The phytotoxin coronatine and methyl jasmonate impact multiple phytohormone pathways in tomato


(fax 405 744 9945; e-mail


Coronatine (COR) is a phytotoxin produced by several pathovars of Pseudomonas syringae and consists of coronafacic acid (CFA), an analog of methyl jasmonic acid (MeJA), and coronamic acid (CMA), which resembles 1-aminocyclopropane-1-carboxylic acid (ACC), a precursor to ethylene. An understanding of how COR functions, is perceived by different plant tissues, and the extent to which it mimics MeJA remain unclear. In this study, COR and related compounds were examined with respect to structure and function. The results indicate that conjugation of CFA to an amino acid is required for optimal activity in tomato, including chlorosis, changes in chloroplast structure, cell wall thickening, accumulation of proteinase inhibitors, induction of anthocyanins, and root growth inhibition. cDNA microarrays were utilized to understand the molecular processes that are regulated by MeJA, COR, CFA and CMA in tomato leaves. A comparison of COR- and MeJA-regulated transcriptomes revealed that COR regulated 35% of the MeJA-induced genes. There was significant overlap in the number of COR and CFA-regulated genes with CFA impacting the expression of 39.4% of the COR-regulated genes. Taken together, the results of biological assays, ultrastructural studies, and gene expression profiling demonstrate that: (1) the intact COR molecule impacts signaling in tomato via the jasmonic acid, ethylene, and auxin pathways; (2) CMA does not function as a structural analog of ACC; (3) COR has a broader range of functions than either CFA or CMA; and (4) COR and MeJA share similar, but not identical activities and impact multiple phytohormone pathways in tomato.


Coronatine (COR) is a non-host-specific phytotoxin produced by several members of the Pseudomonas syringae group of pathovars (Bender et al., 1999; Cintas et al., 2002; Mitchell, 1982). The toxin acts as a virulence factor in P. syringae pv. tomato, allowing the pathogen to obtain higher population densities and develop larger lesions than COR-defective strains (Bender et al., 1987; Brooks et al., 2004; Mittal and Davis, 1995; Peñaloza-Vázquez et al., 2000). The most prominent symptom observed in leaf tissue treated with COR or infected with COR-producing strains of P. syringae is an intense spreading chlorosis (Gnanamanickam et al., 1982); this has been attributed to a loss of chlorophylls a and b in tomato and tobacco (Kenyon and Turner, 1990; Palmer and Bender, 1995). COR was shown to induce the expression of chlorophyllase, the first enzyme in the chlorophyll degradation pathway (Benedetti and Arruda, 2002; Tsuchiya et al., 1999). Although the physiological significance of rapid chlorophyll breakdown is still unknown, it may be correlated with the rapid senescence of leaves exposed to COR (Tsuchiya et al., 1999).

In addition to chlorosis, COR induces a wide array of effects in plants including anthocyanin production, alkaloid accumulation, ethylene emission, tendril coiling, inhibition of root elongation, and hypertrophy (Bender et al., 1999; Feys et al., 1994; Lauchli and Boland, 2003; Weiler et al., 1994). COR also induces volatile production and the accumulation of proteinase inhibitors, which are stress responses commonly associated with herbivory (Feys et al., 1994; Lauchli and Boland, 2003; Palmer and Bender, 1995). Other responses associated with COR include apoptotic cell death and the accumulation of flavonoid phytoalexins, responses that may be unique to grasses and other monocots (Tamogami and Kodama, 2000; Yao et al., 2002). Recently, exogenous methyl jasmonate (MeJA), a compound similar to COR in both structure and function, accumulated in the apoplast of tomato, but in the cytoplasm of barley cells (Bücking et al., 2004); thus it is possible that COR may act differently in monocots and dicots.

COR consists of the polyketide coronafacic acid (CFA) (Parry et al., 1994; Rangaswamy et al., 1998), and coronamic acid (CMA), a cyclized derivative of isoleucine (Mitchell, 1985). CFA and CMA function as biosynthetic intermediates and are joined together by an amide linkage to form the parent compound, COR (Ichihara et al., 1977) (Figure 1a). CMA is a structural analog of 1-aminocyclopropane-1-carboxylic acid (ACC) (Figure 1d), an intermediate in the pathway to ethylene in higher plants (Ecker, 1995). It has also been noted that COR is a structural and functional analog of jasmonic acid (JA) and related signaling compounds such as MeJA and 12-oxo-phytodienoic acid (12-OPDA), the C18 precursor of JA/MeJA (Feys et al., 1994; Weiler et al., 1994). OPDA, JA, MeJA (Figure 1d) and other octadecanoids impact the regulation of diverse plant responses including biotic stress (Farmer et al., 2003), wounding (Howe and Schilmiller, 2002), abscission (Burns et al., 2003), and volatile production (Weber, 2002). The identification of the Arabidopsis coi1 (coronatine insensitive) mutant supports the hypothesis that COR is a functional analog of MeJA (Feys et al., 1994). More recently, a JA-insensitive mutant (jai1) of tomato, which is also insensitive to COR, was identified as a homolog of the Arabidopsis COI1 (Li et al., 2004; Zhao et al., 2003). jai1 plants, like the Arabidopsis coi1 mutant, are insensitive to COR and exhibit resistance to strains of P. syringae that produce COR (Zhao et al., 2003).

Figure 1.

Compounds used in structure-function studies.
(a) CFA, coronafacic acid; CMA, coronamic acid; COR, coronatine.
(b) CPE, 2-[1-oxo-2-cyclopenten-2-ylmethyl]butanoic acid; CFV, coronafacoylvaline.
(c) COR-OH, coronatinol; COR-OAc, acetyl coronatinol; COR-Me, methyl coronatine; COR-OSuc succinyl coronatinol.
(d) MeJA, methyl jasmonate; OPDA, 12-oxo-phytodienoic acid; ACC, 1-aminocyclopropane-1-carboxylic acid.

Although COR is involved in various physiological responses, we do not understand how COR is perceived in different tissues, precisely how it functions, and to what extent it mimics MeJA. Previous reports have documented the production of ethylene in COR-treated tissue (Ferguson and Mitchell, 1985; Kenyon and Turner, 1992), a response that may be attributed to the structural similarities between CMA and ACC. In the present study, we investigate whether each component of COR has biological function in planta. We set out to answer these questions using biological assays and cDNA microarrays to identify the molecular responses associated with COR, CFA, and CMA. Furthermore, we compared the biological activity of naturally occurring compounds derived from COR biosynthesis, chemically modified derivatives of COR, and functional analogs of CFA and CMA (e.g., MeJA and ACC) (Figure 1). The results suggest that CFA mimics MeJA, whereas the coronafacoyl conjugates formed between CFA and other amino acids (including CMA) enhance toxicity by inducing chlorosis and changes in chloroplast structure. Furthermore, biological assays, ultrastructural studies, and gene expression studies suggest that COR induces JA biosynthesis, impacts signaling in tomato via multiple phytohormone pathways, and has a wider range of biological functions than CFA or CMA.


Cytological studies were conducted to determine the response of tomato tissue to COR derivatives, and tomato seedlings were monitored for anthocyanin accumulation and root inhibition in response to COR-related compounds. The molecules investigated for biological activity in tomato include 2-[1-oxo-2-cyclopenten-2-ymethyl]-butanoic acid (CPE), a potential intermediate in the CFA pathway (Mitchell et al., 1995), and coronafacoylvaline (CFV) (Mitchell, 1984), a naturally occurring coronafacoyl conjugate (Figure 1b). Other compounds investigated for biological activity include chemical modifications of the CFA or CMA moiety including: hydroxylated COR (coronatinol, COR-OH), succinylated coronatinol (COR-OSuc), acetylated coronatinol (COR-OAc) and methyl-esterified COR (COR-Me) (Figure 1c). MeJA and ACC, which are mimics of CFA and CMA, respectively, were included for comparative purposes (Figure 1d).

Visual observations of treated tomato leaves

Leaves treated with COR-related compounds (20 nmol) were evaluated for chlorosis 5 days after treatment. COR-treated plants exhibited moderate to severe yellowing with chlorosis spreading 5–10 mm from the application site. In leaves treated with COR-Me, COR-OH, COR-OSuc, COR-OAc, and CFV, the chlorotic area was consistently smaller, ranging 1–5 mm from the inoculation site. Plants treated with CMA, CPE, and ACC showed slight burning at the inoculation site, but were not chlorotic. No chlorosis or burning was observed on plants treated with H2O, CFA, CFA + CMA, MeJA, or ACC (Table 1).

Table 1.  Effects of coronatine (COR) derivatives and analogs on tomato leaf tissue
CompoundaChlorosisbChloroplast ultrastructurecEpidermal cell walldNo. serine proteinase inhibitor I particleseNo. multicystatin particlese
  1. a20 nmol of each compound was applied to tomato leaf tissue as described in Experimental procedures.

  2. b(++), chlorotic zone was 5–10 mm in diameter; (+), chlorotic zone was 1–5 mm; ND, no detectable chlorosis.

  3. cChloroplasts were categorized as being normal in size or smaller than normal and located near the bottom of the palisade mesophyll cells (shrunken and descended). One hundred tomato cells were analyzed from triplicate samples of two independent experiments.

  4. dUpper epidermal cell walls were considered thickened if they were at least two times thicker than those present in control tissue when viewed by light microscopy. One hundred tomato cells were analyzed from triplicate samples of two independent experiments.

  5. eAverage numbers of spherical (serine protease inhibitor I) and polygonal (multicystatin) proteinaceous deposits in the vacuole of tomato cells treated with 1 nmol of the indicated compounds. Values represent the number of deposits present in 100 palisade mesophyll cells; three sets of 100 cells were counted for each treatment and the number represents the mean ± standard deviation.

COR++Shrunken, descendedThickened1594 ± 14360 ± 4
MeJANDNormalNormal401 ± 232 ± 2
CFANDNormalNormal225 ± 301 ± 1
CMANDNormalNormal60 ± 70
CFA + CMANDNormalNormal199 ± 231 ± 0.5
ACCNDNormalNormal84 ± 70
COR-Me+Shrunken, descendedThickened862 ± 3484 ± 10
COR-OH+Shrunken, descendedThickened956 ± 43131 ± 9
COR-OSuc+Shrunken, descendedThickened868 ± 29123 ± 8
COR-OAc+Shrunken, descendedThickened722 ± 7273 ± 14
CFV+Shrunken, descendedThickened888 ± 1381 ± 9
CPENDNormalNormal682 ± 2420 ± 4

Effect on tomato cell ultrastructure

In a previous study, the application of COR to tomato leaves induced the thickening of epidermal cell walls and changes in chloroplast ultrastructure (Palmer and Bender, 1995). In the current study, we observed these responses in tissue treated with COR, which is consistent with our previous study, as well as with the naturally occurring COR analog, CFV, and the COR derivatives COR-Me, COR-OH, COR-OSuc, and COR-OAc (Table 1). In contrast to chloroplasts in healthy tissue, the chloroplasts of tomato leaves treated with COR, CFV, COR-Me, COR-OH, COR-OSuc, and COR-OAc were smaller, stained more intensely, and were generally clustered near the bottom of palisade mesophyll cells (Table 1).

Previously, we reported the accumulation of spherical and polygonal proteinaceous deposits in the vacuoles of COR-treated tomato tissue (Palmer and Bender, 1995; Zhao et al., 2001). Tomato tissue containing these particles contained proteinase inhibitor activity, and immunogold labeling confirmed that the spherical particles were serine proteinase inhibitor I (Zhao et al., 2001). In the present study, particles of serine proteinase inhibitor I (Figure 2a) were observed in tissue treated with all compounds except H2O (Table 1). The number of serine proteinase inhibitor I particles was highest in COR-treated tissue (1594 ± 143), whereas tissue treated with COR-Me, COR-OH, COR-OSuc, COR-OAc, CFV, and CPE contained approximately 50% the number of particles observed in COR-treated tissue. The remaining compounds are listed in descending order with respect to the number of serine proteinase inhibitor particles per 100 cells: MeJA, CFA, CFA + CMA (combined treatment), ACC, and CMA (Table 1).

Figure 2.

Cytology of COR-treated tomato leaf tissue.
(a) COR-treated tomato cell containing spherical particles of serine proteinase inhibitor I (PI-I).
(b) Immunogold labeling of polygonal particles with antiserum to potato multicystatin (MC) in COR-treated tissue.
(c) Enlargement of (b) showing decoration of MC particles with gold. Scale bar = 1 μm. Chl, chloroplast; Va, vacuole.

Immunochemical identification of proteinase inhibitors

In response to wounding by insects or mechanical damage, leaves of tomato and other Solanaceous plants synthesize multiple proteinase inhibitors with a wide range of specificities (Ryan, 2000). COR, CFA, and MeJA induce serine proteinase inhibitors in tomato, and these occur both locally and systemically in tomato leaves (Feys et al., 1994; Palmer and Bender, 1995; Zhao et al., 2001, 2003). In addition to the chymotrypsin inhibitor activity associated with serine proteinase inhibitor I, COR-treated tomato tissue previously exhibited chymopapain inhibitor activity (Palmer and Bender, 1995), which is consistent with the presence of a cysteine proteinase inhibitor (multicystatin). To investigate whether the polygonal vacuolar proteins were particles of multicystatin, tomato tissue treated with 20 nmol COR was incubated with antiserum to potato multicystatin and followed by goat-anti-rabbit (GAR) IgG-gold conjugate. When COR-treated tissue was examined 8 days after inoculation, immunolabeling of the polygonal vacuolar proteins occurred at a high efficiency (275 gold particles μm−2) (Figure 2b,c), indicating that these structures were particles of multicystatin. Labeling of the vacuole, chloroplasts, and particles of proteinase inhibitor I was less than 10 gold particles μm−2 (significantly lower at P = 0.01).

Effect of selected compounds on tomato seedlings

Previously, it was reported that COR inhibits root growth and induces anthocyanin accumulation in Arabidopsis (Bent et al., 1992; Feys et al., 1994). We wondered whether this response was unique to Arabidopsis, or whether it also occurs in tomato. COR, CFA and MeJA each inhibited root growth and induced anthocyanin accumulation in seedlings (Figure 3). COR was more potent than the other compounds and inhibited root growth (50% reduction) at the lowest concentration tested (0.002 nmol) (Figure 3a,c). This amount was 100- and 10 000-fold lower than the amounts of CFA (20 nmol) and MeJA (2 μmol), respectively, required to inhibit root growth by 50% (Figure 3a). It is important to note that the fold differences presented are based on the molar concentrations of the applied compound. The differences in permeability and uptake kinetics of the applied compounds were not taken into account.

Figure 3.

Effects of COR, CFA, MeJA, CMA, and ACC on tomato root growth and anthocyanin production.
(a) Root length of seedlings treated with COR, CFA, MeJA, CMA, and ACC. Each compound was tested at 0, 0.002, 0.02, 0.2, 2, 20 200, or 2000 nmol concentrations (increasing in concentration from left to right).
(b) Percentage anthocyanin accumulation in the seedlings treated with COR, CFA, MeJA, CMA, and ACC. The concentrations of each compound are the same as described in (a).
(c) Visual effects of COR, CFA, MeJA, CMA, and ACC on tomato seedlings. Seedlings were grown on water-saturated filter paper for 4 days in darkness at 24°C and then treated with COR (0, 0.002, 0.02, 0.2, 2, or 20 nmol), CFA (0, 0.002, 0.02, 0.2, 2, or 20 nmol), MeJA (0, 20, 200, or 2000 nmol), CMA (0, 0.02, 0.2, 2, 20, 200, or 2000 nmol), or ACC (0, 20, 200, or 2000 nmol). Photographs were taken 4 days after treatment. Arrows indicate anthocyanin accumulation (COR, CFA, and MeJA-treated tissue). Boxed panel in ACC treatment shows seedlings exhibiting the triple response (reduced elongation, thickened hypocotyl, thickened apical hook).

Similarly, COR was much more active in inducing anthocyanin accumulation in tomato seedlings and was effective in inducing threefold higher amounts of anthocyanin at 0.02 nmol; approximately 1000-fold more CFA and 10 000-fold more MeJA were required to induce similar amounts of anthocyanin (Figure 3b). COR induced visible anthocyanin accumulation in the hypocotyl of treated seedlings at concentrations ranging from 0.2 to 20 nmol (Figure 3c, see arrows). However, higher concentrations of CFA (20 nmol) and MeJA (200 nmol) were required to induce visible anthocyanin (Figure 3c, arrows).

ACC induced a typical ‘triple response’ in tomato seedlings (reduced elongation, thickened hypocotyl, and thickened apical hook) at 0.2 and 200 μmol (Figure 3c; boxed rectangle). To assess whether CMA functions as an ACC analog (Figure 1), we tested the effect of CMA on tomato seedlings. Unlike ACC, CMA did not inhibit root growth or induce a triple response (Figure 3c).

In summary, COR was much more effective in inhibiting root growth and inducing anthocyanin accumulation than CFA and MeJA. In contrast, CMA and ACC were relatively ineffective in stimulating anthocyanin accumulation (Figure 3b,c). ACC (but not CMA) inhibited root growth but only at levels 10 000-fold higher than COR (Figure 3b,c), suggesting that CMA does not behave as a functional analog of ACC in these assays.

Transcript profiling of COR-, CFA-, CMA-, and MeJA-treated tomato leaf tissues

Previous studies, as well as the experiments described above, suggest that COR is a functional mimic of MeJA, and this activity may be mediated by CFA. However, our results suggest that COR is biologically more active than CFA or MeJA in inducing serine proteinase inhibitors, anthocyanin accumulation, and root inhibition (Table 1, Figure 3). Furthermore, CFA and MeJA did not induce the full range of biological activities induced by COR (Table 1). Although CMA showed biological activity (Table 1), it did not function as an analog of ACC. It is important to mention that P. syringae pv. tomato releases or secretes free CFA when fermented in vitro (Bender et al., 1987); consequently it is possible that free CFA and/or CMA are released into the apoplast during pathogenesis. The identification of COR, CFA, and CMA-responsive genes offers an opportunity for studying the potential functions of these compounds; therefore, we conducted experiments to identify gene expression in response to these compounds using cDNA microarrays.

In preliminary experiments, gene expression was compared using RNA isolated from tomato leaves 3, 12, and 24 h after inoculation with purified COR using cDNA microarrays. The 12 h time point was chosen in subsequent experiments because maximal differential expression was observed at this time point when compared with H2O-inoculated control tissue. At 12 h post-treatment (hpt), most COR-induced genes (e.g., LOXD, OPR3, PI-II and LapA) were expressed to the highest level (data not shown).

Differential regulation of gene expression in COR-, CFA-, and CMA-treated tissue was compared using Venn diagrams (Figure 4). The complete list of genes and expression ratios are presented as supplementary material (Table S1). When a twofold induction (P < 0.05) relative to the control was used, 256, 231, and 143 genes were identified as induced by COR, CFA, and CMA, respectively (Figure 4). The largest set of upregulated genes was those induced by COR (256 genes; Figure 4). Fifty-eight genes were induced by both COR and CMA, and 96 genes were upregulated by both CFA and CMA (Figure 4). A group of 48 genes was upregulated by treatment with all three compounds.

Figure 4.

Venn diagrams showing the number of COR-, CFA-, and CMA- (12 h post-treatment) regulated genes. The numbers represent a list of non-redundant genes with ≥twofold induction or repression in response to COR, CFA, or CMA. Numbers within boxes denote the total number of genes regulated by each compound.

The identification of genes downregulated by COR, CFA, or CMA is equally important in understanding the plant responses to these compounds. When a twofold decrease (P < 0.05) in expression was applied as a cut-off, 274, 292 and 176 genes were identified as downregulated by COR, CFA, and CMA, respectively (Figure 4, Table S1). A total of 134 genes were repressed by both COR and CFA, and 95 genes were repressed by both COR and CMA (Figure 4). A total of 136 genes were repressed by both CFA and CMA (Figure 4).

These results indicate the greatest degree of overlap exists between COR- and CFA-regulated genes. However, COR was more active in regulating the genes investigated in this study, as it induced a total of 352 genes and repressed 274 genes. These results further suggest that the COR holotoxin regulates a greater array of genes in tomato than CFA or CMA; furthermore, these results are consistent with our biological assays (Table 1, Figure 3).

Our observations using cell biology demonstrated that COR induces chlorosis and alters the structure of the chloroplast (Table 1, Figure 2). Consistent with our observations, COR downregulated a large number of genes belonging to chloroplast metabolism (e.g., genes encoding chlorophyll a/b binding proteins, NADPH:protochlorophyllide oxidoreductase, thylakoid luminal proteins) (Figure 5, group I; Table S2). Although CFA did not induce visible chlorosis (Table 1), it downregulated a subset of genes involved in chloroplast metabolism that was repressed by COR (Figure 5, group I; Table S2). Interestingly, MeJA did not induce any visible chlorosis (Table 1) and was less active than COR and CFA in repressing genes involved in chloroplast metabolism (Figure 5, group I; Table S2).

Figure 5.

Expression diagram of selected genes that are differentially regulated by MeJA, COR, CFA, or CMA. Functional groups discussed in text include: group I, genes belonging to chloroplast metabolism; group II, genes involved in oxylipin biosynthesis; group III, JA and/or wound-responsive genes; and group IV, genes involved in ethylene biosynthesis. The colored font within groups indicates genes that were either further discussed in the text or selected for analysis by RT-qPCR. Gray tones indicate genes that were not differentially expressed.

Jasmonates and CFA accelerate senescence and chlorophyll breakdown in Arabidopsis, barley and tomato (He et al., 2002; Wasternack et al., 1998; Weidhase et al., 1987). Ethylene stimulates events associated with plant senescence (Bleecker and Kende, 2000; Stall and Hall, 1984); therefore, COR-induced ethylene may function to repress photosynthesis-associated genes and accelerate ethylene-induced senescence. In this study, COR, but not CMA induced ethylene biosynthesis genes (Table S2), and previous reports have substantiated these findings (Ferguson and Mitchell, 1985; Kenyon and Turner, 1992). Thus, we cannot eliminate the possibility that COR-induced ethylene may impact photosynthesis and senescence.

JA biosynthesis is regulated by a JA-mediated positive feedback loop (Sasaki et al., 2001; Stenzel et al., 2003). Consistent with observations in Arabidopsis, MeJA positively stimulated genes involved in JA biosynthesis and JA responsiveness in tomato (Figure 5, groups II and III; Table S2). As it has been suggested that COR is a structural and functional analog of JA, we investigated the effects of COR on genes involved in JA biosynthesis and JA-mediated wound responses. COR was more active than MeJA, CFA, or CMA in upregulating JA biosynthesis genes, including lipoxygenase (LOXD), allene oxide cyclase (AOC), and oxophytodienoate reductase (OPR3) (Figure 5, group II; Table S2). Simultaneous, increased accumulation of COR and MeJA were previously reported during infection of Arabidopsis by P. syringae pv. tomato (Schmelz et al., 2003). In this study, we observed that treatment with COR stimulates the accumulation of endogenous levels of JA in tomato leaves (Figure 6). Thus both COR and endogenous MeJA may modulate genes involved in JA biosynthesis.

Figure 6.

Changes in levels of SA, JA, and IAA in tomato leaf tissue treated with COR (mean ± SD, n = 3).

Several JA/wound-responsive genes were induced by MeJA, COR, CFA, and CMA (e.g., wound-inducible serine proteinase inhibitors I and II) (Figure 5, group III; Table S2). However, CMA was generally less active in inducing this group of genes, especially polyphenol oxidase and multicystatin.

COR stimulates ethylene production in both bean and tobacco, leading to speculation that the CMA portion might stimulate ethylene production (Ferguson and Mitchell, 1985; Kenyon and Turner, 1992). However, in the tomato seedling assays described above, CMA, unlike the ethylene precursor ACC, did not induce a ‘triple response’ (Figure 3c). Consistent with these observations, genes involved in ethylene biosynthesis and/or ethylene responsiveness were not modulated by CMA. These results indicate that CMA does not stimulate ethylene production or functionally mimic ACC. In contrast, COR strongly induced these genes (Figure 5, group IV; Table S2), suggesting that the holotoxin can directly stimulate ethylene production.

Furthermore, our cDNA microarray experiments identified genes that were not previously known to be regulated by COR (Table S2; discussed below). For example, COR induced the expression of a set of auxin-related genes, including IAA-conjugate hydrolases (e.g., IAR3) an auxin-regulated protein (Figure 5), and a member of the ‘no apical meristem’ gene family (NAM) (Table S2). Kunkel et al. (2004) showed that IAR3 was differentially expressed in Arabidopsis tissues inoculated with wild-type P. syringae pv. tomato and cor mutants. Furthermore, we show that the treatment of tomato leaves with COR stimulated the accumulation of endogenous levels of IAA in tomato leaves (Figure 6). These results suggest that COR may increase the free IAA levels within the plant to increase virulence.

RT-qPCR verifies the differential expression of COR, CFA and CMA regulated genes

The microarrays used in this study contained cDNAs from tomato tissues treated with various elicitors and contained redundant cDNAs (Figure S2; Alba et al., 2004). Furthermore, some genes of interest were either not represented in the microarray and/or failed quality control. Consequently, most of the cDNAs regulated by COR were sequenced to confirm their identity. To validate the expression results obtained using microarrays and confirm our results for a subset of genes, we designed gene-specific primers from the available full-length cDNAs and performed real-time quantitative PCR (RT-qPCR). In general, RT-qPCR analyses corroborated the microarray expression data. COR, but not CFA or CMA, strongly induced expression of LOXD, AOS2, AOC, and OPR3, the key enzymes in the JA biosynthesis pathway (Figure 7).

Figure 7.

Overview of (a) the octadecanoid biosynthetic pathway and (b) gene regulation by COR, CFA, and CMA. Transcriptional changes were analyzed using RT-qPCR. The expression data are presented for lipoxygenase (LOXD), allene oxide cyclase (AOC), allene oxide synthase (AOS2), and oxophytodienoate reductase (OPR3); AOS was not represented on the Tom1 array. The primers used for the RT-qPCR are presented in Table S5. The expression data (log2 value) for changes in COR-, CFA- and CMA-treated tissue from cDNA microarrays are transformed into fold change values for comparison with RT-qPCR data. The values represent the average of three independent biological replicates. The α-tubulin gene was used as an internal control.

Results from cell biology studies (Table 1) and microarray analysis (Figures 4 and 5) showed that COR, CFA, and CMA differentially regulate MeJA-responsive genes. In RT-qPCR, COR, CFA, and CMA induced serine proteinase inhibitors, but CMA was not effective in inducing multicystatin (Figure 8). With respect to genes involved in ethylene biosynthesis, COR, but not CMA, strongly induced SAM synthase and ACC oxidase in RT-qPCR (Figure 9). These results are consistent with those obtained in microarray analyses (Figure 5).

Figure 8.

RT-qPCR analysis of JA/wound-inducible genes encoding serine proteinase inhibitor (PI-II) and multicystatin (MCI) by COR, CFA, and CMA. The expression data (fold change values) from transcriptome analysis are presented. The data were analyzed as described in Figure 7. High (relative) fold differences of PI-II and MC in RT-qPCR could be attributed to the low levels of PI-II and MC gene products in untreated controls.

Figure 9.

Overview of the (a) ethylene pathway and (b) gene regulation by COR, CFA, and CMA. Transcriptional changes were analyzed using RT-qPCR for SAM synthase (SAM2), and 1-aminocyclopropane-1-carboxylic acid oxidase (ACO1). The expression data (fold changes) from transcriptome analysis for SAM2, 1-aminocyclopropane-1-carboxylic acid synthase (ACS), and ACO1 are presented. The data were analyzed as described in Figure 7.

Transcriptional profiling of COR and MeJA in tomato leaf tissues

Our observations using cell biology and seedling assays suggested that COR modulates gene expression in a manner similar to MeJA (Table 1; Figures 2 and 3). To investigate this further, transcriptome analysis of MeJA-treated tissues was performed and compared with COR-treated tomato tissue. When a twofold induction (P < 0.05) relative to the control was used as a cut-off value, a total of 256 and 320 genes were identified as induced by COR and MeJA, respectively (Figure 10a; Table S4). Approximately 40% (128) of the MeJA-induced genes were also induced by COR (Figure 10a, Tables S3 and S4). A total of 273 and 403 genes were downregulated by COR and MeJA, respectively (Figure 10a), and 30.7% (124) of the MeJA-repressed genes were also downregulated by COR.

Figure 10.

Comparison of the transcriptional profiles of COR- and MeJA-treated tomato leaf tissues.
(a) A Venn diagram showing the total number of genes regulated by COR and MeJA (12 h post-treatment). The numbers represent a list of genes with ≥twofold induction or repression in response to COR or MeJA. The numbers in the boxes denote the total number of genes regulated by each compound.
(b) Hierarchical clustering of 448 gene expression patterns in COR- or MeJA-treated leaf tissues (12 h post-inoculation). Some of the genes belonging to clusters A–D are discussed in the text. The gene description and the expression values for the clustered data are presented in Table S4.

To facilitate the comparison of COR- and MeJA-induced genes, a total of 448 (non-redundant) genes were subjected to average linkage hierarchical clustering using the default algorithm present in Genesis software (Sturn et al., 2002). Hierarchical clustering trees allowed us to group genes based on the similarities in expression patterns (Figure 10b). Based on this clustering method, we selected four groups for further comparison (Figure 10b, A–D). The clustered data with expression values are presented in Table S4. In addition to identifying many unknown genes as MeJA/COR-responsive (Figure 10a; Tables S3 and S4), our results correlate well with previously reported JA or wound-regulated transcriptomes (Cheong et al., 2002; Goossens et al., 2003; Sasaki et al., 2001; Schenk et al., 2000; Stintzi et al., 2001). Cluster A (A1–A4) consists of genes induced by both COR and MeJA. This group consisted of genes involved in JA responsiveness and JA biosynthesis (Figure 10b, Table S4) and includes genes encoding leucine aminopeptidase, wound-inducible serine proteinase inhibitors I and II, multicystatin, polyphenol oxidase, threonine deaminase, AOC, and OPR3 (Table S4). Interestingly, many genes implicated in wound and/or cellular signaling also clustered with the JA-responsive genes. These included genes encoding systemin precursor, protein phosphatase 2C, serine/threonine protein kinase, calcium binding protein, calcium binding EF-hand family protein, MAPK4, AP2 domain-related protein, and myb-like transcription factor. It is interesting to note that MAPK4 encodes a gene in the JA-dependent pathway and negatively regulates expression of the SA-dependent defense pathway in Arabidopsis (Petersen et al., 2000). This group also included genes involved in stamen and carpel development, including TGA1 and MADS2. Jasmonates may regulate these flower homeotic genes though a SCFCOI/JAI signalosome, and tomato and Arabidopsis mutants that are insensitive to COR/JA were defective in proper floral (stamen and pollen) development (Feys et al., 1994; Li et al., 2004).

Both COR and MeJA induced a gene encoding cysteine protease (Tables S2 and S4, cluster A4), an enzyme implicated in pathogen-induced cell death (Navarre and Wolpert, 1999). This may be relevant in the context of nutrient pools in the apoplast; for example, the release of nutrients from dying cells may facilitate pathogen multiplication in the necrogenic stage of its life cycle. Furthermore, COR and MeJA impacted the expression of genes involved in polyamine biosynthesis (Table S8, cluster A3, A4), and genes encoding the synthesis of putrescine (e.g., arginine decarboxylase) and spermidine (e.g., spermidine synthase1) were strongly upregulated by COR (Tables S2 and S4). Similarly, MeJA alters polyamine metabolism in barley (Walters et al., 2002). Although COR, CFA, and MeJA induced anthocyanin accumulation in tomato seedlings (Figure 3a,c) and in Arabidopsis seedlings and leaves (Feys et al., 1994), neither COR or MeJA-treated tomato leaves accumulated visible levels of anthocyanin (Figure S1). Furthermore, the genes involved in anthocyanin metabolism that were present on the Tom1 array (e.g., dihydroflavonol 4-reductase and anthocyanidin 3-O-glucosyltransferase) were not induced by COR or MeJA.

Cluster B consisted of genes induced by COR that were either suppressed or not differentially expressed in response to MeJA, including genes encoding lipoxygenase, auxin-related protein, NAC domain protein (NAC2), JA transcription factor 2, receptor-like protein kinase, homeobox protein 1, and abnormal inflorescence meristem 1. Cluster C consisted of genes induced by MeJA, but repressed in COR-treated tissues; this cluster included AIR12, which is involved in lateral root development, and a tuberization-related gene (Table S4). Cluster D consisted of genes that were induced by COR and were not differentially regulated by MeJA. This cluster contained genes potentially involved in the ubquitin-proteasome pathway, including ubiquitin-related protein (RUB1) and ubiquitin-conjugating enzyme. Genes involved in ethylene (ACO1) and auxin metabolism (IAR3) were also represented in this cluster (Table S4).

In summary, the transcriptional profiles of MeJA and COR-treated tomato leaf tissues showed substantial overlap with respect to genes involved in JA biosynthesis, JA signaling, ethylene biosynthesis, and auxin metabolism. Functional analysis and determination of the biological relevance of the novel genes identified in this study will help us understand how COR and MeJA function in tomato.


Coronatine regulates an array of biological activities in different plant species (Bender et al., 1999; Feys et al., 1994; Lauchli and Boland, 2003; Palmer and Bender, 1995; Weiler et al., 1994) and functions as a virulence factor in P. syringae pv. tomato (Bender et al., 1987; Brooks et al., 2004; Kloek et al., 2001; Mittal and Davis, 1995; Peñaloza-Vázquez et al., 2000; Zhao et al., 2003). However, the precise mechanism of action of COR is unclear. Using a combination of biological assays, ultrastructural studies, and gene expression profiling, we demonstrate that COR impacts signaling in tomato through multiple phytohormone pathways (JA, ethylene, and auxin). Our data also demonstrate that CMA has biological activity but does not function as an ACC analog. The CFA moiety regulated approximately 40% of the COR-modulated genes, but was less active than COR and did not induce all biological activities associated with COR. A comparison of COR- and MeJA-regulated transcriptomes revealed that COR regulated 35% of the MeJA-induced genes.

Effect of COR and related compounds on tomato cell biology

Treatment of tomato tissue with CFA, CMA, MeJA, and ACC did not induce chlorosis, changes in chloroplast structure, or cell wall thickening; however, all four compounds induced the production of proteinase inhibitor I, with MeJA being the most active (Table 1). The induction of proteinase inhibitors by ACC and CMA was somewhat unexpected; however, Staswick and Tiryaki (2004) recently identified the formation of a JA-ACC conjugate in Arabidopsis. Thus it remains possible that exogenously applied ACC and/or CMA may form conjugates with JA in tomato, and the JA conjugates may induce production of proteinase inhibitors. Although CMA affected the expression of some CFA-regulated genes (Figure 4, Table S1), it is not clear how such structurally different molecules modulate similar genes. Although the existence and activity of JA-CMA conjugates have not been demonstrated, such compounds may occur in planta and could explain this phenomenon.

Our observations from cell biology revealed that COR has a major effect on chloroplast biogenesis and downregulates many genes associated with the photosynthetic apparatus (Figure 5, Table S2). Previous studies have shown that MeJA treatment leads to reduced expression of Rubisco and increased loss of chlorophyll (Parthier, 1990). COR enhances chlorophyll loss in tomato (Palmer and Bender, 1995) and induces the expression of chlorophyllase, the first enzyme in the chlorophyll degradation pathway (Benedetti and Arruda, 2002; Tsuchiya et al., 1999). Immunolocalization experiments with COR-specific antiserum suggest that COR associates with chloroplasts in tomato leaves (Zhao et al., 2001). It is tempting to speculate that COR may translocate to the chloroplast and interact with chloroplast-associated proteins to mediate COR/MeJA-responsive gene signaling.

Structural components of COR required for biological activity

Other biological assays have addressed the structure/function activities of COR, MeJA, and related compounds. Blechert et al. (1999) demonstrated that methylation of the carboxyl group of COR reduced, but did not abolish, tendril coiling in Bryonia dioica. In an earlier study, Shiraishi et al. (1979) demonstrated that methylation of the carboxyl group of COR reduced, but did not eliminate, the induction of hypertrophy in potato tubers. Similarly, COR analogs containing modifications at the carboxyl group retained the ability to induce alkaloid production in Eschscholzia californica cell cultures (Haider et al., 2000). Results obtained in the current study support the contention that the COR molecule is still active when the carboxyl group is methylated. For example, COR-Me was only slightly less active than COR in inducing chlorosis, ultrastructural changes, and proteinase inhibitors in tomato (Table 1).

In a previous study, the keto group of COR was reduced to form coronatinol, which was further modified by addition of acetyl or succinyl groups (Jones et al., 2001). In the current study, we show that modification of COR at the keto group (coronatinol, COR-OAc, COR-OSuc) resulted in reduced chlorosis relative to COR; however, these compounds still elicited all activities associated with the holotoxin with reduced efficiency (as cited above). This is important because COR-OSuc was subsequently conjugated to a carrier protein and used for producing COR-specific monoclonal antisera (Jones et al., 2001). Therefore, the COR-OSuc used for constructing antisera maintained biological activity and will be valuable in future studies designed to isolate the receptor(s) for COR.

Conjugation of CFA to an amino acid moiety was required for the full range of biological activity in tomato, including chlorosis, changes in chloroplast structure, cell wall thickening, and accumulation of both classes of proteinase inhibitors (Table 1). Naturally occurring (CFV) and chemically modified derivatives of COR (COR-Me, COR-OH, COR-OSuc, and COR-OAc) retained biological activity in tomato. Previous studies have shown that coronafacoyl amide conjugates retain biological activity, but the degree of activity varies greatly among the assays utilized, the concentration of compound, and the substitution at the ethylcyclopropyl position (Blechert et al., 1999; Haider et al., 2000; Shiraishi et al., 1979).

Comparison of CFA, CMA, COR, and MeJA: tomato seedling assays and transcript profiling

COR consists of CFA and CMA, compounds that originate from two separate, complicated biosynthetic pathways (Jiralerspong et al., 2001; Mitchell, 1985; Parry et al., 1994). We used a tomato seedling assay to study the biological activity of COR, CFA, CMA, and MeJA (Figure 3). COR inhibited root growth of tomato seedlings at extremely low concentrations, and inhibition occurred in a dose-dependent manner (Figure 3). CFA also inhibited root growth, but the concentration required to see inhibition was much higher than that observed for COR (Figure 3a,c). These results are consistent with previous studies showing that COR, CFA, and MeJA can directly affect root growth (Feys et al., 1994; Staswick et al., 1992; Tung et al., 1996).

In biological assays with tomato leaf tissue and seedlings, CFA and MeJA induced proteinase inhibitors (Table 1), stimulated anthocyanin production (Figure 3b,c), and inhibited root growth in tomato (Figure 3a). In transcript profiling experiments, we observed that MeJA and CFA regulated most of the JA-responsive genes (Figure 5; Table S2), but were generally less active than COR in inducing the above-mentioned activities (Table 1; Figure 3). Although our results with CFA are consistent with previous studies (Bodnaryk and Yoshihara, 1995; Koda et al., 1996; Palmer and Bender, 1995). Wasternack et al. (1998) reported that 250 μm CFA had no effect on the expression of JA-responsive proteins in detached tomato leaves. However, these differences in the biological activity of CFA may be attributed to the assay method and/or the concentration of CFA utilized.

In this study, COR and MeJA, but not CMA, induced genes involved in ethylene biosynthesis and ethylene responsiveness (Table S2). Ethylene plays an important role in the symptoms associated with bacterial speck of tomato; for example, plants that are insensitive to ethylene show impaired disease symptoms (Lund et al., 1998), and ethylene has been implicated in chlorosis and senescence (Bleecker and Kende, 2000; Stall and Hall, 1984). Ethylene production was observed in COR-treated tobacco and bean leaves and was produced when COR-producing bacteria infect susceptible host plants (Ferguson and Mitchell, 1985; Kenyon and Turner, 1992; Weingart et al., 2001). In our transcript profiling experiments, COR induced genes associated with ethylene biosynthesis and responsiveness (Figure 9; Table S2), suggesting that COR may modulate ethylene as a virulence strategy.

COR also induced the expression of a set of auxin-related genes (Figure 5, Table S5), implying that auxin levels also play an important role in pathogenesis (Kunkel et al., 2004). In a study using potato tubers and mung bean hypocotyls, Sakai et al. (1979) concluded that auxin and COR have different primary sites of action but ultimately target the same physiological activities. Similarly, our results suggest that the COR-induced JA pathway may positively regulate auxin responses in tomato. This is consistent with the hypothesis that JA and auxin may function via a common signaling intermediate that modulates response to multiple plant hormones (Devoto and Turner, 2003; Tiryaki and Staswick, 2002).

Our results show that COR modulates genes involved in the pathways to JA, ethylene, and auxin. This raises an interesting question: should COR be considered a phytotoxin or a phytohormone mimic? It is not surprising that COR targets these particular phytohormone pathways, as both ethylene and JA positively regulate susceptible interactions between tomato/Arabidopsis and P. syringae (Kunkel and Brooks, 2002). A popular hypothesis is that COR may act as a suppressor of defense response(s), possibly by suppressing salicylic acid-dependent defenses in Arabidopsis and tomato (Kloek et al., 2001; Zhao et al., 2003). Although suppression of SA-mediated defenses was not observed using exogenously applied COR, genes involved in JA biosynthesis and responsiveness were induced by COR (Table S2). Mutual antagonism between JA- and SA-mediated defense pathways is well documented (Kunkel and Brooks, 2002); consequently, COR may stimulate the JA pathway at the expense of SA-dependent defense responses.

Comparison of COR- and MeJA-regulated transcriptional changes

One outcome of the present study was the identification of gene sets that respond differentially to COR and MeJA, supporting the contention that they have different activities based on the comparison of the expression profiles at a single time point (Figure 10). Interpretation of these changes is complicated due to the differences in the kinetics of induction and by the fact that both primary and secondary transcriptional changes occur following exposure to MeJA or COR. For example, two recent reports document the existence of JA-modifying enzymes, including a MeJA esterase and a JA amino acid synthetase (Staswick and Tiryaki, 2004; Stuhlfelder et al., 2004). Presumably, a MeJA esterase could cleave exogenous MeJA to form JA, which could be further metabolized by a JA amino acid synthetase to form JA amino acid conjugates. These metabolized products of MeJA, along with the other MeJA-induced phytohormones (e.g., ethylene, IAA) could contribute to the secondary transcriptional changes in MeJA-treated leaves. This process may enable plant cells to ‘fine tune’ the chemical signals that regulate plant growth and help maintain jasmonate homeostasis (Staswick and Tiryaki, 2004). However, it remains unclear whether COR is metabolized and forms conjugates with amino acids in planta. There are very striking differences in the structure of COR and MeJA (Figure 1a,d), and these changes might enable COR to ‘evade’ MeJA-modifying enzymes. If COR is not further metabolized, this could lead to perturbations in JA homeostasis and result in phytotoxicity. Clearly, there are many unresolved questions regarding the activity of COR in modulating phytohormone pathways. Experiments are underway to further analyze COR/MeJA-responsive genes and the potential receptors for these compounds, which will help elucidate the mechanism of action for both COR and MeJA.

Experimental procedures

Plant material

Lycopersicon esculentum Mill. cv. Glamour was used in all experiments. Plants were grown from seed in a peat:soil mix in 10 cm diameter plastic pots and maintained in growth chambers (25°C, 40–70% RH, 12-h photoperiod, photon flux density 150–200 μmol m−2 sec−1). Plants were approximately 4 weeks old at the time of treatment.

Isolation and synthesis of coronatine-related compounds

COR, CFA, and CMA were prepared as described previously (Jones et al., 1997). 2-[1-oxo-2-cyclopenten-2-ymethyl]-butanoic acid (CPE) was purified from P. syringae pv. tomato 3362 by column chromatography (Mitchell et al., 1995) and obtained as a homogenous compound as determined by gas chromatography. The purified preparations of CFA, CMA, CPE, and CFV were also analyzed and were shown to be free of any contaminating COR. CFV was isolated from P. syringae pv. atropurpurea as described previously (Mitchell, 1984) and purified by column chromatography (Young et al., 1992). COR-Me was prepared from COR by the addition of excess diazomethane in ether (Shiraishi et al., 1979). COR-OH was produced by the reduction of the carbonyl group of COR to a hydroxyl using sodium borohydride (Jones et al., 1997). COR-OAc was prepared by the treatment of coronatinol with acetic anhydride and purified by reverse-phase HPLC as described previously (Jones et al., 1997). COR-OSuc was prepared as described previously (Jones et al., 2001). Methyl jasmonate was obtained from Bedoukian Research Inc. (Danbury, CT, USA) and ACC was obtained from Sigma (St Louis, MO, USA).

Inoculation, sampling, and tissue processing for microscopy

In treatments where the visual and ultrastructural effects of COR derivatives were evaluated, the underside of the second primary leaf of 4-week-old plants was inoculated within 3 days after full expansion. Compounds (1 or 20 nmol per inoculation site) were suspended in H2O and applied as 2 μl droplets to the undersides of the tomato leaves. Plants inoculated with the volatile MeJA were placed in a room separated from all other plants. All treatments were examined in at least three separate experiments. Chlorosis was evaluated visually 5 and 7 days after inoculation, and leaf tissue was excised and processed for transmission electron microscopy 8 days after inoculation as described previously (Palmer and Bender, 1995).

Immunoelectron microscopy

Immunogold labeling of putative chymopapain proteinase inhibitors (multicystatin) was conducted as follows. Tomato leaves were inoculated with H2O (control) or H2O containing 1 nmol COR as described above. Eight days later, inoculated sites were excised and fixed in 4% paraformaldehyde, 0.5% glutaraldehyde in 0.05 m sodium phosphate buffer (PBS) for 3 h at room temperature. The tissue was then subjected to two consecutive washes (30 min each); once in PBS and a subsequent washing in PBS containing 0.15 m glycine. The tissue was then dehydrated in a graded ethanol series and embedded in L. R. White resin. Thin sections (100 nm) were cut, transferred to nickel grids, and blocked with 1% ovalalbumin in PBS for 5 min. The grids were then incubated on droplets containing anti-potato multicystatin antibodies (diluted 1:200 in PBS) for 3 h at 4°C. The sections were then washed in PBS three times (5 min each), blocked with 1% ovalalbumin in PBS (5 min), and incubated for 40 min in diluted goat-anti-rabbit IgG gold conjugate (1:20 dilution in PBS; Sigma Chemical Co.). Sections were then washed for 5 min in 0.1X PBS (twice) and stained with 2.5% uranyl acetate for 30 min. Sections were rinsed in H2O, dried, and viewed with a JEOL-100CX scanning transmission electron microscope (JEOL, Tokyo, Japan) at 80 kV.

Thin sections of COR-treated leaf tissue were analyzed at a magnification of 29 000X. Gold particles were counted on printed photographs with the aid of 1 cm2 grid. In tissue treated with the anti-potato multicystatin antibodies, a total of four grids were examined, and gold particles were counted in one section per grid and three fields per section. Values are reported as mean number of particles, and the Student–Newman–Keuls test was used to calculate significant differences at P = 0.01.

Seedling assays

Seedlings were grown on water-saturated filter paper for 4 days in darkness at 24°C and then treated with COR (0.002, 0.02, 0.2, 2, or 20 nmol), CFA (0.002, 0.02, 0.2, 2, or 20 nmol), CMA (0.02, 0.2, 2, 20, 200, or 2000 nmol), MeJA (20, 200, or 2000 nmol), or ACC (0, 20, 200, or 2000 nmol). An aliquot of sterile distilled H2O (500 μl) was applied as a control inoculation. Before treatment with these compounds, seedlings with uniform hypocotyl and root lengths were selected and transferred to filter papers saturated with the respective compounds to achieve uniform application. The seedlings were further incubated in light for 4 days at 24°C. Photographs were taken 4 days after treatment and the root length was measured to the nearest 0.1 cm. The mean root lengths of 15 independent seedlings from three independent trials were calculated for each treatment.

Anthocyanins were extracted and quantified using methods similar to those described by Gould et al. (2000). Anthocyanins were extracted from seedlings incubated overnight at 4°C in darkness on a rotary shaker. Seedlings were incubated in a solution containing 1 ml of 3 m HCl:H2O:MeOH (1:3:16 by vol). Anthocyanins in the supernatants were measured spectrophotometrically as A530 − 0.24A653 and expressed as the percentage anthocyanin per gram fresh weight of seedling tissue. Anthocyanins absorb maximally at 530 nm; subtraction of 0.24A653 compensates for the small overlap in absorbance at 530 nm by the chlorophylls (Gould et al., 2000). Anthocyanins were extracted from three seedlings per treatment, and the data shown represent the mean anthocyanin content of nine seedlings from three independent trials.

Plant treatments and RNA extraction for microarray analysis

COR, CFA, and CMA (0.2 nmol per inoculation site) and MeJA (100 μm in 0.001% ethanol) were suspended in H2O, and eight droplets were applied in 2 μl aliquots onto tomato leaves (Figure S1). Sterile distilled H2O was applied to tomato leaves as a mock treatment. Two leaves per plant were harvested 12 h post-treatment (hpt). Approximately 400–500 mg (two leaves) of treated tissue was ground in liquid nitrogen using a mortar and pestle. Total RNA was purified with TRIzolTM reagent (Invitrogen, Carlsbad, CA, USA) according to the instructions of the manufacturer. Total RNA was extracted independently from three separate treated or untreated plants, and following isolation the total RNAs were pooled to represent one biological replicate. Three independent experiments were conducted on three different days. RNA was then further concentrated using Microcon YM100 filters (Millipore Corp., Bedford, MA, USA). Approximately 50 μg of denatured RNA from each biological replicate was reverse-transcribed to synthesize cDNA.

cDNA preparation

Total RNA was reverse transcribed to synthesize cDNA with an oligo d(T) primer, which was tagged with either Cy3- or Cy5-specific 3DNA capture sequences (Genisphere Inc., Hartfield, PA, USA). Briefly, 50 μg of total RNA and 1 μl of tagged oligo d(T) primer (1 pmol) were suspended in a final volume of 10 μl with RNAse-free double-distilled (dd)H2O, heated at 80°C for 10 min to denature the RNA, and cooled on ice and then incubated with 1 μl of RNAseOUTTM (Invitrogen). A master mix (8 μl) containing 5X Superscript RT buffer (Invitrogen), dNTPs (1 μl containing 10 mmol each dNTP), 0.1 m DTT (2 μl), and RNAse-free ddH2O (1 μl) was prepared and gently added to the RNA-RT mix. Each sample was heated for 3 min at 42°C, and 2 μl (400 U) Superscript IITM reverse transcriptase (Invitrogen) was then added. The reaction was incubated at 42°C for 1.5 h, supplemented with 0.5 m NaOH and 50 mmol EDTA (3.5 μl), and then incubated at 65°C for 10 min to stop the reaction and hydrolyze template RNA. The reaction was neutralized by adding 1 m Tris-HCl (5 μl, pH 7.5), and 6.6 μl of Cy3- or Cy5-labeled cDNA solution was used for hybridization. This scaled-up protocol using increased quantities of RNA and low concentrations of tagged oligo dT primer eliminated the need for ethanol precipitation of cDNA and spin column purification of unused primers.

cDNA and 3DNA hybridization

Tomato cDNA microarrays (Tom1 arrays) were obtained from the Center for Gene Expression Profiling, Cornell University (Ithaca, NY, USA). A total of 13 440 elements were printed in 8 × 4 format containing 32 sub-grids. A brief description of the Tom1 array architecture (Figure S2), EST source and the complete list of the spotted genes (gene ID file) are provided (Table S6; Alba et al., 2004). cDNA generated from pooled RNA (representing each biological replicate) was hybridized to individual slides. A ‘two-step’ protocol was adopted for hybridization, and cDNA and 3DNATM reagent were hybridized in succession using the 3DNATM Submicro Kit (Genisphere). A modification of the method reported by Alba et al. (2004) was adopted for hybridization. For the hybridization reaction, tagged cDNA (6.6 μl), salmon sperm DNA (1.5 μg, Invitrogen), oligo d(T) blocker (1 μl, Genisphere), high-end differential enhancer (1 μl, Genisphere), 10% (w/v) SDS (2 μl), and 20X SSC (2 μl) were added to 38.3 μl hybridization buffer (0.25 m NaPO4, 4.5% SDS, 1 mm EDTA, 1X SSC, 2X Denhardt's solution). The mixture (60 μl final volume) was incubated initially at 80°C for 10 min, and then at 60°C for 20 min before applying to the pre-warmed slides. A cover slip was slowly placed on top of the glass slide to uniformly distribute the hybridization solution. The microarrays were placed in a Corning® Hybridization Chamber (Corning Inc., Acton, MA, USA), and hybridized at 60°C in a water bath for 14–16 h. The slide was then removed from the chamber and subjected to the following sequence of washes: 10 min in 2X SSC, 0.2% SDS buffer (at 60°C); 2X SSC (room temperature); and 0.2X SSC (room temperature). Slides were then fixed in 95% ethanol for 2 min, and dried by centrifugation at 1000 g for 2 min.

The hybridized cDNAs were then incubated with Cy3 and Cy5 3DNATM reagent. Briefly, 2.5 μl each of Cy3 and Cy5 3DNATM reagent were mixed with 1 μl of high-end differential enhancer and 54 μl of hybridization buffer containing 1 μl of anti-fade reagent (Genisphere). This mixture was pre-incubated at 80°C (10 min), 60°C (20 min), and then hybridized to the microarrays as explained above. The slides were then washed for 10 min sequentially as described above and dried by centrifugation at 1000 rpm for 2 min.

Scanning and data analysis

The slides were analyzed with ScanArray 3000 (GSI Lumonics, Billerica, MA, USA) and the spots were quantified using GenePix Pro 4.0 (Axon Instruments, Inc., Union City, CA, USA). Local background fluorescence values were automatically calculated by the GenePix program and subtracted from all feature intensities before further calculations, and median values were used in all calculations. Pre-processing of data was accomplished using GenePix Autoprocessor (GPAP) (P. Ayoubi, unpublished data). This analysis included: (1) removal of data points where signal was less than the background plus two standard deviations in both channels; (2) removal of poor quality spots; (3) removal of spots where the ESTs failed quality control; (4) log transformation of the background subtracted Cy3/Cy5 median ratios; and (5) averaging the technical replicates within and across the replicates. Following pre-processing, the expression results were normalized using global LOWESS normalization (Yang et al., 2002) using Bioconductor software package ( Following pre-processing, normalized ratio values for each probe were averaged across valid signals obtained from three or more replicates. For each probe, the fold change, moderated t-statistics (Smyth, 2004) and P-values were determined. A candidate list of differentially expressed genes was then generated using a 5% false discovery rate (FDR) and greater than twofold change between treatments. We used moderated t-statistics and associated P-values after FDR in the limma package of the Bioconductor software package ( A total of six experiments were conducted. The raw data (before normalization) and normalized data with the mean ratios for COR, CFA, CMA, and MeJA-regulated genes, along with a sample of the scanned image, are presented as supplemental data to comply with the MIAME guidelines (Figures S2 and S3; Tables S6–S10). Expression images and average hierarchical gene clustering were generated using Genesis software, Release 1.4.0 (Sturn et al., 2002,

Real-time quantitative PCR analysis

Expression of a subset of selected JA-responsive and ethylene pathway genes were analyzed using RT-qPCR. Primers were designed using the full-length L. esculentum cDNAs available from GenBank (Table S5). α-Tubulin 4 and OPR3 were amplified using the primers described previously (Mysore et al., 2002). Primers were designed using Beacon Designer 2.13 software (Premier Biosoft International, Palo Alto, CA, USA).

The pooled RNA used for microarray analysis and another independent biological replicate were used for PCR. cDNA for RT-qPCR was synthesized from 5 μg total RNA using Superscript IITM reverse transcriptase following DNase I treatment. cDNA (equivalent to 10 ng of initial RNA) was quantified using gene-specific primers (Table S3) and the iCycler iQ Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA) using SYBR Green reagents (Bio-Rad). PCR cycling conditions consisted of an initial polymerase activation step at 95°C for 5 min, 40 cycles of 30 sec at 95°C, and 45 sec at 60°C. Melt-curve analysis was performed to monitor primer–dimer formation and the amplification of gene-specific products. Data quantification and analysis was performed using iCycler software, Ver. 3.06.6070 (Bio-Rad). The average threshold cycles (CT) values calculated from triplicate samples were used to determine the fold expression relative to the controls. Primers specific for α-tubulin 4 were used to normalize small differences in template amounts. Data analysis including normalization and standard deviations were calculated using Gene Expression MacroTM, Ver 1.1 (Bio-Rad) with the algorithms outlined by Vandesompele et al. (2002).

Phytohormone quantification

Tomato leaves (approximately 350 mg) were extracted and analyzed for SA, JA, and IAA using methods described by Schmelz et al. (2004). This method uses a quadropole MS system (5890 GC; Agilent, Palo Alto, CA, USA) connected to a 5989B Mass Selective Detector (Agilent) with electron spray ionization and selective-ion monitoring (selected ion ± 0.5 mass unit). The analytes were separated on a DB-5 column (30 m × 0.25 mm × 0.25 mm, Agilent) using the conditions described by Schmelz et al. (2004). The retention times and mass units of the methyl esters analyzed were: SA-ME, 8.35 min, 152; JA, trans 12.30 min/cis 12.54 min, 224; and IAA, 13.63 min, 189. Internal standards used were: [2H6]SA-ME (8.33 min, 156), dhJA-ME (trans 12.31 min, cis 12.53 min, 226), and [2H5]IAA-ME (13.62 min, 191). The [2H5]IAA-ME was converted to [2H2]IAA-ME to produce a parent ion with a mass unit of 191. Isotopically labeled internal standards were purchased from CDN Isotopes (Pointe-Claire, Quebec, Canada), while dhJA was prepared from methyl dihydrojasmonate (Bedoukian Research Inc.) by alkaline hydrolysis.


We thank Erika Spencer for help with primer design for RT-qPCR, David Jones, Eric Schmelz and Jack Dillwith for help with GC-MS analysis, and Dr Barbara Kunkel and Dr Steve Marek for reviewing the manuscript. The anti-potato multicystatin antibodies were kindly provided by Dr Stephen Gleddie, Agriculture and Agri-Food Canada. C.L. Bender acknowledges support from the National Science Foundation (IBN-0130693), the Oklahoma Center for Advancement of Science (no. AR031-005), and the Oklahoma Agricultural Experiment Station. The OSU Microarray Core Facility is supported by grants from NSF (EOS-0132534) and NIH (1P20RR16478-02 and 5P20RR15564-03).

Supplementary Material

The following material is available from

Figure S1. Experimental design for identification of target genes regulated by COR, CFA, CMA, and MeJA.

Figure S2. A brief description of the architecture and characteristics of Tom1 microarray.

Figure S3. An image of a Tom1 array hybridized with MeJA-treated tissue. Colors indicate the following: red, induced genes; yellow, genes that show no difference in expression; and green, repressed genes.

Table S1 Expression profiles of a non-redundant list of genes that are differentially regulated (RG) by COR, CFA, and CMA in tomato tissues

Table S2 Expression profiles of genes belonging to functional groups (chloroplast metabolism, JA biosynthesis, auxin-related, ethylene biosynthesis, pathogenesis-related genes) that are differentially regulated by MeJA, COR, CFA, and CMA

Table S3 Expression profiles of a full list of genes that are differentially regulated (RG) by COR and MeJA in tomato tissues

Table S4 The gene description and the expression values for the clustered data in Figure 10(b)

Table S5 List of primers and the corresponding clones used for RT-qPCR analysis

Table S6 A gene ID file containing the (a) physical location of the clones, (b) clone ID, and (c) annotation based on best blast hit

Table S7 Raw data from COR-treated tissue

Table S8 Raw data from CFA-treated tissue

Table S9 Raw data from CMA-treated tissue

Table S10 Raw data from MeJA-treated tissue