Plants on early alert: glandular trichomes as sensors for insect herbivores


Author for correspondence:
Gary W. Felton
Tel: +1 814 863 7789


  •  The ability of caterpillar or moth ‘footsteps’ to elicit defenses in the tomato (Solanum lycopersicum) plant was examined. Although touch responses frequently have been observed in plants, the role of herbivore ‘touch’ in eliciting antiherbivore defenses has not been adequately examined.
  •  A combination of methods, including in situ hybridization, reverse transcriptase-polymerase chain reaction, quantitative real-time polymerase chain reaction and gas chromatography-mass spectrometry, was used to determine the role of trichomes in mediating these touch responses. Mutants compromised in jasmonic acid and glandular trichomes were used to test whether both of these were required for these touch responses.
  •  We demonstrated that the rupture of foliar glandular trichomes by caterpillar or moth contact induced the expression of defense transcripts (e.g. proteinase inhibitor 2, or PIN2) regulated by jasmonic acid. Neither chewing nor the release of salivary components was required to initiate this induced response. Jasmonic acid and the genes encoding proteins involved in its biosynthesis were identified in the trichomes.
  •  Using mutants, we showed that both jasmonic acid and trichomes were required for the contact-induced expression of PIN2. In addition, hydrogen peroxide, formed on the leaf surface, was required for PIN2 expression. Because these defenses would be activated before egg hatch, this early detection system for herbivores may be of considerable ecological significance.


Trichomes – the epidermal hairs found on the aerial surfaces of most plants – may serve as the first line of defense against herbivores and pathogens (Wagner, 1991; Goertzen & Small, 1993; Shepherd et al., 2005; Shepherd & Wagner, 2007). Trichomes occur in an array of varied forms and shapes and can interfere with herbivory through several mechanisms. High densities of trichomes on leaves and stems directly impede herbivore feeding by serving as a physical hindrance to their movement and feeding (Wilkens et al., 1996; Chu et al., 2003). Some forms of trichomes produce a sticky exudate that may physically entrap or further hinder the feeding and movement of herbivores (Nihoul, 1993; Van Dam and Hare, 1998a,b; Simmons et al., 2004; Medeiros & Tingey, 2006). Glandular trichomes found on many plants secrete antibiotic compounds that directly intoxicate arthropod herbivores (Ranger et al., 2004; Wang et al., 2004; Hare, 2005), although, in some cases, herbivores may be specifically adapted to glandular trichomes (Van Dam & Hare, 1998b;Hare & Elle, 2002). The source of these antibiotic compounds may result from hyperaccumulation of environmental contaminants and transport to the trichomes (e.g. selenium, Freeman et al., 2006; cadmium, Choi et al., 2004) or through synthesis of defensive compounds [e.g. phenolics, terpenes and proteinase inhibitors (Duffey & Isman, 1981; de Azevedo et al., 2003; Amme et al., 2005; Aziz et al., 2005; Liu et al., 2006]. In some instances, herbivores may utilize these terpenes as attractants for locating their host (Roda et al., 2003). Nevertheless, it is becoming increasingly appreciated that leaf surface traits, such as trichomes, may be as important determinants of herbivore performance as internal leaf chemistry (Mauricio & Rausher, 1997; Traw & Dawson, 2002b; Handley et al., 2005; Valkana et al., 2005; Clauss et al., 2006; Kivimaki et al., 2007). Hence, there has been growing interest in incorporating trichome-based resistance in plant breeding programs (Simmons & Gurr, 2005; Snyder et al., 2005; Simmons et al., 2006). Furthermore, molecular engineering of glandular trichome chemistry has been successfully targeted as a means to enhance pest resistance (Wang et al., 2004; Calo et al., 2006).

In some plant species, herbivory or defoliation triggers the production of increased numbers of trichomes on newly emerging leaves (Agrawal et al., 2002; Traw and Dawson, 2002a,b; Dalin & Bjorkman, 2003; Abdala-Roberts & Parra-Tabla, 2005), although, in other species (e.g. Datura wrightii; Hare & Walling, 2006), trichome phenotype is not inducible. The induction of trichome development has been shown, in some instances, to be under the control of the plant signals jasmonic acid (JA) (Traw & Bergelson, 2003; Li et al., 2004) or methyl jasmonate (MeJA) (Boughton et al., 2005), although not in all examples (Kazama et al., 2004; Hare & Walling, 2006). Because the induction of trichomes in new leaves often requires days to weeks to be observed, this form of induced resistance may be limited to affecting subsequent generations of the herbivore or other herbivore species which later colonize the host (Dalin & Bjorkman, 2003; Valkama et al., 2005).

Glandular trichomes in many plants are an important source of volatile terpenes (Iijima et al., 2004; Li et al., 2004) – compounds frequently implicated as indirect plant defenses because they can attract natural enemies to plants under attack (Thaler et al., 2002; Degenhardt et al., 2003; Kant et al., 2004; Schnee et al., 2006). Alternatively, tufts of nonglandular trichomes, termed ‘domatia’, may enhance biological control because they can serve as protective shelters for harboring natural enemies (Romero & Benson, 2005; English-Loeb & Norton, 2006; Matos et al., 2006). However, there is a significant body of work indicating that glandular trichomes can interfere with the success of natural enemies via entrapment of small parasitoids or predators (Farrar & Kennedy, 1991; Kashyap et al., 1991; Nihoul, 1994; Simmons & Gurr, 2004; Gassmann & Hare, 2005).

In this article, we propose a novel function for the glandular trichomes of the tomato (Solanum lycopersicum) plant – we suggest that, in addition to their well-recognized role in entrapping or impeding small insects (Simmons et al., 2004), glandular trichomes also function as an early detection system for herbivory. Our hypothesis is that glandular trichomes can serve as mechanoreceptors or sensors of herbivores. In our model, we propose that the rupture of the glandular trichomes following insect movement across the leaf releases plant defense signals that rapidly activate the expression of defense genes throughout the leaf. This sensing system could be argued to be nothing more than a curious phenomenon because herbivores would also trigger the defensive response on chewing. However, such an early detection system would have significant ecological meaning if it provides an early and reliable cue for subsequent herbivory. For example, cues triggered by a moth seeking to oviposit on the plant would be particularly valuable if defenses were activated and/or ‘primed’ before the hatching of the herbivore’s eggs, thus ensuring that newly emerging herbivores are then exposed to induced defenses.

We chose to test our hypothesis using the tomato plant, Solanum lycopersicum, where defense signaling has been well studied. JA has emerged as one of the primary signals mediating induced defenses against insects in this plant (Schilmiller & Howe, 2005; Wasternack et al., 2006). JA is formed in a series of reactions initially in the plastids – the initial reaction is the 13-lipoxygenase (LOX)-catalyzed insertion of O2 into the 13 position of linolenic acid to form the 13-hydroxyperoxide of linolenic acid. The hydroperoxide is further processed by allene oxide synthase and allene oxide cyclase to form 12-oxophytodienoic acid (ODPA). ODPA is then transported to the peroxisomes, where it is reduced by an 12-oxophytodienoate reductase 3 (OPR3); JA is then formed by carboxylic acid side-chain shortenings (Wasternack et al., 2006). H2O2, formed in response to wounding, acts as a secondary messenger for the induction of ‘late responding’ defense genes, such as proteinase inhibitors (e.g. PIN2) and polyphenol oxidase (Orozco-Cardenas et al., 2001). Systemin, once thought to be the primary mobile signal, probably functions locally to amplify JA synthesis (Schilmiller & Howe, 2005).

Seven types of trichome occur on various tomato species, including glandular trichomes (types VI and VII) that comprise four- to eight-celled heads (Thipyapong et al., 1997). Insect contact causes the rupture of the heads, releasing their cellular contents, which then polymerize, forming a sticky exudate on the mouthparts and legs of insects (Duffey, 1986; Steffens & Walters, 1991). In the case of small insects, this exudate causes entrapment and death of the insect (Simmons et al., 2004). In this study, we determined whether the rupture of trichomes, in addition to this direct role in plant defense, may play a further role in defense signaling against herbivores.

Materials and Methods

Plants and insects

Tomato (Solanum lycopersicum L.) seeds of cv. Castlemart and the def1 mutant were graciously provided by Gregg Howe, Michigan State University. Seeds of cv. Alisa Craig and the accession LA3610 were obtained from the C.M. Rick Tomato Genetics Resource Center, University of California, Davis, CA, USA. Tomato seeds of cv. Better Boy were obtained commercially. Cultivar Better Boy was used for all experiments, unless stated otherwise. Plants were grown as described previously (Peiffer & Felton, 2005) and were used at the four-leaf stage. Lepidopteran eggs (Helicoverpa zea (Boddie), Heliothis virescens (F.) and Manduca sexta (L.)) were obtained from the insectary at North Carolina State University. Larvae were reared until the desired stage on a wheat germ and casein-based artificial diet (Chippendale, 1970) with ingredients purchased from BIOSERV (Frenchtown, NJ, USA) or Sigma (St. Louis, MO, USA). Insects were kept at 27°C, with a 16 h photoperiod.

Effect of insect contact on PIN2 induction

To determine whether insect contact by crawling or walking induced PIN2 (a gene encoding a JA-regulated proteinase inhibitor; Graham et al., 1985) defense transcripts, fifth instar larvae or moths were allowed to walk on the fourth leaf of the tomato plant (cv. Better Boy) for 10 min. Induction of the PIN2 gene has been well correlated with increased resistance to insects (McGurl et al., 1994). Larvae were closely observed and, if any showed signs of feeding, regurgitating or defecating, they were discarded along with the plant. Plants remained in the glasshouse for the specified time, and then the treated leaves were harvested and immediately frozen in liquid nitrogen.

For quantitative real-time polymerase chain reaction (qRT-PCR), leaves were homogenized in liquid nitrogen and mRNA was purified with a RNeasy Plus Mini-kit (Qiagen, Valencia, CA, USA). One microgram of purified mRNA was used with a High Capacity cDNA Reverse Transcription kit (Applied Biosystems, Foster City, CA, USA) to create cDNA. qRT-PCR primers specific for PIN2 were designed using Primer Quest Software (Applied Biosystems) (Table 1). All reactions used Power SYBR Green PCR Master Mix and were run on a 7500 Fast Real-Time PCR System (Applied Biosystems) using a standard protocol (10 min at 95°C, followed by 40 cycles of 15 s at 95°C, 60 s at 60°C).

Table 1.   Primers used for real-time PCR assays of relative gene expression
Primer nameGene name/accession no.Primer sequence (5′–3′, forward/reverse)
PIN2Wound-inducible proteinase inhibitor 2/K03271GGA TTT AGC GGA CTT CCT TCT G/ATG CCA AGG CTT GTA CTA GAG AAT G

The housekeeping gene ubiquitin (Rotenberg et al., 2006) was used to normalize C(T) values. Relative quantifications, with untreated plants as the reference group, were then calculated using the 2−ΔΔC(T) method (Livak & Schmittgen, 2001). To validate this analysis method, primer efficiency was analyzed by comparing the normalized C(T) values of five serial dilutions of cDNA. Melting curve analysis was conducted to confirm the specificity of the primers.

Trichome purification

Trichomes were purified following well-established methods with slight modifications (Yerger et al., 1992; Slocombe et al., 2008; Besser et al., 2009). Briefly, to isolate glandular trichomes, leaflets were removed from the plant and placed in a 50 ml conical tube containing 1 g of glass beads (diameter, 4 mm) (Kimble Chase, Vineland, NJ, USA). Liquid nitrogen was added and the tube was shaken vigorously to shear trichomes off the leaf. The contents of the tube were then poured into a 100 μm cell strainer and rinsed with liquid nitrogen. The filtrate was collected in liquid nitrogen. When examined by light microscopy, this filtrate was observed to contain glandular trichomes. Because the leaf pieces and some nonglandular trichomes are larger, they remained in the strainer. The purity of the trichome preparation was assessed by microscopy to verify the presence of trichomes and the absence of leaf fragments. The isolated glandular trichomes were then homogenized and the mRNA was purified as described above. The expression of the trichome-specific genes sesquiterpene synthase (SST1) and monoterpene synthase (MTS1) was assayed to further confirm the purity of the trichome preparations (Besser et al., 2009).

To determine whether signaling genes are expressed in glandular trichomes, 1.0 μg of purified mRNA was reverse transcribed with M-MLV (Promega, Madison, WI, USA). Gene-specific primers (Table 1) were then employed to amplify the signaling genes in glandular trichomes using Go-Green Taq Master Mix (Promega). The PCR conditions were as follows: 5 min at 95°C, followed by 30 cycles of 1 min at 94°C, 45 s at 55°C, 1.25 min at 72°C, and then a 7 min extension at 72°C. PCR products were then electrophoresed on a 1.2% agarose gel, and DNA was visualized with SYBR Safe (Invitrogen, Eugene, OR, USA). To quantify gene expression in trichomes, qRT-PCR was performed as described previously.

In situ labeling of the JA signaling gene OPR3

In situ methods were based on the protocol published by Lopez et al. (2007) with minor modifications. OPR3 was cloned from Better Boy tomato leaves, using the TOPO TA cloning kit (Invitrogen). Digoxigenin (DIG)-labeled RNA probes were then generated using the MAXIscript kit (Ambion, Foster City, CA, USA) with DIG purchased from Roche (Penzberg, Germany).

To obtain trichomes for labeling, the fourth leaf on a four-leaf stage Better Boy tomato plant was fixed overnight at 4°C in 4% paraformaldehyde in diethylpyrocarbonate-treated water. Samples were then dehydrated through a graded series of ethanol, then Histosolve (Fisher Scientific, Kalamazoo, MI, USA) and finally embedded in Paraplast (Thermo Scientific, Waltham, MA, USA). All processing was performed in a Shandon Citadel 2000 paraffin processor (Thermo Scientific). Embedded leaves were cut into 10 μm sections with a Shandon Finesse paraffin microtome and dried onto Superfrost Plus slides (VWR Scientific, Pittsburgh, PA, USA).

Leaf sections were dewaxed in Histosolve, and then treated with 1 μg μl−1 of proteinase K (Ambion) for 15 min at 37°C. Sections were then fixed in 4% paraformaldehyde in phosphate-buffered solution for 5 min, dehydrated in an ethanol series and air dried. Sections were hybridized overnight at 42°C in a moist chamber with 400 ng μl−1 of probe in hybridization solution (300 mm NaCl, 10 mm Tris–HCl pH 6.8, 5 mm ethylenediaminetetraacetic acid, 50% formamide, 1 × Denhardt’s, 10% dextran sulfate and 1 mg ml−1 yeast tRNA). For a negative control, some slides were incubated with sense probes or with hybridization solution lacking a probe. Sections were washed twice in 2 × standard saline citrate with 50% formamide for 30 min each. Nonspecific binding sites were blocked with 1% blocking reagent (Roche) in 100 mm Tris–HCl pH 7.5 and 150 mm NaCl for 30 min, and then in 100 mm Tris–HCl pH 7.5, 150 mm NaCl, 1% bovine serum albumin and 0.3% Triton X-100 for 30 min. Sections were incubated with anti-DIG conjugated to alkaline phosphatase (Roche) for 1 h, and then visualized with nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (Vector Laboratories, Burlingame, CA, USA). They were counterstained with methyl green and mounted in Vectamount (Vector Laboratories).

JA extraction and quantification

To measure JA in trichomes, trichomes were purified from either stem or leaves as described previously, and then immediately transferred into FastPrep® tubes (Qbiogene, Carlsbad, CA, USA) containing 1 g of Zirmil beads (1.1 mm; Saint-Gobain ZirPro, Mountainside, NJ, USA), and frozen at −80° C until processing. For leaf samples, 100 mg of leaf tissue was processed in the same way.

To extract and detect JA, we used a previously described method (Tooker & De Moraes, 2005, 2009) which was modified slightly from Schmelz et al. (2003, 2004). Briefly, this method uses gas chromatography-mass spectrometry (GC-MS) with isobutane chemical ionization and selected-ion monitoring to detect JA. Jasmonate was derivatized to MeJA and isolated using a vapor-phase extraction. Quantification of MeJA was based on dihydro-JA, which was added to the samples before processing. By processing samples in the absence of the derivatizing agent and not then recovering MeJA, we verified that MeJA measured in these control samples was derivatized from JA, and was not itself present in the samples.

H2O2 and catalase treatments

H2O2 on the leaf surface was visualized by pressing leaves against filter paper. The paper was then incubated with 1 mg ml−1 diaminobenzadine (DAB; Sigma), pH 3.8 for 20 min to visualize H2O2. To verify that catalase could enzymatically remove H2O2 on the leaf surface, leaves were sprayed with 0.05 μg catalase (Sigma) before pressing onto filter paper and visualization by DAB. Control leaves were sprayed with water.

To study the role of H2O2 from trichome disruption on PIN2 induction, trichomes were first disrupted by rubbing with a latex-gloved finger. This treatment ruptures the delicate glandular trichomes, but does not damage the epidermal cells, as observed under the stereomicroscope. Catalase-treated leaves were then immediately sprayed with 0.05 μg catalase until runoff. After 24 h, treated leaves were harvested and RNA was isolated and quantified as described above.

Dependence of touch-induced responses on JA and trichomes

In order to determine whether the touch-induced induction of PIN2 requires JA, we repeated the touch experiments as described above with the mutant def1, which is compromised in its ability to accumulate JA because of disruption in metabolism between the synthesis of hydroperoxylinolenic acid and 12-oxo-phytodienoic acid (Howe et al., 1996). In these experiments, we compared the wild-type cv. Castlemart with the mutant def.

However, the def1 mutant also has a phenotype with a reduction in type VI glandular trichomes by nearly 65% (Castlemart, 1176 cm−2; def1, 413 cm−2). Therefore, to further uncouple the effects of JA and glandular trichome density, we screened several accessions from the C.M. Rick Tomato Genetics Resource Center, University of California, Davis, CA, USA, which were described as having a reduction in glandular trichomes. One accession, LA3610, had a nearly 37% reduction in type VI trichomes (mean of 742 cm−2) compared with its wild-type parent, cv. Alisa Craig (mean, 1173 cm−2). This accession retained a similar level of wound-induced PIN2 expression compared with its parent line when damaged mechanically with a cork borer (763 ± 154 vs 1063 ± 132, > 0.05). We repeated the touch experiments with this accession and compared the response with its wild-type parent cv. Alisa Craig.

Induction of trichomes and PIN2 induction

Previously, we have shown that MeJA treatment of tomato plants induces increased glandular trichome densities on newly forming leaves for nearly 30 d following treatment (Boughton et al., 2005). In order to determine whether the induction of trichomes could prime or increase the response of the leaves to touch-induced increases in PIN2, we treated plants with MeJA before touch treatments of the plants. Tomato plants (cv. Better Boy) were grown to the four-leaf stage as described previously. The plants were then sprayed to runoff with 7.5 mm MeJA (Sigma) in 0.8% ethanol in water. Control plants were sprayed with 0.8% ethanol. Two weeks after spraying with MeJA, glandular trichomes were disrupted by rubbing with a gloved finger and, after 24 h, PIN2 expression was quantified by qRT-PCR as described previously. To normalize the data and stabilize the variance, data from this experiment were natural-log transformed, and treatment means were compared by analysis of variance (ANOVA).


Insect contact and induction of plant defense genes

To determine whether plants respond to insect contact, we allowed a fifth instar hornworm Manduca sexta to walk on an upper fully developed leaflet (four- to five-leaf stage plants) for 10 min. As a positive control, the experiment was also conducted with plants in which a single leaflet was lightly touched by an investigator using a latex-gloved forefinger and thumb. In a second experiment, fifth instar tobacco budworm larvae Heliothis virescens and the tomato fruitworm moths Helicoverpa zea were allowed to walk for 10 min. In each case, we closely examined the leaves to make sure that no feeding, secretion, defecation or oviposition occurred during the 10 min period. Twenty-four hours after insect contact, leaflets were collected and assayed for PIN2 using qRT-PCR. In all cases, contact by the insects triggered a significant induction of PIN2 transcripts (ANOVA, < 0.05; Fig. 1a,b).

Figure 1.

 Effect of insect contact on the expression of the plant defense gene PIN2. Expression of PIN2 in tomato (Solanum lycopersicum) leaves, 24 h after treatment. Expression is calculated relative to untreated control leaves. (a) Experiment conducted in November with Manduca sexta larvae. Disruption, positive control. (b) Experiment conducted in August with larval Heliothis virescens and Helicoverpa zea moths. The mean fold induction and standard error for five independent biological replicates are shown. (c–f) Time course experiment with H. zea moths and larvae. Moth, squares; caterpillar, circles; researcher, triangles. PIN2, proteinase inhibitor 2 gene; OPR3, oxophytodienoate reductase 3 gene; MTS1, monosesquiterpene synthase 1 gene; SST1, sesquiterpene synthase 1 gene. The mean fold induction and standard error for three independent biological replicates are shown.

In the second set of experiments, performed in August, the levels of induction were considerably higher than experiments conducted in the winter (see Fig. 1b). Because plants were not grown under full-spectrum lighting, we can assume that the UV intensity is lower in winter in Pennsylvania State. Because UV radiation amplifies wound-inducible proteinase inhibitor accumulation (Stratmann et al., 2000), this could account for the observed differences. Alternatively, it may represent differences among insect species.

A third set of experiments was conducted to compare moth and caterpillar walking at earlier time points (Fig. 1c–f). Helicoverpa zea moths and fourth instar caterpillars were allowed to walk on tomato leaves, as described previously. To avoid any possible interaction or contamination from latex gloves, the researcher disrupted plants by gently rubbing with a metal rod without rupturing the leaves. The metal was cleaned with ethanol after each plant. Although PIN2 relative expression increased at 3, 6 and 12 h, differences among treatments were not significant (Fig. 1c; ANOVA, > 0.05). Three hours after treatment, OPR3 was higher in both leaves walked on by caterpillars and moths, but not by those touched by researchers (Fig. 1d; ANOVA, < 0.05). The trichome-specific gene MTS1 also increased 3 h after moths walked on leaves, and then returned to levels similar to untouched controls at 6 and 12 h (Fig. 1e; ANOVA, < 0.05). After caterpillar walking, MTS1 transcripts increased at 3 and 6 h, but were not significantly different from untouched controls (ANOVA, > 0.05). Another trichome-specific gene SST1 was not significantly different from untreated controls in any of the treatments (Fig. 1f; ANOVA, > 0.05).

JA signaling genes in trichomes

Because the induction of PIN2 is known to be regulated by JA (Li et al., 2004), we examined the presence of key JA signaling genes in leaf trichomes. We isolated glandular trichomes from leaves and verified the purity of the preparation using microscopy to ensure that trichomes were present and that other cellular debris, such as leaf fragments, was absent (Fig. 2a). RNA was extracted from the trichome preparation and reverse transcriptase-polymerase chain reaction (RT-PCR) was performed to determine the presence of transcripts encoding JA signaling genes (Fig. 2c). The results of RT-PCR indicated that each of the genes encoding enzymes involved in JA biosynthesis was expressed in the glandular trichomes for both stems and leaves. The prosystemin gene, which encodes for the peptide defense hormone, systemin, was also expressed. Furthermore, WFI1, which encodes a plant respiratory burst oxidase that produces reactive oxygen species, was expressed in trichomes (van de Ven et al., 2000; Sagi et al., 2004). Gene expression was quantified by qRT-PCR in leaves and in trichomes from both stems and leaves. Again, expression for all of the signaling genes tested was verified. Of the signaling genes identified in Fig. 2c, most had comparable expression levels in the trichomes compared with the leaves (Fig. 3a–f). The expression of the allene oxide synthase gene (AOS) was relatively high in the leaf trichomes compared with leaf tissues. Two trichome-specific genes, MTS1 and SST1, were highly expressed in trichomes as expected (Fig. 3g,h). MTS1 was three-fold higher in stem trichomes than in leaves; whereas SST1 was 490-fold higher.

Figure 2.

 Expression of jasmonic acid (JA) signaling genes in glandular trichomes of tomato (Solanum lycopersicum). (a) Light microscopy of trichome extracts indicates that the sample consists of glandular trichome heads and pieces of stalk and is free of leaf debris. (b) In situ labeling indicates that OPR3 is expressed in the head of glandular trichomes. (c) Expression of JA signaling genes as shown by endpoint reverse transcriptase-polymerase chain reaction (RT-PCR). ‘No RT’ samples indicate pure RNA, free of genomic DNA. LOXD, lipoxygenase D gene; AOS, allene oxide synthase gene; AOC, allene oxide cyclase gene; PSYS, prosystemin gene; WFI1, whitefly-induced gp91-phox (a putative NADPH oxidase) gene; OPR3, oxophytodienoate reductase 3 gene.

Figure 3.

 Relative expression of jasmonic acid (JA) signaling genes in trichomes from stem and leaves of tomato (Solanum lycopersicum). (a–h) Trichomes were extracted from stems or leaves and compared with whole stems and leaves with trichomes removed. Expression is calculated relative to leaves with trichomes removed. Trichome samples represent a pooled RNA sample comprising the trichomes extracted from 30 plants. Bars indicate the minimum and maximum relative expression calculated from the standard deviation of three technical replicates. LOXD, lipoxygenase D gene; AOS, allene oxide synthase gene; AOC, allene oxide cyclise gene; PSYS, prosystemin gene; OPR3, oxophytodienoate reductase 3 gene; WFI1, whitefly-induced gp91-phox gene; MTS1, monosesquiterpene synthase 1 gene; SST1, sesquiterpene synthase 1 gene.

We also used in situ hybridization to determine the localization of one of the JA signaling genes: OPR3. As indicated in Fig. 2b, OPR3 is expressed in both types of leaf glandular trichome. The highest expression appears in the cells comprising the glandular ‘heads’ of the trichomes compared with the stalks of the trichomes. These results provide further evidence indicating the expression of JA signaling genes in the glandular trichomes.

Measurement of JA in trichomes

In addition to examining the expression of JA signaling genes in trichomes, we also determined the presence of JA in trichomes of noninduced plants by GC-MS. Our results indicate the positive identification of JA in trichome preparations. Trichomes extracted from stem tissue were found to contain significant amounts of JA, with a mean of two independent isolations of 715.96 ng g−1 trichomes (= 2, SE = 85.5). Trichomes extracted from leaves also contained significant amounts of JA, with a mean of 114.1 ng g−1 trichomes (= 3, SE = 22.9). We also measured JA in two mutants and their wild-type control lines (= 3 for each), and further confirmed the presence of JA in trichomes. Amounts of JA in trichomes across the four lines were statistically similar (ANOVA, F3,12 = 0.98, = 0.45; Castlemart, 1850 ± 27 ng JA g−1 trichomes; def1, 122 ± 42 ng JA g−1; Alisa-Craig, 105 ± 28 ng JA g−1; LA3610, 230 ± 101 ng JA g−1). Extracts from noninduced leaves contained JA with a mean of 1.6 ng g−1 leaves (= 5, SE = 0.7). The levels of JA in unwounded leaves were very similar to values previously reported for tomato (Schmelz et al., 2003; Li et al., 2005; Miersch et al., 2008).

H2O2 and the induction of defense genes

In the light of the observations that ‘rubbing’ tomato stems caused rapid formation of reactive oxygen species, such as H2O2 (Depege et al., 2000), and that H2O2 is a secondary messenger for the induction of PIN2 genes (Orozco-Cardenas et al., 2001), we examined the role of trichome disruption in the formation of H2O2 (Fig. 4a,b). Our results using the DAB stain for the detection of H2O2 demonstrated that ‘touching’ of trichomes rapidly released H2O2 on the leaf surface, and that H2O2 was a key signal in mediating the induction of pin2. The areas of greatest detection of H2O2 occurred in the sections of the leaflet with the highest concentration of glandular trichomes (i.e. near the midrib and basal parts of the leaflet; Fig. 4a). Catalase, an enzyme that catalyzes the decomposition of H2O2 to H2O and O2, effectively prevents PIN2 induction when sprayed on the leaf surface immediately following touching (ANOVA, < 0.05; Fig. 4c). The leaflet illustrated in Fig. 4b shows that pretreatment of the leaf with catalase effectively eliminated most of the H2O2 detected by the DAB stain. These combined experiments indicate that H2O2 and JA are key signals mediating the insect contact-induced expression of defense genes.

Figure 4.

 Hydrogen peroxide (H2O2) is a defense signal produced by ruptured trichomes. (a) Visualization of H2O2 released when trichomes were ruptured by pressing tomato (Solanum lycopersicum) (cv. Better Boy) leaves onto filter paper impregnated with diaminobenzidine (DAB). (b) Leaves treated with catalase before rupturing the trichomes, showing reduced DAB staining. (c) Effect of catalase post-treatment on touch-induced expression of proteinase inhibitor 2 gene (PIN2) in leaves. The mean fold induction and standard error for five independent biological replicates are shown.

Dependence of touch-induced responses on JA and trichomes

In these experiments, we tested the touch response of the def1 mutant which is impaired in JA biosynthesis. The wild-type cv. Castlemart displayed a very strong response to touch, but the def1 mutant was not inducible by this treatment (ANOVA, < 0.05; Fig. 5). These results indicate that the touch-induced PIN2 response is dependent on JA.

Figure 5.

 Touch-induced expression of proteinase inhibitor 2 gene (PIN2) in jasmonic acid (JA) and trichome mutants. Relative expression of PIN2 in tomato (Solanum lycopersicum) leaves, 24 h after trichomes were disrupted by rubbing with a gloved finger. def1 is a mutant impaired in JA signaling (cv. Castlemart is the parent wild-type). LA3610 is an accession with reduced glandular trichome density (in background genotype cv. Alisa Craig). The mean fold induction and standard error for five independent biological replicates are shown.

However, the def1 mutant also displays a reduction in type VI glandular trichomes. Therefore, we conducted further touch experiments with the accession LA3610, which has a nearly 37% reduction in type VI trichomes, and found a significant (< 0.05) reduction in touch-induced PIN2 (Fig. 4) compared with the wild-type parent cv. Alisa Craig. The combined experiments with the def1 and LA3610 mutants indicate that both JA and glandular trichomes are required for touch-induced expression of the defense gene PIN2.

Induction of trichomes and touch-induced PIN2

We used MeJA application to first induce increases in glandular trichomes in new leaves and then to test the responsiveness of induced leaves to touch. Two weeks after plants (cv. Better Boy) were sprayed with MeJA, they showed significantly more trichomes than control plants (control, 219 cm−2; MeJA treated, 1121 cm−2; < 0.001). We then disrupted the trichomes by touching with a gloved finger and quantified PIN2 levels as described previously. Plants varied significantly in their response to MeJA and physical disruption (Fig. 6; overall two-way ANOVA on natural log-transformed data: F3,22 = 16.8, < 0.0001). Plants treated with MeJA showed higher basal levels of PIN2 than untreated or disrupted controls even after the 2 wk treatment (Fig. 6; significant MeJA term: F1,22 = 43.7, P < 0.0001; significant disruption term: F1,22 = 8.5, P = 0.009). Notably, the MeJA-treated plants, when touched, showed PIN2 levels 58-fold higher than those of untreated controls (Tukey’s honestly significant difference test, < 0.05). These MeJA-treated and touched plants also appeared to have three-fold higher levels of PIN2 expression than plants treated only with MeJA, but the high variation rendered the statistical comparison insignificant (Fig. 6; ‘MeJA × disruption’ interaction term: F1,22 = 0.2, = 0.69). Taken together, these results suggest that increases in trichomes from the action of jasmonates may contribute to a greater sensitivity to touch-induced responses.

Figure 6.

 Increased touch-induced proteinase inhibitor 2 gene (PIN2) in tomato (Solanum lycopersicum) plants primed with methyl jasmonate (MeJA). Relative expression of PIN2 in plants treated with MeJA to induce trichomes, with trichomes then disrupted by rubbing with gloved finger. The mean fold induction and standard error for six independent biological replicates are shown. Bars with different letters are significantly different (Tukey’s honestly significant difference test < 0.05; see text for details on statistics).


Most plants possess the ability to perceive and respond to touch or other mechanical stimuli (Braam, 2005; Telewsk, 2006). Greater than 2.5% of Arabidopsis genes are rapidly up-regulated in touch-stimulated plants (Lee et al., 2005). Signaling molecules, such as reactive oxygen species, calcium, ethylene and jasmonates, have been shown to be involved in these touch responses (Lee et al., 2005). However, the role of touch in influencing plant–herbivore interactions has only recently become a topic of interest (Niesenbaum et al., 2006). For instance, physical handling of plants in the field may negatively or positively affect rates of herbivory, but, in these cases, the mechanistic basis for the responses has not been indicated (Cahill et al., 2001; Niesenbaum et al., 2006). Most lepidopteran herbivores initiate contact and attack on plants by laying eggs on them. Egg-laying can induce defenses in cases in which an ovipositor is used to insert eggs into the plant tissue, or when elicitors are present on the egg surface or in other accessory gland secretions (Hilker & Meiners, 2006). Less invasive contact, such as ‘caterpillar footsteps’, has been shown to cause localized cell death, and the production of superoxide radical (O2) and 4-aminobutyrate (Bown et al., 2002; Hall et al., 2004). These responses have been postulated to be caused by the crochets on the caterpillar prolegs which can rupture the leaf surface (Hall et al., 2004). The crochets may be responsible for the responses observed with caterpillars in this study, but cannot account for the responses observed with adult moths which lack crochets and prolegs. However, the impact of insect footsteps on the induction of defense genes has not been reported in this series of papers. The best-known example in the plant kingdom in which plants respond rapidly to insect contact is the mechanical sensing of insects by the carnivorous Venus fly trap Dionaea muscipula (Forterre et al., 2005). The trap closes rapidly when a small clump of leaf hairs on the inside surface of the trap is mechanically stimulated by insects (Forterre et al., 2005).

Although plants are known to respond to both oviposition and feeding by herbivores (Hilker & Meiners, 2002), our findings indicate that the tomato plant can respond to direct contact by herbivores. Herbivore-derived cues, such as saliva and oral secretions, may provide additional feeding cues that mediate plant defensive responses (Zhu-Salzman et al., 2008). In this study, we found that herbivore contact alone was sufficient to directly induce defensive genes regulated by the JA pathway. These experiments excluded the possibility that feeding, oviposition or oral secretions contributed to these responses. As a positive control to demonstrate that herbivore-derived cues (e.g. cuticular chemicals) were not required, we used touch with a gloved hand or a metal rod to similarly induce defense gene expression. We do not know whether natural enemies, such as predators, would similarly induce defenses, although it is likely.

Our contention that the trichomes of tomato are a key component in touch-induced defensive responses is bolstered by the combined evidence of expression profiling of JA signaling genes in the trichomes, the spatial localization of OPR3 in the glandular heads of trichomes, the presence of significant amounts of JA in the trichomes, the release of H2O2 on rupture of the trichomes and the experiments with a glandular trichome mutant. We propose that the trichomes of the tomato plant are poised on contact to release a full complement of defense signals which then trigger a defensive cascade. All of the genes involved in the JA signaling pathway tested were expressed in the trichomes. The use of mutants (compromised in JA biosynthesis or with reduced glandular trichome production) indicated that both trichomes and JA were required for the touch-induced expression of the PIN2 gene. We did not observe significant differences in the levels of JA in trichomes between the def1 mutant and its parent wild-type Castlemart, but this was not unexpected because foliar JA levels in unwounded plants have been shown to be similar in these two lines (Howe et al., 1996). However, it should be pointed out that the number of trichomes varies significantly between lines, so that less JA would be expected to be found as a function of leaf area.

We also showed that H2O2 formed on the leaf surface following rupture of the trichomes was required for PIN2 induction. The results were bolstered by the finding that H2O2 is a downstream, secondary messenger for the regulation of PIN2 (Orozco-Cardenas et al., 2001). The presence of a putative H2O2-forming, respiratory burst oxidase (WFI1) (Fig. 2), which is expressed in the trichomes, provides support that the enzymatic machinery for H2O2 formation is present in the trichomes, although many enzymes may contribute to peroxide formation.

The fact that insect feeding (G. W. Felton, unpublished data) or the application of MeJA induced increases in glandular trichome density in newly emerging leaves indicates a possible mechanism for ‘immunological memory’ or priming of plant defenses (Karban, 2008). Priming is a readying of defenses for enhanced induction of resistance to subsequent attack (Conrath et al., 2006; Ton et al., 2007). Thus, our ‘primed’ plants (those pretreated with MeJA to induce trichomes) showed enhanced induction in response to touch. This is a different proposed mechanism for priming than has been shown in previous studies, where green leafy volatiles are often implicated as the priming agents (Engelberth et al., 2004; Kessler et al., 2006).

The defensive role of glandular trichomes in tomato has been a subject of extensive research for decades (Williams et al., 1980; Liedl et al., 1995; Simmons & Gurr, 2005; Lobato-Ortiz et al., 2007). These trichomes may serve as a physical hindrance to insect movement (Simmons et al., 2004), and may be the source of toxicants (Williams et al., 1980; Lobato-Ortiz et al., 2007), antinutritional proteins (Thipyapong et al., 1997) and other feeding repellents (Frelichowski & Juvik, 2001). More recently, glandular trichomes have been shown to produce terpenes, which may mediate indirect defenses by attracting natural enemies of herbivores (Li et al., 2004). In this study, we identified another defensive role for glandular trichomes in sensing herbivores. In the coevolutionary arms race between plants and herbivores, this represents a newly recognized plant strategy for early alert to herbivory. Moth contact on the leaf was shown to be sufficient to up-regulate defense transcripts; thus, defenses are activated before hatching of the moth’s eggs and would provide extra protection against newly emerging larvae. How broadly this strategy is employed by other plant species remains to be determined.


The support of the USDA NRI is greatly appreciated (2005–35607–15242 and 2007–35302–18218 awarded to GWF and 2006–01823 awarded to JFT). Seeds of the def1 mutant were kindly provided by Gregg Howe, Michigan State University. Seeds for the LA3610 mutant were provided by the C.M. Rick Tomato Genetics Resource Center at the University of California, Davis, CA, USA. All microscopy was performed at the Cytometry Facility, University Park (Huck Institutes of the Life Sciences, Pennsylvania State University). This project was funded, in part, by a grant with the Pennsylvania Department of Health using Tobacco Settlement Funds. The Department specifically disclaims responsibility for any analyses, interpretations or conclusions.