Three unique mutants of Arabidopsis identify eds loci required for limiting growth of a biotrophic fungal pathogen


  • Julia Dewdney,

    1. Department of Molecular Biology, Wellman 10, Massachusetts General Hospital, Boston, MA 02114, USA,
    2. Department of Genetics, Harvard Medical School, Boston, MA, USA
    Search for more papers by this author
  • T. Lynne Reuber,

    1. Department of Molecular Biology, Wellman 10, Massachusetts General Hospital, Boston, MA 02114, USA,
    2. Department of Genetics, Harvard Medical School, Boston, MA, USA
    Search for more papers by this author
    • Present address: Mendel Biotechnology, 21375 Cabot Blvd, Hayward, CA 94545, USA.

  • Mary C. Wildermuth,

    1. Department of Molecular Biology, Wellman 10, Massachusetts General Hospital, Boston, MA 02114, USA,
    2. Department of Genetics, Harvard Medical School, Boston, MA, USA
    Search for more papers by this author
  • Alessandra Devoto,

    1. The Sainsbury Laboratory, Norwich Research Park, Colney, Norwich NR4 7UH, UK
    Search for more papers by this author
    • Present address: School of Biological Sciences, University of East Anglia, Norwich NR4 7TS, UK.

  • Jianping Cui,

    1. Department of Organismic and Evolutionary Biology, Harvard University, Cambridge, MA, USA, and
    Search for more papers by this author
  • Lisa M. Stutius,

    1. Harvard College, Cambridge, MA, USA
    Search for more papers by this author
  • Emma P. Drummond,

    1. Department of Molecular Biology, Wellman 10, Massachusetts General Hospital, Boston, MA 02114, USA,
    Search for more papers by this author
  • Frederick M. Ausubel

    Corresponding author
    1. Department of Molecular Biology, Wellman 10, Massachusetts General Hospital, Boston, MA 02114, USA,
    2. Department of Genetics, Harvard Medical School, Boston, MA, USA
      *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


To identify components of the defense response that limit growth of a biotrophic fungal pathogen, we isolated Arabidopsis mutants with enhanced disease susceptibility to Erysiphe orontii. Our initial characterization focused on three mutants, eds14, eds15, and eds16. None of these is considerably more susceptible to a virulent strain of the bacterial pathogen Pseudomonas syringae pv. maculicola (Psm). All three mutants develop a hypersensitive response when infiltrated with Psm expressing the avirulence gene avrRpt2, which activates resistance via the LZ-NBS/LRR resistance protein encoded by RPS2. The growth of Psm(avrRpt2), while somewhat greater in the mutants than in the wild type, is less than growth of the isogenic virulent strain. These results indicate that resistance mediated via LZ-NBS/LRR R genes is functional. Analysis of the growth of avirulent Peronospora parasitica strains showed that the resistance pathway utilized by TIR-NBS/LRR R genes is also operative in all three mutants. Surprisingly, only eds14 and eds16 were more susceptible to Erysiphe cichoracearum. Analysis of the expression profiles of PR-1, BGL2, PR-5 and PDF1.2 in eds14, eds15, and eds16 revealed differences from the wild type for all the lines. In contrast, these mutants were not significantly different from wild type in the deposition of callose at sites of E. orontii penetration. All three mutants have reduced levels of salicylic acid after infection. eds16 was mapped to the lower arm of chromosome I and found by complementation tests to be allelic to the salicylic acid-deficient mutant sid2.


When plants are attacked by pathogens they mount a battery of defenses, including senescence of infected tissues, reinforcement of cell walls by crosslinking and synthesis of new cell wall components, and production of antimicrobial compounds such as phytoalexins, defensins and enzymes that degrade pathogen cell walls (reviewed in Gan and Amasino, 1997; Glazebrook, 1999; Somssich and Hahlbrock, 1998). If the plant and pathogen express resistance (R) and avirulence (avr) genes, respectively, that interact to trigger a hypersensitive response, the plant is able to successfully resist the pathogen and the relationship is termed incompatible. In contrast, in a compatible interaction the plant fails to resist the pathogen attack and disease ensues. Although many of the same defenses that are induced during an incompatible interaction are also activated during a compatible interaction, the induction is generally slower and/or less extensive (reviewed in Crute et al., 1994 ; Draper, 1997; Van Camp et al., 1998 ; Yang et al., 1997 ), and the plant's defense response is insufficient to prevent colonization by the pathogen. However, in many compatible interactions the plant is nonetheless able to limit pathogen propagation, as demonstrated by the isolation of Arabidopsis mutants that are defective in defense-related processes and consequently allow enhanced growth of virulent pathogen(s) ( Glazebrook et al., 1996 ; Rogers and Ausubel, 1997; Volko et al., 1998 ).

While extensive work on resistance mediated by R-avr gene interactions has revealed that common R gene structures and signaling pathways are involved in resistance to multiple types of pathogen (fungal, bacterial, viral and nematode; reviewed in De Wit, 1997; Gebhardt, 1997), it has also become increasingly apparent that there is more specificity in the responses to different pathogens than previously expected. For example, camalexin, the phytoalexin produced by Arabidopsis, is synthesized following inoculation with virulent or avirulent strains of Pseudomonas syringae pv. maculicola (Psm) ( Glazebrook and Ausubel, 1994), but not by Xanthomonas campestris pv. campestris ( Tsuji et al., 1992 ). Similarly, Arabidopsis PDF1.2, which encodes a defensin, is induced after inoculation with virulent strains of the necrotrophic fungal pathogens Alternaria brassicicola and Fusarium oxysporum ( Epple et al., 1997 ), but not by infection with the virulent biotrophic fungal pathogen Erysiphe orontii ( Reuber et al., 1998 ).

In addition to there being specificity in the defenses activated against distinct pathogens, there is also specificity in the effectiveness of particular defense responses. Although camalexin is synthesized upon infection of wild-type plants by the pathogens P. syringae and A. brassicicola ( Glazebrook and Ausubel, 1994; Thomma et al., 1999b ), it appears to play a significant role in limiting pathogen growth only in the A. brassicicola interaction ( Glazebrook and Ausubel, 1994; Thomma et al., 1999b ). Similarly, among 13 enhanced disease susceptibility (eds) mutants that are more susceptible to the bacterial pathogen P. syringae, only four are also more susceptible to the fungal pathogen E. orontii ( Reuber et al., 1998 ; Volko et al., 1998 ). Therefore it is necessary to analyze host interactions with multiple pathogens in order to gain a full understanding of the complexity of the defense response and the roles of each of the defense-response components.

Powdery mildews infect numerous plant species and cause extensive crop loss ( Agrios, 1997). As obligate biotrophic fungal pathogens, they require specific host signals for development. Consequently, they are useful organisms for studying both the host factors that facilitate disease development and host defense responses that limit disease. Three species of powdery mildew that infect Arabidopsis have been identified: Erysiphe cichoracearum ( Adam and Somerville, 1996), E. cruciferarum ( Koch and Slusarenko, 1990), and E. orontii ( Plotnikova et al., 1998 ).

Both compatible and incompatible interactions between Erysiphe species and Arabidopsis ecotypes have been characterized ( Adam and Somerville, 1996; Plotnikova et al., 1998 ; Reuber et al., 1998 ; Xiao et al., 1997 ). A number of ecotypes exhibit R-avr gene-mediated resistance, although in some instances the resistance conferred by a single R gene is weak, and the synergistic action of multiple R genes is required for an effective resistance response ( Adam and Somerville, 1996; Xiao et al., 1997 ; J. Dewdney and F.M. Ausubel, unpublished results). In addition, resistant mutants have been isolated from susceptible parental lines. One of these, edr1, de-represses defense responses to give a phenotype similar to that of late-acting resistance in cereals ( Frye and Innes, 1998). Two other classes of resistant mutants, consisting of pmr1 and pmr4, and pmr3, respectively, also show enhanced activation of some defense responses ( Vogel and Somerville, 2000). An additional resistant mutant, pmr2, does not appear to be caused by heightened defenses, and has been suggested to be deficient in a susceptibility factor required by the fungus for development ( Vogel and Somerville, 2000). In addition to these mutants of Arabidopsis, an important class of mutant resistant to powdery mildew has been identified in barley. mlo mutants confer broad-spectrum resistance against multiple races of E. graminis through de-repression of cell death and defense-response pathways that act at an early stage of fungal infection ( Büschges et al., 1997 ).

Our laboratory studies E. orontiiArabidopsis interactions. On mature Arabidopsis leaves, asexual conidia of E. orontii germinate within 1–2 h, appressoria begin to form by 5 h, and by 24 h development of haustoria is initiated ( Plotnikova et al., 1998 ). When infected by E. orontii, Arabidopsis expresses the pathogenesis related genes PR-1, PR-2 (BGL2), and PR-5 ( Reuber et al., 1998 ). It has previously been shown that the induction of these PR genes occurs at least partially via an SA (salicylic acid)-dependent pathway (reviewed in Yang et al., 1997 ). The importance of this pathway in limiting E. orontii growth was demonstrated by an analysis of mutant lines that exhibit increased susceptibility to the bacterial pathogen P. syringae ( Reuber et al., 1998 ). Mutants compromised in SA accumulation: pad4 ( Glazebrook et al., 1996 ; Jirage et al., 1999 ; Zhou et al., 1998 ) and eds5 ( Nawrath and Métraux, 1999; Rogers and Ausubel, 1997; Volko et al., 1998 ), and the SA-deficient transgenic line expressing the bacterial nahG gene ( Gaffney et al., 1993 ), are more susceptible to E. orontii, as is the SA-unresponsive mutant npr1 ( Cao et al., 1994 ; Delaney et al., 1995 ; Shah et al., 1997 ). Furthermore, analysis of PR gene expression in mutant and transgenic lines suggested that all of the PR-1 mRNA accumulation that is elicited by E. orontii infection occurs via SA-dependent pathway(s), whereas both SA-dependent and SA-independent pathways contribute to BGL2 and PR-5 expression ( Reuber et al., 1998 ). An additional signal transduction pathway that is instrumental in defense against some pathogens requires the signaling molecules jasmonic acid (JA) and ethylene ( Staswick et al., 1998 ; Thomma et al., 1999a ; reviewed in Chang and Shockey, 1999; Creelman and Mullet, 1997; Wasternack and Parthier, 1997), and leads to the production of the antimicrobial proteins defensin and thionin ( Epple et al., 1995 ; Penninckx et al., 1996 ). In contrast to SA-inducible PR genes, the defensin gene PDF1.2 and the thionin gene THI2.1 are not expressed in E. orontii-infected Arabidopsis ( Reuber et al., 1998 ).

To enhance our understanding of the mechanisms by which plants restrict growth of a biotrophic fungal pathogen, our laboratory has isolated a collection of Arabidopsis mutants that are specifically attenuated in defense responses. In this paper, we describe a set of mutants that were identified on the basis that they are more susceptible to infection by E. orontii. Three of these eds mutants correspond to defense-response genes that had not been identified previously in our laboratory, and have been characterized with regard to susceptibility to multiple pathogens and induction of defense responses that are mediated by different signaling pathways. These three mutants further illustrate the complexity and specificity of plant defense-response pathways.


Isolation of mutants with enhanced susceptibility to E. orontii

Approximately 13 000 4-week-old M2 plants from ecotype Columbia seed that had been mutagenized by ethyl-methane sulfonate (EMS) treatment or fast neutron bombardment were inoculated with E. orontii isolate MGH. The amount of fungal growth was visually assessed 12–14 days after inoculation. After one round of screening, 376 (2.8%) plants were selected for retesting. Fifty-two lines were confirmed to have an eds phenotype. Using a scoring system that ranks disease symptoms according to the amount of leaf area covered by powdery mildew ( Reuber et al., 1998 ), the mutants were found to vary in the severity of symptoms, from slightly more susceptible than Columbia to considerably more so ( Figure 1). Sixteen mutants with the most reproducible phenotypes were chosen for further characterization.

Figure 1.

Powdery mildew disease symptoms in eds mutants and wild type.

(a) Erysiphe orontii-infected plants photographed at 17 dpi.

(b) Disease scores in side-by-side comparison with wild type. Plants were scored at 10 days after inoculation using a scoring system developed by Reuber et al. (1998) . 0, no visible symptoms of infection; 4, approximately 100% coverage of infected leaves. For each line, mutant and wild-type (Col-0) plants were intermingled in the same cell to minimize inoculum variability. M4 plants were used for the photographs, but similar results were observed in mutant F2 progeny after backcrossing each line to Col-0.

Prioritization of mutants for further analysis

Each of the 16 mutants chosen for further analysis was backcrossed to the parental line, Col-0, and the F1 progeny were scored for enhanced disease susceptibility. For all 16 mutants, the F1 progeny were wild-type in their susceptibility to E. orontii, indicating that all of the corresponding eds mutations were recessive (data not shown).

To ascertain whether the 16 eds mutants represented new alleles of previously identified defense-response mutants which were known to be more susceptible to E. orontii ( Reuber et al., 1998 ; Volko et al., 1998 ), a series of assays was performed ( Table 1). To determine whether any were allelic to pad4, which is deficient in induction of the phytoalexin camalexin in response to P. syringae infection ( Glazebrook et al., 1996 ), the mutants were assayed for camalexin levels in infected tissue. Because E. orontii induces only low levels of camalexin accumulation in wild-type plants ( Reuber et al., 1998 ), plants were inoculated with Psm ES4326 for these assays. Three of the mutants (J1, J2 and R1) produced markedly lower levels of camalexin. Two of the mutants were not assayed for camalexin, as other tests suggested that they were not pad4 alleles ( Table 1). To further test whether any of the 16 mutants were alleles of pad4, all but S5 (see below) were crossed to pad4 and the F1 progeny scored for susceptibility to E. orontii. The three phytoalexin-deficient mutants failed to complement pad4 and were therefore eliminated from further analysis in this study. As two of these pad4 mutants were from the same batch of mutagenized seed, they are likely to be siblings, so we concluded that a total of two new pad4 mutants were isolated in this screen. These are described in a separate publication ( Jirage et al., 1999 ). Another class of mutants with attenuated defenses against some pathogens have lesions in ethylene signaling ( Knoester et al., 1998 ; Thomma et al., 1999a ). In particular, ein2 is more susceptible to E. orontii (T.L. Reuber and F.M. Ausubel, unpublished results). To test for deficiencies in ethylene signaling, the remaining 13 mutants were tested for sensitivity to ACC, an ethylene precursor. One of the 13 mutants, S5, was ethylene insensitive. Because this mutant could be allelic to components of the ethylene signaling pathway which have already been identified, we did not include it in the current study.

Table 1.  Assays for allelism with known defense-related genes
LinePsm induction
of camalexin
SA induction
of PR-1
on ACC
Complementation tests
  1. +, Complementation; –, failure to complement; wt, wild type; nt, not tested.

C2 (eds14) wtwtwt++ntnt+
Q1 (eds15) ∼wtwtwt++nt++
R2 (eds16) ∼wtwtwt+++++

The product of the NPR1 gene has also been shown to have a vital role in plant defenses, as mutations in NPR1 cause plants to become more susceptible to a variety of pathogens, including E. orontii ( Cao et al., 1994 ; Delaney et al., 1995 ; Glazebrook et al., 1996 ; Reuber et al., 1998 ; Volko et al., 1998 ). One of the characteristics of npr1 plants is their inability to induce PR-1 expression after treatment with salicylic acid ( Cao et al., 1994 ; Delaney et al., 1995 ; Shah et al., 1997 ). Mutants other than those identified as pad4 or ACC-insensitive were assayed for PR-1 mRNA accumulation after treatment with 0.5 m m SA, with the exception of one mutant (J5) which failed to grow on SA plates. In all the lines tested, PR-1 levels were comparable to those in the wild type ( Table 1), and significantly greater than in npr1 (data not shown). Therefore we concluded that none of the mutations were in NPR1. In addition to pad4, npr1, and ein2, the mutant eds1 ( Parker et al., 1996 ) is also more susceptible to E. orontii (T.L. Reuber and F.M. Ausubel, unpublished observation). After discontinuing the analysis of the pad4 and ethylene-response mutants, plus one mutant (K16) that did not have a strong phenotype in the backcrossed line, the remaining 11 mutants were crossed to eds1. According to the disease susceptibility phenotype in the F1 progeny, none of the mutants that we tested was an allele of eds1.

We chose three of the mutants to study in detail for this current publication. Mutants C2, Q1, and R2 appeared to be novel based on the assays described above; maintained a moderate to strong phenotype after being backcrossed to Col-0; and segregated in the F2 progeny of the backcross as single recessive mutations ( Table 2). We confirmed that these three mutants were not allelic to each other by pairwise crosses. In all cases, E. orontii susceptibility in the F1 progeny was no greater than in the wild type, demonstrating that these mutants are not allelic to each other (data not shown).

Table 2.  Segregation analysis of the eds phenotype
  1. Segregation was scored in E. orontii infected F2 progeny from the backcross to Col.

C2 (eds14) 88221.720.2 > P > 0.05
Q1 (eds15) 65220.00380.99 > P > 0.95
R2 (eds16) 39120.0610.95 > P > 0.80

Next, we tested C2, Q1, and R2 for allelism with three additional eds mutants, eds5 ( Rogers and Ausubel, 1997), eds10 ( Volko et al., 1998 ), and eds13 ( Volko et al., 1998 ), which are also more susceptible to E. orontii ( Reuber et al., 1998 ; Volko et al., 1998 ). Based on the phenotype of C2 in a bacterial growth assay, we concluded that C2 does not correspond to eds5, eds10, or eds13. Q1 and R2 were each crossed to eds10 and eds13. The F1 progeny were scored for E. orontii growth, and in all cases the F1 susceptibility was wild-type ( Table 1), indicating that Q1 and R2 are not allelic to these eds loci. On the basis of defense-related gene-induction patterns, we reasoned that of the three mutants being analyzed only R2 might be allelic to eds5, as both show a significant deficiency in PR-1 induction. The wild-type phenotype of the F1 from an R2 × eds5 cross indicates that they represent different loci.

Based on these analyses, we concluded that C2, Q1, and R2 represent new loci that play a role in defense of Arabidopsis against some fungal pathogens, and have renamed them eds14 (C2), eds15 (Q1), and eds16 (R2). Mutations in eds15 and eds16 were generated by fast neutron bombardment, and in eds14 by EMS.


To create mapping populations for each of the mutants, eds14, eds15, and eds16 were crossed to the Arabidopsis ecotype Landsberg erecta. However, the eds phenotype did not segregate 1 : 3, as expected, in any of the three crosses. In the eds14 × La-er and eds15 × La-er crosses, fewer than one in four F2s had an eds phenotype, although they segregated 1 : 3 in a backcross to Columbia. Without a correlative molecular marker such as PR gene expression to verify the genotype of the F2s, we felt that they did not constitute a good mapping population. We are currently working on developing an alternative mapping line that will provide a more compatible genetic background for detection of the eds phenotype. In the eds16 mapping population, the F2 progeny segregated approximately 1 : 3 for enhanced susceptibility. However, when these F2s were checked for loss of PR-1 induction, which co-segregated with the eds phenotype in the backcross to Columbia, only approximately 50% were found to express this trait, implying that only 50% were homozygous for eds16. Twenty F2s that were homozygous were scored for a set of 22 cleaved amplified polymorphic sequence (CAPS) markers distributed throughout the genome ( Drenkard et al., 1997 ; Konieczny and Ausubel, 1993). The results indicated that the mutation was located on the lower arm of chromosome 1. To obtain a finer map position, we scored 65 F2s that were homozygous (based on RNA blot analysis to score PR-1 expression) with the CAPS marker PAB5 and the simple sequence length polymorphism ( SSLP) markers nga111 and AthATPase ( Bell and Ecker, 1994). Markers PAB5 and AthATPase showed 1.54 and 1.56% recombination, respectively, with eds16, while no recombination was observed between eds16 and nga111 (data not shown). The mutation in eds16 was also mapped in collaboration with Cho and co-workers in order to test a newly developed microarray-based mapping technique using biallelic simple nucleotide polymorphism (SNP) markers ( Cho et al., 1999 ). Results from the two methods localized eds16 to the same region of chromosome 1.

Backcrossed lines were generated for eds14, eds15, and eds16, and used in all subsequent experiments.

Eds14, eds15, and eds16 show differential susceptibility to other pathogens

To determine whether the deficiencies in eds14, eds15, and eds16 defenses are specific to E. orontii or also compromise defenses against other pathogens, we assayed the growth of both compatible and incompatible pathogens. The bacterial pathogen Psm ES4326 causes disease characterized by water-soaked lesions and chlorosis on multiple Arabidopsis ecotypes, including Col-0 ( Dong et al., 1991 ; Whalen et al., 1991 ). Plants of eds14, eds15 and eds16 (4.5 weeks old) were inoculated with Psm at a dose of 103 cfu cm−2 and the infected tissue was analyzed for bacterial densities at 0, 1, 2, and 3 days post-inoculation (dpi) ( Figure 2a). All three mutants are slightly but reproducibly more susceptible to Psm proliferation than wild-type plants. However, the difference was statistically significant only in eds16, and then not reproducibly in all experiments. In the four F3 families that were scored, the enhanced Psm growth phenotype co-segregated with the E. orontii eds phenotype when eds16 was backcrossed to wild-type plants (data not shown), indicating that the same mutation in eds16 is responsible for enhanced susceptibility to both Psm and E. orontii.

Figure 2.

Growth of virulent and avirulent bacterial pathogens in the eds mutants.

(a,b) Leaves were inoculated with Psm at a dose of 103 cfu cm−2 (virulent) or 104 cfu cm−2 (avirulent), and leaf disks harvested from five or six plants per line for growth assays at each of the times indicated. Col, ○; eds14, □; eds15, ⋄; eds16, ×. Experiments were performed twice with similar results.

(c) Hypersensitive response in wild-type and mutant plants inoculated with 105 cfu cm−2Psm ES4326/avrRpt2. Leaves were photographed at 20 hpi.

An isogenic incompatible strain of Psm expressing the avr gene avrRpt2 was also tested for growth in the mutants to determine whether any were compromised in R-gene mediated resistance. AvrRpt2 interacts with RPS2, a resistance protein of the LZ-NBS/LRR class present in wild-type Columbia. eds14, eds15, and eds16 were inoculated with a dose of 104 cfu cm−2Psm/avrRpt2, and analyzed for bacterial growth at 3 dpi ( Figure 2b). Although growth of this avirulent pathogen is not limited as much in the mutants (growth of 2–2.5 logs) as in wild-type (growth of about 1 log), it is still restricted relative to the growth of the virulent strain (4.5–5 logs; Figure 2a), indicating that R-avr-mediated resistance via the NDR1 pathway ( Aarts et al., 1998 ; Century et al., 1995 ; Century et al., 1997 ) is functional in all three mutant lines. Using the same incompatible bacterial pathogen, all three lines were also tested for their ability to mount a hypersensitive response (HR). As shown in Figure 2(c), all of the mutants are capable of responding to avirulent Psm with an HR.

An alternative signaling pathway for R-avr gene-mediated resistance is utilized by R genes in the TIR-NBS/LRR class ( Aarts et al., 1998 ). One of the components of this pathway is EDS1 ( Aarts et al., 1998 ). Mutations in EDS1 cause increased susceptibility to virulent E. orontii ( Reuber et al., 1998 ) as well as to some avirulent pathogens ( Parker et al., 1996 ). Therefore it was of interest to know if eds14, eds15, or eds16 were attenuated in resistance mediated by the EDS1 pathway. Growth of two different Peronospora parasitica isolates with avirulence genes that trigger resistance via TIR-NBS/LRR R genes in Columbia was assayed on the three mutants. In all cases, growth on the mutants was the same as on wild-type Columbia (E. Holub, personal communication).

Erysiphe cichoracearum is closely related to E. orontii and, like E. orontii, is a biotrophic fungal pathogen that infects Arabidopsis. To assess the specificity of the effects of mutations in eds14, eds15, and eds16, they were tested for sensitivity to E. cichoracearum. eds14 and eds16 were both more susceptible than wild-type plants. Surprisingly, however, eds15 was no more susceptible than the wild-type control.

Defense responses are altered in eds mutants

Induction of pathogenesis-related genes

The defense-related genes PR1, BGL2, PR5, and GST1 are induced by E. orontii infection in wild-type Columbia, whereas PDF1.2 and Thi2.1 are not ( Reuber et al., 1998 ). Expression of PR1, BGL2, and PR5 in eds14, eds15, and eds16 was monitored by Northern blot analysis. No significant differences were seen between the mutant lines and Columbia in uninoculated samples. However, as shown in Figure 3(a–c), all three mutants showed alterations in PR gene expression in infected leaves. In eds14, PR-1 was reproducibly hyperinduced ( Figure 3a), whereas in eds16 PR-1 induction was dramatically reduced ( Figure 3a). BGL2 and PR-5 expression were also reduced in eds16 ( Figure 3b,c), by approximately 61 and 57%, respectively. In eds15, the induction of all three PR genes was low, ranging between 46 and 65% of wild-type levels ( Figures 3a–c).

Figure 3.

Northern blot analysis of PR-1, BGL2, PR-5 and PDF1.2 mRNA accumulation.

Plants were inoculated 4.5 weeks after planting and harvested at 7 dpi. Results from two samples of uninfected leaves and three samples of infected leaves were averaged. Each sample represents leaves from three or four plants. Expression was quantified by phosphorimager and the ratio of defense-related gene expression to UBQ5 expression calculated. That ratio is expressed as a percentage of the ratio in the wild type. The experiment was performed once with backcrossed lines, and the results were consistent with those in M4 plants. (a) PR-1; (b) BGL2; (c) PR-5; (d) PDF1.2.

PDF1.2 mRNA accumulation was also assayed. Although not expressed at detectable levels in E. orontii-infected leaves of wild-type plants, PDF1.2 was induced to relatively high levels in infected eds16, but not eds14 or eds15 leaves ( Figure 3d). In an independent experiment, expression of PDF1.2 was as high in E. orontii-infected eds16 as in Columbia tissue infected with Botrytis cinerea, a pathogen which is known to induce this defense-response gene (data not shown). The significance of this finding is that it suggests E. orontii is capable of activating the JA/ethylene defense-response pathway in addition to the salicylic acid pathway, which was not apparent in previous studies with wild-type Arabidopsis.

Accumulation of salicylic acid

The importance of salicylic acid (SA) as a signaling molecule in defense responses has been demonstrated in numerous experiments (reviewed in Dong, 1998; Draper, 1997; Glazebrook, 1999; Wobbe and Klessig, 1996). To determine if SA levels were low in any of the mutants, extracts from E. orontii-infected leaves were analyzed by HPLC ( Figure 4). In Col-0, the concentration of total SA at 7 dpi was 5.6 ± 1.0 µg g−1 FW. Total SA in E. orontii-infected leaves of each of the three mutants was lower than in wild-type plants. The deficiency was greatest in eds16, which accumulated only 0.5 µg g−1 FW total SA, less than 10% of wild-type levels. In eds14, SA concentrations were also significantly diminished relative to wild-type, with accumulations of approximately 2.6 µg g−1 FW. Although the deficiency in SA accumulation was not as strong in eds15 as in the other two mutants, it too consistently accrued less than wild-type plants.

Figure 4.

HPLC analysis of salicylic acid in mutants infected with Erysiphe orontii.

Plants were inoculated at 4.5 weeks old and harvested at 7 dpi. Results for three samples, each representing several plants, were averaged for each line. The experiment was done once with M4 plants and once with backcrossed plants, with similar results.

Production of camalexin

Previous analysis indicated that E. orontii infection did not elicit the accumulation of the phytoalexin camalexin ( Reuber et al., 1998 ). However, using a more sensitive HPLC assay, we have found that small amounts of camalexin are synthesized in E. orontii-infected Col-0 leaves. None of the mutants in this study consistently produced less camalexin than wild-type plants, and in fact eds16 reproducibly accumulated more (data not shown).

Callose deposition

Formation of cell wall appositions, or papillae, at sites of attempted penetration has been correlated with resistance to powdery mildew infections ( Bayles et al., 1990 ; Kobayashi et al., 1997 ; Stanghellini et al., 1993 ). eds14, eds15, and eds16 were evaluated for the formation of papillae during E. orontii infection by staining for callose, which is a major component of these cell wall reinforcements. Leaves were assayed at 1, 2, 3, and 5 dpi. No substantial differences between the mutants and wild type were seen. All of the mutant lines accumulated callose in discrete papillae beginning by 24 hours post inoculation (hpi), as in the wild type, and continuing through 5 dpi. The response at 2 dpi is shown in Figure 5. Measurement of cell wall apposition size at 2 dpi indicated that the average size did not vary by more than 8% from the wild type in any of these mutant lines (data not shown).

Figure 5.

Papillae formation in response to Erysiphe orontii.

Plants were inoculated with E. orontii at 4.5 weeks old and leaves harvested 2 days later. Samples were stained with aniline blue to detect callose and then trypan blue to stain fungal structures. Bar, 0.1 mm; c, cell wall apposition; h, fungal hypha.

The phenotypes of eds14, eds15, and eds16 are summarized in Table 3.

Table 3.  Summary of phenotypic analysis of eds14, eds15 and eds16
  • wt, Wild type.

  • a

    Susceptibility relative to Col-0, with +++ being the most susceptible.

  • b

    Log difference in growth relative to Col-0.

  • c

    Percentage of levels in wild type.

Susceptibility to virulent pathogens
 E. orontiia+++++
 E. cichoracearuma+wt++
 Psm ES4326 b+0.2+0.2+0.5
Resistance to avirulent pathogens
 LZ-NBS-LRR R gene mediatedyesyesyes
 TIR-NBS-LRR R gene mediatedyesyesyes
Defense responses
 E. orontii induction of pathogenesis-  related genes c
 E. orontii-induced accumulation of   salicylic acid c∼50%∼75%<10%
 Callose depositionyesyesyes

Complementation analysis with sid2

The map position and low salicylic acid levels in infected tissue suggested that eds16 might be allelic to sid2 ( Nawrath and Métraux, 1999). F1 and F2 progeny from a cross between the two mutants all had an eds phenotype when scored for E. orontii susceptibility (data not shown). In addition, total SA levels in E. orontii-infected leaves from F1 progeny were greatly reduced relative to concentrations in infected wild-type leaves (data not shown). Based on the lack of complementation, we concluded that eds16 is allelic to sid2, and accordingly have renamed it sid2-2.


Isolation and characterization of mutants

Multiple factors determine the outcome of attempted pathogen infection of a plant. Defense responses, including expression of PR genes and formation of cell wall appositions, are critical in limiting infection. Biotrophic pathogens may require specific host susceptibility factors for development and colonization. Additionally, other aspects of the host environment, such as nutrient content, may influence the growth of a biotrophic pathogen. Therefore mutants that support enhanced growth of a biotrophic pathogen may be different from the wild type in any of these traits. In a screen for mutants that have enhanced susceptibility to E. orontii, we isolated both new alleles of genes that have previously been shown to function in defense-response pathways (pad4 and sid2), and novel genes (eds14 and eds15) that have a role in defense. Nine mutants that were identified in the screen have not been sufficiently characterized to determine whether they are defense-related, or affect other traits that influence the outcome of an interaction with a biotrophic pathogen. Although the three mutants, eds14, eds15, and sid2/eds16, described in detail in this work are all aberrant in some aspect of defense responses, each has a unique phenotype (summarized in Table 3).

In sid2/eds16, the loss of PR-1 induction indicates that this mutation affects the salicylic acid signal transduction pathway, as all induction of PR-1 during a compatible interaction with E. orontii occurs via an SA-dependent pathway ( Reuber et al., 1998 ). Furthermore, the low SA levels and the ability of sid2/eds16 to respond to exogenous SA suggest that the sid2/eds16 product functions upstream of SA. The level of BGL2 and PR-5 expression in infected sid2/eds16 leaves is similar to the portion of induction of these PR genes indicated by previous studies to be SA-independent ( Reuber et al., 1998 ).

Similarly, eds15 appears to act upstream of SA, as it fails to accumulate wild-type levels of PR-1 mRNA and salicylic acid. The resemblance between eds15 and eds16 suggests that both may function in the same pathway, with eds15 having a less critical role. However, in eds15 the attenuation of BGL2 and PR-5 expression is almost as great as in eds16, while there are large differences between the two mutants in PR-1 induction.

eds14 is unusual in several aspects. First, it is more susceptible despite accumulating PR-1 mRNA to higher levels than in the wild type. One possible explanation for the increase in PR-1 mRNA accumulation is that more cells are infected in eds14 than in wild-type plants. However, we do not favor this interpretation because levels of the signaling compound required for PR-1 induction, salicylic acid, are reduced in eds14 relative to the wild type. This is the second unusual feature of eds14, namely that PR-1 induction appears to be uncoupled both from SA accumulation and from induction of BGL2 and PR-5.

Role of SA in response to E. orontii infection

The isolation of eds14, eds15, and sid2/eds16 reinforces the theory that SA signaling has a critical role in limiting E. orontii growth, as previously suggested by Reuber and co-workers ( Reuber et al. , 1998 ). pad4, eds5, npr1, and NahG plants are all deficient in SA signal transduction ( Cao et al., 1994 ; Gaffney et al., 1993 ; Nawrath and Métraux, 1999; Zhou et al., 1998 ) and are all more susceptible to E. orontii ( Reuber et al., 1998 ). In pad4, eds5, NahG, eds15, and sid2/eds16 the accumulation of SA is attenuated, whereas npr1 fails to activate responses downstream of SA. As a result of SA signaling deficiencies, the production of PR-1 mRNA in these lines is reduced. Several studies have shown that PR-1 has antifungal activity ( Niderman et al., 1995 ; Rauscher et al., 1999 ) and can associate with fungal cell walls ( Cordier et al., 1998 ). On the other hand, the phenotype of eds14 indicates that PR-1, although it may be an important defense against E. orontii, is not the only determinant of susceptibility. However, the importance of SA-mediated defenses is not refuted by this mutant, as SA concentrations in infected eds14 leaves are lower than in infected wild-type plants, in contrast to PR-1 levels being higher.

Role of JA/ethylene pathways in response to E. orontii

In sid2/eds16 but not the wild type, PDF1.2 is induced by E. orontii infection. Therefore there must be a mechanism by which E. orontii activates the JA/ethylene signaling pathway, which was not apparent from previous studies. A second implication is that PDF1.2 is only induced when SA levels are low, suggesting that there is suppression of the JA/ethylene signaling pathway by SA. Similar antagonistic effects have been reported by other groups. PDF1.2 expression is higher in NahG plants than in the wild type ( Penninckx et al., 1996 ), and PDF1.2 mRNA accumulation is higher in the mutant ssi1 in the absence of functional NPR1 ( Shah et al., 1999 ). In addition, the biosynthesis of both jasmonic acid and ethylene are reportedly inhibited by exogenous acetyl salicylate and SA, respectively ( Leslie and Romani, 1988; Peña-Cortés et al., 1993 ). On the other hand, there is also evidence for synergistic interactions between the two pathways. Ethylene may potentiate SA-mediated induction of PR-1 ( Lawton et al., 1994 ; Lawton et al., 1995 ; Xu et al., 1994 ), and there are at least two reports of genes (EREBP1 in tobacco and an ACC oxidase gene in Nicotiana glutinosa) that are inducible by both SA and ethylene ( Horvath et al., 1998 ; Kim et al., 1998 ).

Intimations of an additional defense-response pathway

Loss of suppression could also be the basis for the elevated levels of PR-1 in eds14. However, BGL2 and PR-5, which are induced via both SA-dependent and SA-independent pathways ( Reuber et al., 1998 ), are not hyperinduced, so it seems unlikely that the eds14 mutation results in inactivation of a pathway or compound that normally inhibits the SA signaling pathway. Alternatively, there might be limitations on PR-1 but not BGL2 or PR-5 expression that are released in eds14. Because the JA/ethylene pathway does not appear to be activated during E. orontii infection unless SA concentrations are abnormally low, it is unlikely that it is the source of this hypothesized inhibition. The existence of an additional defense response pathway has been suggested by the isolation of the pmr3 mutant ( Vogel and Somerville, 2000). In this mutant, which is more resistant to E. cichoracearum and E. orontii, PR-1 expression is suppressed ( Vogel and Somerville, 2000). The lesion in eds14 may lie in this pathway, but may have the opposite effect to the pmr3 mutation – increased susceptibility and elevated PR-1.

Salicylic acid and the hypersensitive response

There are conflicting reports about the requirement for SA in the hypersensitive response ( Glazebrook, 1999; Lam et al., 1999 ; Rate et al., 1999 ; Asai et al., in press ) . Although SA levels are severely reduced in sid2/eds16, it is still capable of producing an HR. In this work we have measured SA only in leaves infected with a virulent pathogen, but data on the sid2-1 mutant indicate that SA levels are also low during an incompatible interaction ( Nawrath and Métraux, 1999). However, in both our experiments and those reported by Nawrath and Métraux, SA was not measured until 2 dpi ( Nawrath and Métraux, 1999) or 7 dpi (our work). In other work, avirulent pathogens have been shown to trigger a rapid and transient rise in SA, beginning as early as 1 hpi, in addition to a later, more sustained increase (reviewed in Draper, 1997). This avr-stimulated early increase in SA may not be attenuated in sid2/eds16. It is logical to think that the same determinants of recognition of a virulent pathogen are also present during interaction with an isogenic avirulent pathogen, and that the response to an avirulent pathogen is a combination of R-avr gene-mediated responses and responses that would be stimulated by the corresponding virulent pathogen. Both may contribute to limitations on pathogen growth. In the sid2/eds16 mutant only the second pathway may be altered, causing increased susceptibility to a virulent pathogen but leaving R-avr gene-mediated resistance intact. According to this model, the phenotype of sid2/eds16 is not inconsistent with SA being required for the HR.

Specificity of mutations on susceptibility to various pathogens

In none of the three mutants, eds14, eds15, and sid2/eds16, are the lesions specific to interactions with a biotrophic fungal pathogen. In sid2/eds16, growth of Psm is also less restricted than in wild-type plants, although the significance of this difference varied among experiments. Similarly, eds14 allows more growth of Psm, although in this mutant the effect on Psm growth is minor, while the effects on Erysiphe species are significant. There is some inconsistency in the data for eds15, indicating that restrictions on growth are slightly relaxed for Psm but not for E. cichoracearum, which is closely related to E. orontii. It may be that the phenotype of eds15 is not strong enough for differences in growth of E. cichoracearum to be apparent, whereas the assay used to quantify Psm growth may be more sensitive to small variations. In summary, the lesions in all three mutants affect defenses against more than one class of pathogen, but susceptibility to different pathogens is enhanced to different extents.

The unique phenotype of each of the mutants described in this paper indicates that each can provide new information about the controls that govern plant defense responses, and about the defenses that are effective in limiting growth of specific pathogens.

Experimental procedures

Growth of A. thaliana

For most experiments, Arabidopsis plants were grown in Metro-Mix 200 (Scotts-Sierra Horticultural Products Co., Marysville, OH, USA) under a 12 h light–dark cycle, either in a greenhouse with supplemental fluorescent lighting (19 ± 2°C) or in a Percival AR-601 growth chamber (20°C, 80% relative humidity, illumination approximately 100 µE m−2 s−1). For assays of E. cichoracearum susceptibility, plants were grown in Levington Multipurpose compost (Ipswich, UK) in growth chambers with an 8 h photoperiod (150 µE m−2 s−1) and a temperature of 22°C. Arabidopsis accession Col-0 was obtained from G. Redei (Arabidopsis Information Service). EMS and fast neutron-mutagenized seed were obtained from Lehle Seeds (Round Rock, TX, USA).

To create backcrossed lines, the mutants were crossed to the parental line Col-0 and the F2 progeny scored for E. orontii susceptibility. F3 families from F2 progeny with an eds phenotype were scored to confirm that the F2s were homozygous for the eds mutation. All experiments were performed at least once in lines from the first backcross, although some experiments were done first with M4 progeny and repeated in the backcrossed lines.

Fungal, oomycete, and bacterial inoculations

Erysiphe orontii inoculations were done as described by Reuber et al. (1998) , except that all inoculations for experiments were done using a settling tower. For scoring susceptibility, a light inoculum (conidia from one infected leaf) was used per 206 cm2 box of plants. A heavier inoculum (conidia from three infected leaves) was used for analysis of defense-related gene expression. Erysiphe cichoracearum was propagated as described by Vogel and Somerville (2000). Arabidopsis plants were inoculated by dusting with conidia from 10–12-day-old E. cichoracearum cultures. Peronospora parasitica inoculations were done as described in Warren et al. (1999) . Assays for growth of Psm ES4326 and Psm ES4326(avrRpt2) were done as described by Volko et al. (1998) .

Camalexin assays

Two methods were used to assess camalexin levels. For determination of allelism with pad4, camalexin was assayed in Psm ES4326-infected leaves by visualization on TLC plates, as described by Glazebrook and Ausubel (1994). In the characterization of mutants eds14, eds15, and sid2/eds16, camalexin was quantified by HPLC as outlined in the procedure for SA analysis.

ACC sensitivity assay

Mutants were tested for ACC sensitivity following the procedure used by Van Der Straeten et al. (1993) . Mutants were compared with wild-type plants germinated on the same plate, and with the mutants ein2 and etr1.

Salicylic acid induction of PR-1

Seeds were sterilized and plated on 0.5 × MS (0.5 × MS salts, Gibco 500-1117; 1 × Gamborg's B-5 vitamins, Sigma G1019; 2% sucrose; 1 g l−1 2-[N-morpholino]ethane-sulfonic acid; 0.8% Phytagar, Gibco-BRL 10675-031). At 9 days after sowing, seedlings were transferred to 0.5 × MS plates containing 0.5 m m SA. 7 days after transfer seedlings were harvested for RNA extraction.


DNA extractions and detection of CAPS markers were done according to the method described by Drenkard et al. (1997) . Primers and amplification conditions for the SSLP markers are detailed by Bell and Ecker (1994).

RNA analysis

Plants 4.5 weeks old were inoculated via a settling tower and harvested at 7 dpi. RNA was prepared and RNA gel blots were performed as described by Reuber and Ausubel (1996). Probes were prepared as described by Rogers and Ausubel (1997). Expression was quantified by phosphorimager, and is stated as a ratio to UBQ5. mRNA accumulation in the mutant lines is shown as a percentage of wild-type levels.

Salicylic acid analysis

Salicylic acid was extracted and analyzed by HPLC using a modification of the methods described by Meuwly and Métraux (1993). For SA analysis, 0.3–0.5 g FW leaf tissue from E. orontii-infected leaves (7 dpi) was frozen in glass tubes with liquid nitrogen, and either used directly or stored at −80°C. The frozen tissue was ground in liquid nitrogen to a fine powder using a chilled glass rod. 3 ml of 90% methanol and 250 ng o-anisic acid (internal standard) were added to each sample. Samples were vortexed, sonicated for 20 min, and centrifuged for 20 min at 1700 g in a table-top centrifuge. The supernatant was transferred to a new tube, and the pellet re-extracted with 2 ml 90% methanol. The two supernatants were combined, divided into two portions of equal volumes (for total SA and free SA measurements), vacuum dried, and frozen at −80°C. For total SA, 500 µl β-glucosidase (80 U ml−1 in 100 m m sodium acetate pH 5.2; Sigma, St Louis, MO, USA) was added to each sample. The samples were sonicated for 5 min, vortexed, covered with foil, and incubated for 90 min at 37°C. For both total and free SA samples, 2.5 ml 5% trichloroacetic acid was added, and the samples vortexed, sonicated for 5 min, and centrifuged at 1700 g for 15 min. The supernatant was extracted twice with 2.5 ml of a 1 : 1 (v/v) mixture of ethyl acetate : cyclopentane. The organic phases were combined, vacuum dried, and frozen at −80°C. Just prior to loading samples on the HPLC, each was resuspended in 250 µl of 20% methanol, vortexed, sonicated for 5 min, and filtered through a 0.22 µm nylon filter.

HPLC separation of o-anisic acid (oANI) and SA was performed on a Waters 600 system equipped with Waters 474 scanning fluorescence detector and 996 photodiode array detector (Waters Corp., Milford, MA, USA). A 5 µm, 15 cm × 4.6 mm ID Supelcosil LC-ABZPlus column (Supelco, Bellefonte, PA, USA) preceded by a LC-ABZPlus guard column was maintained at 27°C and equilibrated in 15% acetonitrile with 25 m m KH2PO4 pH 2.6 at a flow rate of 1.0 ml min−1. 50 µl of each sample was manually injected. The elution program began with an isocratic flow of 15% acetonitrile with 25 m m KH2PO4 pH 2.6 for 1 min, followed by a linear increase to 20% acetonitrile over 5 min, isocratic flow at 20% for 10 min, a linear increase from 20 to 55% acetonitrile over 17.5 min, and to 90% in 5 min. For the second sample set, subsequent to the linear increase to 55% acetonitrile a 1 min 50 : 50 (v/v) acetonitrile : water wash followed by 5 min 100% acetonitrile wash was utilized to reduce possible salt precipitation. The column was then rinsed with water for 1–2 min prior to equilibration in 15% acetonitrile with 25 m m KH2PO4 for 15 min, and injection of the subsequent sample. oANI and SA were quantified using the fluorescence detector programmed to 305 nm excitation/365 emission for oANI and 305/407 emission for SA. Calibration curves for SA and oANI (Sigma) were y = 4.26x + 3.46 (R2 = 1.0) for oANI and y = 3.11x + 6.76 (R2 = 0.99) for SA, where y is in ng and x is in area units × 10−4. oANI eluted at approximately 11.4 min and SA eluted at approximately 25.2 min; the detection limit for both compounds was 4 ng, similar to those reported by Meuwly and Métraux (1993). Using this extraction and HPLC method we were also able to resolve and quantify benzoic acid (PDA detection), trans-cinnamic acid (PDA detection), o-coumaric acid (PDA detection), and camalexin (fluorescence detection).

Callose detection

Leaves were stained and examined microscopically as described by Reuber et al. (1998) . To assay the formation of cell wall appositions over time, leaves were harvested from two plants per line at each time point. Multiple colonies on each leaf were observed. For measurement of cell wall apposition size, one leaf per plant was harvested from seven to nine plants per line at 2 dpi. The diameter of the fluorescent area at the level of focus where the papillae outline was sharpest was measured for 44–70 papillae per line, using the single largest papilla associated with each E. orontii colony.


We thank Eric Holub for assaying resistance of the mutants to avirulent isolates of P. parasitica, Simone Ferrari for RNA from B. cinerea-infected Arabidopsis leaves, and Antony Buchala for sharing his expertise on salicylic acid HPLC analysis. This work was funded in part by NIH grant GM48707.