Arabidopsis local resistance to Botrytis cinerea involves salicylic acid and camalexin and requires EDS4 and PAD2, but not SID2, EDS5 or PAD4


  • Simone Ferrari,

    1. Department of Genetics, Harvard Medical School, and Department of Molecular Biology, Massachusetts General Hospital, Boston, MA 02114, USA, and
    Search for more papers by this author
  • Julia M. Plotnikova,

    1. Department of Genetics, Harvard Medical School, and Department of Molecular Biology, Massachusetts General Hospital, Boston, MA 02114, USA, and
    Search for more papers by this author
  • Giulia De Lorenzo,

    1. Dipartimento di Biologia Vegetale, Università di Roma ‘La Sapienza’, Piazzale Aldo Moro 5, 00185 Rome, Italy
    Search for more papers by this author
  • Frederick M. Ausubel

    Corresponding author
    1. Department of Genetics, Harvard Medical School, and Department of Molecular Biology, Massachusetts General Hospital, Boston, MA 02114, USA, and
      For correspondence (fax +1 617 726 5949; e-mail
    Search for more papers by this author

For correspondence (fax +1 617 726 5949; e-mail


Salicylic acid (SA) is an important regulator of plant defense responses, and a variety of Arabidopsis mutants impaired in resistance against bacterial and fungal pathogens show defects in SA accumulation, perception, or signal transduction. Nevertheless, the role of SA-dependent defense responses against necrotrophic fungi is currently unclear. We determined the susceptibility of a set of previously identified Arabidopsis mutants impaired in defense responses to the necrotrophic fungal pathogen Botrytis cinerea. The rate of development of B. cinerea disease symptoms on primary infected leaves was affected by responses mediated by the genes EIN2, JAR1, EDS4, PAD2, and PAD3, but was largely independent of EDS5, SID2/ICS1, and PAD4. Furthermore, plants expressing a nahG transgene or treated with a phenylalanine ammonia lyase (PAL) inhibitor showed enhanced symptoms, suggesting that SA synthesized via PAL, and not via isochorismate synthase (ICS), mediates lesion development. In addition, the degree of lesion development did not correlate with defensin or PR1 expression, although it was partially dependent upon camalexin accumulation. Although npr1 mutant leaves were normally susceptible to B. cinerea infection, a double ein2 npr1 mutant was significantly more susceptible than ein2 plants, and exogenous application of SA decreased B. cinerea lesion size through an NPR1-dependent mechanism that could be mimicked by the cpr1 mutation. These data indicate that local resistance to B. cinerea requires ethylene-, jasmonate-, and SA-mediated signaling, that the SA affecting this resistance does not require ICS1 and is likely synthesized via PAL, and that camalexin limits lesion development.


Plants are constantly exposed to a wide range of microbial pathogens that use disparate strategies to attack their hosts. At the early stages of their development, biotrophic pathogens do not kill plant cells, but rather subvert the host metabolism to favor their own growth. In contrast, necrotrophic pathogens need to kill host cells to metabolize their contents. The crucial role of salicylic acid (SA) in plant defense response against different pathogens is well documented. Tobacco and Arabidopsis plants transformed with the bacterial gene nahG, encoding an SA hydroxylase, are unable to accumulate SA and show enhanced susceptibility to both compatible and incompatible pathogens, and reduced expression of the SA-inducible pathogenesis-related protein PR1 (Delaney et al., 1994; Gaffney et al., 1993).

It is commonly believed that SA can be synthesized in plant cells through the phenylpropanoid pathway, the first step of which is catalyzed by phenylalanine ammonia lyase (PAL; Coquoz et al., 1998). More recently, an alternative pathway involving isochorismate synthase (ICS) has been shown to be required for SA accumulation, PR1 induction, and local and systemic acquired resistance (SAR) responses in Arabidopsis (Wildermuth et al., 2001). Wildermuth et al. (2001) postulated that SA biosynthesis mediated by PAL occurs in cells undergoing cell death at the site of infection, whereas SA synthesized via ICS1 is required for local and SAR responses occurring in adjacent and distal plant cells. The role of these two pathways in response to necrotrophic fungi has not been determined.

The key role of SA in plant resistance is demonstrated by Arabidopsis npr1/nim1/sai1 mutants, which are defective in PR1 induction, have enhanced susceptibility to Pseudomonas syringae, Peronospora parasitica, and Erysiphe orontii, and are unable to mount an effective SAR response (Cao et al., 1994; Reuber et al., 1998; Ryals et al., 1997; Shah et al., 1997). It appears that NPR1/NIM1/SAI1 (henceforth NPR1) most likely functions in a process that involves SA perception, as npr1 mutants are unable to respond to exogenous SA (Cao et al., 1994; Ryals et al., 1997; Shah et al., 1997). In addition to npr1 mutants, other Arabidopsis mutants with enhanced susceptibility to pathogens are also impaired in SA perception or accumulation. For example, mutations in the EDS5 (Nawrath et al., 2002) and PAD4 (Jirage et al., 1999) genes, which encode a protein with similarity to MATE transporters and a lipase, respectively, prevent the accumulation of SA in plants infected with P. syringae, E. orontii, and P. parasitica, resulting in more extensive colonization (Glazebrook et al., 1996, 1997; Nawrath and Metraux, 1999; Reuber et al., 1998; Rogers and Ausubel, 1997; Zhou et al., 1998). The eds4 mutant, isolated in a screen for plants more susceptible to P. syringae (Glazebrook et al., 1996), also exhibits reduced SA accumulation in response to bacterial infection (Gupta et al., 2000). Conversely, Arabidopsis mutants such as cpr1, cpr5, cpr6, cim9, and cim13, which constitutively accumulate high levels of SA in absence of infection, are also more resistant to P. syringae and a variety of biotrophic pathogens including P. parasitica (cpr1, cpr5, cpr6) and Erysiphe species (Bowling et al., 1994, 1997; Clarke et al., 1998; Maleck et al., 2002).

Salicylic acid signaling does not occur solely through NPR1. There is accumulating evidence for SA-dependent, NPR1-independent signaling pathways, as exemplified by defense responses such as the accumulation of the antimicrobial compound camalexin, which requires SA but is not mediated by NPR1 (Glazebrook and Ausubel, 1994; Thomma et al., 1999b). There is also evidence that activation of the pathogenesis-related genes PR2 (BGL2) and PR5 can occur through SA-dependent, NPR1-independent pathways. For instance, an eds5 npr1 double mutant shows reduced PR gene expression and enhanced susceptibility to E. orontii, compared to the single eds5 and npr1 mutants (Reuber et al., 1998). Similarly, double cpr6 npr1 mutants exhibit constitutive PR gene expression and resistance to P. parasitica similarly to the cpr6 mutant (Clarke et al., 1998).

Interestingly, SA-dependent signaling appears to activate defense-response pathways that primarily confer resistance to bacterial or biotrophic fungal pathogens. In contrast, several lines of evidence suggest that SA-independent responses are primarily involved in conferring resistance against necrotrophic fungi, including Botrytis cinerea, the causal agent of gray mold. B. cinerea causes soft rot symptoms in more than 200 different plant species, including Arabidopsis, and genetically defined resistance against B. cinerea has not been described. Infection of Arabidopsis plants with B. cinerea causes the induction of a subset of defense genes that are not induced by SA, including PDF1.2, which encodes an antifungal defensin-like peptide (Penninckx et al., 1996; Zimmerli et al., 2001). Induction of PDF1.2 by B. cinerea is blocked in ein2 and coi1 mutants (Penninckx et al., 1996; Zimmerli et al., 2001), which are defective in ethylene or jasmonic acid (JA) signal-transduction pathways, respectively (Feys et al., 1994; Guzman and Ecker, 1990), and ein2 and coi1 plants are highly susceptible to B. cinerea infection (Thomma et al., 1998, 1999a). Consistent with these data, B. cinerea infection fails to induce SA accumulation or SAR in Arabidopsis (Govrin and Levine, 2002). On the other hand, several reports show that SA or its analog BTH can induce resistance to B. cinerea in several plant species, including bean, tobacco, and tomato (Audenaert et al., 2002; De Meyer et al., 1999; Murphy et al., 2000). Moreover, Arabidopsis plants expressing the nahG gene and infected with B. cinerea show larger lesions compared to the parental plants (Govrin and Levine, 2002). To reconcile some of the apparently conflicting data in the literature about the role of SA in response to B. cinerea infection, we hypothesized that defense responses mediated by SA affect the rate of colonization of the primary inoculated leaves during the early steps of the infection at the site of the lesion, but play relatively minor roles in preventing systemic infection. In this paper, we report on experiments in which we analyzed the effect on B. cinerea symptoms of previously described Arabidopsis mutations in genes involved in the transduction of SA, ethylene, and JA-defense signals. We also determined the effect of exogenous SA on lesion development in wild-type and mutant plants. Our data show that the rate of B. cinerea lesion formation at the primary site of infection depends on ethylene-, JA-, and SA-signaling pathways as well as on the synthesis of the phytoalexin camalexin. Interestingly, SA mediating localized resistance appears to be synthesized via PAL and not ICS, and to act independently of NPR1.


Local and systemic resistance to Botrytis cinerea in Arabidopsis mutants impaired in defense responses

Isolated leaves of wild-type Arabidopsis plants and of a set of mutants previously shown to exhibit enhanced susceptibility to various pathogens were inoculated with a B. cinerea spore suspension, and lesion size was measured 3 days after inoculation. The results for all the genotypes tested are summarized in Table 1. Infection of ein2-1 (Figure 1a) or etr1-1 (data not shown), which are insensitive to ethylene (Chang et al., 1993; Guzman and Ecker, 1990), as well as jar1-1 (see below), coi1-1, and eds8-1 plants (data not shown), which are insensitive to JA (Feys et al., 1994; Staswick et al., 1992; Ton et al., 2002), resulted in lesions about twice as large as those formed on wild-type leaves, confirming the importance of ethylene- and JA-mediated signaling in local resistance to B. cinerea. Infection of nahG transgenic plants or eds4-1 mutant plants also resulted in lesions about twice the size as those observed on wild-type plants. In contrast, mutations in the NPR1, EDS5, SID2, and PAD4 genes did not significantly affect the size of the lesions (Figure 1a). Mutations in the NDR1 and EDS1 genes, which are required for gene-for-gene resistance against avirulent pathogens (Century et al., 1997; Falk et al., 1999), also did not significantly affect the development of symptoms caused by B. cinerea (data not shown).

Table 1.  Summary of Arabidopsis lines used in this work and their local susceptibility to Botrytis cinerea
GenotypeaAffected pathwaysbSusceptibilitycReferences
  • a

    All lines are in Columbia-0 background, except where differently indicated.

  • b

    For references to the description of the phenotype of the mutants, see text.

  • c

    Size of lesions at site of infection caused by Botrytis cinerea in rosette leaves of mutant or transgenic plants compared to the parental line, as measured 3 days after inoculation. =: lesions not significantly different from the wild type; +: lesions about twofold larger than in the wild type, ++: lesions about threefold larger than in the wild type; +++: lesions more than threefold larger than in the wild type; –: reduced lesion size.

  • d

    The authors only examined systemic susceptibility to B. cinerea.

  • e Insensitivity to SA or ethylene was previously reported (Gupta et al., 2000; Ton et al., 2002), although in this work, PDF1.2 and PR1 appeared normally expressed during infection.

  • f

    Landsberg erecta background.

nahGSA (accumulation)+This work; Govrin and Levine (2002)
npr1-1SA (response)=This work; Thomma et al. (1998)d
ein2-1Ethylene+This work; Thomma et al. (1999b)d
etr1-1Ethylene+This work
jar1-1JA+This work
coi1-1JA+This work; Thomma et al. (1998)d
npr1-1 jar1-1SA and JA+This work
npr1-1 ein2-1SA and ethylene++This work
ein2-1 jar1-1Ethylene and JA++This work
npr1-1 ein2-1 jar1-1SA, ethylene and JA++This work
eds4-1SA and/or ethylenee+This work
eds5-1SA=This work
eds8-1JA+This work
sid2-2SA (biosynthesis)=This work
pad2-1SA, camalexin biosynthesis+++This work
pad3-1Camalexin biosynthesis++This work
pad4-1SA=This work
eds1-2fR gene-mediated resistance=This work
ndr1-1R gene-mediated resistance=This work
cpr1-1SA (constitutive)This work
Figure 1.

Botrytis cinerea local and systemic symptoms in Arabidopsis lines impaired in defense responses.

(a) Size of lesions formed in leaves of wild-type (WT), nahG transgenic, and mutant plants 3 days after inoculation with B. cinerea. Data represent average ± SE of at least 10 lesions. Asterisks represent data sets significantly different from the wild-type data set, according to anova analysis, followed by Bonferroni's multiple comparison test (P > 0.99). This experiment was repeated twice with similar results.

(b) Percentage of dead plants 9 days after inoculation of the lower leaves with B. cinerea. Data represent the average of three independent experiments ± SD (n > 10 for each experiment).

Interestingly, pad2-1 and pad3-1 mutants, which accumulate abnormally low levels of the phytoalexin camalexin in response to pathogen infection (Glazebrook et al., 1997), displayed the highest level of susceptibility to B. cinerea, with lesions four and three times larger than in the parental line, respectively (Figure 1a). Previous reports have failed to show an inhibitory effect of camalexin on B. cinerea growth (Thomma et al., 1999b). However, Figure 2(a) shows that purified camalexin inhibited in vitro B. cinerea growth in a dose-dependent manner, and completely blocked it at a concentration of 50 µg ml−1. The same dose also completely inhibited conidiospore germination in liquid culture (data not shown). Furthermore, HPLC analysis showed that after 48 h of infection with B. cinerea, relatively high levels of camalexin accumulated in wild-type plants, but that accumulation was significantly reduced in inoculated pad2 leaves (about 25% of wild-type levels), completely blocked in pad3 leaves (no detectable induction), but only moderately affected in pad4 leaves (Figure 2b). As lesion development in pad4 and wild-type leaves was comparable, whereas pad2 and pad3 were more susceptible (Figure 1a), it is likely that camalexin plays a role in determining the severity of local symptoms.

Figure 2.

Effect of camalexin on Botrytis cinerea growth.

(a) Round plugs of B. cinerea mycelium were transferred on potato dextrose agar plates supplemented with different concentrations of purified camalexin. The diameter of the fungal colony was measured after 3 days. Data represent average ± SE of three experiments.

(b) Camalexin levels were determined at the indicated time points in leaves inoculated with B. cinerea. Bars represent the percentage of induction, compared to the levels observed in Col-0 (WT) at 2 days post-infection (dpi; 100%). Empty bars, wild type (WT); striped bars, pad2-1; black bars, pad3-1; dotted bars, pad4-1.

When susceptibility to B. cinerea was measured by the percentage of plants that developed a systemic infection and died following inoculation of few lower leaves, nahG and eds4 plants were as susceptible as the wild-type, and pad2 and pad3 plants were only slightly more susceptible. In contrast, ein2 plants were dramatically more susceptible (Figure 1b). Of interest, the degree of systemic susceptibility (dead plants) among the pad mutants correlated with camalexin levels in the inoculated leaves, suggesting that camalexin may play a minor role in limiting systemic infection.

Taken together, these results indicate that defense responses important for local resistance to B. cinerea are not only dependent on JA and ethylene, but also on SA, and require the EDS4 gene product. They also confirm that ethylene-dependent, but not SA-dependent, defense responses play a major role in limiting the systemic spread of B. cinerea to distal portions of the plant. Consistent with these data and confirming a recent report (Govrin and Levine, 2002), PR1 expression, which is SA-dependent, was strongly induced in inoculated leaves, but not in upper, non-inoculated leaves (Figure 3a). Similarly, when leaves of transgenic plants harboring a PR1::GUS construct were inoculated with B. cinerea, a strong increase of GUS activity was detected in cells immediately around the necrotic lesions, but not in more distal regions of the leaves (Figure 3b), nor in the upper, non-inoculated leaves (not shown). In contrast, PDF1.2 expression dramatically increased in both inoculated and non-inoculated leaves with similar kinetics (Figure 3a), consistent with a previous observation that the PDF1.2 promoter is systemically activated by B. cinerea infection (Manners et al., 1998).

Figure 3.

Local and systemic defense gene expression during Botrytis cinerea infection.

(a) Wild-type (WT) intact plants were inoculated with B. cinerea, and local, infected leaves or systemic, non-inoculated leaves were harvested at the indicated time points (days). RNA blot was hybridized with the indicated probes.

(b) GUS staining of leaves from transgenic PR1::GUS plants after infection with B. cinerea. Plants were inoculated as in (a) and, after 48 h, leaves were stained for GUS activity. The picture shows a representative leaf.

The extent of local symptom development is independent of PR1, PDF1.2, or SID2/ICS1 expression

The rate of lesion development observed in nahG, npr1, eds5, and sid2 plants suggested that local resistance to B. cinerea is, at least in part, SA-dependent, but is independent of NPR1, PAD4, EDS5, and SID2 (Figure 1a). This suggests that the SA mediating lesion development is not synthesized via ICS1. However, PR1 expression has been correlated with SA synthesized via ICS1 (Wildermuth et al., 2001). To further examine the role of SA in lesion development, we examined (i) PR1 and ICS1 expression during infection, (ii) the effect of exogenous application of SA, and (iii) the effect of a PAL inhibitor on symptom development.

The role of SA and ICS1/SID2 in conferring B. cinerea resistance was explored by measuring the levels of PR1, PDF1.2, and ICS1/SID2 mRNA after inoculation of various mutants. As shown in Figure 4(a),B. cinerea induced PR1 expression in wild-type plants, but PR1 expression was significantly reduced in nahG, eds5-1, sid2-2, and, to a lesser extent, in npr1-1 plants. These data confirmed several previous reports demonstrating a high degree of correlation between SA levels and PR1 expression and the dependence of PR1 expression on SID2 (Nawrath and Metraux, 1999; Yalpani et al., 1991). As expected, B. cinerea-induced PDF1.2 expression was not affected in the npr1, eds5, or sid2 mutants, but was activated earlier in nahG plants (Figure 4a). Surprisingly, however, ICS1 transcripts were not detectable after 2 days of infection with B. cinerea even in wild-type plants, even though PR1 was clearly induced at the site of the lesion (Figure 3) and PR1 expression was reduced in the sid2-2 mutant (Figure 4a). Perhaps in response to B. cinerea infection, ICS1 expression is highly localized, as is PR1, and not very abundant (in contrast to PR1), and therefore was not detected in these samples. Indeed, an increase of ICS1 mRNA levels has been observed by DNA microarray analysis in tissues immediately surrounding B. cinerea lesions (J.M. Plotnikova and F.M. Ausubel, unpublished data). PR1 expression was also impaired in pad2 and pad4 leaves and, surprisingly, also in pad3, whereas PDF1.2 induction in pad2, pad3, and pad4 was even stronger than in wild-type plants (Figure 4b). No significant difference in PR1 or PDF1.2 mRNA levels was observed between inoculated wild-type and eds4 leaves (Figure 4b). As a positive control for SID2-mediated PR1 expression, wild-type and sid-2 plants were infected with E. orontii and PR1, PDF1.2, and ICS1 expression levels were determined. As expected, PR1 and ICS1 expression was not detected in the sid2-2 plants, but defensin expression was significantly increased (Figure 4a; Dewdney et al., 2000; Wildermuth et al., 2001). Taken together, the results in this section show that the degree of local susceptibility observed in different genotypes is not directly correlated to the levels of PR1 expression and that activation of SA-dependent defense responses important for local resistance to B. cinerea does not appear to require ICS1. Furthermore, high PDF1.2 expression levels are not sufficient to confer wild-type levels of resistance.

Figure 4.

Defense gene expression in Botrytis-infected mutants.

(a) Leaves of wild-type (WT) and mutant plants were inoculated with B. cinerea and harvested at 0, 1, and 2 days post-infection (dpi), or inoculated with E. orontii and harvested after 7 days.

(b) Leaves of wild-type (WT) and mutant plants were inoculated with B. cinerea and harvested at 0 and 2 dpi. RNA was extracted from infected leaves and RNA blots were hybridized with the indicated probes.

Effect of exogenous salicylic acid on Botrytis cinerea local growth and defense gene induction in Arabidopsis mutants

To further investigate a role for SA in local B. cinerea lesion development, the effect of exogenous SA was examined in wild-type, ein2, and nahG plants, as well as in a variety of mutants impaired in SA-signal transduction. Treatment with SA before B. cinerea inoculation resulted in a significant reduction in the size of lesions in wild-type plants as well as in ein2, eds5, sid2, pad4, and eds4 plants (Figure 5a). PR1 expression in all of these mutants is responsive to exogenous SA or its analogs (Nawrath and Metraux, 1999; Zhou et al., 1998), with the exception of eds4, which is insensitive to low, but not high concentrations of SA (Gupta et al., 2000). In contrast, no significant reduction of lesion development was observed either in nahG or in npr1 leaves treated with SA (Figure 5a). PR1 and PDF1.2 expression was analyzed before inoculation with B. cinerea. As expected, PR1 was strongly induced in SA-treated wild-type plants as well as in SA-treated eds5, sid2, eds4, ein2, and pad4 plants, whereas no detectable expression was observed in SA-treated nahG or npr1 plants (Figure 5b).

Figure 5.

Effects of exogenous salicylic acid (SA) on lesion development and gene expression.

(a) Adult plants were sprayed with a 5-mm solution of SA (black bars) or with a control solution (white bars). After 36 h, leaves were detached and inoculated with Botrytis cinerea and lesion size was measured after 3 days. Asterisks indicate data sets significantly different from the corresponding control data set, according to Student's t-test (P > 0.95). Data represent average ± SE of at least 12 lesions.

(b) RNA from leaves of plants treated with SA (+) or control solution (–) as described in (a) was extracted before fungal inoculation, and RNA blots were hybridized with the indicated probes. The reduced induction, compared to the wild type, of PR1 in ein2 plants treated with SA was not observed in an independent experiment and is not considered significant.

(c) Adult wild-type (WT), pad2-1, and pad3-1 plants were treated as in (a), and lesion size was measured 2 days after inoculation. Asterisks indicate data sets significantly different from the corresponding control data set, according to Student's t-test (P > 0.95). Data represent average ± SE of at least 12 lesions.

(d) RNA from leaves of wild-type (WT) or pad2-1 plants treated with SA (+) or control solution (–) as described in (a) was extracted after 36 h, and RNA blots were hybridized with the indicated probes.

(e) Wild-type (WT) and nahG plants were sprayed with a 5 mm catechol solution (+) or with a control solution (–), and leaves were harvested after 48 h. RNA gel blot was hybridized with the indicated probes.

(f) Leaves of wild-type plants were sprayed with a solution containing 0.01% Silwett L77, 0.01% methanol, and 5 mm catechol, or with a solution containing 0.01% Silwett L77 and 0.01% methanol (control) and inoculated with B. cinerea 48 h after the treatment. Lesion size was measured 3 days after inoculation. Data represent average ± SE of at least 12 lesions.

We also tested whether SA had an effect on development of symptoms in pad2 and pad3 mutants. Lesion size was measured after 2 days because at later timepoints, most mutant leaves showed extensive maceration. SA pre-treatment resulted in significantly enhanced resistance in pad3 and, to a lesser extent, in pad2 plants (Figure 5c). Furthermore, pad2 plants accumulated PR1 transcripts in response to SA (Figure 5d), indicating that this mutation does not affect SA perception or transduction. PDF1.2 transcript levels in both mock- and SA-treated pad2 plants were higher than in wild-type plants (Figure 5d), although not in untreated pad2 plants (Figure 4b), suggesting that the use of surfactant in the control solution or some other unidentified stress may induce some PDF1.2 expression in this mutant.

Unexpectedly, SA-treated nahG plants exhibited a significant increase in PDF1.2 mRNA steady-state levels, whereas no significant induction was detected in the wild type or any other of the mutants analyzed (Figure 5b). Previous reports failed to show PDF1.2 transcript accumulation in nahG plants in response to SA analogs (Clarke et al., 1998). As nahG plants convert SA in catechol, we reasoned that the accumulation of high levels of this compound in response to exogenous SA might be responsible for the induction of defensin expression. Indeed, catechol treatments had no effect on PR1 expression, but resulted in a dramatic induction of PDF1.2 in both wild-type and nahG plants (Figure 5e). Catechol was ineffective in inducing PDF1.2 in ein2 plants, however, showing that an intact ethylene signal-transduction pathway is required for the expression of PDF1.2 in response to catechol (data not shown). Interestingly, wild-type plants treated with catechol showed reduced lesion size, compared to control plants, suggesting that catechol accumulation is not the direct cause of the enhanced susceptibility observed in nahG plants (Figure 5f).

The above results show that SA-dependent defense responses can play a role in local restriction of B. cinerea growth. Defense responses induced by exogenous SA appear to be NPR1-dependent, whereas localized resistance in B. cinerea, in the absence of exogenous SA, appears to be NPR1-independent (Figure 1a). If NPR1 really plays a role in mediating local resistance to B. cinerea, we speculated that its importance might become apparent if other defense responses were blocked. To test this hypothesis, lesions were measured in inoculated ein2, npr1, and jar1 single, double, and triple mutant lines. As expected, infection of wild-type and npr1 plants resulted in similar symptoms, whereas jar1 and ein2 mutants showed increased susceptibility (Figure 6). Lesion size in the jar1 npr1 double mutant was comparable to the lesion size in the jar1 single mutant, but the ein2 npr1 double and the ein2 jar1 npr1 triple mutants were significantly more susceptible than either ein2 or jar1 mutant (Figure 6). Lesion size in the jar1 ein2 double mutant was also greater than in the single ein2 or jar1 mutants (Figure 6). These results demonstrate that NPR1-mediated responses may, in fact, play a role in conferring resistance to B. cinerea infection, but that the effect is relatively small and can only be observed when the defense responses mediated by EIN2 are impaired.

Figure 6.

Effects of ein2, jar1, and npr1 on Botrytis cinerea local symptoms.

Size of lesions formed in leaves of wild-type (WT) and npr1, ein2, and jar1 single, double, and triple mutants 3 days after inoculation with B. cinerea. Data represent average ± SE of at least 14 lesions. Different letters represent data sets significantly different, according to anova analysis followed by Bonferroni's multiple comparison test (P > 0.985); j1n1, jar1 npr1; j1e2, jar1 ein2, e2n1, ein2 npr1; e2j1n1, ein2 jar1 npr1.

Under our experimental condition, most B. cinerea infections of untreated plants resulted in rapidly expanding, water-soaked lesions, typical of soft rot diseases. In a limited number of cases, however, the lesions were dry and did not spread beyond a size of 2.5–3 mm, resembling the localized necrosis observed during incompatible interactions. The number of inoculations resulting in an expanding water-soaked lesion was dramatically reduced in wild-type plants treated with SA (Figure 7a). In contrast, no significant difference in the number of spreading lesions versus necrotic spots was observed in control and SA-treated nahG plants (Figure 7a). Zimmerli and colleagues have previously shown that the size of spreading lesions in Arabidopsis leaves inoculated with B. cinerea correlates to the extent of fungal growth (Zimmerli et al., 2001). As most inoculations in untreated wild-type plants resulted in spreading lesions, the enhanced lesion size observed in untreated mutant plants was likely as a result of a greater fungal growth rate rather than an increased number of successful infections. To confirm this hypothesis, we analyzed the expression of the B. cinerea polygalacturonase gene Bcpg1, which was previously shown to be correlated with active fungal growth in planta (ten Have et al., 2001), in infected leaves of Arabidopsis wild-type and mutant plants. Two days after inoculation, Bcpg1 was detected in all samples, but transcript levels were highest in pad3 and enhanced in nahG, coi1, pad3, and eds4 plants, compared to wild-type and npr1 plants (Figure 7d). This indicates that larger lesions in susceptible genotypes correlate with enhanced fungal growth.

Figure 7.

Effect of salicylic acid (SA) levels on the number of spreading lesions.

(a) Adult wild-type (WT) and nahG plants were sprayed with a 5-mm solution of SA (black bars) or with a control solution (white bars). After 36 h, leaves were detached and inoculated with Botrytis cinerea and the percentage of inoculations resulting in spreading lesions was determined after 3 days. Bars represent average ± SD of three independent experiments (n > 10 for each experiment). Asterisk indicates data set significantly different from the control, according to Student's t-test (P > 0.99).

(b) Diameter of lesions formed in leaves of wild-type (WT) and cpr1-1 plants 3 days after inoculation with B. cinerea. Data represent average ± SE of at least 10 lesions. Asterisks represent data sets significantly different from the wild-type data set, according to Student's t-test (P > 0.99). This experiment was repeated twice with similar results.

(c) Percentage of inoculations resulting in spreading lesions in leaves of wild-type (WT) and cpr1-1 plants 3 days after inoculation with B. cinerea. Bars represent average ± SD of three independent experiments (n > 10 for each experiment). The asterisk represents significant difference from the wild-type data set, according to the Student's t-test (P > 0.99).

(d) RNA was extracted from infected leaves of wild-type (WT), nahG, coi1, npr1, pad3, and eds4 plants inoculated with B. cinerea and harvested at 0 and 2 days post-infection (dpi). RNA blots were hybridized with the indicated probes.

We also analyzed the symptoms caused by B. cinerea in the cpr1 mutant, which constitutively accumulates high levels of SA (Bowling et al., 1994). The average lesion size in cpr1 leaves was significantly reduced, compared to that in the wild type (Figure 7b). As observed in SA-treated wild-type plants, the increased resistance in cpr1 plants appeared to be caused by a dramatic reduction of the number of expanding lesions (Figure 7c).

Effect of PAL inhibition on Botrytis cinerea local growth

Salicylic acid biosynthesis may occur in Arabidopsis through two separate pathways, one requiring PAL activity (Coquoz et al., 1998) and the other one requiring ICS1 (encoded by SID2; Wildermuth et al., 2001). The latter pathway seems not to be involved in defense against B. cinerea, as symptoms in wild-type and sid2 plants were comparable (Figure 1a). It is therefore likely that SA synthesized by the PAL pathway, rather than by ICS1, is important for the induction of defense responses that are effective against B. cinerea at the site of infection. To test this hypothesis, leaves of wild-type and nahG plants were infiltrated with 2-aminoindan-2-phosphonic acid (AIP), a specific inhibitor of PAL activity (Zon and Amrhein, 1992), and subsequently inoculated with B. cinerea. It was previously reported that AIP could enhance lesion development in response to B. cinerea (Govrin and Levine, 2002). If the reason why PAL inhibition enhanced Botrytis lesion development was because of a reduction of SA levels, we would predict that AIP would not further increase susceptibility in nahG plants. However, if other phenylpropanoids synthesized downstream of PAL were responsible for PAL-dependent localized resistance, inhibition of PAL and depletion of SA should have an additive effect on lesion size, and AIP would increase susceptibility in nahG plants. As shown in Figure 8(a), AIP strongly reduced localized resistance to B. cinerea in the wild type, but had little effect in nahG plants, indicating that a reduction of SA levels in response to AIP is likely responsible for the observed increase in lesion size. In contrast, AIP caused enhanced susceptibility in sid2 plants to a degree similar to that observed in wild-type plants (Figure 8b), supporting the hypothesis that SA is synthesized via PAL in the inoculated cells.

Figure 8.

Effects of phenylalanine ammonia lyase (PAL) inhibition on Botrytis cinerea symptom development.

Arabidopsis wild-type (WT) and nahG(a) or sid2-2(b) leaves were infiltrated with 54 µm 2-aminoindan-2-phosphonic acid (AIP; black bars) or a control solution (empty bars), detached and inoculated with B. cinerea. Lesion diameter was measured after 48 h. Asterisks indicate data sets significantly different from the corresponding control data set, according to Student's t-test (P > 0.97). Data represent average ± SE of at least 12 lesions.


The results described in this paper indicate that different signaling pathways mediate the activation of defense mechanisms determining local and systemic resistance to B. cinerea. In particular, responses mediated by SA and by the EDS4, PAD2, and PAD3 gene products appear to be important in limiting fungal growth at the site of infection, but are less important in systemic resistance. In contrast, as previously described (Thomma et al., 1998, 1999a), ethylene and JA signaling is involved in mediating both local and systemic defense responses. The enhanced lesion size in nahG leaves suggests that SA-dependent defense responses reduce the severity of symptoms observed at the site of infection. However, several well-studied mutants in which SA signaling is affected –npr1, eds5, pad4, and sid2– exhibited wild-type levels of resistance to B. cinerea.

The results we obtained with the sid2 mutant were particularly informative. The ICS1 gene, which is mutated in sid2, is required for inducible SA accumulation, SAR induction, and PR1 expression in response to infection by P. syringae and E. orontii, but does not affect SA levels in untreated plants (Wildermuth et al., 2001). As loss of ICS1 function (sid2-2 is a deletion allele and most likely a null mutation; Wildermuth et al., 2001) had no obvious impact on local B. cinerea disease development, it can be concluded that production of SA through the ICS1-dependent biosynthetic pathway is not involved in B. cinerea resistance. In contrast, nahG plants show enhanced disease symptoms, suggesting that SA produced independently of ICS1 may play an important role in determining the rate of fungal growth at the site of infection.

The facts that sid2 mutants are less susceptible to B. cinerea, but that treatment with AIP results in enhanced susceptibility, suggest a role for PAL in the local production of SA at the site of inoculation, where host tissues undergo necrosis. On the other hand, the highly localized induction of PR1 following B. cinerea infection (Figure 3a,b) appears to be dependent on SA synthesized via the ICS pathway as this induction is impaired in the sid2 mutant. ICS activity during B. cinerea infection may be restricted to a relatively small number of cells surrounding the necrotic lesion similar to the localization of PR1 expression. If ICS1 expression is highly localized and not very abundant, this could explain our inability to detect SID2/ICS1 mRNA following B. cinerea infection. Although ICS1 and PR1 are expressed in cells surrounding the lesions, they do not appear to play a significant role in limiting lesion size in response to B. cinerea as evidenced by the wild-type level of local susceptibility of the sid2 mutant and the lack of correlation of PR1 expression with resistance for a number of different mutants. These results are consistent with the distinct roles of the two SA-biosynthetic pathways as postulated by Wildermuth et al. (2001). The observation that inhibition of PAL by AIP greatly enhances susceptibility in wild-type and sid2 plants, but has a small impact on nahG plants, suggests that the effects of AIP treatments can be ascribed predominantly to a reduction in SA levels, rather than other compounds derived from phenylalanine via PAL. However, it cannot be ruled out that the loss of phenolic compounds with a defensive role against B. cinerea may be responsible for the enhanced sensitivity of AIP-treated plants. As previously suggested, the phenolic profile could also be altered in nahG plants, contributing to their enhanced susceptibility to pathogens (Cameron, 2000). A recent report suggests that impaired non-host resistance to P. syringae pv. phaseolicola in nahG Arabidopsis plants is likely because of oxidative stress caused by catechol accumulation (Van Wees and Glazebrook, 2003). On the other hand, exogenous catechol enhanced, rather than decreased resistance of Arabidopsis plants to B. cinerea (Figure 5f), suggesting that catechol accumulation is not responsible for the enhanced susceptibility of nahG plants.

Despite the wide use of PR1 and PDF1.2 as markers for the induction of SA- and ethylene-/JA-dependent defense pathways, respectively, no clear correlation between the levels of expression of either gene and the level of susceptibility to B. cinerea was observed. For instance, npr1, eds5, sid2, and pad4 plants showed disease symptoms similar to the wild type, despite a dramatic reduction of PR1 expression. Similarly, induction of PDF1.2 in SA-treated nahG plants (apparently mediated by catechol) was not sufficient to increase resistance. Finally, eds4 plants displayed enhanced susceptibility, despite the fact that PR1 and PDF1.2 expression was comparable to that observed in the parental plants. It is therefore likely that other defense responses, alone or in combination with PR1 and defensins, are required for wild-type levels of resistance against this pathogen.

One of these defense responses possibly is camalexin, as it is able to limit B. cinerea growth in vitro (Figure 2a), and pad3 mutants, impaired in an enzyme required for camalexin biosynthesis (Zhou et al., 1999), fail to accumulate detectable levels of camalexin in response to B. cinerea (Figure 2b) and display greatly enhanced disease symptoms (Figure 1a). Relatively low levels of camalexin production may also be the cause for the enhanced susceptibility observed in pad2. However, it is known that this mutant is impaired in other defense responses in addition to camalexin biosynthesis. For example, pad2, but not pad3, plants are more susceptible to P. syringae (Glazebrook et al., 1997) and Phytophthora porri (Roetschi et al., 2001). SA accumulation and PR1 expression in response to P. porri are affected in pad2 plants (Roetschi et al., 2001). We observed that PR1 expression was also reduced in Botrytis-infected pad2 leaves (Figure 4b). However, reduced SA levels cannot be the only cause of the enhanced pathogen susceptibility of pad2, because resistance to either P. porri (Roetschi et al., 2001) or B. cinerea (Figure 1a) was significantly lower in pad2 plants than in nahG or npr1 plants. It does not appear that PAD2 is required for the response to exogenously supplied SA. PR1 induction by exogenous SA in pad2 plants was even greater than in wild-type plants (Figure 5d), and a significant increase in B. cinerea resistance was observed in SA-treated pad2 plants (Figure 5a). On the other hand, PR1 expression was also reduced in inoculated pad3 leaves. As the pad3 mutant is almost definitely not affected in SA biosynthesis or sensitivity, it is possible that very rapid tissue maceration observed in pad2 and pad3 leaves may prevent normal PR1 induction, as previously suggested for the esa1 mutant, which is specifically affected in defense responses required for resistance against necrotrophic fungi (Tierens et al., 2002).

The results obtained with the eds4 mutant are also difficult to explain. The eds4 mutant accumulates reduced levels of SA during bacterial infection and also displays reduced sensitivity to exogenous SA treatments (Gupta et al., 2000). However, Ton and colleagues recently showed that eds4 is also resistant to 1-aminocyclopropane-1-carboxylate (ACC)-induced growth inhibition and fails to accumulate PDF1.2 transcript in response to ACC or methyl jasmonate (MeJA), suggesting that EDS4 is required for responses mediated by ethylene (Ton et al., 2002). Under our experimental conditions, PR1 and PDF1.2 expression in eds4 and wild-type-inoculated plants were comparable. Furthermore, exogenous SA pre-treatments reduced lesion size in eds4 plants. These data suggest that SA and ethylene are not involved in EDS4-mediated defense responses against B. cinerea. However, because reduced SA sensitivity in eds4 is observed under specific growth conditions using low SA concentrations (Gupta et al., 2000), it is still possible that some SA-dependent defense responses are also impaired in this mutant. A defect in camalexin is unlikely to be responsible for the enhanced susceptibility of eds4, as camalexin accumulation after P. syringae infection of eds4 is undistinguishable from that in wild-type plants (Gupta et al., 2000). However, it will be necessary to measure camalexin levels in B. cinerea-infected eds4 plants in order to definitely rule out an effect of EDS4 on phytoalexin production.

It is interesting that treatments with exogenous SA, which alone cannot induce camalexin accumulation in Arabidopsis (Thomma et al., 1999b), can increase resistance to B. cinerea through an apparently different mechanism than that involved in establishing the localized resistance observed in untreated plants. Exogenous SA-induced resistance required NPR1 and correlated with PR1 expression; furthermore, exogenous SA had a major impact on the early stages of infection, as the number of inoculated sites displaying soft-rot symptoms was significantly reduced. In contrast, no significant difference in the number of expanding lesions could be observed between wild-type and nahG plants, despite the fact that the rate of lesion development appeared enhanced in the latter. Similarly, infection of cpr1, which constitutively accumulates high levels of SA, resulted in a reduced number of spreading lesions. This phenomenon is unlikely because of unspecific toxicity of SA, as treatment of npr1 plants with the same concentration of SA had no obvious impact on local resistance. These results suggest that elevated concentrations of SA obtained either by exogenous treatment or as a consequence of the cpr1 mutation activate some defense responses, including PR1 expression, which are not normally induced by B. cinerea, or are induced to a lesser extent, under normal conditions.

The fact that B. cinerea lesion development is essentially the same on wild-type and npr1 mutant plants indicates that defense responses mediated by NPR1 have a secondary role during B. cinerea. On the other hand, the fact that the double ein2 npr1 mutant was more susceptible than the single ein2 mutant suggests that NPR1 does play a role in resistance to B. cinerea, but that it only becomes evident when the EIN2-dependent responses are not activated. Similarly, mutations in EIN2 and JAR1 appear to have an additive effect on lesion size, indicating that EIN2 and JAR1 may be separately required for some defense responses. In contrast, mutations in NPR1 and JAR1 do not have an additive effect on symptoms development, suggesting that the products of both genes act in the same pathway.

To summarize, our results suggest that the ability of Arabidopsis to counteract B. cinerea growth at the site of inoculation is determined by defense mechanisms requiring not only ethylene and JA, but also SA. Localized resistance at the site of infection does not require biosynthesis of SA through the ICS1-mediated pathway and does not correlate with PR1 or defensin expression. Furthermore, local resistance is compromised by mutations in the EDS4, PAD2, and PAD3 genes, and is at least partially dependent on camalexin accumulation. In addition to this localized endogenous resistance, local disease symptoms can be further reduced by application of exogenous SA through an NPR1-dependent mechanism. NPR1 also appears to be involved in the activation of defense responses mediated by JAR1, but not by EIN2, although NPR1-dependent responses most likely have a minor role in determining the local susceptibility to B. cinerea. In conclusion, the B. cinerea–A. thaliana interaction is a suitable system to study both SA-dependent and SA-independent defense responses and should provide further insights into the mechanisms used by plants to defend themselves against necrotrophic pathogens.

Experimental procedures

Plant lines and growth conditions

Arabidopsis thaliana accession Columbia (Col-0) was obtained from G. Redei (Arabidopsis Information Service, Frankfurt, Germany). Generation of the Col-0 nahG line is described by Reuber et al. (1998). Seeds of ein2-1 (Guzman and Ecker, 1990) and jar1-1 (Staswick et al., 1992) were obtained from the Arabidopsis Biological Resource Center (Columbus, OH, USA). Isolation of the npr1-1 (Cao et al., 1994), pad2-1, pad3-1, pad4-1 (Glazebrook and Ausubel, 1994; Glazebrook et al., 1997), eds4-1, eds5-1 (Glazebrook et al., 1996; Rogers and Ausubel, 1997), sid2-2 (eds16-1; Dewdney et al., 2000), and ein2-1 npr1-1, ein2-1 jar1-1, npr1-1 jar1-1, and ein2-1 jar1-1 npr1-1 (Clarke et al., 2000) mutants has been previously described.

Plants were grown in a greenhouse, in Metro-Mix 200 (Scotts-Sierra Horticultural Products Co., Marysville, OH, USA), at 19 ± 2°C under a 12-h light/dark cycle.

Fungal growth and plant inoculation

Botrytis cinerea was isolated from Brassica oleracea (J. Plotnikova, unpublished results) and grown on 20 g l−1 malt extract, 10 g l−1 proteose peptone n.3 (Difco, Detroit, USA), and 15 g l−1 agar for 7–10 days at +24°C with a 12-h photoperiod before collection of spores. Rosette leaves from 4-week-old soil-grown Arabidopsis plants were placed in Petri dishes containing 0.8% agar, with the petiole embedded in the medium. Inoculation was performed by placing 5 µl of a suspension of 5 × 105 conidiospores ml−1 in 24 g l−1 potato dextrose broth (PDB; Difco, Detroit, USA) on each side of the middle vein. The plates were incubated at +22°C with a 12-h photoperiod. High humidity was maintained by covering the plates with a clear plastic lid. Under these experimental conditions, most inoculations resulted in rapidly expanding water-soaked lesions of comparable diameter. Lesion size was determined by measuring the diameter or, in case of oval lesions, the major axis of the necrotic area. For whole-plant infections, three to four fully expanded leaves per plant were inoculated, and the plants were covered with a clear plastic lid and incubated as above. Inoculation of leaves attached to the plants resulted in a less reproducible rate of lesion development, but, after 9 days, all inoculated leaves showed complete maceration.

Plant treatments

Salicylic acid and catechol treatments were performed by spraying adult plants with a solution containing 5 mm SA or 5 mm catechol and 0.01% Silwet L-77 (OSi Specialties, Inc., Sistersville, WV, USA), 0.01% MeOH. Plants were incubated for 48 h at +22°C with a 12-h photoperiod. Leaves were then harvested for RNA analysis or fungal inoculation. A 0.01% Silwet L-77, 0.01% MeOH solution in distilled water was used as a control.

2-Aminoindan-2-phosphonic acid was a kind gift from N. Amhrein (Swiss Federal Institute of Technology, Zurich, Switzerland). Rosette leaves were infiltrated with a solution containing 54 µm AIP in 10 mm sodium acetate. After 1 h, leaves were harvested for fungal inoculation. Control leaves were infiltrated with the buffer alone.

RNA analysis

RNA was prepared using the Trizol reagent (Life Technologies, Inc., Gaithersburg, MD, USA), following the manufacturer's instructions. Five to 10 µg of total RNA was separated on a formaldheyde–agarose gel, blotted to GeneScreen membrane (New England Nuclear, Boston, MA, USA), and hybridized overnight at 42°C with different 32P-labeled probes as previously described by Reuber and Ausubel (1996). Blots were washed twice with 1% SDS, 2× SSC at 65°C for 45 min. Images were taken with a Phosphorimager (Molecular Dynamics, Sunnyvale, CA, USA). Single-stranded radioactive probes were prepared from a linear double-stranded DNA template by polymerase chain reaction. Specific PR1 and UBQ5 probes were prepared as previously described by Rogers and Ausubel (1997). The PDF1.2 probe template was prepared based on Penninckx et al. (1996); the ICS1 probe was prepared as in Wildermuth et al. (2001). The Bcpg1 template and primers (5′-ATCGGGAGTCTATGGTTCAACTTCTCTCAATG-3′ and 5′-ATCGGGTCGACTTAACACTTGACACCAGATGG-3′) were a kind gift from Francesca Sicilia.

Determination of camalexin content and antifungal activity

Three or four fully expanded leaves of 4-week-old plants were inoculated with a B. cinerea spore suspension as described above. The inoculated leaves (0.3 g FW) were harvested at 0 and 2 days after infection, and camalexin was extracted and analyzed by HPLC as described by Dewdney et al. (2000; retention time for camalexin was approximately 32.3 min).

Purified camalexin was kindly provided by W.A. Ayer (University of Alberta, Edmonton, Canada) and dissolved in dimethylsulfoxide (DMSO) before use. To determine the effect of camalexin on B. cinerea radial growth on solid medium, round plugs with a diameter of 1 cm were collected from the mycelium of a 1-week-old culture on full strength potato dextrose agar (Difco, Detroit, USA) and transferred to plates supplemented with camalexin or to control plates containing 0.1% DMSO. Colony diameter was measured after 48 h of incubation at +25°C in light.

Upon request, all novel materials described in this article will be available on request for non-commercial purposes.


We are grateful to the Arabidopsis Biological Resource Center and Drs X. Dong and J.G. Turner for providing seeds, Dr W.A. Ayer for purified camalexin, and Dr N. Amhrein for AIP. We also thank Dr M.C. Wildermuth for help with HPLC analysis, for helpful discussions, and for critical reading of the manuscript. S.F. is the recipient of an Institute Pasteur-Fondazione Cenci Bolognetti research fellowship. This research was funded by NIH Grant GM48707 awarded to F.M.A. and by the Giovanni Armenise Harvard Foundation.