An EDS1 orthologue is required for N-mediated resistance against tobacco mosaic virus

Authors


* For correspondence (fax +44 1603 450011; e-mail david.baulcombe@bbsrc.ac.uk).

Summary

In Arabidopsis, EDS1 is essential for disease resistance conferred by a structural subset of resistance (R) proteins containing a nucleotide-binding site, leucine-rich-repeats and amino-terminal similarity to animal Toll and Interleukin-1 (so-called TIR-NBS-LRR proteins). EDS1 is not required by NBS-LRR proteins that possess an amino-terminal coiled-coil motif (CC-NBS-LRR proteins). Using virus-induced gene silencing (VIGS) of a Nicotiana benthaminana EDS1 orthologue, we investigated the role of EDS1 in resistance specified by structurally distinct R genes in transgenic N. benthamiana. Resistance against tobacco mosaic virus mediated by tobacco N, a TIR-NBS-LRR protein, was EDS1-dependent. Two other R proteins, Pto (a protein kinase), and Rx (a CC-NBS-LRR protein) recognizing, respectively, a bacterial and viral pathogen did not require EDS1. These data, together with the finding that expression of N. benthamiana and Arabidopsis EDS1 mRNAs are similarly regulated, lead us to conclude that recruitment of EDS1 by TIR-NBS-LRR proteins is evolutionarily conserved between dicotyledenous plant species in resistance against bacterial, oomycete and viral pathogens. We further demonstrate that VIGS is a useful approach to dissect resistance signaling pathways in a genetically intractable plant species.

Introduction

Many examples of disease resistance in plants involve proteins with a predicted nucleotide-binding site (NBS) domain and carboxy-terminal leucine-rich-repeats (LRRs) (vanderBiezen and Jones, 1998). There are two groups of NBS-LRR proteins depending on the amino-terminal sequences. In one group this region (the TIR domain) is similar to the cytoplasmic signaling domains of Drosophila Toll and mammalian Interleukin-1 transmembrane receptors (O'Neill and Greene, 1998; vanderBiezen and Jones, 1998). The other group of NBS-LRR proteins possesses an amino-terminal portion that likely forms a coiled coil domain (CC) (Pan et al., 2000). We refer to these groups as TIR-NBS-LRR and CC-NBS-LRR proteins, respectively.

It is thought that the amino-terminal domains of the NBS-LRR proteins are involved in signaling to disease resistance (Hammond-Kosack and Jones, 1997). This idea is supported by mutational analyses in the model plant Arabidopsis that have identified components of the signaling pathways required for disease resistance. One of these components, NDR1, is required for resistance specified by CC-NBS-LRR proteins encoded by RPM1, RPS2 and RPS5 (Century et al., 1995). In contrast, EDS1 mediates resistance conferred by the TIR-NBS-LRR proteins encoded by RPP1, RPP2, RPP4, RPP5 and RPS4 (Aarts et al., 1998). It was suggested that R proteins of the CC-NBS-LRR class signal through NDR1 while the TIR-NBS-LRR class of R proteins utilize EDS1 (Aarts et al., 1998). However, there are exceptions to this model. For example, the CC-NBS-LRR protein RPP8 has no strong requirement for either NDR1 or EDS1 (Aarts et al., 1998; McDowell et al., 2000).

To investigate the relationship of NBS-LRR proteins and resistance signaling components we have examined the role of a Nicotiana EDS1 orthologue in N-mediated resistance against tobacco mosaic virus (TMV). N is a TIR-NBS-LRR protein of Nicotiana spp. (Whitham et al., 1994) and, based on the Arabidopsis studies, we hypothesized that it would depend on EDS1 function. We further investigated the role of the Nicotiana EDS1 orthologue in resistance signaling pathways that are not conditioned by TIR-NBS-LRR proteins.

To investigate the function of EDS1 homologues in Nicotiana we used gene silencing systems that are activated and given specificity by virus vector constructs. With a virus vector carrying a fragment of a host gene, the virus-induced gene silencing (VIGS) is targeted against the corresponding host mRNA (Baulcombe, 1999). If this gene is required for disease resistance we anticipated that the virus vector-infected plant would become partially or completely susceptible to the pathogen. VIGS has been used to test the function of candidate cellulose synthase genes (Burton et al., 2000) but has not been used previously to investigate disease resistance pathways. We show that Nicotiana EDS1 is essential for N-mediated resistance against TMV but is not required for the disease resistance phenotypes of the CC-NBS-LRR Rx protein. EDS1 is also not required for resistance against Pseudomonas syringae that requires the Pto kinase. Thus, a conserved EDS1 signaling pathway in dicotyledonous plants is specifically associated with TIR-NBS-LRR proteins and provides resistance against viruses, fungi and bacteria.

Results

To identify components of the N signaling pathway using VIGS it was not possible to utilize Nicotiana tabacum. This species is commonly used as the host plant to investigate N, but does not support VIGS efficiently (M. Teresa Ruiz and D. C. Baulcombe, unpublished data). It was therefore necessary to transfer N as a transgene to N. benthamiana, a good host for VIGS. In addition, because PTGS is dependent on a high degree of sequence similarity in the virus and the target RNA (Mueller et al., 1995), it was necessary to identify the Nicotiana orthologue of EDS1.

N-mediated resistance in N. benthamiana

A genomic DNA fragment containing N (see Experimental procedures) was transformed into N. benthamiana. Primary N. benthamiana transformant lines were assessed for resistance against a GFP-tagged strain of TMV (TMV:GFP). This vector derivative was based on the U1 strain of TMV (Donson et al., 1991) with the coat protein open reading frame (ORF) replaced by the corresponding ORF of odontoglossum ringspot tobamovirus. Out of 19 independent lines, 16 were resistant to TMV:GFP. Of these, one line, 310 A, is homozygous for a single copy of N and was used in this study.

To characterize N-mediated resistance in N. benthamiana we inoculated TMV:GFP onto 310 A plants and monitored virus accumulation by GFP fluorescence. On inoculated leaves of 310 A plants the foci of GFP reached a maximum diameter of 0.25 mm but, as shown in Figure 1(a), had disappeared by 6 days post inoculation (dpi). In contrast, on non-transgenic N. benthamiana, the foci of GFP expanded to 3 mm diameter and persisted beyond 6 dpi (Figure 1a). RNA gel blot analysis confirmed that levels of GFP fluorescence were an indicator of TMV:GFP RNA accumulation. On N-transgenic plants, TMV:GFP RNA failed to accumulate beyond 2 dpi and was restricted to the inoculated leaves. In contrast, TMV:GFP RNA accumulated for at least 8 dpi on inoculated leaves of non-transgenic N. benthamiana and was readily detectable in systemic leaves by 10 dpi (Figure 1b). Potato virus X (PVX) and tobacco rattle virus (TRV) caused symptoms to develop as rapidly and extensively on 310 A plants as on non-transgenic N. benthamiana plants (data not shown) indicating that the resistance conferred by the N transgene in line 310 A was specific to TMV:GFP. PVX induced a mosaic whereas TRV-infected plants exhibited moderate stunting.

Figure 1.

N-mediated resistance operates in N. benthamiana.

(a) Leaves from non-transgenic (NT) or N-transgenic 310 A plants (NN) were manually inoculated with TMV:GFP sap. The photograph was taken under UV light at 7 dpi.

(b) RNA gel blot showing accumulation of TMV:GFP RNA over a period of 8 days on inoculated leaves and at 10 dpi in systemic leaves (*) of non-transgenic (NT) or N-transgenic 310 A plants (NN). TMV:GFP inoculation was by agroinfiltration (see methods). Genomic-(gRNA) and subgenomic-(sgRNA) TMV:GFP RNAs are shown. Equal loading in each lane is indicated by ethidium bromide staining of rRNA in the gel.

Similar results were obtained in three independent experiments.

From these data, we concluded that the recognition and resistance signaling capabilities of N were similar in N. benthamiana and the natural genetic backgrounds of N. tabacum and N. glutinosa. However, the resistance was not identical since local necrotic lesions that are the hallmark of N resistance in N. tabacum and N. glutinosa were only occasionally observed on N. benthamiana (data not shown). Normally the resistance to TMV:GFP was associated instead with chlorosis at the point of inoculation. Moreover, the wild type U1 TMV strain moved systemically in the 310 A plants causing an N-dependent necrotic response (data not shown). All of our other N. benthamiana lines carrying the N transgene also exhibited this systemic necrosis after inoculation with TMV U1. We attribute the N-responses to differences in the virulence of TMV:GFP and TMV U1 on N. benthamiana. On non-transgenic plants the U1 strain causes wilting and severe stunting at approximately 8 dpi whereas TMV:GFP causes a mild mosaic that first appears at approximately 15 dpi (data not shown). Thus, from these data, it seems that N-mediated resistance is influenced by species background and is weaker in N. benthamiana than in N. tabacum. Nevertheless, because the N-mediated responses were specific to TMV, we considered that the resistance against TMV:GFP in the 310 A plants was a suitable system for the identification of signaling components required for N-mediated resistance using VIGS.

The N. benthamiana orthologue of EDS1

To identify the N. benthamiana EDS1 orthologue (NbEDS1), we used primers based on a tomato EST (DbEST Accession no. TC8977) in regions that were most similar to A. thaliana EDS1 (AtEDS1). PCR amplification with these primers produced a 560-bp fragment of N. benthamiana cDNA that was similar to both TC8977and AtEDS1. The sequence of this fragment was then used to design primers for overlapping 3′ and 5′ RACE (rapid amplification of cDNA ends) PCR. Assembly of sequences from these RACE products generated a 1824-bp cDNA sequence predicted to encode a 608-aa putative NbEDS1 protein with a molecular mass of 69.8 kDa that is 41% identical and 59% similar to AtEDS1.

Figure 2 shows a sequence alignment of NbEDS1 and AtEDS1 highlighting conserved serine (S), aspartic acid (D) and histidine (H) residues that form a putative hydrolase catalytic triad within the amino-terminal lipase-like region (Falk et al., 1999). Consistent with previous alignments made between AtEDS1 and eukaryotic lipases, residues surrounding the serine and aspartic acid of the putative catalytic triad are more highly conserved than those surrounding the histidine (Falk et al., 1999). In addition to the putative lipase motif, the carboxy-termini of NbEDS1 and AtEDS1 show a high degree of similarity in the EDS1 and PAD4-specific (EP) domain (Figure 2; Feys et al. 2001). Within the EP domain, the five amino acid ‘KNEDT’ motif (Figure 2) is conserved in all EDS1 orthologues cloned to date (BJF and JEP, unpublished data). Importantly, this motif distinguishes EDS1 from PAD4, another lipase-like protein required for resistance mediated by TIR-NBS-LRR genes (Feys et al., 2001; Jirage et al., 1999). It also discriminates EDS1 from the product of Arabidopsis SAG101, a gene belonging to the EDS1/PAD4 family that is highly expressed during plant senescence (Feys et al., 2001; He et al., 2001). A glutamate at position 466 of the Arabidopsis EDS1 protein, which is substituted for a lysine in the loss of function eds1–1 mutant (Falk et al., 1999), is conserved at NbEDS1 residue 487 (Figure 2). Database searches using NbEDS1 returned AtEDS1 as the closest match. From these various features of the protein sequence and from DNA gel blot data indicating that there are at most two NbEDS1-homologous genes in the N. benthamiana genome (data not shown), we conclude that NbEDS1 is an orthologue of AtEDS1.

Figure 2.

Sequence alignment of predicted NbEDS1 and AtEDS1 protein sequences.

Sequences were aligned using ClustalW (http://www2.ebi.ac.uk/clustalw/) and align ments were shaded using Boxshade (http:// http://www.ch.embnet.org/software/BOX_form.html). Each protein is indicated to the left of the alignment. Identical residues are shaded black and similarly charged residues shaded gray. The serine (S), aspartate (D) and histidine (H) residues of the putative lipase catalytic triad are marked by arrowheads. The EP domain extends from AtEDS1 residues 395–602 and from NbEDS1 residues 416–591. The EDS1-specific KNEDT motif is boxed and the conserved glutamate (E) at AtEDS1 residue 466 and NbEDS1 residue 487 that is substituted for a lysine in the loss of function Ws eds1–1 mutant is marked (*). The fragment of NbEDS1 cDNA used in the TRV:EDS construct is shown by a solid black overline. The fragment of NbEDS1 cDNA used as a probe for RNA gel blot analysis of NbEDS1 transcript levels is marked by a dashed black overline.

VIGS of NbEDS1 compromises N-mediated resistance

Figure 3(a) shows how VIGS was used to test the requirement of NbEDS1 for N-mediated resistance. First, VIGS was initiated by inoculation of a TRV construct to 310 A seedlings. Second, approximately 21 days after this first inoculation, we challenge-inoculated upper leaves with TMV:GFP and monitored GFP fluorescence over a period of 10 days. If the VIGS vector had caused silencing of a gene required for N-mediated resistance, the GFP would accumulate and persist. The TRV:EDS construct carried a 560-bp PCR fragment of NbEDS1 (Figure 2). As a control for loss of resistance against TMV:GFP, we used TRV:N carrying a 508-bp exon fragment encoding the region of N between the NBS and LRR. We used the empty TRV vector (TRV:00) as a second control.

Figure 3.

VIGS of NbEDS1 compromises N-mediated resistance to TMV.

(a) Schematic representation of the VIGS procedure to test the requirement of NbEDS1 for N-mediated resistance. Four to five week old N-transgenic seedlings were inoculated with TRV vectors by agroinfiltration and, approximately 21 days later, upper leaves were challenge-inoculated with TMV:GFP. Accumulation of TMV:GFP was monitored by GFP fluorescence under UV illumination.

(b) TRV:00, TRV:N or TRV:EDS plants were challenge inoculated with TMV:GFP sap and accumulation of TMV:GFP monitored by GFP fluorescence at 5 dpi on inoculated leaves (upper panels) and at 15 dpi in systemic organs (lower panels). White arrows on the lower panels indicate foci of GFP.

(c) RNA gel blot analysis. Upper panel: TMV:GFP RNA accumulation at 5 dpi on inoculated leaves of TRV:00, TRV:N or TRV:EDS plants. Inoculation of TMV:GFP was by agroinfiltration. Genomic-(gRNA) and subgenomic-(sgRNA) TMV:GFP RNAs are shown. Middle panel: NbEDS1 RNA levels. Lower panel: ethidium bromide staining of rRNA to confirm equal loading of the gel in each lane. Lanes 1 and 2 show unchallenged TRV:00 plant material. Lanes 3–11 show TRV:00, TRV:N or TRV:EDS plant material after challenge with TMV:GFP.

Similar results were obtained in several independent experiments.

At 5 days post TMV:GFP inoculation, GFP foci (up to 2.5 mm diameter) were present on inoculated leaves of TRV:N and TRV:EDS plants but absent on the TRV:00 plants (Figure 3b, upper panels). Approximately 15 days after TMV:GFP inoculation, GFP foci could be seen on systemic organs of TRV:N and TRV:EDS plants but not on systemic organs of TRV:00 plants (Figure 3b, lower panels). Similarly, by RNA gel blot analysis of TMV:GFP inoculated leaves at 5 dpi, we detected high levels of TMV:GFP RNA on TRV:EDS and TRV:N plants but not on TRV:00 control plants (Figure 3c, upper panel). By comparison with dilution standards we estimate that TMV:GFP RNA levels were between 200 and 1000 times higher on TRV:EDS or TRV:N tissue than in the TRV:00 control tissue (data not shown).

VIGS is normally associated with a reduction in the level of the target mRNA. We therefore carried out RNA gel blot analysis in line 310 A to establish whether TRV:EDS caused a reduction of NbEDS1 mRNA. To ensure that the EDS1 hybridization signal was from the cellular mRNA rather than the viral sequence, the probe was derived from a region of the NbEDS1 mRNA that was not present in TRV:EDS1 (Figure 2). We found that, prior to TMV:GFP challenge, NbEDS1 mRNA was present at low levels in all plants infected with the TRV vector derivatives, as shown for TRV:00 plants (Figure 3c, middle panel, lanes 1, 2). However, five days after inoculation of TMV:GFP, the TRV:00 and TRV:N plants exhibited a significant increase in NbEDS1 mRNA levels (Figure 3c, middle panel, lanes 3–8). This increase was not dependent on N as similar results were obtained in-non-transgenic N. benthamiana after TMV:GFP inoculation (data not shown). Consistent with the VIGS of NbEDS1 there was no increase of NbEDS1 mRNA in TRV:EDS plants after challenge with TMV:GFP (Figure 3c, middle panel, lanes 9–11).

Our main conclusion from these analyses is that silencing of NbEDS1 mRNA caused loss of N-mediated resistance against TMV:GFP and therefore that NbEDS1 is a key component of the signaling pathway. A second conclusion is that NbEDS1 mRNA levels increase in an N-independent manner following infection with TMV:GFP. The observed up regulation of NbEDS1 mRNA is reminiscent of the response of AtEDS1 mRNA to either virulent or avirulent pathogens or treatment with an SA analogue that activates resistance signaling pathways (benzo(1,2,3)-thiadiazole-7-carbothioic acid S-methyl ester: (BTH) (Falk et al., 1999; Feys et al., 2001).

In order to characterize the expression of NbEDS1 further, we monitored its mRNA after mechanical wounding and challenge with virulent Pseudomonas syringae pv tabaci (hereafter Ps. tabaci) or disarmed A. tumefaciens. Treatment with BTH was also investigated. The results in Figure 4 show that NbEDS1, like AtEDS1, is induced by pathogen infection and by BTH treatment but not by wounding. We also examined NbEDS1 expression after challenge with avirulent pathogens. Recognition in N. benthamiana of PVX by Rx or of Ps. tabaci expressing AvrPto (Ps. tabaci/AvrPto) by Pto also caused an increase in NbEDS1 mRNA levels (data not shown). From these results it seems likely that expression of NbEDS1 is regulated by a similar mechanism to AtEDS1.

Figure 4.

NbEDS1 is induced by pathogen and BTH treatment.

RNA gel blot analysis. Upper panel: NbEDS1 transcript levels 48 h after no treatment (–), wounding (rubbing with carborundum), inoculation with virulent Ps. tabaci or A. tumefaciens, and treatment with BTH or a water control. Lower panel: ethidium bromide staining of rRNA on the gel to show equal loading. Plants used for these experiments were N-transgenic except for those inoculated with virulent Ps. tabaci which were transgenic for Pto.

Independent experiments gave similar results.

NbEDS1 is not required by Rx or Pto

We found that NbEDS1 is not a general cofactor of disease resistance because, as shown below, it is not required for Pto-mediated resistance against Ps. tabaci/AvrPto or the Rx-mediated response against PVX.

The lack of a requirement for NbEDS1 in Pto function was established by inoculating TRV vectors to Pto-transgenic N. benthamiana plants (line 38–12; Rommens et al., 1995) and subsequently infiltrating leaves with Ps. tabaci/AvrPto. The results in Figure 5 show that the TRV:EDS and TRV:00 plants were equally resistant against Ps. tabaci/AvrPto. Disease symptoms were absent at 7 days post bacterial inoculation (Figure 5a) and bacterial growth was similar in both sets of plants (Figure 5b). As a control to demonstrate that Pto-mediated resistance could be suppressed by VIGS we used a TRV construct carrying a 675-bp fragment of Prf (TRV:Prf). Prf is a CC-NBS-LRR protein that is required for Pto-mediated resistance (Salmeron et al., 1996) and, on TRV:Prf infected 38–12 plants, necrotic symptoms of Ps. tabaci (Figure 5a) developed as quickly as on non-transgenic plants (data not shown). Corres pondingly, the growth of Ps. tabaci/AvrPto was approximately 10 times higher on TRV:Prf plants than on the TRV:00 and TRV:EDS plants (Figure 5b).

Figure 5.

NbEDS1 is not required by Rx or Pto.

(a) Disease symptoms on Pto transgenic TRV:00, TRV:Prf or TRV:EDS plants 7 days after infiltration with Ps. tabaci/AvrPto. Disease is manifested in TRV:Prf plants as necrosis with a chlorotic halo surrounding the infiltrated patch.

(b) Growth of Ps. tabaci/AvrPto on Pto transgenic TRV:00, TRV:Prf or TRV:EDS plants. Leaves were infiltrated with Ps. tabaci/AvrPto and growth of bacteria in inoculated leaves monitored over a 4-day period. Each data point represents the mean ± SE of three replicate samples.

(c) Accumulation of PVX:GFP six days after agroinfiltration of Rx transgenic TRV:00, TRV:Rx or TRV:EDS plants. Accumulation of PVX:GFP was monitored under a UV dissecting microscope.

Independent infection assays gave similar results.

N. benthamiana transgenic for Rx (line Rx18) are resistant to PVX expressing GFP (PVX:GFP; Bendahmane et al., 1999). We assessed the requirement for NbEDS1 in Rx-mediated resistance by infecting Rx18 plants with TRV:EDS and, 21 days later, challenge inoculating with PVX:GFP. As controls, the Rx18 plants were inoculated with TRV carrying a fragment of Rx (TRV:Rx) or with TRV:00. In the TRV:Rx plants there were GFP foci 6 days after PVX:GFP inoculation, indicative of PVX replication (Figure 5c). PVX:GFP RNA was detected in these plant samples by RNA gel blot analysis (data not shown). There were no GFP foci on PVX:GFP-inoculated leaves of TRV:EDS1 or TRV:00 plants (Figure 5c) and PVX:GFP RNA was not detected on these plants by RNA gel blot analysis (data not shown). Thus, the Rx resistance response does not require NbEDS1.

VIGS of NbEDS1 does not cause enhanced disease susceptibility

A. thaliana eds1 mutants are compromised for resistance against avirulent pathogens and exhibit enhanced susceptibility to virulent pathogens (Aarts et al., 1998; Parker et al., 1996). To address whether NbEDS1 also influenced susceptibility to virulent pathogens, we inoculated TMV:GFP, PVX:GFP or Ps. tabaci onto susceptible N. benthamiana that had been pre-inoculated with TRV:00 or TRV:EDS. Viral RNA accumulation was then monitored over a period of 7 days by assessment of GFP fluorescence (data not shown) and on RNA gel blots. Susceptibility to Ps. tabaci was monitored by symptom expression and growth of the bacteria.

In all instances the susceptible response was not affected in the TRV:EDS plants. Thus, in TMV:GFP and PVX:GFP infected leaves, GFP fluorescence (data not shown) and viral RNAs accumulated to the same level on TRV:EDS and TRV:00 N. benthamiana, as shown in Figure 6(a,b). Similarly, Ps. tabaci disease symptoms developed at similar rates on TRV:EDS, TRV:00 and TRV:Prf plants (Figure 6c). The rate of bacterial growth was similar in infiltrated leaves of TRV:EDS, TRV:00, and TRV:Prf plants (Figure 6d). These data show that VIGS of NbEDS1, unlike the AtEDS1 loss of function eds1 mutations, does not increase disease susceptibility.

Figure 6.

VIGS of NbEDS1 does not cause enhanced disease susceptibility.

(a) RNA gel blot analysis. Upper panel: TMV:GFP RNA accumulation over 7 days after agroinfiltration of non-transgenic TRV:00 and TRV:EDS plants. Genomic-(gRNA) and subgenomic (sgRNA) TMV:GFP RNAs are shown. Lower panel: ethidium bromide staining of rRNA to confirm equal loading of the gel.

(b) RNA gel blot analysis. Upper panel: PVX:GFP RNA accumulation over 7 days after agroinfiltration of non-transgenic TRV:00 and TRV:EDS plants. Genomic-(gRNA) and subgenomic-(sgRNA) PVX:GFP RNAs are shown. Lower panel: ethidium bromide staining of rRNA to confirm equal loading of the gel.

(c) Disease symptoms in Pto transgenic TRV:00, TRV:Prf or TRV:EDS plants 7 days after infiltration with virulent Ps. tabaci. Disease symptoms are manifested as necrosis with a chlorotic halo.

(d) Growth of virulent Ps. tabaci in leaves of Pto transgenic TRV:00, TRV:Prf or TRV:EDS plants. Bacterial growth was measured over 4 days. Each data point represents the mean ± SE of three replicate samples.

Independent infection assays gave similar results.

Discussion

A conserved role for EDS1 in resistance signaling by TIR-NBS-LRR proteins

Here we have shown that TRV:EDS infection of N. benthamiana results in loss of N-mediated resistance against TMV (Figure 3). The level of NbEDS1 mRNA was specifically reduced in the TRV:EDS plants indicating that loss of resistance was due to VIGS. We further demonstrate that VIGS operates on an induced mRNA species, since NbEDS1 transcript levels increased substantially after TMV:GFP inoculation of TRV:00 and TRV:N plants but were almost undetectable in the TRV:EDS background (Figure 3c, middle panel).

In principle, the target of TRV:EDS accounting for loss of N mediated resistance could be NbEDS1 or other highly sequence-related homologues in the genome of N. benthamiana. Based on Arabidopsis data, Nicotiana homologues of SAG101 or PAD4 would be only distantly related to the insert of TRV:EDS and it is unlikely that they would be a VIGS target in these experiments. However, we cannot rule out the possibility that silencing of close NbEDS1 homologues, resulting from aneuploidy of the N. benthamiana genome, could be responsible for the suppression of N-mediated resistance. There are 19 chromosomes (1n value) in contrast to diploid members of the Solanaceae that have 12 chromosomes (Smith, 1979). DNA gel blot analysis is consistent with the proposed duplication of NbEDS1 (data not shown).

A conclusive identification of the VIGS target accounting for loss of N-mediated resistance will need further characterization of the NbEDS1 homologues so that VIGS can be specifically targeted at regions where their sequences are most likely to be different such as the 3′-non-coding part of the mRNA. Even allowing for ambiguity about the precise VIGS target, we can conclude that N-mediated resistance against TMV requires a protein or proteins that are EDS1-like and therefore that the N signaling pathway in Nicotiana is similar to that of other TIR-NBS-LRR proteins in Arabidopsis. This finding has several implications for our understanding of disease resistance in plants. It shows, for example, that R gene-mediated signaling can be similar irrespective of whether the pathogen target is a virus (TMV) as shown here, or a bacterium or fungus as demonstrated previously (Aarts et al., 1998; Parker et al., 1996). It also identifies the EDS1-type protein(s) as a conserved, most likely invariant, feature of resistance signaling pathways associated with TIR-NBS-LRR proteins. Moreover, because VIGS of NbEDS1 did not compromise disease resistance responses of Rx and Pto (Figure 5) we conclude that EDS1-independence of CC-NBS-LRR protein mediated resistance is also a conserved feature in dicot plants as diverse as Arabidopsis and Nicotiana spp.

EDS1 function in compatible plant–pathogen interactions

VIGS is a relatively new approach to the analysis of genes required for disease resistance that has clear advantages over mutagenesis when there is functional redundancy of homologous genes (Baulcombe, 1999). However, this approach may have limitations and its impact remains to be fully established. One illustration of these potential limitations is from the conflicting analyses of EDS1 in compatible plant–pathogen interactions. It appears from the VIGS results in N. benthamiana that NbEDS1 is not required to limit colonisation by the virulent pathogens PVX, TMV and Ps. tabaci (Figure 6). In contrast, Arabidopsis null eds1 mutations confer enhanced susceptibility to Ps. tomato and Peronospora parasitica (Aarts et al., 1998; Feys et al., 2001; Parker et al., 1996).

These conflicting results could indicate differences between N. benthamiana and Arabidopsis. However, NbEDS1 mRNA was induced during compatible interactions (Figure 4) suggesting that, as in Arabidopsis, the role of the encoded EDS1 protein in N. benthamiana may not be restricted to R protein-mediated resistance responses. This idea is further supported by the responsiveness of both AtEDS1 (Falk et al., 1999; Feys et al., 2001) and NbEDS1 (Figure 4) to treatment with the SA mimic, BTH. It is possible that NbEDS1 activity in limiting virulent pathogen growth was not apparent in the TRV:EDS plants because there was incomplete suppression of the NbEDS1 mRNA. An alternative explanation is that the TRV vector activated defense responses that compensated for the enhanced susceptibility.

VIGS as a general tool for the analysis of disease resistance in plants

We have also used VIGS to assess the role of 20 genes, in addition to NbEDS1 in N-mediated resistance against TMV (J. R. Peart, unpublished data). These genes include WIPK, SIPK and EREBP1 that were selected because their enzyme or mRNA levels increase in N-genotype plants infected with TMV (Horvath et al., 1998; Zhang and Klessig, 1998). Our survey also included the Nicotiana homologue of Arabidopsis NPR1 that modulates resistance downstream of SA (Cao et al., 1997) and several sequences that are implicated in a TMV-induced hypersensitive response by over-expression studies (Karrer et al., 1998). However, in none of these examples did VIGS produce a consistent loss of resistance against TMV. One interpretation of these results is that the target genes are not required for N-mediated resistance. However, as with the analyses of compatible interactions, it is possible that residual low level expression of the target gene or interference by the virus vector may have masked an effect of VIGS on resistance signaling. It is particularly likely that these factors influenced the analysis of EREBP1 because, in several experiments, VIGS of this gene resulted in partial loss of N-mediated resistance.

Notwithstanding these limitations, the results presented here validate the use of VIGS for identification of genes required for disease resistance. NbEDS1 is the first gene in the N signaling pathway to be identified by a functional assay and, because of possible complications associated with ploidy, as discussed above, it is unlikely that it would have been identified by mutagenesis. Other investigations, to be described elsewhere, have used VIGS to identify novel genes required for resistance signaling mediated by N, Rx and Pto (J. R. Peart and Rui Lu, unpublished data). We have also identified genes required for resistance against P. infestans using this approach (I. Malcuit, unpublished data). It is likely that VIGS used in conjunction with other approaches will help extend the currently incomplete picture of resistance signaling in plants.

Experimental procedures

Plant material and cultivation conditions

The N-transgenic N. benthamiana line 310 A has been described previously (Bendahmane et al., 1999). It was generated by Agrobacterium tumefaciens leaf disc transformation (Horsch et al., 1985) using the AGL1 strain carrying the binary plasmid pTG34, kindly provided by B. Baker (University of California, Berkeley, CA, USA; Whitham et al., 1996). The pTG34 plasmid contains N as a 12.3kb genomic DNA fragment with 3.4 kb coding sequence, 3.2 kb introns, 4.3 kb of 5′ flanking sequence with the presumed promoter and 1.3 kb of 3′ flanking sequence. DNA gel blot analysis showed that the 310 A line contained a single copy of N. The Rx-transgenic N. benthamiana line used, Rx-18, has been described previously (Bendahmane et al., 1999). The Pto-transgenic N. benthamiana line 38–12 was kindly provided by BJ Staskawicz (University of California, Berkeley, CA, USA) and has been described previously (Rommens et al., 1995). Plants were grown in containment glasshouses under MAFF license PHL 24B/3654 (3/2001). N. benthamiana plants were germinated on a 1 : 1 mixture of compost and peat, then grown individually in pots. Conditions were 25°C by day and 20°C by night. Supplementary winter lighting from halogen quartz iodide lamps provided a 16-h day length.

Viral pathogen isolates

The TMV:GFP construct was derived from the TMV vector (TMV/odontoglossum ringspot virus TB2) described previously (Donson et al., 1991). TMV coding sequence under the CaMV 35S promoter was transferred from pTB2 into a BIN19 based binary vector (Bevan, 1984) to form a double 35S promoter driven TMV construct with a nopaline synthase (Nos) terminator. GFP4 (Haseloff et al., 1997) was inserted into the TMV coding sequence behind a duplicated TMV coat protein promoter to form TMV:GFP. The PVX:GFP construct used here, pGr208, is based on pGr106 (Takken et al. 2000) that was modified to express GFP5 (Haseloff et al., 1997) driven by a double PVX coat protein promoter. The bipartite TRV vector and its application for silencing has been described previously (Ratcliff et al., 2001). The TRV RNA1 clone is in pBINTRA6 and the RNA2 clone for insertion of target sequences is in pTV00.

Ps. tabaci inoculation and growth determination

Cultures of Ps. tabaci (containing an empty vector or expressing AvrPto) and preparation of inoculum have been described (Thilmony et al., 1995). Bacterial suspensions were inoculated onto leaves using a needleless syringe. Inoculation of 104 cfu ml−1 was used for disease symptom and bacterial growth assays. To measure in planta growth of Ps. tabaci, three replicates each containing six 0.3 cm2 leaf discs were collected from plants inoculated with each TRV construct. Discs were ground in 10 mm MgCl2, serially diluted and plated onto L medium containing Kanamycin (25 µg ml−1) and Rifampicin (100 µg ml−1).

BTH treatments

An aqueous solution of 300 µm BTH (Lawton et al., 1996) with 0.01% silwet was sprayed onto leaves to imminent run-off as a fine mist. A solution of 0.01% silwet alone was used as the water control. Following application, water- or BTH-treated plants were kept in the same glasshouse chamber and separated by 2 m or more.

Construction of gene silencing vectors

N. benthamiana cDNA was obtained as previously described (Ruiz et al., 1998). PCR primers (5′-GCA GGG CAC TCG TCG GGT GGC GCT ATA-3′ and 5′-GTT TGC AAT ATC CAA GGG CTC AAC TTG CCT-3′) to amplify NbEDS1 cDNA were designed based on tomato EST TC8977 in regions that are most similar to AtEDS1. PCR was performed as described (Ruiz et al., 1998) except using annealing temperatures 5°C to 10°C lower than those predicted to be optimal for each primer pair. PCR products of the expected size were subcloned into the pGEM-T Easy vector (Promega), sequenced and cloned into the TRV vector. The 508 bp N fragment used to produce TRV:N was generated by PCR on a plasmid containing N genomic DNA, using primers (5′-GGT GAC TGC ACT ACC CGA TC-3′) and (5′-CGA GCC ATA ATC TGC TAC GTT C-3′). The 675 bp N. benthamiana Prf fragment (Genbank accession number AF479624) used to form TRV:Prf was generated by PCR on N. benthamiana cDNA using the following primers: (5′-GTT GGC ATG CCA GGA TTG GGC-3′) and (5′-ACA AGG CTT AAG ATA GTG TGG T-3′). The resulting fragment is 90% identical at the nucleotide level to the corresponding region of tomato Prf and two tomato Prf homologues but does not share significant nucleotide homology with any other sequences in the Genbank database. The TRV:Rx construct was made using a fragment of Rx cDNA derived from within the LRR encoding region (A Bendahmane, Génoplante, Evry, unpublished data).

Virus inoculations and Agrobacterium tumefaciens-mediated transient expression (Agroinfiltration) in N. benthamiana

Infection of plants with TRV derivatives was by agroinfiltration as previously described (Ratcliff et al., 2001). However, cultures containing pTV00-derived constructs were mixed with those containing pBINTRA6 (RNA1) in a 10 : 1 ratio prior to infiltration. Two expanded leaves of 4- to 5-week-old N. benthamiana plants were infiltrated with the mixture. TMV:GFP was inoculated by two methods. For visual assessments of TMV:GFP accumulation, manual inoculation of TMV:GFP sap using a light dusting of carborundum was performed and accumulation monitored under UV light. For RNA gel blot analysis TMV:GFP was agroinfiltrated at a 50 fold dilution from an OD600 = 1. PVX:GFP was inoculated by agroinfiltration at a 1000 fold dilution from an OD600 = 1.

For NbEDS1 mRNA induction, Agrobacterium containing pBIN19 (Bevan, 1984) was infiltrated at a 50-fold dilution from OD600 = 1. Agrobacterium strains were GV3101 (Holsters et al., 1980) for TRV and PVX inoculation and UIA143 (Farrand et al., 1989) for TMV:GFP inoculation and NbEDS1 induction.

Cloning procedures

The 5′ and 3′ ends of NbEDS1 were determined by rapid amplification of cDNA ends (RACE) by using the MARATHON cDNA amplification kit (Clonetech Laboratories, Palo Alto, CA). To obtain specific RACE products, a nested amplification step was performed as recommended by the manufacturer. Two sets of oligonucleotides were designed, based on the TRV:EDS insert sequence for use with the adapter primers (AP1 and AP2) of the kit: EDS 3′R (5′-ATG ATG CTT GCT CCC CTT TCA TCG ATT C-3′) and EDS 3′RN (5′-CAT CCA AAA TCC CGG AAT TAT CAG CAC G-3′) for 3′ RACE of NbEDS1 and EDS 5′R (5′-TGC AGA ACA GCA TCT GGA TTC TCA ACG AC-3′) and EDS 5′RN (5′-CCC ATT TCC AGT GCA GAA GAT GTA AGT TCC-3′) for 5′ RACE of NbEDS1. RACE products were cloned into the pGEM-T Easy plasmid (Promega), and sequences of six independent 3′ end clones and eight independent 5′ end clones were determined. RACE PCR products overlapped with the NbEDS1 insert sequence of the TRV:EDS construct. The SeqMan (DNASTAR) contig assembly program was used to build a consensus sequence of full length NbEDS1 (Figure 2 and Genbank accession number AF479625).

RNA gel blot analysis

Total RNA was extracted using Tri-reagent (Sigma) according to the manufacturer's instructions and approximately 15 µg was fractionated in a 1% W/V agarose-formaldehyde gel, transferred to nylon membrane (Hybond-NX) and cross-linked with UV illumination. Membranes were prehybridized, hybridized and washed as described previously (Jones et al., 1998). A PCR amplified fragment of GFP was 32P-labeled and used as a probe to determine TMV:GFP and PVX:GFP levels. A 32P-labeled fragment of NbEDS1 (Figure 2) was used to determine NbEDS1 mRNA levels on gel blots (Figure 3).

GFP imaging

Visual detection and close-up imaging of GFP fluorescence were done as described (Ratcliff et al., 1999; Voinnet et al., 1999).

Note added in proof

VIGS was recently used to demonstrate a role for calcium-dependent kinases (CDPKs) in HR mediated by CO-4 and CO-9 (Romeis et al., 2001 EMBO20, 5556–5567).

Acknowledgements

We are grateful to The Gatsby Charitable Foundation for support of this work and to our colleagues at the Sainsbury Laboratory. Thanks to M. Teresa Ruiz and Lu Rui for the TRV:Prf construct and to Abdelhafid Bendahmane for the TRV:Rx construct and the Rx-18 N. benthamiana line. Barbara Baker kindly provided the pTG34 N-gemonic binary plasmid and Brian Staskawicz kindly provided the 38–12 Pto-transgenic plants and the P. syringae strains. Work with recombinant viruses and transgenic plants was performed in containment greenhouses under MAFF license PHL 24B/3654 (3/2001). GC and BJF were supported by grants from The Biotechnology and Biological Sciences Research Council.

GenBank Accession numbers AF479624and AF47925.

Ancillary