific part of the oxylipin cascade.


Plant resistance to herbivory is mediated by a diverse set of defense traits that can be constitutively expressed or are elicited by herbivore attack. Many inducible defense responses are activated by the octadecanoid signaling cascade (Blee, 2002; Reymond and Farmer, 1998). The plant hormone, jasmonic acid (JA), and its precursor, 12-oxo-phytodienoic acid (OPDA), collectively called jasmonates (JAs), are induced by mechanical wounding and herbivory. In native tobacco species, Nicotiana sylvestris (McCloud and Baldwin, 1997) and N. attenuata (Schittko et al., 2000), and in corn (Schmelz et al., 2003b), wound-induced JA accumulation is dramatically amplified by attack from host-specific herbivores via the introduction of constituents in herbivore oral secretions and regurgitant (R) into wounds (Halitschke et al., 2001; Schmelz et al., 2003a).

The JA biosynthetic pathway involves: regio- and stereospecific dioxygenation of linolenic acid by a lipoxygenase (LOX); formation of an epoxide by allene oxide synthase (AOS); ring formation by allene oxide cyclase (AOC); reduction by OPDA reductase (OPR); and side-chain shortening by three consecutive β-oxidation steps. Enzymes involved in the biosynthesis of JAs have been characterized in several plant species, and their genes have been overexpressed or silenced in transgenic plants (Berger, 2002; Schaller, 2001; Turner et al., 2002). Manipulation of different enzymes in the octadecanoid cascade, such as AOS (Laudert et al., 2000; Park et al., 2002), AOC (Stenzel et al., 2003), or a specific OPR (OPR3; Sanders et al., 2000; Stinzi and Browse, 2000) by antisense or overexpression of endogenous genes, as well as with insertional mutants, has demonstrated the importance of these enzymes in JA biosynthesis. Additionally, a mutant (fad3 fad7 fad8 triple mutant) impaired in the biosynthesis of linolenic acid, the fatty acid precursor of JA biosynthesis, has been characterized (McConn and Browse, 1996) and found to be impaired in wound-induced JA accumulation (McConn et al., 1997).

Plant LOXs occur in a gene family and can be grouped into two main subfamilies (classes 1 and 2), according to their primary sequence similarities (Feussner and Wasternack, 2002; Shibata et al., 1994). According to this classification, the LOXs catalyzing dioxygenation of fatty acids at the 13C position (13-LOX) are members of the type 2 class, which possess a chloroplast transit peptide. Biosynthesis of JA requires a 13-LOX to produce the initial intermediate of the cascade, 13(S)-hydroperoxy linolenic acid. A specific lipoxygenase (AtLOX2), which is involved in JA biosynthesis, has been identified in Arabidopsis thaliana. Antisense expression of AtLOX2 reduced wound-induced JA accumulation (Bell et al., 1995). In tomato plants, expression of a specific chloroplast-targeted 13-LOX (TomLOXD) is rapidly and transiently induced by wounding (Heitz et al., 1997), whereas other LOX genes are expressed only in seeds or fruits (TomLOXA and TomLOXB; Ferrie et al., 1994) and are not induced by wounding (TomLOXC; Heitz et al., 1997). Three distinct classes of LOXs have been defined in potato (Royo et al., 1996): LOX1 is expressed in roots and tubers and is involved in tuber development (Kolomiets et al., 2001; Royo et al., 1996); LOX2 isozymes are expressed in leaves; and LOX3 is expressed in leaves and roots (Royo et al., 1996). Both LOX2 and LOX3 possess putative chloroplast transit peptides, but show distinct expression patterns in response to wounding (Royo et al., 1996). Potato LOX2 (LOX-H1) transcript accumulation increases steadily for 24 h after wounding, whereas transcripts of LOX3 (LOX-H3) are induced much faster and transiently with highest accumulation 30 min after wounding (Royo et al., 1996). Co-suppression of LOX-H1 reduced the release of green leaf volatiles (GLVs), C6 alcohols, and aldehydes that are products of the hydroperoxide lyase (HPL) reaction, but did not influence basal and wound-induced JA levels (Leon et al., 2002). Surprisingly, antisense expression of LOX-H3 influenced neither the release of GLVs (Leon et al., 2002) nor the basal and wound-induced JA accumulation (Royo et al., 1999).

Three systems of transgenic plants impaired in JA biosynthesis have been used to study the octadecanoid signaling cascade in eliciting resistance to herbivory. These studies, while providing tantalizing evidence for the importance of octadenaoids in mediating plant resistance, lack either a detailed characterization of the genetic manipulations or the phenotypic consequences of manipulation. In the first system, the resistance of Arabidopsis against the fungal gnat Bradysia impations has been studied in two mutants, opr3 mutant and fad3 fad7 fad8 triple mutant, both of which lack wound-induced JA accumulation. However, while the fad3 fad7 fad8 triple mutant has impaired resistance to the fungal gnat, the opr3 mutant exhibits resistance comparable to wild-type (WT) plants (McConn et al., 1997; Stintzi et al., 2001). Therefore, oxylipins derived from linolenic acid other than JA are likely to mediate resistance against the fungal gnat. The defense traits elicited by the unknown oxylipin have not been identified. In the second system, tomato def1 mutants, deficient in wound-induced JA accumulation, show lower resistance to feeding Manduca sexta (Howe et al., 1996) and Spodoptera exigua (Thaler et al., 2002a) larvae and spider mites (Li et al., 2002). However, the def1 mutation remains to be understood in terms of its function in JA biosynthesis. The def1 mutation is thought to result in a biosynthetic defect between the LOX and OPR reactions (Howe et al., 1996), but is not a mutation in the genes coding for either AOS or AOC catalyzing the intermediate biosynthetic steps (Li et al., 2002). In a third system, the induction of protease inhibitor (PI) activity was abolished in antisense LOX-H3 potato plants, enhancing the performance of Colorado potato beetle and beet armyworm larvae feeding on these plants. However, as no reduction in JA accumulation was observed (Royo et al., 1999), the mechanism of LOX-H3-mediated herbivore-resistance is uncertain. In short, although the importance of oxylipins in mediating herbivore-resistance has been demonstrated, both the signals and the resistance mechanisms remain to be characterized.

In this study, we isolated and characterized three classes of the lipoxygenase gene families in the wild tobacco plant N. attenuata, and analyzed their tissue-specific and wound-induced expression patterns. Antisense expression-mediated silencing of one specific isoform reduced accumulation of JA in response to wounding and application of M. sexta R in three independent antisense (as-lox) lines, but the silencing did not reduce the release of GLVs. The reduction in JA signaling capacity decreased N. attenuata's resistance to herbivory by M. sexta larvae. The deficiency in resistance could be rescued by methyl-JA (MeJA) treatment. We analyzed the expression of several direct and indirect defense traits in response to M. sexta feeding and demonstrated reductions in induced nicotine accumulation, PI activity, and release of R-induced terpenoid volatiles, known to function as defenses against Manduca larvae attack (Glawe et al., 2003; Kessler and Baldwin, 2001; Voelckel et al., 2001). With a cDNA microarray which was specifically designed to detect M. sexta-induced changes in transcript accumulation in N. attenuata, we analyzed the expression of 240 genes in M. sexta attacked as-lox plants in comparison to attacked WT plants. We demonstrate the reduced expression of genes involved in defense activation and altered expression of growth-related genes.


Isolation and characterization of LOX cDNAs

A partial 1.98-kbp genomic DNA fragment of N. attenuata lipoxygenase was used to screen cDNA libraries prepared from N. attenuata root and shoot tissue induced by feeding of M. sexta larvae (Hermsmeier et al., 2001). Three of the 26 initial positive shoot cDNA clones with expected fragment sizes were sequenced. The isolated cDNAs, 2832, 2740, and 2740 bp, designated as NaLOX1a, NaLOX1b, and NaLOX1c, respectively, differed only in the polyadenylation site and two nucleotide substitutions in NaLOX1b, which might have resulted from the fact that the cDNA library was prepared from plants of six different N. attenuata populations (Winz and Baldwin, 2001). Sequence comparisons showed 81 to 82% nucleotide identity with potato lipoxygenase genes PotLOX1 and PotLOX2 (Kolomiets et al., 1996) and tomato lipoxygenase gene TomLOXA (Ferrie et al., 1994).

A second lipoxygenase gene was identified by the random sequencing of a flower cDNA library. The cDNA sequence of the 2836-bp clone, designated as NaLOX2, showed 84 and 83% identity with potato LOX2 (clone H1; Royo et al., 1996) and tomato lipoxygenase TomLOXC (Heitz et al., 1997), respectively.

A third class of LOX genes was isolated by screening the N. attenuata shoot and root cDNA libraries with a probe derived from tomato lipoxygenase gene TomLOXD (Heitz et al., 1997). Initial sequencing of four leaf and three root cDNA clones revealed high sequence similarity; only the longest root and leaf cDNA clones were sequenced completely. The two sequences were identical and showed only a 40-bp truncation in the 5′ end of the clone obtained from the root cDNA library. The 3042-bp leaf cDNA clone, designated as NaLOX3, showed 87 and 83% nucleotide identity with potato LOX3 (clone H3; Royo et al., 1996) and tomato lipoxygenase TomLOXD (Heitz et al., 1997), respectively.

Individual LOXs show distinct expression patterns

Specific probes for detecting individual LOX transcripts were designed and cross-hybridized against plasmids containing the different cDNAs (Figure 1a). Northern analysis of different tissues from N. attenuata plants grown in hydroponic culture (flowers, stem, leaves, and roots) revealed low to undetectable constitutive expression of all three LOX genes. Only weak constitutive expression of NaLOX2 was detected in leaf tissue of untreated plants (data not shown). To investigate the potential involvement of the three enzymes in biosynthesis of wound- and R-induced JA, we analyzed that expression levels of NaLOX1, NaLOX2, and NaLOX3 in N. attenuata leaves at different times after wounding and treatment of the wounds with either water or R. NaLOX1 transcripts were not detectable in leaves of unwounded control plants and were not induced by water and R treatment of mechanical wounds (Figure 1b). Northern analysis of root RNA from soil-grown plants showed clear transcript accumulation of NaLOX1 in root tissue (data not shown). Transcript accumulation of NaLOX2, which was detectable in control and treated plants, did not show a clear expression pattern after water or R treatment of plant wounds (Figure 1b). In marked contrast, NaLOX3 showed a rapid and transient increase in transcript accumulation, starting at 15 min and peaking between 45 and 60 min after induction by water or R treatment (Figure 1b). Transcript levels declined to basal level 2–3 h after the treatment (Figure 1b). These expression patterns suggest that NaLOX3 is involved in induced JA biosynthesis in N. attenuata, which follows a similar kinetic in response to wounding (Kahl et al., 2000; Schittko et al., 2000).

Figure 1.

NaLOX3 is rapidly and transiently wound- and herbivore-induced.

(a) Plasmids (0.1–100 ng) containing full-length cDNAs of NaLOX1, NaLOX2, and NaLOX3 were slot-blotted and hybridized with specific probes for the individual LOXs.

(b) Total leaf RNA (10 µg) of individual N. attenuata plants harvested at the indicated times after treating single leaves by wounding and application of water (W) or M. sexta oral secretion and regurgitant (R) was hybridized with specific probes for NaLOX1, NaLOX2, and NaLOX3. Control plants (C) remained untreated. Ethidium bromide-stained 18S rRNA is shown as loading control.

Antisense expression of NaLOX3 specifically reduces elicited JA accumulation

Transgenic plants expressing NaLOX3 in an antisense orientation (as-lox) were produced by Agrobacterium-mediated transformation (Krügel et al., 2002), and transgene incorporation was verified by Southern analysis (Figure S1). Three independently transformed lines (A363, A211, and A300) showing the largest reduction in JA accumulation in a high throughput phenotype screen (Krügel et al., 2002), were further characterized. The basal JA levels of the as-lox lines as measured in unwounded leaves of plants from each line did not differ significantly from WT levels (Figure 2a; anova, F3,13 = 0.445; P = 0.7249). The wound-induced JA concentrations in the as-lox lines A363, A211, and A300, which were measured 35 min after mechanical wounding and water treatment, were significantly reduced by 45, 54, and 65%, respectively, compared to the wound-induced WT level (Figure 2a; anova, F3,16 = 10.703; P = 0.0004; Fisher's protected least significant difference (PLSD) ≤ 0.0051). A similar reduction of 32% (A363), 22% (A211), and 50% (A300) in comparison to WT JA levels was observed in plants induced by wounding and by application of larval R (Figure 2a; anova, F3,14 = 13.062; P = 0.0002; Fisher's PLSD ≤ 0.0492), which amplified wound-induced levels in WT plants by 38%. Real-time PCR analysis of NaLOX3 expression demonstrated a reduction of wound- and R-induced NaLOX3 transcript accumulation by 73–83% in A363 plants and 83–92% in A300 plants, respectively (Figure 3). The greater silencing efficiency of NaLOX3 transcripts in the as-lox line A300 is accompanied by a larger accumulation of as-lox transcripts, which are not detectable in WT plants (Figure 3).

Figure 2.

Silencing of NaLOX3 specifically suppresses JA accumulation.

(a) Jasmonic acid was analyzed in single leaves of five replicate WT N. attenuata plants and three independently transformed antisense-lox lines (A363, A211, and A300) after wounding with a pattern wheel and immediately suppling water (W) or M. sexta oral secretion and regurgitant (R). Control plants (C) remained untreated. The treated leaves were harvested 35 min after treatment (time of maximum JA accumulation) for the analysis of endogenous JA levels (mean ± SE).

(b) Mean (±SE) hexanal and (Z)-3-hexenal release from five replicate plants in the first minute after mechanical wounding of a single leaf of the same genotypes described in (a). Unwounded plants had no detectable emissions (data not shown).

Figure 3.

Antisense expression suppresses wound- and herbivore-induced NaLOX3 transcript accumulation.

Accumulation of endogenous NaLOX3 (circle) and as-lox (triangle) transcripts was analyzed by real-time PCR. cDNAs were prepared from 20 ng total leaf RNA of four replicate plants harvested at the indicated times for each of WT and as-lox lines (A363 and A300). At time 0, plants were elicited by wounding single leaves and applying water (open symbol, dotted line) or M. sexta oral secretion and regurgitant (closed symbols, solid line). Expression is shown as mean value (±SE) of four independent experiments in arbitrary units resulting from a calibration with a 10× dilution series of cDNAs prepared from RNA samples containing NaLOX3 and as-lox transcripts.

As 13-LOX catalyzes the formation of fatty acid hydroperoxides and thereby supplies the substrate for an additional downstream and potentially competitive reaction, namely the formation of GLVs, we analyzed the emissions of hexanal and (Z)-3-hexenal, which are released exclusively after wounding and not from undamaged control plants (data not shown). Emissions of hexanal (Figure 2b; anova, F3,16 = 0.557; P = 0.6511) and (Z)-3-hexenal (Figure 2b; anova, F3,16 = 2.670; P = 0.0826) from wounded as-lox lines did not differ from levels released by WT plants. These findings demonstrate that NaLOX3 supplies the substrates for induced JA biosynthesis but not for the release of GLVs.

Antisense expression of NaLOX3 reduces resistance to M. sexta attack

Manduca sexta larvae feeding on as-lox lines A363 and A300 gained more mass than did caterpillars feeding on WT plants by 72 and 63%, respectively (Figure 4a; anova, F3,20 = 3.306; P = 0.0412; Fisher's PLSD ≤ 0.0290). Larvae feeding on plants of line A211 did not differ significantly in their mass from larvae feeding on WT plants (Figure 4a; Fisher's PLSD = 0.6350). Treatment with MeJA recovered the resistance of as-lox plants and diminished the increase in mass gain of larvae feeding on as-lox plants (Figure 4a; anova, F3,19 = 2.505; P = 0.0900). Larvae feeding on plants of line A300 consumed significantly more leaf tissue than did larvae feeding on WT plants (Figure 4b; anova, F3,21 = 4.261; P = 0.0169; Fisher's PLSD ≥ 0.0185), whereas no significant difference in leaf area consumption was observed on plants of as-lox lines A363 and A211 in comparison to WT plants (Figure 4b; Fisher's PLSD ≥ 0.2613). Leaf area consumption on MeJA-treated plants of all three as-lox lines did not show significant differences in comparison to MeJA-treated WT plants (Figure 4b; anova, F3,23 = 0.880; P = 0.4664). The concentrations of two secondary metabolites known to function as direct defenses were analyzed in plants damaged by continuous larval feeding in order to understand differences in larval mass gain.

Figure 4.

Silencing of NaLOX3 reduces herbivore resistance and direct defense activation.

The second and third fully expanded leaves of eight replicate plants of WT N. attenuata plants and three independently transformed antisense-lox lines (A363, A211, and A300) were treated with lanolin paste (Lanolin) or 224 µg MeJA in 20 µl of lanolin paste (MeJA), and a freshly eclosed M. sexta larva was placed on the second leaf. Larvae were allowed to feed for 7 days after which larval mass was recorded and plants were harvested to analyze defense metabolites and leaf area consumed.

(a) Mean (±SE) mass of individual M. sexta larvae was recorded after 7 days of feeding on separate plants.

(b) Mean (±SE) square centimeter leaf area consumed was calculated from image analysis of digitized pictures of leaves damaged by M. sexta feeding for 7 days on the replicate plants from each genotype.

(c) Mean (±SE) nicotine concentration was analyzed in the first fully developed leaf, which remained unattacked, of the replicate plants from each genotype.

(d) Mean (±SE) TPI activity was analyzed in the damaged third fully developed leaf of the replicate plants from each genotype. Statistical comparisons (anova) were performed within treatment groups, and stars represent significant differences in the as-lox lines compared to WT plants (Fisher's PLSDs, P < 0.05).

Antisense expression of NaLOX3 reduces elicitation of direct defenses

The Manduca-induced accumulation of nicotine was significantly reduced in plants of all three as-lox lines by 33–42% (Figure 4c; anova, F3,22 = 7.125; P = 0.0016; Fisher's PLSD ≤ 0.0025) compared to WT plants. Treatment with MeJA restored nicotine induction, and MeJA-treated WT and as-lox plants accumulated similar levels of nicotine (Figure 4c; anova, F3,22 = 1.536; P = 0.2333). In contrast, the reduction in Manduca-induced trypsin protease inhibitor (TPI) activity was significant only in as-lox line A211 (Figure 4d; anova, F3,22 = 3.511; P = 0.0331; Fisher's PLSD = 0.0165) and not in as-lox lines A363 and A300 (Figure 4d; Fisher's PLSD ≥ 0.0772). However, treatment with MeJA elicited significantly higher TPI activity in WT plants than in as-lox plants A363 and A300 (Figure 4d; anova, F3,21 = 4.935; P = 0.0095; Fisher's PLSD ≤ 0.0153). These results demonstrate the involvement of endogenous JA biosynthesis in the elicitation of both direct defense responses but with distinct elicitation signals or mechanisms.

Antisense expression of NaLOX3 inhibits elicitation of indirect defenses

To investigate the involvement of the octadecanoid signaling cascade in elicitation of the volatile release, we analyzed the wound- and R-induced cis-α-bergamotene emission in WT and the as-lox plants with the lowest R-induced JA accumulation (A300 and A363). Wounding alone did not induce the release of cis-α-bergamotene in WT and as-lox plants (Table 1; t-test, P ≥ 0.1384) in comparison to unwounded control plants, whereas the addition of R to the wounded leaf elicited three- to eightfold increases in the emissions of cis-α-bergamotene in WT plants (Table 1; t-test, P ≤ 0.0033) but not in as-lox plants (Table 1; t-test, P ≥ 0.4068). MeJA elicitation induced 7- to 17-fold increases in emissions of cis-α-bergamotene in all WT and as-lox plants (Table 1; t-test, P ≤ 0.0038), demonstrating that terpenoid biosynthesis in all plants was unimpaired and that the inability of as-lox plants to release volatiles after R-elicitation results from a defect in JA signaling.

Table 1.  Inhibition of terpenoid volatile emissions in as-lox plants
Linecis-α-Bergamotene release (ng h−1 per plant)
  1. Mean (±SE) cis-α-bergamotene release from individual WT and as-lox (A363, A300) N. attenuata plants 24 h after treating the second fully expanded leaf by wounding with a pattern wheel and immediately supplying water (W; n = 6) or M. sexta oral secretion and regurgitant (R; n = 6) or applying 224 µg MeJA in 20 µl lanolin (MeJA; n = 4). Control plants (C; n = 4) remained untreated. Significant differences compared to the volatile release of untreated control plants (unpaired t-tests; P < 0.005) are shown in bold numbers.

WT1.98 ± 0.922.22 ± 0.4415.53 ± 2.4333.55 ± 3.70
A3631.84 ± 0.460.77 ± 0.212.92 ± 0.8613.13 ± 6.10
WT5.60 ± 1.035.23 ± 0.5214.68 ± 1.9542.49 ± 13.80
A3005.00 ± 1.145.50 ± 1.234.22 ± 1.5055.48 ± 5.23

Antisense expression of NaLOX3 alters JA-dependent gene expression

We compared transcript accumulation in response to herbivore feeding in as-lox plants (A300 and A363) and WT plants to identify JA-dependent transcriptional responses by cDNA microarray analysis. A complete list of the spotted genes can be found in Table S1. Presented are mean values of the two expression ratios (ERs) calculated from two PCR fragments (ER1 and ER2), spotted in quadruplicate on the array for each analyzed gene (Figure 5).

Figure 5.

Silencing of NaLOX3 suppresses herbivore-induced transcriptional reorganization.

A cDNA microarray designed to analyze changes in gene expression (up- and downregulation) elicited by M. sexta attack was used to identify genes with altered expression in antisense-lox lines. Fluorescencently labeled cDNA derived from M. sexta attacked leaves of as-lox plants A300 and A363 (Cy3) were hybridized against cDNA derived from similarly attacked WT N. attenuata plants (Cy5, see experimental scheme and Experimental procedures for ER criteria). Mean ER of genes that fulfilled both criteria for significant regulation for both ER1 and ER2 individually are depicted as closed bars. Open bars represent mean ERs of genes for which the significance criteria were not fulfilled in both experiments. The arbitrary thresholds for up- or downregulation (1.25 and 0.75, respectively), one of the criteria for significant expression, are shown as gray shaded areas. The analysis of as-lox line A300 cDNA was repeated in an independent hybridization experiment to determine experimental variation between microarray experiments.

To evaluate the criteria applied to detect significant changes in transcript accumulation, we analyzed the reproducibility of results obtained from repeated hybridizations of A300 against WT-derived cDNAs (Figure 5). Two independent hybridization experiments identified identical expression patterns for 201 genes (84%). Moreover, for an additional 11% (26) of the genes, expression patterns were significant (meeting both criteria) in one of the experiments while failing for only one criterion in the second hybridization. The remaining 5% (13) of the genes showed a significant response in one experiment but not in the other.

The Manduca-induced transcript accumulation was significantly increased for 30 genes and significantly decreased for 24 genes in both as-lox lines. Additionally, 31 genes showed significantly increased (23 genes) or decreased (8 genes) transcript accumulation in one of the as-lox lines but a non-significant trend in the second analyzed line. We discuss genes with known or putative function that showed altered expression in response to M. sexta feeding in the as-lox lines (Figure 5). ERs of all genes including genes without significant similarities to known genes in the databases are available in Table S2.

As expected, the NaLOX3 probe on the microarray, which was designed from the region of NaLOX3 that was transformed into the as-lox plants, detected higher NaLOX3 transcript accumulation in the as-lox lines. This high ER was because of the constitutive expression of the transgene (see also Figure 3) under the control of the 35S promoter of cauliflower mosaic virus (CaMV). AOS transcripts, which are induced by M. sexta feeding on N. attenuata WT plants (Ziegler et al., 2001), showed mixed responses. One allele (Ziegler et al., 2001) did not show as-lox-dependent regulation (data not shown), while the second (Hui et al., 2003) exhibited a slight upregulation (Figure 5). Manduca-induced accumulation of HPL transcripts was strongly reduced in both as-lox lines, whereas transcript accumulation of α-dioxygenase (α-DOX) was strongly amplified in as-lox plants in response to M. sexta feeding (Figure 5), demonstrating differential JA-mediated modulation within the oxylipin biosynthetic pathways.

Many defense-related genes that include threonine deaminase (TD), xyloglucan endotransglucosylase/hydrolase (XTH), and TPI are strongly induced by M. sexta feeding (Glawe et al., 2003; Halitschke et al., 2003; Hui et al., 2003). As expected, transcripts of these genes were strongly suppressed in as-lox plants (Figure 5), demonstrating the importance of a LOX3-derived oxylipin for their elicitation after herbivore attack. While not strictly associated with defense, two additional genes, a major intrinsic protein (MIP2) and a RNA-binding glycine-rich protein (RGP-1a), which have been found to be upregulated by M. sexta feeding (Hui et al., 2003), were both downregulated in as-lox plants, suggesting a similar dependence on a LOX3-mediated oxylipin for their elicitation.

In addition to the strong upregulation of genes thought to be involved in herbivore resistance, attack from M. sexta larvae and elicitation by MeJA also result in a dramatic downregulation of photosynthesis- and growth-related genes in N. attenuata (Halitschke et al., 2003; Hermsmeier et al., 2001; Hui et al., 2003). The Manduca-induced downregulation of two photosynthetic genes (light-harvesting complex protein LHCII and photosystem II (PSII) oxygen-evolving complex polypeptide) was suppressed (ER > 1; Figure 5) in as-lox plants, suggesting LOX3-dependent downregulation in WT plants. Surprisingly, additional three genes involved in photosynthesis (RUBISCO small subunit (SSU), and two peptides of PSII), which are also downregulated by herbivore attack in WT plants (Halitschke et al., 2003; Hermsmeier et al., 2001; Hui et al., 2003), showed amplified downregulation in M. sexta-attacked as-lox plants (ER < 1; Figure 5), suggesting that for these genes, a LOX3-derived oxylipin attenuates an even larger downregulation in WT plants. A similar response was found for three genes that are thought to be involved in the metabolic re-configuration elicited by herbivore attack, and have been found to be upregulated after M. sexta attack: sulfite reductase, α-amylase, and ferridoxin-dependent (FD) glutamate synthase. These genes were all upregulated in as-lox plants and hence a LOX3-derived oxylipin likely suppresses upregulation in WT plants.


We identified three different lipoxygenases in N. attenuata, which we designated as NaLOX1, NaLOX2, and NaLOX3 to correspond to the potato LOX classification (Royo et al., 1996). The sequences and expression patterns of the identified enzymes strongly resemble previously described LOXs in other solanaceous plants. NaLOX1, which is expressed only in root tissue and most likely possesses 9-LOX activity, is unlikely to be involved in wound-induced production of JA. While both NaLOX2 and NaLOX3 putatively possess 13-LOX activity, the inducibility and expression kinetic of NaLOX3 (Figures 1 and 3) suggested that it is involved in JA biosynthesis. A function in JA biosynthesis and signal transduction has been suggested for the potato (LOX-H3; Royo et al., 1996) and tomato (TomLOXD; Heitz et al., 1997) homologs of NaLOX3. Antisense expression of NaLOX3 reduced the wound- and R-induced transcript accumulation of NaLOX3 by up to 92% (Figure 3) and JA levels in the three as-lox lines by as much as 65% (Figure 2a). These results clearly demonstrate that JA biosynthesis in N. attenuata is mediated by NaLOX3. In contrast, antisense expression of the potato homolog (LOX-H3) in potato did not result in reduced accumulation of wound-induced JA levels (Royo et al., 1999).

While antisense expression of NaLOX3 reduced the wound- and herbivore-elicited increases in JA, no effects on the wound-induced release of GLVs (Figure 2b) were found. This demonstrates that NaLOX3 specifically supplies fatty acid hydroperoxide substrates to the octadecanoid pathway but not to the HPL reaction. In potato, LOX-H1, a homolog of NaLOX2 characterized in this study, supplies the substrate for the HPL reaction required for the release of GLVs, but does not supply the hydroperoxide to the octadecanoid pathway (Leon et al., 2002). While the wound-induced release of GLVs was unaffected by antisense expression of NaLOX3, the normal wound-induced levels of HPL transcripts are clearly reduced in as-lox plants (Figure 5). The apparent disconnect between transcript accumulation and enzyme activity may reflect the high constitutive HPL activity detected in N. attenuata leaves (Ziegler et al., 2001).

Antisense expression of NaLOX3 decreased the plant resistance and increased the performance of M. sexta larvae (Figure 4a), which in turn, was correlated with reduced expression of two direct defenses: nicotine accumulation and TPI activity. In previous research, inhibition of both of these direct defenses individually increased the performance of M. sexta larvae (Glawe et al., 2003; Voelckel et al., 2001). Here, we find that reductions in nicotine accumulation correlate more strongly with caterpillar performance than do the reductions in TPI activity (Figure 4c,d). Our laboratory study measured only the effects of direct defenses and excluded indirect defenses that possibly enhance the defensive function of particular direct defenses. Under natural conditions, the defensive function of TPIs that slow the growth of herbivores and keep them for longer in smaller life stages that are more vulnerable to predators, may be much greater when coordinately expressed with indirect defenses, such as the release of volatiles that attract predators (Glawe et al., 2003; Kessler and Baldwin, 2001).

A LOX3-derived oxylipin was demonstrated to elicit numerous transcriptional and secondary metabolite responses in N. attenuata. The induction of resistance responses is reduced in as-lox plants despite the larger amount of damage caused by M. sexta feeding, which positively correlates with JA accumulation and defense activation in WT plants (Baldwin et al., 1997; Halitschke et al., 2000). The observation that deficiencies in both induced nicotine accumulation (Figure 4c) and cis-α-bergamotene emissions (Table 1) in as-lox plants could be fully restored by MeJA treatment is consistent with an important role for JA in eliciting these defensive responses (Halitschke et al., 2000; Winz and Baldwin, 2001). In WT plants, M. sexta attack and MeJA treatment elicit dramatic increases in defense-related transcripts and decreases in transcripts of the light-harvesting complex protein and oxygen-evolving complex polypeptide of PSII (Hermsmeier et al., 2001; Hui et al., 2003). In M. sexta-attacked as-lox plants, the defense-related transcripts, PI, TD, XTH, and HPL, were decreased, while the transcripts of the two photosynthetic genes were increased, demonstrating that an oxylipin produced via the LOX3-mediated octadecanoid pathway regulates these genes.

However, the involvement of signals in addition to LOX3-derived oxylipins is implicated in the elicitation of other herbivore-induced responses in N. attenuata. For example, TPI activity was not fully recovered by MeJA treatment (Figure 4d), suggesting the involvement of an additional herbivore-induced signal in TPI activation. Possible signals could be ethylene, which is induced by M. sexta feeding (Kahl et al., 2000) and has been shown to synergize PI elicitation in tomato (O'Donnell et al., 1996), or intermediates in the octadecanoid cascade upstream of JA, such as OPDA, which has been shown to elicit the production of numerous secondary metabolites (Blee, 2002).

In addition to the strong evidence for a LOX3-dependent oxylipin in upregulating several defense responses and downregulating some photosynthetic genes, our results also demonstrate an antagonistic effect of the LOX3-dependent signal on the activity of other signaling pathways. For example, an additional branch of the complex oxylipin signaling network in plants is mediated by α-oxygenation of fatty acids (Hamberg et al., 1999). α-DOX transcripts are known to increase in response to wounding and to be further amplified when R is added to mechanical wounds in N. attenuata (Schittko et al., 2001). Here, we found the Manduca-induced α-DOX transcript levels to be amplified in as-lox plants in comparison to WT plants (Figure 5), suggesting that a LOX3-mediated oxylipin normally suppresses the amplification process. Pathogen attack is known to elicit α-DOX (de Leon et al., 2002), and signal cross-talk between pathogen- and herbivore-induced responses (Reymond and Farmer, 1998) may be involved in the α-DOX elicitation. In addition to suppressing amplification of α-DOX transcripts, LOX3-derived oxylipins suppress the downregulation of a set of photosynthetic genes in response to M. sexta feeding and MeJA treatment in WT plants (Hermsmeier et al., 2001; Hui et al., 2003). In as-lox plants, the transcript accumulation of RUBISCO SSU and two enzymes of PSII were further reduced in response to M. sexta feeding (Figure 5). In contrast to the photosynthetic genes that showed strong JA-dependent downregulation, the downregulation of these genes is negatively modulated by the oxylipins of the LOX3-dependent octadecanoid cascade, again suggesting cross-talk between JA-dependent and JA-independent signaling cascades.

The traits responsible for herbivore resistance have long been studied by ecologists and are known to involve complex adaptations that include direct and indirect defenses, as well as tolerance responses (Karban and Baldwin, 1997). The majority of molecular work on plant–herbivore resistance has been conducted in model plant systems in which a detailed understanding of the phenotypic traits responsible for herbivore resistance is lacking or only recently being examined (Thaler et al., 2002). In this study, we generated transgenic plants impaired in a JA-biosynthetic enzyme with reduced wound- and R-induced JA accumulation and reduced resistance against herbivore attack. Because these transformations were conducted in a species in which the traits responsible for herbivore resistance have been thoroughly characterized (Baldwin, 2001), the transformants provided valuable insights into the complex signaling processes involved in activating induced herbivore resistance. In particular, the laboratory-based experiments presented here have established causal associations among a signal cascade, particular direct defenses, and herbivore resistance. We anticipate that by examining herbivore resistance of these plants in the complex habitats of their natural environment, we will eventually understand how direct and indirect defenses function together to provide herbivore resistance.

Experimental procedures

Plant material

Wild-type N. attenuata plants were grown from seeds of a field collection from a native population (DI Ranch, Santa Clara, UT, USA) after subsequent selfing for 11 generations. Seeds were germinated as described by Krügel et al. (2002). After 10 days, seedlings were transferred into potting soil and grown for 2–3 weeks under 16 h light (28°C)/8 h dark (24°C). Two- to three-week-old rosette-stage plants were used in all experiments.

Isolation of lipoxygenase genes

A genomic DNA fragment (L7a2) of N. attenuata LOX gene was synthesized by PCR with primers derived from highly conserved regions of tomato TomLOXA and TomLOXB (Ferrie et al., 1994) and potato LOX1 (Royo et al., 1996) genes, cloned into the pCR®2.1-TOPO vector (Invitrogen, Carlsbad, CA, USA), and sequenced. An exon-specific fragment of L7a2 was PCR amplified, and served as a probe for a cDNA library screen. A second probe was derived from tomato TomLOXD (Heitz et al., 1997) by PCR.

Two hundred thousand plaque-forming units (pfu) of cDNA libraries (λZAP II; Stratagene, La Jolla, CA, USA) prepared from root (Winz and Baldwin, 2001) and shoot (Hermsmeier et al., 2001) mRNA of N. attenuata plants exposed to M. sexta feeding for 24 h were plated, blotted, and screened according to the manufacturer's instruction. The PCR-derived probes described above were labeled with 32P using a random prime labeling kit (RediPrime II; Amersham-Pharmacia, Little Chalfont, UK) and purified on G50 columns (Amersham-Pharmacia). Blots were washed four times with 2× SSC, 0.1% SDS at 65°C for 30 min after hybridization at 65°C and analyzed by autoradiography. Initial positive clones were analyzed for full length by PCR, and selected plaque-pure clones were excised in vivo according to the λZAP II protocol (Stratagene), and sequenced on an ABI310 sequencer using the Big Dye terminator kit (Applied Biosystems, Darmstadt, Germany). Additionally, 96-flower cDNA containing plasmids were sequenced with plasmid-specific primers after in vivo excision from a cDNA library prepared by directional cloning of poly(T)-primed flower cDNA into the EcoRI/XhoI site of the Uni-ZAP XR vector (λZAP II; Stratagene). Two cDNA clones with similarities to known LOX sequences were sequenced completely.

Generation and characterization of transgenic plants

The Agrobacterium tumefaciens (strain LBA 4404)-mediated transformation procedure and the transformation vector pNATLOX1 are described by Krügel et al. (2002). Transformation was confirmed by PCR and resistance screening of potentially transformed plants and progeny of verified transformants was screened for the desired phenotype, namely reduced wound-induced JA accumulation in a high throughput screen (Krügel et al., 2002). Progeny of homozygous plants were selected by nourseothricin (NTC) resistance screening and used for further experiments.

Nucleic acid analysis

Extraction of total RNA and Northern blot analysis were performed as described by Winz and Baldwin (2001). Probes for NaLOX1, NaLOX2, and NaLOX3 were synthesized by PCR with gene-specific primers and labeled with 32P using a random prime labeling kit (RediPrime II; Amersham-Pharmacia). Blots were washed after overnight hybridization at 42°C three times with 2× SSPE and one time with 2× SSPE/2% SDS at 42°C for 30 min and analyzed on a phosphoimager (model FLA-3000; Fuji Photo Film Co. Ltd, Tokyo, Japan). The specificity of the designed probes was analyzed by slot-blotting (model PR 648; Hoefer Scientific Instruments, San Francisco, CA, USA) a dilution series (0.1, 1, 10, and 100 ng) of the plasmids containing the full-length cDNA of NaLOX1a, NaLOX2, and NaLOX3, respectively, onto a nylon membrane (GeneScreen plus; NEN, Boston, MA, USA) according to the manufacturer's protocols, and hybridization with the radioactively labeled probes. Additionally, NaLOX3 and as-lox transcript accumulation was analyzed by real-time PCR (ABI PRISM™ 7000; Applied Biosystems, Darmstadt, Germany). Total RNA was extracted as described above, and cDNA was prepared from 20 ng RNA with MultiScribe™ reverse transcriptase (Applied Biosystems), and amplified using the qPCR™ core reagent kit (Eurogentec, Seraing, Belgium) and gene-specific primers and probes (Figure S2).

Analysis of M. sexta performance

Freshly eclosed M. sexta larvae (North Carolina State University, Raleigh, NC, USA) were placed on the second fully developed leaf (position +2) of eight replicate plants from each analyzed line. An additional eight plants of each line received 1 µmol MeJA to the second (+2) and third (+3) fully developed leaves prior to the larva being placed on the former. Larvae were allowed to feed on the plant for 7 days after which larval mass was recorded. Leaf area consumed was calculated from digital pictures of the damaged leaf.

Analysis of direct defense traits

Samples of the harvested +1 leaves were analyzed by HPLC as described by Keinanen et al. (2001) with the following modification of the extraction procedure: approximately 100 mg frozen tissue was homogenized in 1 ml extraction buffer utilizing the FastPrep® extraction system (Savant Instruments, Holbrook, NY, USA). Samples were homogenized in FastPrep® tubes containing 900 mg lysing matrix (BIO 101, Vista, CA, USA) by reciprocating shaking at 6 m sec−1 for 45 sec.

Trypsin protease inihibitor activity was analyzed by radial diffusion activity assay in the harvested +3 leaves as described by van Dam et al. (2001).

Analysis of JA accumulation and GLV release

Five replicate plants of each genotype were treated by wounding the second fully expanded leaf with a pattern wheel and immediately applying 20 µl water or 20 µl M. sexta oral secretion and regurgitant, diluted 1 : 1 (v/v) with water to the puncture wounds. Control plants received no treatment. Leaves were harvested 35 min after elicitation, and endogenous JA levels were analyzed by gas chromatography–mass spectrometry (GC–MS) with a 13C-labeled internal standard as described by Krügel et al. (2002).

GLV emissions were analyzed from individual leaves of five replicate plants per genotype before and in the first minute after wounding with a pattern wheel with a portable gas analyzer (zNose™; EST, Newbury Park, CA, USA; Krügel et al., 2002). Released amounts of hexanal and hexenal were calculated by calibration curves and normalized to the fresh mass of the analyzed leaf. The zNose™ instrument was calibrated with gas-phase samples using a calibration system (model 3100; EST). Absolute amounts (0.2–200 ng) of hexanal and (E)-2-hexenal in 0.2 µl methanol were vaporized and analyzed. Linearity of response was demonstrated by regression analysis (correlation coefficient of 0.9957 and 0.9914 for hexanal and (E)-2-hexenal, respectively). The chromatographic separation of (Z)-3-hexenal and (E)-2-hexenal did not provide sufficient resolution to definitely identify the released isomer. Therefore, wound-induced volatiles were analyzed by proton-transfer reaction mass spectrometry (PTR-MS; Ionicon Analytik, Innsbruck, Austria) to identify the released hexenal isomer. Excised leaves were wounded as described above and placed into a 100-ml glass chamber. The chamber was flushed with a stream of purified air (200 ml min−1), of which a portion (15 ml min−1) was introduced continuously into the PTR-MS. Fragmentation pattern of the headspace sample of freshly wounded leaves from WT plants was dominated by the mass ion 81 and showed low abundance of mass ions 57 and 99. This fragmentation pattern is identical to the fragmentation pattern of (Z)-3-hexenal and does not match the fragmentation pattern of (E)-2-hexenal, which in turn, is dominated by the mass ion 55 (Fall et al., 1999). This fragmentation was confirmed by the analysis of synthetic standards (Bedoukian, Danbury, CT, USA) and clearly identified the plant-released hexenal as the (Z)-3-isomer.

Analysis of terpenoid volatile release

We analyzed volatile emission of WT and as-lox-transformed plants from the A300 and A363 lines. Plants (four to six replicate plants per treatment) were treated by wounding the second fully expanded leaf with a pattern wheel and immediately supplying 20 µl water (W; n = 6) or 20 µl M. sexta oral secretion and regurgitant, diluted 1 : 1 (v/v) with water (R; n = 6) or applying 1 µmol MeJA in 20 µl lanolin (MeJA; n = 4) to undamaged leaves. Control plants (C; n = 4) remained untreated. Plants were enclosed in open-top volatile collection chambers 24 h after elicitation, and volatiles were collected for 8 h and analyzed by GC–MS as described by Halitschke et al. (2000).

cDNA microarray analysis

Total RNA was extracted from pooled +2 leaves of the replicate plants of each line used in the M. sexta performance experiment described above. Isolation of mRNA and cDNA synthesis was performed as described by Halitschke et al. (2003). cDNA derived from WT plants was labeled with Cy5 and cDNA of as-lox lines was labeled with Cy3 fluorescent dye (Halitschke et al., 2003). The microarray, described extensively by Halitschke et al. (2003), contains PCR fragments of 240 N. attenuata genes identified by differential display analysis of M. sexta-induced gene expression (Halitschke et al., 2003; Hermsmeier et al., 2001; Hui et al., 2003). The criteria for detection of up- and downregulated expression are as follows. The quadruplicate spotting of each PCR fragment allowed a statistical analysis of the ERs. Therefore, log-transformed ER was analyzed for significant differences from a hypothesized mean of ‘0’, corresponding to an untransformed ER of ‘1’, by one-sample t-test. A gene was considered to be up- or downregulated if the following two criteria were fulfilled for the ER of both spotted PCR fragments (ER1 and ER2): (i) both individual ER were significantly different from ‘1’ (one-sample t-test, P = 0.05) and (ii) both individual ERs were equal to or exceeded an arbitrary threshold of 0.75 or 1.25, representing 25% down- or upregulation, respectively.

Availability of materials

Upon request, all novel material described in this publication will be made available in a timely manner for non-commercial research purposes. Sequence data were submitted to GenBank under accession numbers AY254345 (NaLOX1a), AY254346 (NaLOX1b), AY254347 (NaLOX1c), AY254348 (NaLOX2), and AY254349 (NaLOX3).


We thank Clarence A. Ryan (Washington State University, Pullman, WA, USA) for kindly providing the tomato TomLOXD cDNA clone, Thomas Hahn, Susan Kutschbach, and Matthias Held for sequencing and technical assistance in microarray analysis, Anja Paschold for technical assistance in the real-time PCR analysis, and Axel Knop-Gericke, Michael Haevecker, and Robert Schlögl (Fritz-Haber-Institut, Berlin Germany) for the use of the PTR-MS instrument. Supported by the Max-Planck-Gesellschaft.

Supplementary Material

The following material is available from

Figure S1. Transgene insertion in plants of three independently transformed antisense-lox lones.

Genomic DNA (10 μg) from two individual plants of each as-lox line (A363, A211, A300) and WT Nicotiana attenuata plants was digested with SspI or EcoRI, respectively, and blotted onto nylon membrances (Winz and Baldwin, 2001). The blots were hybridized with a PCR fragment of the region of NaLOX3 used for the antisense construct. The digestion with SspI, having a recognition site inside the NaLOX3 coding region, and EcoRI (without a recognition site in NaLOX3), indicate a single copy of the transgene in as-lox lines A211 and A300 and potentially two insertions in the genome of A363, in addition to the endogenous NaLOX3 gene, which was also detected in the WT.

Figure S2. Real time PCR assay for analysis of endogenous NaLOX3 and antisense-lox (as-lox) transgene expression.

Real time PCR analysis was performed on a ABIPRISM™ 700 Sequence detection System (Applied Biosystems, Darmstadt, Germany). A) Specific amplicons (green box), a PCR primer pair and a double fluorescent dye-labeled probe, were designed for the detection of endogenous NaLOX3 and as-lox transgene transcripts. The NaLOX3 assay amplified the 5′ region of NaLOX3 and spans the transition to the region of NaLOX3 used for the construction of the transformation vector, pNATLOX1. The as-lox assay amplifies the 3′ region of the expressed transgene with a gene specific primer annealing to the antisense-lox sequence and a second transformation vector-specific primer and the probe annealing to the transcribed cauliflower mosaic virus 35S terminator (TCAMY) sequence. The specificity of the PCR reaction was verified by gel analysis of the reaction products and the quality of the assay was evaluated by analyzing previously characterized RNA samples, analyzed by Northern analysis, and comparing the results of the two independent methods. Expression levels of NaLOX3 in WT plants 45 (R45) and 60 (R60) minutes after wounding and application of Manduca sexta oral secretions and regurgitant were analyzed by real time PCR and compared to Northern blot analysis (left panel, see Figure 1b lanes 45 and 60 min). Expression of NaLOX3 and as-lox was analyzed in plants treated with lanolin or methyl jasmonate and attacked by a single M. sexta larva (right panel, see Figure 4). n.d. represents a sample not determined due to lack of RNA.

Table S1 Description of all genes spotted on the DNA microarray

Table S2 Mean (±SE) expression ratios of all genes on the cDNA microarray

Accession numbers: AY254345 (NaLOX1a), AY254346 (NaLOX1b), AY254347 (NaLOX1c), AY254348 (NaLOX2), and AY254349 (NaLOX3).