Deficiencies in Jasmonate-Mediated Plant Defense Reveal Quantitative Variation in Pathogenesis

Summary

Despite the described central role of jasmonate signaling in plant defense against necrotrophic pathogens, the existence of intraspecific variation in pathogen capacity to activate or evade plant jasmonate-mediated defenses is rarely considered. Experimental infection of jasmonate-deficient and jasmonate-insensitive Arabidopsis thaliana with diverse isolates of the necrotrophic fungal pathogen Botrytis cinerea revealed pathogen variation for virulence inhibition by jasmonate-mediated plant defenses and induction of plant defense metabolites. Comparison of the transcriptional effects of infection by two distinct B. cinerea isolates showed only minor differences in transcriptional responses of wild-type plants, but notable isolate-specific transcript differences in jasmonate-insensitive plants. These transcriptional differences suggest B. cinerea activation of plant defenses that require plant jasmonate signaling for activity in response to only one of the two B. cinerea isolates tested. Thus, similar infection phenotypes observed in wild-type plants result from different signaling interactions with the plant that are likely integrated by jasmonate signaling.

Introduction

Jasmonate-mediated signaling controls diverse aspects of plant growth and defense. In particular, jasmonate signaling exerts a major influence on plant response to wounding, chewing insects, and necrotrophic pathogens such as Botrytis cinerea, Alternaria brassicicola, Plectosphaerella cucumerina, and Sclerotinia sclerotiorum [1]–[6]. Appropriate plant responses to these diverse stimuli are believed to be tailored by cross-talk between jasmonate and other hormone signals, such as salicylic acid (SA), ethylene, and abscisic acid (ABA) [7]–[14]. Jasmonate signaling therefore does not mediate plant defense in isolation, but as part of a network of signals with the potential for positive and negative interactions. These signals include inputs from the pathogen that may influence the plant's defense response with positive or negative outcomes for the plant.

Two major pathogen classes are roughly delineated by the pathogen's “lifestyle”: biotrophic pathogens infect living host cells and necrotrophic pathogens kill cells prior to consuming them [15]–[17]. This difference in the pathogen's mode of attack strongly influences which signaling networks mediate the plant response. Plant responses to biotrophic pathogens are largely mediated by salicylate signaling with an emphasis on specific recognition of pathogen effectors by the products of plant resistance (R) genes, often characterized by nucleotide binding sites and leucine-rich repeats [18], [19]. Plant responses to necrotrophic pathogens appear to be mediated by a complex web of signaling dominated by jasmonates and ethylene [20]–[22]. Specific recognition of necrotrophic pathogens by the products of plant R genes is currently unknown, although recent identification of a gene possessing structural similarities to R-genes as the molecular basis of a quantitative trait locus (QTL) affecting resistance of Arabidopsis thaliana to multiple necrotrophic and hemibiotrophic pathogens has been suggested to link mechanisms of defense against biotrophic and necrotrophic pathogens [23]. While plants respond to biotrophic and necrotrophic pathogens via different signaling systems, these systems activate common defense responses, such as the production of the A. thaliana defense metabolite, camalexin. Thus, common responses may be controlled by distinct regulatory networks.

The simplified statement that biotrophic and necrotrophic pathogens activate distinct, but overlapping, defense signaling pathways is largely based on observation of single genotypes of the respective pathogens. Yet biotrophic pathogen species exhibit considerable variation in activation of plant defense signaling. This biotroph variation is largely associated with diversity in the R-gene mediated specificity of plant-pathogen recognition, a phenomenon not documented for necrotrophic pathogens [24]–[26]. Examples of naturally occurring intraspecific pathogen variation affecting plant defense against necrotrophs include variation in toxin production by pathogens and variation in pathogen tolerance or detoxification of plant-produced defense compounds [27]–[31].

While activating plant defense signaling should logically hinder infection, pathogens may manipulate plant defense signaling to improve pathogenesis by diverting plant resources toward defense strategies that are less effective against, or actually increase sensitivity to, the pathogen. Pathogens are known to produce plant hormones or analogues such as coronatine, gibberellins or ABA, and the production of these compounds has been associated with virulence [32]–[34]. Interestingly, the ability to produce these compounds may vary among isolates of the same pathogen species as shown by a survey of 95 strains of Pseudomonas syringae where only 15% assayed positively for coronatine production [35]. While production of ABA by pathogenic fungi has not been as extensively assayed, ABA-overproducing and ABA-deficient B. cinerea strains have been described [36]. In addition, some B. cinerea isolates produce ethylene [37]. Thus, while elements of plant defense signaling may be associated with resistance to particular pathogens, pathogen variation in activation, manipulation, and response to plant defense signaling may alter these associations. Despite available literature suggesting that B. cinerea natural diversity could impact plant defense signaling, this diversity has not been routinely integrated into studies of plant—pathogen interaction.

Unlike many pathogens that possess shorter or longer biotrophic stages, B. cinerea is identified as an unambiguously necrotrophic pathogen [17], [20]. This ascomycete fungus occupies broad geographic and host ranges and exhibits a high degree of genetic and phenotypic variability [38]–[40]. However, this variation has been little explored in the context of plant defense signaling. Testing the interaction between a collection of B. cinerea isolates and A. thaliana mutant genotypes with defined deficiencies in jasmonate signaling revealed significant variation in plant response to B. cinerea isolates that was not apparent in wild-type plants. This included variation in lesion phenotype, altered mRNA transcript accumulation responses, and variation in accumulation of the A. thaliana defense metabolite camalexin. An unexpected dependency of camalexin accumulation in response to B. cinerea infection on intact jasmonate signaling was also revealed. The results presented here, while not contradicting the accepted view that jasmonate-mediated defense is vital for plant resistance to B. cinerea, suggest that additional pathways modulate A. thaliana—B. cinerea interactions. Finally, the architecture of plant defense signaling networks that provide resistance to necrotrophic pathogens is not static, and will vary with the pathogen genotype investigated.

Results

Pathogen variation in jasmonate-dependent infection phenotypes

To test effects of jasmonate-mediated plant defense on diverse B. cinerea isolates, A. thaliana leaves of the aos genotype (deficient in jasmonate biosynthesis) and its corresponding wild-type were inoculated with 10 diverse B. cinerea isolates, two abiotic elicitors (acifluorfen and AgNO₃), or a mock inoculation (Table 1) [41]. Visible initiation of leaf necrotic lesions was observed between 24 and 48 hours post inoculation with B. cinerea. While tissue necrosis of aos plants initiated within a time frame similar to wild-type plants, lesions expanded more rapidly in aos plants, with near total consumption of the leaf by B. cinerea between 72 and 96hpi. aos mutant leaves failed to develop the zone of chlorosis surrounding the developing lesion that is often observed in B. cinerea infections (Figure 1A).

A comparison of camalexin accumulation in wild-type versus aos leaves induced by 10 B. cinerea isolates revealed significant diversity (Figure 2). Among the B. cinerea isolate treatments tested, camalexin accumulation in aos leaves ranged from 5% to 50% of camalexin accumulation in wild-type leaves, with a median camalexin accumulation among B. cinerea infections of 14% wild-type levels. Mock treatment, acifluorfen, and AgNO₃ induced camalexin in aos leaves at 5–7% wild-type levels. In no case was the absence of jasmonate synthesis associated with increased camalexin accumulation. To explore the observed pathogen variation in interaction with jasmonate-deficient genotypes and activation of metabolic defense, the two B. cinerea isolates inducing camalexin accumulation in the aos leaves at the highest and lowest levels relative to wild-type leaves, BcGrape (5%) and Bc83-2 (50%) were used for further experiments.

One hypothesis that could explain the differential accumulation of camalexin in BcGrape and Bc83-2 infected jasmonate-deficient plants is that one of the B. cinerea isolates produces a molecule that stimulates the intact jasmonate perception in the A. thaliana aos mutant. To determine whether plant deficiencies in jasmonate synthesis and jasmonate perception create similar infection phenotypes and show fully overlapping effects on plant defense signaling, we generated a double mutant containing both aos and the coronatine-insensitive 1 (coi1) mutation that confers deficiency in jasmonate perception [42]. A population segregating both coi1 and aos mutations was experimentally infected with BcGrape and Bc83-2. coi1 aos double mutant plants displayed infection phenotypes for both tested isolates that did not differ significantly from those observed in either the single mutant coi1 or aos plants (Figure 3). Both the coi1 and aos mutations appear recessive for these phenotypes, as infection phenotypes of plants heterozygous for either or both mutations tested did not differ significantly from homozygous wild-type plants (data not shown). The similarity of coi1 and aos phenotypes suggested that camalexin accumulation in jasmonate-deficient plant genotypes infected with Bc83-2 is not likely mediated by isolate-specific production of a metabolite with jasmonate-like coi1 dependent activity similar to coronatine [43].

Testing this segregating population also showed that the glabrous (gl1) mutation, present in the aos mutant background and thus segregating in the aos×coi1 F2 population, had no significant effect on lesion size or camalexin accumulation [44]. We additionally tested a downstream component of the JA pathway, utilizing JAZ1Δ3 mutant plants. These plants produce a modified version of the JAZ1 protein that confers a dominant jasmonate-insensitive phenotype. The JAZ1Δ3 mutant plants showed defects in B. cinerea mediated camalexin induction similar to aos and coi1 plants, but with a less-dramatic increase in lesion size (Figure S1). These defense responses showed similar B. cinerea isolate dependency to that observed in aos and coi1. Thus, B. cinerea isolates vary in their stimulation of signaling networks within A. thaliana as demonstrated by the ability of Bc83-2 to induce moderate camalexin levels in the absence of a functional jasmonate signaling pathway (Figure 3).

The interaction of jasmonate-mediated defense with camalexin biosynthesis

The A. thaliana Phytoalexin Deficient 3 (PAD3) locus encodes a cytochrome P450 enzyme catalyzing the final steps of camalexin biosynthesis [45]. The increased susceptibility of pad3 mutants to necrotrophic pathogens has supported the conclusion that camalexin is an important defense against these pathogens [28], [46]. We showed that camalexin accumulation depends in part upon an intact jasmonate signaling pathway (Figures 2 and 3). To evaluate the extent that increased susceptibility of jasmonate-insensitive A. thaliana genotypes is due to decreased camalexin accumulation in these mutants, we measured development of necrotic lesions and camalexin accumulation in experimentally-infected Col-0 (wild-type), coi1, pad3, and coi1 pad3 double mutant plants (Figure 4). Lesion size at 72hpi did not differ between coi1 and coi1 pad3 plants, but both of these genotypes developed significantly larger lesions than pad3 single mutants, indicating that camalexin deficiency explains a significant fraction of, but not the entire increase in, susceptibility of jasmonate mutants to B. cinerea (Figure 4A). As anticipated, pad3 and coi1 pad3 plants did not accumulate measurable amounts of camalexin (Figure 4B). This observation shows that camalexin accumulation in jasmonate mutants infected with Bc83-2 is not due to a previously-undescribed camalexin biosynthetic capacity in B. cinerea. Further, the similarity in lesion size between pad3 mutant plants infected with BcGrape and Bc83-2 suggests that the difference in susceptibility of jasmonate mutants to these two isolates is not explained by camalexin accumulation in jasmonate mutants infected with Bc83-2.

While BcGrape and Bc83-2 induced similar levels of necrosis on wild-type and pad3 plants, lesions produced by Bc83-2 on coi1 and coi1 pad3 plants were significantly smaller than those produced by BcGrape, supporting our observations that jasmonate deficiency had comparatively less impact on plant susceptibility to Bc83-2 (Figures 2 and 4). Consistent with previous experiments, coi1 plants infected with BcGrape accumulated extremely low levels of camalexin that did not significantly differ from levels accumulated in pad3 mutants, and coi1 plants infected with Bc83-2 accumulated camalexin at levels significantly lower than wild-type but significantly greater than pad3 mutant plants (Figure 4). In combination, this shows that while camalexin is a large component of the jasmonate-mediated defense against B. cinerea, its accumulation does not explain the differential virulence of Bc83-2 and BcGrape on jasmonate-deficient A. thaliana.

Camalexin accumulation in wild-type and jasmonate-insensitive leaves over a time course of B. cinerea infection

To determine whether observed differences in camalexin accumulation and lesion growth between B. cinerea treatments were associated with differences in the timing of plant response, time course experiments were conducted using wild-type (COI1/COI1) and coi1 mutant plants (Figure 5). B. cinerea isolates BcGrape and Bc83-2 produced similarly-sized necrotic lesions on wild-type leaves at 48 hpi, but lesions produced by BcGrape infection of coi1 leaves rapidly expanded starting at 40–48 hpi. Bc83-2 showed an increase in induced necrosis on coi1 leaves that was less dramatic than shown by BcGrape but still significantly larger than necroses formed on wild-type leaves. By 32 hpi, camalexin was significantly induced in wild-type but not coi1 leaves (Figure 5B). Camalexin accumulation at all time points after 24 hpi was highest in wild-type leaves infected with Bc83-2. coi1 infected with Bc83-2 showed consistently higher levels of camalexin than coi1 infected with BcGrape. Thus, the difference in camalexin response or virulence between Bc83-2 and BcGrape does not appear to be solely an issue of infection timing but rather variation in pathogen interaction with the plant.

Transcription of camalexin biosynthetic genes

To explore mechanisms controlling altered accumulation of camalexin in jasmonate deficient plants as well as differences between B. cinerea treatments, we examined transcript levels of PAD3 and CYP71A13. These genes encode enzymes which catalyze respectively the first committed step and the final steps in camalexin biosynthesis [45], [47], [48]. Relative levels of PAD3 and CYP71A13 transcripts were measured at 24 and 48 hours post-inoculation, time points flanking the observed onset of camalexin accumulation (Figure 5B). PAD3 transcript levels were low but detectable at 24 hours post inoculation (Figure 6A). At 48 hpi, all B. cinerea treated samples showed significantly increased PAD3 transcript accumulation compared to mock treatments. Samples from coi1 mutants showed less induction of PAD3 than wild-type samples but the reduction was not commensurate with the observed decrease in metabolite accumulation. While camalexin accumulation was nearly abolished in coi1 infected with BcGrape, PAD3 transcript was reduced by only half. Further, Bc83-2 infection is associated with relatively higher camalexin accumulation in coi1, but significantly lower PAD3 transcript accumulation in coi1 compared to BcGrape infected coi1. CYP71A13 transcript accumulation showed a similar pattern (Figures S2 and S4). Lack of correlation between PAD3 transcript accumulation and camalexin accumulation measured from the same tissue pool contrasts with previous reports that PAD3 transcript and camalexin accumulation are highly correlated (Figure 6B) [45]. B. cinerea infection with diverse isolates thus reveals evidence of additional regulation of camalexin biosynthesis, beyond transcriptional regulation of known biosynthetic genes.

As camalexin accumulation during B. cinerea infection occurs primarily within the plant tissue immediately bordering the developing lesion, it is possible that the spatial distribution of camalexin biosynthetic transcript within an infected leaf may be more relevant to camalexin accumulation than total transcript accumulation within a leaf [30]. To visualize effects of jasmonate insensitivity and B. cinerea isolate differences on the pattern of transcript accumulation of the camalexin biosynthetic enzyme CYP79B2, we crossed a CYP79B2 promoter-GUS fusion transgene into a coi1 background. CYP79B2 catalyzes the conversion of tryptophan to indole-3-acetaldoxime during camalexin biosynthesis in planta [45], [49]. Leaves from homozygous wild-type and coi1 plants showing GUS activity were inoculated with B. cinerea isolates BcGrape and Bc83-2. Wild-type leaves infected with either B. cinerea isolate showed blue staining indicative of GUS activity in a narrow zone bordering the lesion, consistent with previous studies showing that camalexin accumulates primarily within this zone (Figure 1) [30]. coi1 leaves showed a dramatic difference in staining pattern between BcGrape and Bc83-2 infections, with BcGrape-infected coi1 leaves showing patterns of GUS activity similar to those seen in wild-type plants, and Bc83-2 infected coi1 leaves showing intense blue staining within the area visually defined as the necrotic lesion. This intense staining was not associated with increased accumulation of CYP79B2 transcript in coi1 leaves infected with Bc83-2 (Figure S3). The presence of the ProCYP79B2:GUS transgene did not significantly affect camalexin accumulation compared to plants without the transgene from the same segregating F2 population.

A possible explanation for the above observation is that there is less cell death within the Bc83-2 lesion in comparison to BcGrape. We stained infected leaves with a vital stain, Trypan Blue, to compare patterns of cell death associated with infection by the two isolates on wild-type and coi1 leaves. This showed similarly sized halos of plant cell death surrounding the BcGrape and Bc83-2 lesions on both wild-type and coi1 leaves that was a lighter color in the coi1 lesions (Figure 7). Interestingly, these areas contained no detectable fungal cells, suggesting that plant cell death can be caused by mobile plant or fungal signals. No living or dead plant cells were visible within the hyphal mass, suggesting that B. cinerea rapidly consumes material in this region and that the observed difference in camalexin accumulation is not due to differential presence of plant cells. These results suggest that the observed GUS staining pattern is caused by persistence of plant-produced protein within the Bc83-2 lesion, rather than active transcription and translation from the plant genome within the Bc83-2 lesion, implying that the absence of a functional jasmonate signaling network alters the ability of Bc83-2 to degrade or disperse proteins. Trypan Blue staining also showed that the two isolates have different growth habits independent of the plant genotypes tested. Bc83-2 hyphae grew at higher density with a well-defined boundary to the hyphal mass, while BcGrape hyphae grew more sparsely with isolated probing hyphae that grow into the surrounding plant issue.

We further compared the infection phenotypes of BcGrape and Bc83-2 using staining for H₂O₂ accumulation (DAB). On wild-type A. thaliana leaves, infection by either tested B. cinerea isolate was associated with diffuse H₂O₂ generation within and around the lesion, suggesting that both the plant and fungus generate H₂O₂. In contrast, Bc83-2 caused a strong halo of H₂O₂ surrounding the developing lesion on coi1 whereas the BcGrape lesions were associated with a H₂O₂ accumulation pattern similar to that observed in wild-type leaves (Figure 7). As generation of reactive oxygen species, including H₂O₂, is associated with production of camalexin, the observed pattern of H₂0₂ accumulation supports our earlier observation that Bc83-2 induces camalexin via a jasmonate-independent mechanism that is lacking in BcGrape infections. Interestingly, this staining also showed that infection by BcGrape is associated with a systemic accumulation of H₂O₂ in trichomes that was independent of plant jasmonate perception and not seen in leaves infected with Bc83-2 (Figure 7). These B. cinerea isolates elicit distinct defense responses from plants that include both jasmonate-dependent and jasmonate-independent phenotypes, suggesting both the danger of oversimplifying models of plant—“B. cinerea” interaction and the rich potential of intraspecific studies of this pathogen.

Transcriptional profiling

To identify additional differences in plant transcriptional response to these two B. cinerea isolates and build hypotheses regarding the molecular basis of differences in infection phenotype, whole-genome transcriptional profiles of A. thaliana leaves inoculated with B. cinerea isolates BcGrape or Bc83-2 were compared to each other and to control leaves using both wild-type and jasmonate-insensitive (coi1) plants. Based on directed transcript measurements, where induction of camalexin biosynthetic and other defense-associated transcripts was not detected until 48hpi, transcriptional profiling was performed on samples from this 48hpi time point (Figures 6 and S2). Additionally, both B. cinerea isolates had initiated lesions by 48hpi, but lesions at this time point, arising from a single inoculation droplet per leaf, occupy only a small portion of the total leaf area and do not show the large differences in lesion size observed on coi1 leaves at later time points (Figure 5). Estimates of transcript accumulation obtained from arrays were highly consistent with targeted transcript measures obtained via quantitative RT-PCR, with significant Pearson correlation coefficients ranging from 0.76 to 0.91 (Figure S4). Array data are provided as Dataset S1.

A. thaliana transcriptional responses to B. cinerea infection

Of 22810 transcripts represented on the arrays, over half (12,999) showed significant effects for the model transcript = genotype + treatment + (genotype × treatment) even after false-discovery adjustments. The majority (11,989) of these statistically significant transcript changes were associated with treatment where most of these transcripts differed between B. cinerea-infected and control leaves, rather than between leaves infected with the two B. cinerea isolates. We therefore describe statistically significant plant responses consistent between both pathogen isolates as responsive to “B. cinerea”. Transcript accumulation from 1458 genes of the B. cinerea-responsive loci identified above showed greater than 2-fold increase in response to B. cinerea infection, while transcripts from 1602 genes showed more than 2-fold decrease relative to control samples. Differences in transcript abundance between wild-type and coi1 plants as well as between B. cinerea-inoculated and control plants showed overlap with previous studies [50], [51].

All known enzymes of the camalexin biosynthetic pathway were upregulated by B. cinerea infection, with CYP71A13 and PAD3 respectively showing 124-fold and 67-fold increases in B. cinerea infected leaves. An additional five transcripts contributing to biosynthesis of the camalexin precursor, tryptophan, were also upregulated in response to B. cinerea, but less dramatically than camalexin biosynthetic genes (Table S1). Other transcripts showing greater than 2-fold transcriptional effects of B. cinerea infection that have been previously identified as contributing to plant defense against fungal pathogens included a camalexin regulator (PAD4), the MYB transcription factor botrytis-susceptible 1 (BOS1), phenylalanine ammonia lyase (PAL1), polygalacturonase-inhibiting protein (PGIP1), and pathogenesis response proteins (PR1, PR4, and PR5) (Table S1). Transcripts of PDF1.2a and VSP2, considered markers for jasmonate signaling, were detected only at extremely low levels in both B. cinerea-infected and control leaves from coi1 plants, further supporting our conclusion that camalexin accumulation in jasmonate mutants infected with Bc83-2 is not attributable to isolate-specific pathogen-mediated jasmonate signaling independent of coi1 and aos (Figures S2 and S4) [52], [53].

Pathway responses

To associate biological activities with the numerous transcriptional changes caused by B. cinerea infection, genes showing >2-fold transcriptional changes in B. cinerea-infected leaves were grouped by annotated associations with metabolic pathways [54]. In addition to upregulation of camalexin and tryptophan biosynthetic genes, described above, transcripts associated with ascorbate-glutathione metabolism, including 10 glutathione transferases, were strongly upregulated by B. cinerea infection (Table S1). Genes associated with lignin biosynthesis and jasmonate synthesis and response, including six genes encoding JAZ proteins, also showed positive transcriptional responses to B. cinerea infection. Pathways downregulated in B. cinerea-infected leaves primarily control core metabolic functions such as biosynthesis of chlorophyll and starch, but transcripts linked with the biosynthesis of aliphatic glucosinolates, metabolites primarily associated with plant defense against insect herbivores, were an exception to this pattern. Aliphatic glucosinolate-associated transcripts, including three regulatory MYB transcription factors, were strongly decreased in B. cinerea-infected leaves (Table S1). This is consistent with previously documented local repression of aliphatic glucosinolate biosynthesis by B. cinerea infection [55].

Identification of putative response networks

We used the A. thaliana co-expression database ATTED-II to investigate patterns of co-expression for genes transcriptionally affected by B. cinerea infection that are not currently associated with described metabolic pathways. Microarray data have successfully identified genes controlling A. thaliana—B. cinerea interactions [50], [56]–[59]. Among the transcripts lacking prior pathway associations, we identified three groups of co-regulated loci that may represent undescribed B. cinerea responsive networks (Table 2). These proposed groups, described below, represent hypothesized contributions of these genes to A. thaliana—B. cinerea interaction, requiring experimental validation.

DETOX: Transcript from the At2g04050 locus, encoding a Multidrug and Toxin Extrusion (MATE) efflux family protein, has been previously documented to increase in response to elevated soil concentrations of boron, tri-nitro toluene, and NaCl [60]–[62]. 16 B. cinerea responsive transcripts were identified as co-regulated with At2g04050 These transcripts were induced by both B. cinerea isolates, but accumulated to lower levels in coi1 leaves infected with BcGrape than wild-type leaves infected with BcGrape, while Bc83-2 infected leaves showed an opposite pattern (Figure 8). This suggests that jasmonate signaling activates this putative network in response to BcGrape but represses it in response to the Bc83-2 isolate, indicating that intraspecific pathogen diversity can affect the outcome of jasmonate signaling. The genes in this network included several multidrug transporters that may act in response to fungal toxins, and UGT74E2, a glucosyltransferase implicated in detoxification (Table 2) [60], [63]. This network showed an overrepresentation of ABA response elements (ABRE) suggesting a possible influence of ABA [64], [65].

Glucosinolate turnover: Another co-expressed cluster of 15 B. cinerea-induced transcripts includes loci encoding enzymes hypothesized to function in catabolism of glucosinolates (Table 2). Glucosinolate turnover may play a role in fungal defense by allowing redistribution of cellular resources stored in glucosinolates to antifungal metabolites [55]. Alternatively, accumulation of these transcripts in response to B. cinerea may relate to the function of glucosinolate activation products in pathogen defense and signaling [66], [67]. This hypothesized upregulation of glucosinolate catabolism contrasts with downregulation of transcripts involved in the biosynthesis and activation of aliphatic glucosinolates in B. cinerea-infected leaves. Transcripts involved in both the synthesis of aliphatic glucosinolates and their hypothesized catabolism were detected at higher levels in wild-type leaves than coi1 leaves (Figure 8).

MATE: A group of 40 transcripts induced by B. cinerea infection were identified by association with the highly B. cinerea-responsive locus At3g23550, encoding another MATE transporter (Table 2 and Table S2). MATE proteins are associated with resistance to toxins, but may also be involved in transport of plant-produced metabolites required for defense [68]–[70]. This group of transcripts also contains several likely biosynthetic genes, such as acyltransferases, oxidoreductases, and cytochromes P450. Promoter analysis showing an over-representation of two elements, ABRE and GC box, supports coordinated transcriptional response of these genes [64], [65]. Considering inclusion of transcripts previously associated with plant defense against necrotrophic pathogens, such as PDF1.2 defensins and the ethylene and jasmonate responsive transcription factor ORA59 (a member of a secondary metabolite regulatory gene family), in this group we hypothesize that these genes contribute to biosynthesis and transport of a currently unknown defense-associated metabolite [71].

Comparison of transcriptional effects: BcGrape vs. Bc83-2

Differences in transcript accumulation after infection by BcGrape or Bc83-2 were generally similar in direction of effect between wild-type and coi1 leaves but of greater magnitude in coi1 leaves. Of 824 transcripts showing differential accumulation in response to the two tested B. cinerea isolates, 787 show larger differences in coi1 leaves than wild-type (Table S2). While this correlates with lesion development at later time points, lesion sizes at 48 hours do not significantly differ among genotype×isolate combinations (Figure 5). To identify patterns in these transcript differences that might enhance our understanding of the biology of A. thaliana response to B. cinerea, we clustered these transcripts by similarity of normalized transcript levels. This identified two large groups of transcripts, those showing relative increases in transcript level in response to B. cinerea (clusters 1–3) and those relatively decreased in B. cinerea-infected leaves (clusters 4–6) (Figure 9, Table S2). Subsequent clustering of transcript profiles for these loci by genotype and treatment suggested that BcGrape and Bc83-2 infections exert similar transcriptional effects on wild-type leaves. This contrasts with a dramatic difference in transcript patterns observed in coi1 samples, where Bc83-2 infected coi1 leaves showed transcript patterns similar to mock-inoculated samples while BcGrape infected coi1 were more transcriptionally similar to infected wild-type samples. This echoes the pattern observed for the putative DETOX network for A. thaliana response to B. cinerea, and suggests that jasmonate has opposing transcriptional effects on a set of genes in response to these two B. cinerea isolates (Figure 8).

Differential transcriptional response to BcGrape

In wild-type plants, the transcript with the greatest increase in accumulation in leaves infected with BcGrape relative to Bc83-2 differed only by 1.75-fold. In coi1 leaves, however, a greater than 10-fold difference was observed between the two B. cinerea treatments. This transcript, At3g23550 from the above MATE network, and associated transcripts (Table 2), showed greater accumulation in leaves infected with BcGrape relative to leaves infected with Bc83-2 with a differential coi1 dependence between the two isolates (Table S2, Figure 9 (clusters 3, 4, and 6)).

Differential transcriptional response to Bc83-2

The locus associated with the largest transcript difference where Bc83-2 infected leaves showed higher transcript levels than BcGrape infected leaves was At1g52690, encoding a late embryogenesis abundant (LEA) protein. LEA proteins are associated with seed maturation, but are also suggested to play important roles in stress tolerance in vegetative tissues [72]. This transcript has not been previously described as pathogen-responsive, but is upregulated in response to exogenous application of ABA and osmotic stress (Genevestigator Response Viewer; www.genevestigator.ethz.ch). In Bc83-2 inoculated wild-type leaves, this transcript accumulated to levels 3-fold greater than those observed in BcGrape-infected wild-type leaves, and more than 11-fold greater than observed in BcGrape-infected leaves from coi1 plants. However, in the initial analysis of B. cinerea effects on transcript abundance, At1g52690 transcript was identified as downregulated by B. cinerea infection, suggesting that this difference represents a failure of Bc83-2 infection or associated plant defense response to reduce transcript accumulated from this locus.

16 associated transcripts showed expression patterns similar to At1g52690. All of these loci showed higher relative transcript accumulation in Bc83-2 infected leaves compared to those infected with BcGrape (Table S2). These transcripts were also detected at higher levels in coi1 leaves infected with either B. cinerea isolate than similarly-treated wild-type leaves, suggesting that these transcripts are repressed by jasmonate-mediated response to B. cinerea infection (Figure 8). The expression pattern displayed by these transcripts in Bc83-2 infected leaves is similar to the pattern observed in uninfected leaves, further supporting the hypothesis that BcGrape infection represses accumulation of these transcripts, but that this does not occur during a similar stage of infection with Bc83-2 (Figure 9). As such, BcGrape and Bc83-2 differ in induction of both positive and negative transcriptional responses in planta.

Discussion

Jasmonate signaling plays a vital role in plant defense against the highly variable necrotrophic fungal pathogen B. cinerea but its molecular effects may differ with pathogen diversity. We show that the genetic diversity contained within B. cinerea generates quantitative variation in plant response to the pathogen in the absence of jasmonate synthesis or perception. This variation in plant response was most readily observed as differential accumulation of the A. thaliana defense metabolite camalexin, which did not directly correspond with changes in associated biosynthetic transcripts. Analyses of A. thaliana transcriptional responses to B. cinerea isolates BcGrape and Bc83-2 revealed highly similar changes in transcript levels induced by infection of wild-type plants, yet dramatic differences in transcript profiles between A. thaliana infected with these two pathogen isolates when jasmonate signaling is impaired by mutation of COI1.

Regulation of camalexin accumulation

Jasmonate signaling controls a substantial portion of camalexin accumulation

Camalexin and jasmonate-mediated defenses have been presented as separate elements of A. thaliana resistance to necrotrophic pathogens because jasmonate mutants had no detected effect on the accumulation of camalexin biosynthetic transcripts [51], [73]. We show that, in response to B. cinerea, camalexin accumulation was substantially decreased by deficiencies in jasmonate signaling. A. thaliana mutants deficient in jasmonate synthesis or jasmonate perception showed significantly lower camalexin accumulation than wild-type plants for all 10 B. cinerea isolates tested. Observation of similarly dramatic decreases in camalexin accumulation for jasmonate deficient plants treated with the abiotic elicitors AgNO₃ and acifluorfen suggests that decreased camalexin accumulation observed in response to B. cinerea infection in these genotypes is not due to an active pathogen repression of camalexin biosynthesis or accumulation in the absence of jasmonate signaling (Figure 2). The similarity of responses shown by three mutants deficient in distinct aspects of jasmonate signaling, aos, coi1, and JAZ1Δ3, suggests that decreased camalexin induction represents a requirement for the entire jasmonate pathway. Yet previous study suggested that camalexin accumulation in response to B. cinerea infection does not require MYC2 [74], a transcriptional regulator of jasmonate signaling, which is repressed by physical interaction with JAZ proteins [75]–[77]. This suggests that jasmonate signaling controls B. cinerea-induced camalexin production via an unidentified transcription factor that is repressed by JAZ proteins in a manner similar to MYC2.

Camalexin biosynthesis is regulated at multiple functional levels

Previous studies report accumulation of camalexin biosynthetic transcripts in coi1 mutant plants at levels greater than or equal to wild-type in response to B. cinerea and oligogalacturonides, leading to the conclusion that jasmonate and camalexin responses are not connected [50], [51]. In light of the common assumption that camalexin transcript levels predict the level of metabolite accumulation, the metabolite is infrequently measured. Our data showed a dramatic decrease in camalexin metabolite accumulation in the coi1 mutant without an equivalent decrease in transcript levels for the first and last enzymatic steps, PAD3 and CYP71A13 (Figures 3 and 6). In addition, variance in camalexin-associated transcripts measured in transcriptional profiling experiments was explained primarily by treatment (B. cinerea infection), rather than genotype, despite an obvious effect of plant genotype on camalexin accumulation (Table S1, Figures 3-5). Thus, observed accumulation of these camalexin biosynthetic transcripts is not a reliable surrogate for measurement of metabolite accumulation. These results suggest that additional regulation of camalexin biosynthesis exists, either post-transcriptional regulation or transcriptional regulation involving an unidentified pathway intermediate. While an alternative explanation, that B. cinerea degrades camalexin, remains plausible, this is not supported by the observation that camalexin accumulation was also lower in jasmonate-deficient leaves treated with abiotic elicitors of camalexin. The relative performance of BcGrape and Bc83-2 on camalexin deficient pad3 plants suggests that these isolates are not camalexin-insensitive, and camalexin insensitivity documented in B. cinerea is associated with export, rather than degradation, of the metabolite [78]. These data support a deficiency in camalexin biosynthesis by the plant, rather than active degradation by the pathogen that functions only in the absence of jasmonate-mediated plant defense (Figure 2). Thus, jasmonate signaling likely plays a complex regulatory role in camalexin synthesis.