Tomato leaves or cotyledons expressing the Cf-2 or Cf-9 Cladosporium fulvum resistance genes induce salicylic acid (SA) synthesis following infiltration with intercellular washing fluid (IF) containing the fungal peptide elicitors Avr2 and Avr9. We investigated whether SA was required for Cf gene-dependent resistance. Tomato plants expressing the bacterial gene nahG, encoding salicylate hydroxylase, did not accumulate SA in response to IF infiltration but remained fully resistant to C. fulvum. NahG Cf0 plants were as susceptible to C. fulvum as wild-type Cf0. Neither free nor conjugated salicylic acid accumulated in IF-infiltrated Cf2 and Cf9 NahG leaves and cotyledons but conjugated catechol did accumulate. The Cf-9-dependent necrotic response to IF was prevented in NahG plants and replaced by a chlorotic Cf-2-like response. SA also potentiated Cf-9-mediated necrosis in IF-infiltrated wild-type leaves. In contrast, the Cf-2-dependent IF response was retained in NahG leaves and chlorosis was more pronounced than in the wild-type. The distribution of cell death between different cell types was altered in both Cf2 and Cf9 NahG leaves after IF injection. IF-induced accumulation of three SA-inducible defence-related genes was delayed and reduced but not abolished in NahG Cf2 and Cf9 leaves and cotyledons. NahG Tm-22 tomato showed increased hypersensitive response (HR) lesion size upon TMV infection, as observed in TMV-inoculated N gene-containing NahG tobacco plants.
At sites of attempted pathogen invasion, plants rapidly induce an array of defence mechanisms that limit pathogen growth. In many plant/pathogen interactions, specific host and pathogen genes control resistance. These ‘gene-for-gene’ ( Flor, 1971) interactions are frequently used to investigate the recognition and signalling mechanisms by which plants resist infection. Four classes of ‘gene-for-gene’ resistance (R) genes have been isolated, exemplified by the Pto, Cf-9, Xa21 and N genes ( Hammond-Kosack & Jones, 1997). Much remains to be discovered about how R gene products work and whether or not they act through the same or different pathways.
One molecule known to play a major role in defence signalling is salicylic acid (SA). Dramatic de novo increases in free and conjugated SA accumulation are associated with resistance following pathogen challenge. For example, SA levels increase at least 20-fold in N gene-containing tobacco leaves following tobacco mosaic virus (TMV) inoculation ( Malamy et al. 1990 ).
SA is required for the full function of several R genes ( Delaney et al. 1994 ; Gaffney et al. 1993 ), as shown using plants transformed with nahG, a gene encoding salicylate hydroxylase that removes endogenous SA by converting it to catechol. Expression of nahG in both tobacco and Arabidopsis significantly weakens resistance to several pathogens. SA is required for full function of the tobacco N gene (conferring TMV resistance), RPS2 and RPM1 from Arabidopsis (conferring Pseudomonas syringae resistance) and several Arabidopsis RPP genes (conferring Peronospora parasitica resistance) ( Delaney et al. 1994 ; Glazebrook et al. 1996 ). These R genes all encode R proteins of the nucleotide binding site leucine-rich repeat (NBS-LRR) class ( Hammond-Kosack & Jones, 1997). In addition, nahG expression compromises the resistance phenotype of tomato plants engineered to over-express Prf, an NBS-LRR protein required for Pto function ( Oldroyd & Staskawicz, 1998). Collectively, these data indicate an absolute requirement for SA in resistance operating through NBS-LRR proteins.
The interaction between the biotrophic fungus Cladosporium fulvum and tomato (Lycopersicon esculentum) has been extensively characterized. Cf resistance genes in tomato encode products predicted to recognize the corresponding Avr gene products in C. fulvum ( De Wit, 1992). Cf-9 and Cf-2 encode predominantly extra-cytoplasmic leucine rich repeat transmembrane (eLRR-TM) glycoproteins ( Dixon et al. 1996 ; Jones et al. 1994 ). Investigations of Cf-9/Avr9 and Cf-2/Avr2 gene-mediated incompatibility, by infiltration of tomato cotyledons with intercellular washing fluids (IF) containing the Avr2 and Avr9 peptides, identified considerable Cf/Avr gene- dependent accumulation of free SA within 8–12 h ( Hammond-Kosack et al. 1996 ).
To determine whether salicylic acid is required for Cf gene-mediated resistance, transgenic tomato lines constitutively expressing nahG were generated. Here we report that resistance of tomato to Cladosporium fulvum conferred by the Cf-2 and Cf-9 genes, in striking contrast to other characterized R genes, does not require SA accumulation. Using identical NahG tomato lines, we also demonstrated that the phenotype of Tm-22-mediated TMV resistance in tomato is altered in the same way as N gene resistance in tobacco ( Gaffney et al. 1993 ), indicating that nahG in tomato can alter a resistance pathway that does require SA. This paper presents the first example of a ‘gene-for-gene’ interaction in which pathogen resistance is SA-independent. Possible mechanisms by which R proteins of the eLRR-TM type might function without requiring SA are discussed.
Figure 1(a) shows the gene fusion constructed between nahG and the 35S promoter. Thirteen independent Cf0 tomato primary transformants were generated. Single-copy lines were selected and crossed to near-isogenic Cf0, Cf2 and Cf9 stocks to generate lines heterozygous for both nahG and Cf genes.
A necrotic phenotype co-segregated with nahG. Lesions normally appeared on older leaves of 6–7-week-old plants as grey/brown necrotic patches that expanded concentrically and were often associated with vascular tissue. Lesions later appeared on younger leaves. High light conditions (240 μE m−2 sec−1) accelerated necrosis, causing severe lesions even in young cotyledons and primary leaves. Lesion severity varied in different NahG lines, correlating with mRNA expression levels (data not shown). The identical SLJ 7321 nahG construct in tobacco caused no lesions despite comparable expression of nahG mRNA. It also compromised N gene function (data not shown) as reported by Gaffney et al. (1993) and Delaney et al. (1994) .
The necrotic phenotype did not occur at lower light (70 μE m−2 sec−1) until later in development. NahG tomato plants under 6-weeks-old remained lesion-free and indistinguishable from wild-type. Such plants, with no molecular or biochemical symptoms of stress (see below), were used in all experiments.
The Tm-22 resistance phenotype is altered in NahG tomato
We tested whether NahG lines can compromise a resistance gene in tomato by inoculating wild-type and NahG Tm-22 tomato plants with TMV and comparing their resistance responses. Both wild-type and NahG lines carried Tm-22, a gene conferring TMV resistance. In wild-type leaves, there was no macroscopic response associated with Tm-22-mediated resistance. Tm-22 restricts TMV in the initial stages of infection before cell-to-cell movement ( Nishiguchi & Motoyoshi, 1987) and HR is only observed in Tm-22 heterozygotes or at high temperatures ( Hall, 1980). In NahG leaves, small brown specks appeared 3–4 days after inoculation and enlarged steadily over the 18 days of measurement as shown in Fig. 1(b,c). The lesions were often surrounded by a chlorotic halo. NahG upper leaves developed no mosaic symptoms, indicating that virus had not spread systemically despite increased HR lesion size on inoculated leaves. Wild-type tomato plants lacking Tm-22 developed typical systemic chlorotic mosaic symptoms. Wild-type and NahG mock-inoculated leaves showed no responses.
The infectivity of sap from NahG Tm-22 leaves was assayed on Samsun NN tobacco leaves. The number of subsequent HR lesions provided a crude measure of viral titre. Only sap from within necrotic NahG Tm-22 lesions was highly infective (data not shown). Green tissue adjacent to the lesion margins was of low infectivity and sap from leaf tissue distal to NahG Tm-22 lesions rarely caused HR lesions on tobacco (data not shown). Clearly, most viral particles were contained within the NahG tomato HR lesions, with only low levels of virus just outside the necrosis.
Cf2 and Cf9 NahG tomato plants fail to accumulate SA in response to intercellular fluid (IF) containing Avr2 and Avr9
We measured both free and total SA accumulation in wild-type and NahG leaves and cotyledons 24 h after IF infiltration ( Table 1). Measurements were performed at low light intensity (70 μE m−2 sec−1) to prevent necrotic lesion development in NahG plants. Free SA levels increase in IF-infiltrated wild-type Cf2 and Cf9 cotyledons, with maximal levels at 24 h ( Hammond-Kosack et al. 1996 ). As expected, both free and total SA levels increased significantly in wild-type Cf2 and Cf9 leaves and cotyledons, with highest accumulation always in Cf9. SA levels were very similar for leaves and cotyledons. In contrast, SA was not detected in NahG leaves or cotyledons of Cf0, Cf2 or Cf9 ( Table 1) or in uninfiltrated control leaves and cotyledons of each genotype (data not shown). Thus, nahG expression prevented IF-induced SA accumulation in Cf2 and Cf9.
Table 1. Salicylic acid and catechol levels in IF a-infiltrated wild-type and NahG tomato under low b light conditions
a IF, intercellular fluid (race5/Cf0, twofold dilution) containing Avr2 and Avr9. b Low light (70 μE m −2 sec−1). cFW, fresh weight. dFree salicylic acid or catechol. eFree plus conjugated salicylic acid or catechol. fND, not detected (≤ 100 ng g −1 FW). g ± standard deviation between four independent samples.
Catechol accumulation was also assessed in both free and total extracts to determine whether salicylate hydroxylase causes accumulation of free or conjugated catechol. Other groups failed to detect catechol in nahG-expressing plant tissue ( Bi et al. 1995 ; Mur et al. 1997 ). Free catechol was not detected in IF-infiltrated NahG Cf0, Cf2 or Cf9 leaf and cotyledon extracts, although the true levels of free catechol are unknown as recovery in spiked controls was very low (100 μg catechol spikes: 2–3% recovery). However, total extracts from IF-infiltrated Cf2 and Cf9 NahG leaves and cotyledons contained significant levels of a catechol conjugate (CC) ( Table 1). The total SA extraction method also resulted in extremely low catechol levels (100 μg spikes: not detected). Therefore, actual CC levels in IF-infiltrated Cf2 and Cf9 NahG leaves must have been considerably higher than in Table 1. CC was not detected in Cf0 NahG samples. Therefore, significant SA increases must be required for detectable levels of CC to be generated in IF-infiltrated NahG leaves.
Growth of C. fulvum in Cf0, Cf2 and Cf9 NahG tomato
The requirement for SA in Cf-2- and Cf-9-mediated resistance was tested in NahG and wild-type leaves infected with C. fulvum race 4 (avirulent on wild-type Cf2 and Cf9 but virulent on Cf0). Race 4 also expresses β-glucuronidase (GUS), permitting easy assessment of fungal ingress ( Ashfield et al. 1994 ; Hammond-Kosack & Jones, 1994). In wild-type interactions, at high humidity (see Experimental procedures), C. fulvum sporulates on Cf0 about 14 days after inoculation but causes no infection or HR on Cf2 or Cf9 leaves.
Cf0 NahG leaves were unaltered in either timing or severity of C. fulvum infection compared with wild-type Cf0 leaves. Sporulation was visible in both genotypes 14 days after inoculation. Cf2 and Cf9 NahG leaves responded identically to wild-type leaves and were macroscopically normal 14 days after inoculation (data not shown). Leaf discs cut from inoculated leaves at various time points were stained with 5-bromo-4-chloro-3-indolyl-beta- d-glucuronic acid (X-gluc). GUS staining patterns were macroscopically identical between NahG and wild-type Cf0, Cf2 or Cf9 leaf discs over the infection time course ( Fig. 2).
To confirm these data, microscopic analysis was performed on race 4 (GUS)-inoculated wild-type and NahG Cf2 and Cf9 cotyledons. Cotyledons, being of uniform age, size and cell number, are more reproducible than leaves for quantifying fungal infection. The nahG transgene eliminated SA as effectively in cotyledons as in leaves ( Table 1). Multiple independent samples (n = 21) were assessed 14 days after inoculation. Figure 3 shows the distribution frequencies for both lateral and vertical hyphal ingress extending from penetrated sub-stomatal cavities. Despite a tendency for slightly longer hyphae, Cf2 and Cf9 NahG cotyledons clearly retained full resistance to C. fulvum because hyphae very rarely extended further than four cell lengths from the penetration site either laterally or vertically. To confirm that penetration in cotyledons was comparable to that in leaves, a smaller number of race 4 (GUS)-inoculated wild-type and NahG Cf2 and Cf9 leaf discs were examined microscopically 8 days after infection. As shown in Table 2, at least 100 penetration events were scored for each genotype, with no significant differences observed between wild-type and NahG leaves.
Table 2. C. fulvuma hyphal ingress in wild-type and NahG Cf2 and Cf9 leaves
Distance scored as cell length/layer equivalents. Leaves were harvested 8 days after inoculation. At least six independent leaf discs were analysed per genotype.
Wild-type and NahG Cf2 and Cf9 leaf discs and cotyledons were also stained with trypan blue at the time of Cf0 sporulation to look for cell death associated with resistance (data not shown). The failure of cells to exclude trypan blue dye is an indicator of cell death ( Hammond-Kosack & Jones, 1994; Keogh et al. 1980 ). Microscopic analysis in the wild-type confirmed that cell death was not essential for pathogen arrest. In wild-type Cf2 and Cf9, hyphae were usually arrested in the sub-stomatal cavity without cell staining, as reported previously ( Ashfield et al. 1994 ; Hammond-Kosack & Jones, 1994). There was more cell death in NahG Cf2 and Cf9 cotyledons and leaves, but this was unrelated to fungal penetration.
NahG compromises Cf-9-dependent but not Cf-2-dependent macroscopic responses to Avr peptides
Wild-type and NahG Cf0, Cf2 and Cf9 leaves were infiltrated with IF containing Avr2 and Avr9. Figure 4(a) shows the macroscopic responses. Wild-type Cf9 leaves developed grey necrosis within 24 h. This was invariably contained within the injected area. Cf9 NahG leaves did not develop necrosis even after 48 h, but chlorosis was visible 3–5 days after injection both within and beyond the injected panel. Wild-type Cf2 leaves developed chlorosis inside the IF-injected panel within 3–5 days. Chlorosis was also observed in NahG Cf2 leaves after 3–5 days; this was often more pronounced than in wild-type.
Figure 4(b) shows that NahG Cf2 and Cf9 cotyledons were indistinguishable from wild-type in their IF responses at the dilution which differentiated wild-type and NahG Cf9 leaves. Both wild-type and NahG Cf9 cotyledons developed necrosis by 24 h, and both wild-type and NahG Cf2 cotyledons showed chlorosis after 3–5 days. However, necrosis was not elicited in NahG Cf9 cotyledons at a lower IF dilution (1/16) that still elicited necrosis in wild-type Cf9 cotyledons (data not shown). Wild-type and NahG Cf0 cotyledons did not respond to IF infiltration.
The failure of NahG Cf9 tomato leaves to develop grey necrosis suggests SA may potentiate this reaction. To test this hypothesis, leaflets of excised Cf9 seedlings pretreated with 250 μm SA were injected with IF at a concentration too low to normally cause any macroscopic response (1/128 dilution). SA-pretreated leaflets developed confluent necrosis after 16 h, whereas water pretreated controls only developed confluent necrosis with a 1/64 dilution of IF. This result, shown in Table 3, supports the conclusion that SA potentiates necrosis development.
Table 3. The effect of SA pretreatment a on the necrosis-inducing activity of IF b at various dilutions
a 21-day-old seedlings were excised at soil level and placed in glass vials containing either 100 or 250 μm sodium salicylate or distilled water for 8 h. The liquid was then replaced with 20 ml distilled water before leaf 1 was injected with IF. bIF from a compatible Cf0/race 5 C. fulvum interaction. cIF dilutions 1, 2, 3 and 4 were 1/32, 1/64, 1/128 and 1/256, respectively. d–, no symptoms; +, patchy grey necrosis by 16 h; G, confluent grey necrosis by 16 h. The experiment was repeated four times using at least six plants per treatment and genotype.
Progression of Cf-2- and Cf-9-dependent cell death in NahG plants
The cellular responses of wild-type and NahG Cf2 and Cf9 were investigated in leaf sections cut at various times after IF infiltration. Cell viability was assessed by trypan blue staining. Figure 5(a) shows cell staining in untreated tomato leaves. After IF infiltration, both wild-type and NahG Cf0 leaves retained 100% cell viability except at injection wound sites (data not shown).
Figure 5(b) shows the progression of cell death in wild-type and NahG leaves after IF infiltration. Trypan blue staining in Cf9 leaves began earlier in wild-type than in NahG, although the spatial pattern of cell death was similar. By 12 h, some grey necrosis was visible in wild-type leaves, corresponding with cell staining in the vascular parenchyma cells around minor veins. As necrosis developed, cell death spread out from these veins mainly in lower mesophyll cells. By 48 h, necrosis was dry and cells appeared granular, no longer retained trypan blue dye and were surrounded by cell debris.
In NahG Cf9 leaves, cell death was not evident until 24 h when vascular parenchyma cells were stained as in wild-type leaves at 12 h. By 48 h, NahG leaves were still macroscopically normal despite considerable cell death in vascular tissue and in both lower and upper mesophyll cells adjacent to veins. By 72 h, when chlorosis became visible in NahG leaves, most cells around minor veins were heavily stained, and lower and upper mesophyll cells were dying in interveinal regions.
Figure 5(c) shows the spatial and temporal progression of cell death in wild-type and NahG Cf2 leaves after IF infiltration. Veins in both wild-type and NahG leaves began staining after 24 h. Individual interveinal upper mesophyll (palisade) cells were also stained at 24 h in both genotypes. More of these cells were stained in NahG leaves. By 48 h, equivalent numbers of palisade cells were stained in wild-type and NahG leaves; samples were indistinguishable. At 72 h, as chlorosis appeared, stained palisade mesophyll cells disappeared in wild-type leaves, to be replaced by granular cells and debris. However, the proportion of stained palisade cells in NahG leaves remained constant even 7 days after IF infiltration.
In summary, IF-infiltrated Cf9 NahG mesophyll cells did not die confluently. SA therefore appears to potentiate cell death in Cf9 IF responses. The effect of nahG on Cf2 cell death is unclear, although failure of stained cells to disappear in NahG Cf2 suggests that SA plays some role. The heightened chlorosis in IF-infiltrated NahG Cf2 leaves is probably not due to increased cell death.
Cf-2- and Cf-9-dependent defence gene induction in NahG plants
To determine whether IF-infiltrated NahG tomato is compromised in defence gene induction, we compared RNA transcript accumulation in wild-type and NahG leaves for three SA-inducible defence genes, acidic β-1,3-glucanase, basic β-1,3-glucanase and PR1a. Figure 6(a) shows the RNA gel blots. Transcript accumulation was similar for leaves and cotyledons. Transcript was very low in all Cf0 samples but detected in wild-type and NahG Cf2 and Cf9 samples for each defence gene. Accumulation was delayed and reduced in NahG.
RNA transcript accumulation was also examined for these defence genes at various developmental stages, as shown in Fig. 6(b). This was performed under high light conditions (240 μE m−2 sec−1) to induce NahG lesion development. Defence gene transcripts were not detected in wild-type or NahG RNA from 14-day-old cotyledons or second leaves of 28-day-old plants. All experiments in this paper used plants of these ages. However, increased transcript abundance was detected in leaf 6 of 7-week-old plants for all three defence genes. Transcript levels were low in wild-type RNA but high in NahG RNA. In these 7-week-old NahG plants, leaf 6 had a few necrotic lesions while leaf 3 had severe lesions. Defence gene transcript levels were high in NahG leaf 3 RNA compared with wild-type leaf 3. The NahG necrotic lesions therefore appear to induce defence gene expression via an SA-independent pathway.
SA and catechol levels in NahG plants
SA and CC levels were analysed in the same plants used for RNA analysis. Table 4 shows that SA levels in healthy wild-type cotyledons and leaves grown at 240 μE m−2 sec−1 were greater than for equivalent material grown at 70 μE m−2 sec−1 ( Table 1). SA was not detected in NahG samples, whereas CC was detected in all NahG material. CC levels were low in healthy NahG cotyledons and leaf 2 of 28-day-old plants, whereas considerable CC accumulation was detected in both leaf 6 and leaf 3 of 7-week-old NahG plants with lesions. Conceivably, accumulation of CC in NahG tomato triggers lesion development. Both free and total SA were detected in leaf 6 and leaf 3 extracts of 7-week-old wild-type plants, but levels were not significantly higher than in younger leaves.
Table 4. Salicylic acid and catechol levels in wild-type and NahG Cf0 tomato under high a light conditions
a High light (240 μE m −2 sec−1). bFW, fresh weight. cFree salicylic acid or catechol. dFree plus conjugated salicylic acid or catechol.
e –, not measured. f ± standard deviation (≥ 4 independent samples). g ND, not detected (≤ 100 ng g −1 FW). hLeaf 6 (lesions just appearing). iLeaf 3 (severe lesions in nahG). jLeaf 3 of 7-week-old-plants, 14 days after inoculation with C. fulvum race 5.
To determine whether CC in NahG leaves can inhibit C. fulvum, the lesioned 7-week-old Cf0 NahG plants were inoculated with C. fulvum race 5 alongside wild-type controls. After 14 days, C. fulvum sporulation was comparable on both genotypes. Conjugated SA and catechol levels were analysed in the inoculated leaves ( Table 4). High CC levels were still detected in NahG leaf 3. SA levels in wild-type leaves were not significantly higher than before inoculation. These data demonstrate that high CC levels permit C. fulvum growth in planta.
To test whether catechol inhibits C. fulvum growth in vitro, the toxicity of SA and catechol towards C. fulvum races 4 and 5 was compared. Catechol was 10 times more inhibitory to C. fulvum vegetative growth than SA and four times more inhibitory to spore formation (conidiation). However, catechol levels below 2 m m did not affect C. fulvum, and physiological concentrations are unlikely to be so high in planta. Thus, CC accumulation in NahG plants is unlikely to inhibit C. fulvum.
Consequences of SA removal in NahG tomato for Tm-22and Cf-9/Cf-2 function
Several experiments proved that the nahG construct (SLJ 7321) functions effectively. We studied TMV resistance in SLJ 7321-transformed NahG tobacco carrying the N gene (data not shown) and obtained indistinguishable results from those of Gaffney et al. (1993) . We also demonstrated that Tm-22-mediated resistance to TMV was altered by nahG expression, in a manner strongly resembling the effects of nahG on N function in tobacco. Resistance caused by over-expression of Prf in tomato is also compromised in SLJ 7321 NahG tomato lines ( Oldroyd & Staskawicz, 1998). Analysis of SA accumulation in SLJ 7321 NahG tomato provided further proof of SLJ 7321 function because SA removal was very efficient. SA did not accumulate in IF-infiltrated Cf2 or Cf9 NahG leaves, compared with significant increases in wild-type controls ( Table 1). The nahG construct used by Gaffney et al. (1993) permitted two- to threefold increases in SA accumulation in TMV-inoculated NahG tobacco lines, yet N-mediated resistance was still compromised. Therefore, the SLJ 7321 NahG tomato lines should have compromised Cf-2- and Cf-9-mediated resistance if SA was required. The altered IF responses of Cf9 and Cf2 ( Figs 4 and 5) provide further evidence that nahG expression does affect SA-dependent defence functions in tomato despite C. fulvum resistance being retained. Our data clearly show that SA is not required for Cf-2- or Cf-9-mediated resistance to C. fulvum even though all other known race-specific disease resistance genes do require SA.
Although CC accumulated in IF-infiltrated and older uninfiltrated leaves prior to lesion development, CC was not detected in uninfiltrated controls of experimental age. Thus, although catechol inhibits C. fulvum at concentrations above 2 m m, CC levels in NahG leaves would probably be too low to inhibit fungal growth. Furthermore, older lesioned Cf0 NahG plants with high CC levels ( Table 4) still permitted normal C. fulvum sporulation. Young NahG plants of experimental age were not stressed because PR gene expression was identical to that of controls ( Fig. 6b). Since Cf0 NahG plants retain full susceptibility, resistance of Cf2 and Cf9 NahG tomato to C. fulvum is unlikely to be related to the necrosis in older NahG plants.
Cf-mediated resistance without cell death: possible explanations for SA-independence
In IF-infiltrated Cf9 NahG leaves, both the onset and completion of cell death were delayed. Cell death was also affected in NahG Cf2 IF responses. SA may potentiate the Cf-9-mediated IF response because SA pretreatment of Cf9 leaves increased sensitivity to IF too dilute to cause necrosis in untreated leaves ( Table 3). It is unclear whether SA potentiates Cf2 IF responses despite the altered cell death patterns in NahG leaves. The failure of NahG Cf2 palisade cells to stop staining with trypan blue after 48 h may mean that NahG cells cannot complete cell death without SA. However, IF-infiltrated NahG Cf2 leaves became more chlorotic than wild-type, so prolonged staining may result from continual triggering of cell death. The cause of the enhanced chlorosis in NahG Cf2 leaves is also uncertain. SA removal could increase jasmonic acid (JA) and ethylene-mediated signalling due to antagonism between the JA and SA pathways ( Reymond & Farmer, 1998). Both JA and ethylene can affect chlorosis development ( Bent et al. 1992 ; Feys et al. 1994 ). Alternatively, interveinal chlorosis may result from cell death in vascular tissue. Both Cf2 and Cf9 NahG leaves developed necrosis around the major veins after IF infiltration. Blockage of vascular tissue could explain the chlorosis in Cf9 NahG leaves and the heightened chlorosis in Cf2 NahG leaves 3–5 days after IF infiltration, because inhibition of sucrose loading into phloem causes chlorosis in tobacco leaves ( von Schaewen et al. 1990 ).
Several research groups have proposed that SA potentiates defence responses ( Ryals et al. 1996 ). For example, SA potentiates HR cell death in soybean cell cultures undergoing an oxidative burst ( Levine et al. 1994 ), and SA pretreatment of parsley cell cultures enhances elicitor-induced oxidative burst responses ( Kauss & Jeblick, 1995). SA pretreatment of tobacco before viral or bacterial inoculation also enhances AoPR1 and PAL-3–GUS expression ( Mur et al. 1996 ) and physiological SA concentrations amplify pathogen resistance signals in soybean cell cultures ( Shirasu et al. 1997 ). Thus if SA potentiates cell death, and cell death is not important for Cf-dependent resistance, then it follows that Cf-2 and Cf-9 do not require SA for resistance.
The fact that nahG expression altered IF responses but not C. fulvum resistance highlights the differences between these experimental situations. The elicitor quantity and number of cells involved after IF infiltration are greater than in response to pathogen inoculation. The flood of elicitor from IF may cause a massive oxidative burst followed by high SA induction, which, in Cf9 plants at least, may contribute to the HR-like necrotic response. During incompatible C. fulvum interactions with Cf2 or Cf9 at high humidity, HR does not occur ( Ashfield et al. 1994 ; Hammond-Kosack & Jones, 1994). Therefore, SA may not accumulate significantly during normal Cf-mediated resistance.
Comparison of Cf-dependent and other R gene-dependent signalling pathways
Does Cf-mediated resistance differ mechanistically from resistances conferred by other R genes? The type of resistance pathway triggered by a pathogen may depend upon the structure of the corresponding R gene. Pathogens that deliver intracellular ligands appear to be recognized by different classes of R gene than extracellular pathogens such as C. fulvum. All the NBS-LRR resistance proteins currently studied and Pto protein kinase recognize internally delivered ligands. Therefore, similar SA-mediated signalling pathways may be triggered by each of these R proteins upon ligand binding. An Avr9 receptor is located in the plasma membrane ( Kooman-Gersmann et al. 1996 ). Signalling through Cf-9 may therefore be more similar to other pathways triggered by extracellular ligands than to the NBS-LRR and Pto classes of R protein. Elicitor/receptor responses with potential similarity to Cf-mediated signalling include those of parsley suspension cultures elicited with Pep-13 and of tobacco suspension cultures elicited with cryptogein ( Jabs et al. 1997 ; Pugin et al. 1997 ). The role of SA in these two systems has not yet been determined. As in these systems, Piedras et al. (1998) reported the calcium-dependent release of reactive oxygen species (ROS) within 5 min of Avr9 addition to transgenic Cf-9-expressing tobacco cell cultures as well as alkalinization of the culture medium. Phosphorylation and activation of two protein kinases was also observed in Cf-9-expressing tobacco within 2–5 min of Avr9 elicitation ( Romeis et al. 1999 ), again resembling the Pep-13 and cryptogein responses.
What HR- and SA-independent mechanisms might exist in tomato to limit C.fulvum growth? In incompatible interactions, hyphal growth is arrested but the fungus is not killed ( Hammond-Kosack & Jones, 1994). Recognition of C. fulvum avirulence products by Cf proteins causes an oxidative burst (May et al. 1996; Piedras et al. 1998 ), and ROS may attack C. fulvum directly and limit its growth. Alternatively, ROS may oxidatively cross-link cell-wall structural proteins resulting in cell-wall fortification ( Lamb & Dixon, 1997). Reduced permeability of reinforced cell walls would minimize nutrient leakage to the apoplast. Joosten et al. (1990) suggested that C. fulvum probably relies on extracellular sucrose, released from host cells, as a carbon source.
Several defence molecules are specifically targeted to the apoplast. These include defensins, many of which are potent fungal inhibitors that may be important for resistance. Penninckx et al. (1996) showed that induction of an Arabidopsis defensin was not mediated by SA but instead required both JA and ethylene. SA-independent defensin induction may be required for C. fulvum resistance because JA is rapidly induced in Cf9 tobacco cell cultures by Avr9 (Piedras, Farmer and Jones, personal communication). However, in Cf2 and Cf9 tomato Never ripe (Nr) ethylene-insensitivity mutants, C. fulvum resistance was unaltered ( Brading, 1997).
Cf-9 and Cf-2 may activate both a SA-dependent response and a JA/ethylene-dependent response. This may seem paradoxical since these pathways have been reported to be antagonistic. The necrotic response of Cf9 compared with the chlorosis in Cf2 may be due to higher relative stimulation of the SA pathway by Cf-9 than by Cf-2. The JA/ethylene-dependent response may be sufficient for C. fulvum resistance and the same may be true for the SA-dependent response since the Nr mutants were also still resistant ( Brading, 1997). Thus, losing either the SA or the JA/ethylene pathway would not abolish resistance.
It has been proposed that all classes of R protein may eventually converge into a central signalling cascade that co-ordinates a multi-faceted and locally induced plant defence response. However, the data presented in this paper suggest that this proposition should be re-examined. We conclude that, unlike other R proteins, the Cf proteins restrict pathogen growth through resistance mechanisms that do not require SA.
DNA construction, plant transformation and growth conditions
The Pseudomonas putida salicylic acid hydroxylase gene (nahG) was excised from the plasmid pH22 (kindly provided by Brian Staskawicz) as a 3.1 kb HindIII fragment, and ligated into the bluescript KS+ vector previously cut with HindIII to create plasmid SLJ 7307. SLJ 7307 was cut with HpaI and SspI and the 1.5 kb fragment recovered. This fragment was ligated into the vector pSLJ 4K1 ( Jones et al. 1992 ), which had been cut with ClaI and BamHI and treated with T4 DNA polymerase. The resulting SLJ 7311 plasmid carries a 35S:omega leader:nahG gene with a nopaline synthase (nos) termination sequence. The entire nahG gene was then subcloned into binary vector SLJ 7292 as a 3 kb HindIII + EcoRI fragment. In the final plasmid, SLJ 7321, the neomycin phosphotransferase gene and the 35S:nahG:nos gene are transcribed in parallel towards the right border (RB) of the T-DNA ( Fig. 1a).
The binary T-DNA construct SLJ 7321 was mobilized into Agrobacterium tumefaciens strain LBA4404 ( Hoekema et al. 1983 ) and transformed into Moneymaker (Cf0) plants. Primary transformants were regenerated as described by Fillatti et al. (1987) . T-DNA locus number was assessed for each NahG tomato transformant according to the ratio of kanamycin-resistant to kanamycin-sensitive seedlings from selfed seed. This was later confirmed by DNA gel blot analysis. The NahG transformants were crossed with Cf2 and Cf9 near-isogenic stocks (carrying the Cf-2 and Cf-9 resistance genes, respectively) to generate experimental material. For all experiments, tomato plants were grown in Levington's M3 compost under growth room conditions at 22–24°C with a 16 h photoperiod and approximately 70% relative humidity (RH). For most experiments, plants were grown at a light intensity of 70 μE m−2 sec−1.
Transgenic plant selection techniques
Plants were selected for the nahG gene using the neomycin phoshotransferase II (NPT II) selectable marker also on construct SLJ 7321. Seedlings expressing NPT II grew normally on Murashige and Skoog media containing 300 μg ml−1 kanamycin, whereas wild-type seedlings did not grow branched roots or true leaves. NPT II activity was detected in older plant extracts using an assay that measures transfer of radiolabel from a γ32P-ATP co-factor to neomycin by binding the radiolabelled product to phosphocellulose paper ( McDonnell et al. 1987 ). This assay was used to distinguish between nahG-expressing F1 seedlings and wild-type siblings.
Pathogen material and inoculations
Cladosporium fulvum race 5 and race 4 (GUS) were obtained from Richard Oliver (University of East Anglia, Norwich, UK). C. fulvum was grown in vitro on ¼-strength potato dextrose agar as described previously ( De Wit, 1977). Seedling cotyledons or the second and third leaves of 4–5-week-old tomato plants were spray-inoculated with a conidial suspension in water (3 × 105 conidia ml−1). Plants were transferred to propagators with air vents closed. Four days after inoculation, the vents were opened to promote stomatal opening, although humidity remained almost 100%. Inoculation experiments were performed in a controlled growth cabinet (light intensity = 240 μE m−2 sec−1) but experimental plants were grown at 70 μE m−2 sec−1 and transferred to the experimental cabinet 3 days before inoculation.
Tobacco mosaic virus (TMV) strain U1 was passaged through susceptible tobacco plants and stored as infected tissue at −20°C. Tissue was homogenized in 50 m m sodium phosphate buffer pH 7.0. Infections were performed by rubbing diluted homogenates onto tomato leaves with the aid of carborundum powder. After inoculation, plants were showered with water to run-off to remove surface inoculum.
To assess virus levels, TMV-inoculated tomato leaves were washed in 70% ethanol for a few seconds before rinsing in distilled water. This aimed to destroy remaining virus particles on leaf surfaces before cutting sections to make homogenates. Equivalent tissue quantities were homogenized in 100 μl of 50 m m sodium phosphate buffer pH 7.0. A 10 μl aliquot of homogenate was used to inoculate each tobacco leaf.
Isolation and injection of race-specific elicitors
Intercellular washing fluids (IF) containing C. fulvum race-specific elicitors were isolated from heavily infected tomato leaves when the fungus was sporulating ( De Wit & Spikman, 1982). The crude intercellular fluid was heated to 100°C for 10 min for sterilization and precipitation of large proteins. The precipitate was removed by centrifugation at 4°C. The supernatant was stored at −20°C.
IF was injected into the apoplastic spaces of interveinal panels of healthy leaves or cotyledons using a needleless 1 ml disposable syringe ( Hammond-Kosack & Jones, 1994). For most experiments, crude IF was diluted once with water before use.
For the SA pretreatment experiment, 21-day-old Cf0 and Cf9 seedlings were excised at soil level and placed for 8 h in glass vials containing either 100 or 250 μm sodium salicylate or distilled water. The liquid was replaced with 20 ml distilled water before leaf 1 was infiltrated with IF from a compatible Cf0/race 5 C. fulvum interaction. IF was injected at dilutions of 1/32, 1/64, 1/128 and 1/256.
Staining and microscopy techniques
The histochemical localization of β-glucuronidase (GUS) activity was performed essentially as described by Jefferson et al. (1987) . To examine growth of C. fulvum race 4 (GUS), whole cotyledons or leaf discs were excised from inoculated plants and vacuum-infiltrated with a solution containing 0.5 mg ml−1 5-bromo-4-chloro-3-indolyl-beta- d-glucuronic acid (X-gluc) sodium salt, 0.05% (v/v) Triton-X100, 1 m m EDTA and 50 m m phosphate buffer, pH 7.0. Samples were incubated at 37°C in the dark overnight and then destained in 2.5 g ml−1 chloral hydrate saturated solution. C. fulvum growth in leaves and cotyledons was quantified under phase contrast optics by scoring the number of cell length/layer equivalents that hyphae grew from each penetrated stomate ( Hammond-Kosack & Jones, 1994). For lateral growth, a cell length equivalent represented the length averaged from 10 epidermal cells measured with an eyepiece graticule. Vertical growth was measured by scoring the number of cell layer planes from the penetrated stomate in which hyphae were observed ( Hammond-Kosack & Jones, 1994).
Lactophenol–trypan blue staining ( Keogh et al. 1980 ) was used to assess cell viability after IF injection and C. fulvum inoculation. Non-viable cells accumulated stain. Samples were destained and viewed in 2.5 g ml−1 chloral hydrate solution on a Zeiss Axiophot microscope under either bright field illumination or phase contrast. Photomicrographs were prepared using Kodak Ektachrome EPT 160T film.
RNA extraction and gel blot analysis
RNA was isolated from frozen leaf material by extraction with phenol/chloroform/isoamyl alcohol and guanidine hydrochloride followed by precipitation with acetic acid and ethanol ( Logemann et al. 1987 ). Aliquots (10 μg) of total mRNA were separated by electrophoresis through formaldehyde agarose (1.4%) gels. Gels were blotted onto nylon membrane (Hybond-N, Amersham, UK) as described by Maniatis et al. (1982) . Ethidium bromide (10 μg ml−1) was included in the sample loading buffer, allowing photography of the gel after electrophoresis to confirm equal RNA loading. The RNA was UV-crosslinked to the nylon membrane (Stratagene Autolinker) and hybridization and washes were performed approximately according Feinberg & Vogelstein 1983) . Probe DNA was labelled with 32P-dCTP using a standard kit procedure (Pharmacia). Blots were washed to a final stringency of 1× SSC, 0.1% SDS at 65°C. Blots were exposed to imaging plates (Fuji).
Extraction and quantification of salicylic acid and catechol
Free and total SA levels in tomato leaves and cotyledons (approximately 100–500 mg tissue) were determined by a method adapted from a procedure described by Raskin et al. (1989) . HPLC separation was achieved using a ColumbusTM 5 μm C18 reverse phase column (25 cm × 6 mm; Phenomonex Ltd, Macclesfield, UK). SA concentrations were assessed from absorbance spectra at 300 nm using a Spectra Focus scanning UV detector (Spectra Physics) and were corrected for SA recovery by comparison with spiked controls.
Assessment of salicylic acid and catechol toxicity to C. fulvum
C. fulvum race 4 and race 5 conidial suspensions (0.5 ml; 1 × 105 spores/ml) were spread onto PDA (¼-strength) plates also containing SA or catechol at various concentrations between 0 and 30 m m. Plates were incubated for 7 days at 22°C before assessment of fungal growth.
We thank James Keddie for making nahG construct SLJ 7321 and Kate Harrison for transforming it into tomato and tobacco. Richard Oliver provided the C. fulvum races and original seeds of the Moneymaker tomato near-isogenic lines. We also thank Sara Perkins and Jonathan Darby for their excellent horticultural assistance. The Gatsby Charitable Foundation supported this work.