Introgression of resistance trait Cf-4 from wild tomato species into tomato cultivar MoneyMaker (MM-Cf0) has resulted in the near-isogenic line MM-Cf4 that confers resistance to the fungal tomato pathogen Cladosporium fulvum. At the Cf-4 locus, five homologues of Cladosporium resistance gene Cf-9 (Hcr9s) are present. While Hcr9-4D represents the functional Cf-4 resistance gene matching Avr4, Hcr9-4E confers resistance towards C. fulvum by mediating recognition of the novel avirulence determinant Avr4E. Here, we report the isolation of the Avr4E gene, which encodes a cysteine-rich protein of 101 amino acids that is secreted by C. fulvum during colonization of the apoplastic space of tomato leaves. By complementation we show that Avr4E confers avirulence to strains of C. fulvum that are normally virulent on Hcr9-4E-transgenic plants, indicating that Avr4E is a genuine, race-specific avirulence determinant. Strains of C. fulvum evade Hcr9-4E-mediated resistance either by a deletion of the Avr4E gene or by production of a stable Avr4E mutant protein that carries two amino acid substitutions, Phe82Leu and Met93Thr. Moreover, we demonstrate by site-directed mutagenesis that the single amino acid substitution Phe82Leu in Avr4E is sufficient to evade Hcr9-4E-mediated resistance.
The biotrophic fungal pathogen Cladosporium fulvum causes leaf mould on its only host, tomato. During colonization of host leaf tissues, C. fulvum secretes several proteins into the apoplastic space of tomato leaves, including race-specific elicitor proteins encoded by avirulence (Avr) genes. These Avr proteins induce defence-related responses, including a hypersensitive response (HR), in tomato plants that carry the corresponding Cf resistance (R) genes (Joosten and De Wit, 1999). Cf genes encode extracellular, plasma membrane-anchored, receptor-like proteins. The highest degree of amino acid variability is present in the N-terminal, solvent-exposed residues of the leucine-rich repeats, which are thought to determine resistance specificity. Experimental data obtained so far support the hypothesis that Cf proteins function according to the guard model, which proposes that the resistance protein recognizes the modification of a so-called virulence target by the Avr protein (Van der Hoorn et al., 2002).
The Cf genes of tomato are organized in clusters of resistance gene homologues, which have been designated Hcr2s and Hcr9s for homologues of Cladosporiumresistance gene Cf-2 and Cf-9 respectively. The Cf-2 and Cf-5 loci map at the short arm of chromosome 6, at identical locations (Balint-Kurti et al., 1994), whereas the Cf-4 and Cf-9 loci map at the short arm of chromosome 1 (Jones et al., 1993). At the Cf-2 locus, the nearly identical Cf-2.1 and Cf-2.2 genes confer resistance towards strains of C. fulvum that carry the matching Avr2 gene, whereas a third homologue (Hcr2-2A) is not functional (Dixon et al., 1996). Four other Hcr2 homologues are found at the Cf-5 locus, of which Hcr2-5C is the functional Cf-5 gene that is thought to mediate recognition of the not yet characterized Avr5 gene product (Dixon et al., 1998). The Cf-4 and Cf-9 gene clusters, which originate from wild species of the genus Lycopersicon, have been introgressed in tomato cultivar MoneyMaker (MM-Cf0), resulting in near-isogenic lines MM-Cf4 and MM-Cf9, respectively (Boukema, 1980; Stevens and Rick, 1988). Each locus comprises five Hcr9s, of which Hcr9-4D and Hcr9-9C are the functional Cf-4 and Cf-9 genes, respectively (Parniske et al., 1997; Thomas et al., 1997), that confer resistance to C. fulvum through recognition of Avr4 and Avr9, respectively (Van Kan et al., 1991; Joosten et al., 1994).
Within the Hcr9 gene family, resistance specificity conferred by genes other than Cf-4 and Cf-9 has been identified. At the Cf-9 locus, Hcr9-9B and Hcr9-9E were found to provide weak resistance towards C. fulvum infection, in particular towards C. fulvum races 4, 5 and 5.9, suggesting that these homologues mediate recognition of avirulence determinants other than Avr4 and Avr9 (Parniske et al., 1997; Laugéet al., 1998). Moreover, this weak resistance was only observed in adult plants and is clearly different from Cf-9-mediated immunity resulting from recognition of Avr9, as some disease symptoms, visible as patches of sporulating mycelium, are still present on the leaves of these plants (Parniske et al., 1997; Laugéet al., 1998).
Takken et al. (1999) found that MM-Cf4 plants carrying a Ds transposon insertion within homologue Hcr9-4D or a deletion of the open reading frame (ORF) of Hcr9-4D are still resistant against certain strains of C. fulvum, indicating that additional Hcr9s at the Cf-4 locus have the ability to mediate disease resistance. As Thomas et al. (1997) had already shown that Hcr9-4A, -4B and -4C are not functional, the only candidate that remains is Hcr9-4E. Indeed, by performing complementation experiments, it appeared that this homologue conferred resistance towards C. fulvum, independent of Hcr9-4D, through recognition of a novel Avr determinant (Takken et al., 1999).
In adult transgenic MM-Cf0 plants, Hcr9-4E confers the same level of resistance towards C. fulvum as Hcr9-4D (Takken et al., 1998; 1999). The Hcr9-4E-expressing plants do not recognize Avr4 (Thomas et al., 1997), but still respond with a clear chlorosis and/or necrosis to apoplastic fluid (AF) isolated from C. fulvum race 5-infected MM-Cf0 leaves (Takken et al., 1998; 1999). This implies that Hcr9-4E mediates recognition of a secreted avirulence determinant that is different from Avr4 and which has been designated Avr4E. Thus, for virulence of C. fulvum on plants carrying the Cf-4 locus, the pathogen has to evade recognition of both Avr4 (Joosten et al., 1994; 1997) and Avr4E.
To clone the Avr gene matching Hcr9-4E, Avr4E was purified from AF of C. fulvum-infected tomato leaves. The Avr4E gene, which was subsequently cloned by reverse genetics, was found to encode a cysteine-rich protein of which the mature form contains 101 amino acids, which is secreted into the apoplastic space of tomato leaves. Complementation showed that Avr4E confers avirulence to strains of C. fulvum that are normally virulent on plants that carry Hcr9-4E, demonstrating that Avr4E is a genuine avirulence determinant. It appeared that C. fulvum circumvents Hcr9-4E-mediated resistance either by deletion of the Avr4E gene or by production of a stable Avr4E mutant protein that carries two amino acid substitutions: Phe82Leu and Met93Thr. Moreover, by site-directed mutagenesis we could show that only one amino acid substitution (Phe82Leu) in Avr4E is sufficient to evade Hcr9-4E-mediated resistance. To evade host resistance, point mutations (Avr2 and Avr4; Joosten et al., 1994; Luderer et al., 2002a), gene deletions (Avr9; Van Kan et al. 1991) and transposon insertions (Avr2; Luderer et al., 2002a) have been found in several strains of C. fulvum. However, the production of a stable, non-functional isoform of an avirulence protein adds a fourth mechanism to the arsenal of strategies of C. fulvum to evade host resistance.
Purification of the Avr4E protein from C. fulvum-infected leaves and isolation of an Avr4E cDNA and the encoding gene
To isolate Avr4E, proteins present in AF from MM-Cf0 tomato leaves infected by C. fulvum strain 19, a strain that is avirulent on MM-Cf4 plants because of the production of Avr4 and Avr4E (Takken et al., 1999) (further denoted as race 5), were fractionated by gel filtration. The protein fractions were assayed for necrosis-inducing activity (NIA) by injection into leaves of Hcr9-4E-transgenic plants. Proteins present in fractions that exhibited NIA were subsequently subjected to anion exchange chromatography and the resulting active fractions were further purified by reverse-phase chromotography and again tested for NIA (see Experimental procedures). In this way, we identified a protein of about 12 kDa that exhibited specific NIA in Hcr9-4E-transgenic plants and which we tentatively designated Avr4E.
From the purified protein, an N-terminal and an internal amino acid sequence were obtained. These protein sequences allowed the design of degenerated oligonucleotides used to amplify the Avr4E gene from C. fulvum race 5 genomic DNA, by anchored polymerase chain reaction (PCR) (see Experimental procedures). In this way, the 5′ end of Avr4E was obtained and subsequently used to screen a cDNA library derived from RNA of leaves of MM-Cf0 plants infected by C. fulvum race 5. This screening resulted in the isolation of a full-length cDNA of Avr4E. The Avr4E ORF encodes a predicted protein of 121 amino acids, which contains six cysteine residues. The signalp program (Nielsen et al., 1997) identified a 20-amino-acid N-terminal signal sequence for extracellular targeting (Fig. 1). The predicted cleavage site of the signal peptide is consistent with the sequence information obtained from N-terminal sequencing of Avr4E purified from AF, demonstrating that the N-terminus of Avr4E is not further processed by fungal and/or plant proteases in the apoplastic space of tomato. Assuming that C-terminal processing does not occur, which is however not proven, the mature Avr4E protein consists of 101 amino acids, which matches with the molecular weight of 12 kDa as determined by SDS-PAGE (results not shown). tblastn searches with the Avr4E sequence using the public databases, including the COGEME database of phytopathogenic fungi (Soanes et al., 2002) and the sequenced genomes of the fungi Aspergillus nidulans, Fusarium graminiarum, Magnaporthe grisea and Neurospora crassa (Galagan et al., 2003), did not result in hits with E-values lower than 0.05, showing that no sequences with significant homology are known. A genomic library of C. fulvum strain 16 (race 4, virulent on Hcr9-4E-carrying plants) was screened to isolate an Avr4E genomic sequence, because a race 5 (avirulent on Hcr9-4E-carrying plants) genomic library was not available. The screening resulted in the identification of a genomic clone of which the sequence of the promoter and terminator region was used to isolate a genomic Avr4E sequence of race 5 by PCR with a promoter–terminator primer set (Fig. 1). In addition to a small intron (Fig. 1), the promoter region contains seven typical consensus sequences (TA)GATA, which are putative binding sites of regulatory proteins that are involved in nitrogen metabolite (de-)repression (Marzluf, 1997).
Avr4E is the avirulence determinant matching Hcr9-4E
To determine whether the gene that we identified is the avirulence determinant of C. fulvum on tomato plants carrying the Hcr9-4E resistance gene, a complementation experiment was carried out. Four strains virulent on Hcr9-4E-transgenic plants, strain 113 (further denoted as strain SA), race 4, race 22.214.171.124 and race 126.96.36.199 were used for complementation with the genomic Avr4E clone of the avirulent strain, race 5. In contrast to race 4, strain SA is virulent on Hcr9-4E-transgenics but is not virulent on Hcr9-4D-transgenics, indicating that SA lacks a functional Avr4E gene, but has a functional Avr4 allele, thereby behaving as a race 4E (Fig. 2A). Races 4, 188.8.131.52 and 184.108.40.206 are virulent on both Hcr9-4D- and Hcr9-4E-transgenic plants and on MM-Cf4 plants carrying the complete Cf-4 locus, indicating that these races do not produce functional Avr4 and Avr4E elicitor proteins (Fig. 2C; only shown for race 4).
For the virulence assays, four independent transformants of each strain were inoculated onto the differential set of tomato lines shown in Fig. 2. As expected, introduction of the functional Avr4E gene from avirulent race 5 into the virulent isolates SA and races 4, 220.127.116.11 and 18.104.22.168 resulted in transformants that all became avirulent on tomato carrying Hcr9-4E (Fig. 2B and D; only shown for isolate SA and race 4). To show that the transformants secrete the Avr4E protein in sufficient quantities for recognition, susceptible MM-Cf0 plants were inoculated with the transformants and the parent strains, and AF was isolated from the infected leaves. Only AF obtained from transformant-infected tomato showed NIA when injected into leaves of Hcr9-4E-transgenic plants (data not shown). These results demonstrate that the isolated Avr4E gene encodes the race-specific avirulence determinant Avr4E, which confers avirulence to strains that are normally virulent on plants that carry Hcr9-4E.
Cladosporium fulvum uses two mechanisms to avoid Hcr9-4E-mediated recognition
To further investigate the mechanism conditioning virulence of C. fulvum on plants carrying Hcr9-4E, the expression of Avr4E during growth on susceptible tomato plants was investigated. For strains of C. fulvum that are avirulent (race 5) as well as for strain SA and race 4, which are virulent on these plants, Avr4E transcripts accumulated during colonization of tomato leaves (groups I and II; Fig. 3A). However, in another group of strains that are also virulent on Hcr9-4E plants, no Avr4E transcripts could be detected (group III; Fig. 3A). Transformation of strain SA and races 4, 22.214.171.124 and 126.96.36.199 with the genomic clone of Avr4E resulted in accumulation of Avr4E transcripts in planta (group IV; Fig. 3A; not shown for race 188.8.131.52) and avirulence on Hcr9-4E transgenics (Fig. 2; not shown for races 184.108.40.206 and 220.127.116.11). Transformation with a plasmid containing the genomic Avr4E clone resulted in a higher accumulation of Avr4E transcripts compared with the recipient strains, which is probably caused by multiple ectopic insertions of the Avr4E-carrying plasmid into the genome of the fungus (Van den Ackerveken et al., 1992).
As demonstrated by C. fulvum race 4, various strains of C. fulvum have developed the ability to evade both Hcr9-4D- and Hcr9-4E-mediated resistance (Fig. 2C). Previously, it has been demonstrated that, in most cases, strains circumvent Hcr9-4D-mediated resistance by production of Avr4 isoforms that carry different single amino acid substitutions (Joosten et al., 1994; 1997). To investigate whether loss of Hcr9-4E-mediated resistance is also accomplished by production of mutated Avr4E isoforms, we PCR-amplified the Avr4E ORF from the genome of 25 strains of C. fulvum that are virulent and from 10 strains that are avirulent on Hcr9-4E-transgenic plants (see Experimental procedures). Sequence analysis of the PCR products revealed that all avirulent strains contain an Avr4E ORF which is identical to that of the Avr4E gene present in C. fulvum race 5 (Fig. 1). From the 25 virulent strains, eight strains, among which are the races that were classified in group III (Fig. 3A) did not give a PCR product (see below), whereas the other 17 strains gave a fragment. In the Avr4E ORF of all these strains, including strain SA and race 4 (Fig. 3A), a double point mutation, resulting in amino acid substitutions Phe82Leu and Met93Thr (avr4ELT), was detected. This mutation does not affect transcription or stability of the Avr4E mRNA, because transcripts are still detected in these strains during colonization of tomato leaves (Fig. 3A). To analyse whether avr4ELT still exhibits NIA, we used the PVX-based expression system (Chapman et al., 1992). Although the native signal peptide of Avr2, Avr4 and Avr9 mediates proper extracellular targeting of these proteins when expressed in tomato using PVX (Takken et al., 2000), the native signal peptide of Avr4E appeared not effective (data not shown). Therefore, PVX::Avr4E and PVX::avr4ELT were obtained by cloning the sequence encoding the mature protein downstream of the PR-1a signal peptide sequence of Nicotiana tabacum (Hammond-Kosack et al., 1995). Tomato plants were inoculated with the recombinant PVX vectors and subsequently examined for the development of systemic HR (Fig. 4). MM-Cf4 and Hcr9-4E-transgenic plants inoculated with PVX::Avr4E developed a clear systemic HR, which eventually resulted in plant death. However, when these plants were inoculated with PVX::avr4ELT, only systemic mosaic symptoms developed, similar to the symptoms visible on the inoculated MM-Cf0 and Hcr9-4D-transgenics (Fig. 4; see asterisks). Thus, unlike Avr4E, avr4ELT does not exhibit NIA on tomato that carries Hcr9-4E, demonstrating that these 17 strains of C. fulvum evade Hcr9-4E-mediated resistance by only a slight modification of Avr4E.
To elucidate why the other eight strains did not yield a PCR fragment containing the Avr4E ORF, a PCR was performed with the promoter–terminator primer set used for race 5 (PROM-B and TER-X; see Experimental procedures), which should allow amplification of the full-length Avr4E gene. From each of the 17 strains carrying avr4ELT, a PCR product of ≈2 kb was obtained. However, from the eight strains again no PCR product was obtained. Additional promoter–terminator primer sets also did not yield PCR products using genomic DNA of these eight strains (results not shown). Therefore, DNA gel blots of races 4 and 5 and four of the eight strains that did not give a PCR product (the strains of group III) were made and probed with the Avr4 and Avr4E ORF. As expected, the DNA of all tested strains hybridized with the Avr4 probe, whereas Avr4E only appeared to be present, as a single-copy gene, in races 4 and 5 (Fig. 3B). The other four strains did not show any hybridization and therefore lack the Avr4E gene, confirming the results obtained by PCR.
Altogether, these data demonstrate that strains of C. fulvum evade Hcr9-4E-mediated resistance either by the production of an Avr4E mutant protein that carries two amino acid substitutions, Phe82Leu and Met93Thr, or by a deletion of the encoding gene.
The single amino acid substitution Phe82Leu is sufficient for loss of Avr4E avirulence function
As Agrobacterium-mediated transient expression in tobacco has previously been demonstrated to facilitate successful expression of extracellular elicitors and membrane-anchored Cf proteins (Van der Hoorn et al., 2000), we used this method to study the elicitor activity of mutants of Avr4E. To this aim, Agrobacterium cultures carrying Avr- and Hcr9-4-encoding genes were mixed in a 1:1 ratio and infiltrated into tobacco leaves (Fig. 5A). In the presence of Hcr9-4E, but not Hcr9-4D, Avr4E induced a necrotic response, whereas Avr4, which was included as a control, only induced necrosis in the presence of Hcr9-4D (Fig. 5A).
To investigate whether the amino acid substitutions Phe82Leu and Met93Thr are both required to evade recognition, the NIA of avr4ELT and AvrE mutant proteins carrying the single amino acid substitution Phe82Leu (avr4EL) or Met93Thr (avr4ET) was examined by agroinfiltration. A quantitative analysis (Van der Hoorn et al., 2000) was performed by co-infiltration of a dilution series of Agrobacterium cultures carrying Avr4E (mutant)- and Hcr9-4E-encoding genes. As a control, the NIA of mutant avr4 proteins, produced by strains of C. fulvum evading Hcr9-4D-mediated recognition, was also quantitatively analysed in combination with Agrobacterium carrying Hcr9-4D. The percentage of infiltrated leaf area showing necrosis was measured 7 days post inoculation and plotted against the percentage of Agrobacterium cultures carrying Avr (mutant) genes. For all Avr4 mutant proteins tested, a significant reduction in NIA was observed compared with the Avr4 wild-type protein (data not shown). These observations confirmed results obtained previously, using PVX-mediated expression of the various Avr4 mutants that occur in nature (Joosten et al., 1997). Agrobacterium cultures carrying avr4ELT induced necrosis in 50% of the infiltrated leaf area at a significantly lower dilution than Agrobacterium cultures carrying wild-type Avr4E (Fig. 5B). Thus, opposed to PVX-mediated expression in tomato (Fig. 4), avr4ELT still exhibited some NIA when agroinfiltrated in tobacco leaves, although to a much lower extent than the wild-type Avr4E protein. Interestingly, no difference in NIA was observed between avr4EL and avr4ELT and between avr4ET and Avr4E (Fig. 5B), indicating that the single amino acid substitution Phe82Leu in Avr4E is sufficient to evade Hcr9-4E-mediated recognition.
The virulent avr4E proteins are properly secreted and stable
Having established that avr4EL and avr4ELT both show reduced NIA, we analysed whether the single and double amino acid substitutions affected the efficiency of secretion and the stability of the protein. To test this, we transiently expressed Avr4E, avr4ELT, as well as avr4EL and avr4ET by agroinfiltration into tobacco leaves, without the matching resistance gene, and analysed AFs from the infiltrated leaf sectors for the presence of the expressed proteins. We also included Agrobacterium expressing the Avr4 elicitor and some avr4 mutants that occur in nature. To facilitate detection of the various mutant proteins by Western blotting, the proteins, which were fused to a PR-1a signal peptide, were provided with an N-terminal c-Myc-tag (see Experimental procedures). Co-infiltration of Agrobacterium cultures carrying these constructs with cultures carrying the matching Hcr9-4 resistance gene demonstrated that the tag did not affect the NIA of Avr4 and Avr4E (data not shown).
Western blot analysis of AF isolated from tobacco leaves agroinfiltrated with the various Avrs revealed that the same amount of protein accumulated irrespective whether Agrobacterium encoded the mutant or wild-type Avr4E (Fig. 5C). However, a significantly lower amount of protein was detected for the Avr4 mutant proteins compared with wild-type Avr4 (Fig. 5C). These data confirm that circumvention of Hcr9-4D-mediated resistance is accomplished by production of Avr4 isoforms that are unstable (Joosten et al., 1997). Furthermore, we now also show that loss of Hcr9-4E-mediated resistance is achieved by production of an inactive avr4E elicitor that is properly secreted and as stable as Avr4E.
Hcr9-4E confers resistance upon recognition of the Avr4E elicitor
Resistance trait Cf-4 has been introgressed from wild tomato species into tomato cultivar MoneyMaker (MM-Cf0) (Stevens and Rick, 1988), resulting in the near-isogenic line MM-Cf4 that confers resistance to C. fulvum through recognition of the Avr4 gene product (Joosten et al., 1994). Within the introgressed Cf-4 region, five homologues of Cladosporium resistance gene Cf-9 (Hcr9s) are present of which homologue Hcr9-4D represents the functional Cf-4 resistance gene that matches Avr4. It has previously been reported that homologue Hcr9-4E is also a functional Cf gene (Takken et al., 1998) that confers resistance towards C. fulvum race 5 through recognition of a novel avirulence determinant, designated Avr4E (Takken et al., 1998; 1999). Here, we demonstrate that homologue Hcr9-4E confers the same level of resistance towards C. fulvum race 5 as Hcr9-4D and we identified the matching Avr4E gene. The genomic Avr4E sequence of race 5 was found to confer avirulence to strains originally virulent on Hcr9-4E-transgenic plants, indicating that Avr4E functions as a genuine avirulence determinant.
The Avr4E protein is a cysteine-rich, secreted protein of 101 amino acids that is stable in the apoplastic space of tomato leaves. As found for other small and stable, secreted cysteine-rich elicitor proteins of C. fulvum, the six cysteine residues present in Avr4E are thought to be involved in disulphide bonding (Van den Burg et al., 2001; Van den Hooven et al., 2001; Luderer et al., 2002b). The Avr4E promoter region contains various putative (TA)GATA binding sites. In A. nidulans and N. crassa, for example, the AREA and NIT-2 transcription factors, respectively, contain a zinc finger motif that facilitates binding to promoter domains containing (TA)GATA sequences. Such (TA)GATA sequences are also present in the promoter of Avr9, a gene that appears to be regulated by the GATA-transcription factor NRF1 of C. fulvum (Pérez-García et al., 2001). Indeed, derepression of Avr9 was found to occur under conditions of nitrogen limitation (Van den Ackerveken et al., 1994), a phenomenon that might also hold for Avr4E.
Circumvention of Hcr9-4D- and Hcr9-4E-mediated resistance
In order to infect MM-Cf4 plants containing the complete Cf-4 locus successfully, C. fulvum has to evade both Hcr9-4D- and Hcr9-4E-mediated resistance. We have analysed 17 natural strains of the fungus that are virulent on Hcr9-4E-transgenic plants and still produce transcripts that hybridize with Avr4E. For strains of C. fulvum virulent on Hcr9-4D-transgenic plants, a variety of single point mutations in the ORF of the Avr4 gene has been found (Joosten et al., 1997). It appeared that these modifications resulted in unstable Avr4 isoforms, indicating that protein stability plays a crucial role in evasion of Hcr9-4D-mediated resistance. In contrast, circumvention of Avr4E recognition by plants that carry Hcr9-4E is mediated by production of a secreted and stable mutant avr4E protein that carries two amino acid substitutions, Phe82Leu and Met93Thr (avr4ELT). No additional modifications were identified in the Avr4E ORF of strains virulent on Hcr9-4E-transgenic plants. PVX-mediated expression of avr4ELT in tomato plants that express Hcr9-4E showed absence of NIA of the mutant protein, whereas Agrobacterium-mediated expression of avr4ELT in tobacco leaves, in combination with Hcr9-4E, revealed a strongly reduced NIA. The higher residual level of NIA in tobacco is thought to result from a more efficient Cf-mediated defence-signalling pathway compared with tomato (Hammond-Kosack et al., 1998; Kamoun et al., 1999). Nevertheless, production of the avr4ELT mutant protein in tomato by natural virulent strains of C. fulvum does not trigger Hcr9-4E-mediated defence responses that restrict colonization by the fungus.
We further analysed whether amino acid substitutions Phe82Leu and Met93Thr are both required for loss of Avr4E avirulence function. Interestingly, the Avr4E mutant protein carrying Met93Thr (avr4ET) appeared to be as active as wild-type Avr4E, whereas the mutant protein carrying Phe82Leu (avr4EL) shows the same reduction in NIA as avr4ELT, suggesting that the single amino acid substitution Phe82Leu is sufficient to evade Hcr9-4E-mediated resistance. Thus, this substitution appears to result in an essential conformational or qualitative change in the Avr4E protein. Moreover, as demonstrated for avr4ELT, the avr4EL and avr4ET mutant proteins are as stable as the wild-type Avr4E protein.
In eight strains of C. fulvum virulent on Hcr9-4E-transgenic plants that we analysed, the Avr4E gene is absent and no Avr4E transcript and Avr4E protein could be detected. Transformation of two of these strains with the genomic Avr4E sequence of C. fulvum race 5 resulted in accumulation of Avr4E transcripts and avirulence on plants that carry Hcr9-4E. This strategy of evasion of resistance resembles that of strains of race 9, which also evade Hcr9-9C-mediated resistance by losing the Avr9 gene (Van Kan et al., 1991).
Could the Avr4E protein also have a function in virulence? Strains of C. fulvum that carry avr4ELT or strains that lack the Avr4E gene do not show a visible reduction in virulence on susceptible tomato plants when compared with strains that carry wild-type Avr4E, suggesting that Avr4E does not significantly contribute to virulence of C. fulvum. Moreover, overexpression by transformation with the genomic Avr4E sequence of C. fulvum race 5 did not visually increase virulence (Figs 2 and 3). On the other hand, both Avr4E and avr4ELT might still beneficial to C. fulvum outside the plant, as the proteins might, for example, contribute to the saprophytic fitness of the fungus outside the growing season. As only one type of modification was found in the Avr4E ORF, mutations other than Phe82Leu and Met93Thr might affect the stability of Avr4E and thereby the virulence of C. fulvum. In the strains lacking Avr4E, its absence might be compensated by functional homologues that share no sequence homology to Avr4E. Therefore, Avr4E gene disruption will be necessary to unambiguously proof whether the encoded protein contributes to virulence.
Sequence polymorphism in the ORF of the AvrE gene in the C. fulvum population
As avr4EL exhibits a very much reduced elicitor activity, similar to avr4ELT, one would expect that in a natural situation the single amino acid substitution Phe82Leu would be abundant, as it is sufficient for loss of Avr4E avirulence function. However, thus far, no strains of C. fulvum virulent on Hcr9-4E-transgenic plants have been identified that carry avr4EL. Moreover, although avr4ET is as active as Avr4E, no strains avirulent on Hcr9-4E transgenics were found that carry this mutation. This implies that strains carrying avr4ELT originated from strains carrying wild-type Avr4E as a result of one single evolutionary event. Alternatively, a more favourable hypothesis would be that two independent evolutionary events occurred, where initially strains evolved that carried avr4ET, which are still avirulent on plants carrying Hcr9-4E, followed by emergence of strains that carry avr4ELT, which are virulent on these plants.
Within the C. fulvum population, several strains of C. fulvum have been identified that evade more than one Cf-mediated resistance trait, including Cf-2, Cf-4 (Hcr9-4D), Hcr9-4E, Cf-5 and Cf-9 (Hcr9-9C). The ORFs of Avr2 and Avr4 contain more sequence variation than the Avr4E gene (Joosten et al., 1997; Luderer et al., 2002a), suggesting that C. fulvum has endured more extensive selection pressure to evade Cf-2- and Hcr9-4D-mediated resistance than to evade Hcr9-4E-mediated resistance. However, a total of 17 breeding lines that carry Cf-4 were also found to carry Hcr9-4E ( Haanstra et al., 2000), implying that selection pressure to evade Avr4 recognition is similar to that of Avr4E. Moreover, little genetic variability was found within the C. fulvum population, as determined by AFLP analysis, implying that the C. fulvum population consists of a single clonal lineage (Joosten and De Wit, 1999). Thus, strains of C. fulvum that evade Hcr9-4D-mediated resistance might have evolved from a common ancestor strain that was already virulent on plants carrying Hcr9-4E.
The stable Avr4E and avr4ELT proteins can further be used to study whether a high-affinity binding site (HABS) for these proteins is present in tomato plants, similar to the Avr9 HABS (Kooman-Gersmann et al., 1996). Moreover, by screening wild species belonging to the genus Lycopersicon, Hcr9-4E alleles that still mediate recognition of the Avr4ELT mutant protein may be identified. Alternatively, accelerated evolution of Hcr9-4E may be performed by gene shuffling or domain swaps with known Hcr9s (Van der Hoorn et al., 2001; Wulff et al., 2001). These studies should allow the identification of critical determinants in tomato, which are involved in the recognition of Avr4E.
Plants and strains of C. fulvum
Tomato (Lycopersicon esculentum) cultivar MoneyMaker, carrying no known genes for resistance against C. fulvum (MM-Cf0), near isogenic line MM-Cf4 (containing an introgression segment that carries the Cf-4 locus), transgenic tomato MM-Cf0 plants that carry the functional Cf homologues Hcr9-4D or Hcr9-4E (Thomas et al., 1997), Nicotiana clevelandii and tobacco (N. tabacum cv. Petite Havana SR1) plants, were grown under standard greenhouse conditions. The described strains of C. fulvum were cultured according to De Wit and Flach (1979).
Cladosporium fulvum inoculation procedure and isolation of apoplastic washing fluids
Cladosporium fulvum was inoculated onto tomato plants as described by De Wit (1977). Briefly, suspensions of conidia of C. fulvum (≈ 5 × 106 conidia per ml) were used to inoculate the lower leaf surface of 5-week-old tomato plants. Apoplastic washing fluids (AFs) were isolated from C. fulvum-infected tomato leaves 3 weeks after inoculation, as described by De Wit and Spikman (1982).
All plasmid manipulations were carried out essentially as described by Sambrook et al. (1989). PCRs were performed with Pfu (Stratagene) or with the Expand High Fidelity PCR system (Roche Diagnostics), according to the manufacturer's instructions. Restriction enzymes, T4 DNA ligase and Escherichia coli DH5α cells were obtained from Invitrogen (Breda). Oligonucleotides were synthesized by Amersham-Pharmacia and probes were labelled with the Random Primers DNA Labelling System (Promega) using [α-32P]-dATP (Amersham-Pharmacia).
Purification of the Avr4E elicitor protein
Proteins present in AF isolated from a compatible interaction between C. fulvum strain 19 (race 5) and MM-Cf0 were separated by gel filtration on a Sephadex G-50 column (Amersham-Pharmacia), as described by Joosten et al. (1990). The fractions that showed specific chlorosis- and/or necrosis-inducing activity (NIA) when injected into leaves of Hcr9-4E-transgenic plants were subjected to anion exchange chromatography on a Resource Q column (Amersham-Pharmacia), according to the protocol described by Laugéet al. (2000). Biologically active fractions were subsequently loaded on a ProRPC HR 5/10 reverse-phase column (Amersham-Pharmacia), as described by Joosten et al. (1994). Eventually, a pure protein was obtained, running as a single 12 kDa band on SDS-PAGE gels and exhibiting specific NIA in leaves of Hcr9-4E plants. The protein was sequenced from the N-terminus and sequencing of a tryptic digest provided an internal amino acid sequence of the Avr4E protein.
Isolation of a cDNA and genomic clone of Avr4E
To facilitate isolation of the Avr4E gene by reverse genetics, degenerated oligonucleotides were designed. Two forward primers N1 (5′-gtnggnaaycargcngartgg-3′) and N2 (5′-gtngayggnacngcnathcc-3′) were synthesized, based on the underlined residues of the N-terminal amino acid sequ-ence (N-DFSRDXPPGSVGNQAEWSARGVDGTAIPRE-C) respectively. Reverse primer X-3b (5′-gcnccnacytcdatytc cca-3′) was designed based on the sequence of a tryptic digestion product of Avr4E (N-WEIEVGG-C). Anchored PCR was performed by using BstYI and ApoI to digest 100 ng of genomic DNA of C. fulvum race 5, followed by ligation of the corresponding double-stranded oligonucleotide anchors (Stuurman et al., 1996). A series of nested PCRs were performed with N1 and N2, together with primers VECT24 (5′-agcactctccagcctctcaccgcc-3′, which anneals to the BstYI-digested anchor site) and RH24 (5′-agcactctccagcctctctcaccgca-3′, which anneals to the ApoI-digested anchor site), and the PCR products were sequenced. Sequence analysis revealed that the putative translation products of the PCR fragments corresponded to the Avr4E peptide sequences. Additional 5′ sequence information was obtained by PCR using primer RH24 together with reverse primer X-3b, as well as by PCR using RH24 together with the two homologous reverse primers N3 (5′-cccagcacgttcaagaac-3′) and N4 (5′-gaagcagtcacagaggctg-3′), which anneal to the 5′ end of Avr4E.
In order to isolate the full-length Avr4E sequence, a cDNA library derived from a compatible interaction between C. fulvum race 5 and tomato (Van Kan et al., 1991) was screened using the Avr4E PCR fragments as a probe, according to a method described by Van den Ackerveken et al. (1992). This screening resulted in the isolation of a clone carrying the full-length Avr4E ORF. By screening a genomic library derived from a C. fulvum race 4 (Bussink and Oliver, 2001), we obtained promoter and terminator sequence information, which allowed subsequent isolation of the genomic Avr4E sequence of C. fulvum race 5 by PCR, using the following primers: forward primer PROM-B (5′-cgcggatccctaactct agggtctacc-3′, BamHI-site underlined) and reverse pri-mer TER-X (5′-gatcctcgaggccacctatgcataacttg-3′, XhoI-site underlined). PCR on genomic DNA of race 5 resulted in a fragment that contains a promoter region of 1475 bp and a terminator region of 190 bp, flanking the Avr4E ORF. The PCR product was digested with BamHI and XhoI and ligated into BglII/XhoI-digested site of pAN7-1, a plasmid carrying the hygromycin B selection marker (Punt et al., 1987). The obtained plasmid was used to transform strains of C. fulvum that are virulent on Hcr9-4E-transgenic plants.
Transformation of C. fulvum
The procedure for preparation of C. fulvum protoplasts was adapted from the method described by Harling et al. (1988) and Van den Ackerveken et al. (1992). Mycelium of C. fulvum strain SA, race 4, race 18.104.22.168 and race 22.214.171.124 was harvested from 3-day-old cultures that were grown in liquid B5 medium. Protoplasts were obtained upon digestion of the mycelium in 20 mM 2-(N-morpholino)-ethanesulphonic acid, pH 5.8, 1 M MgSO4, containing 25 mg of Glucanex (Novo Nordisk) and 1 mg of kitalase (Wako Pure Chemical Industries) per ml. Transformation of the protoplasts was achieved with polyethylene glycol 6000 (Merck), according to Oliver et al. (1987). Transformants resistant to hygromycin B (Sigma), obtained 3–4 weeks after culture on potato dextrose agar medium, were subcultured to obtain mono-spore isolates. For virulence assays, four independent transformants were inoculated onto a differential set of tomato lines.
Sequencing of the Avr4E ORFs in virulent and avirulent strains of C. fulvum
For standard PCR amplification, genomic DNA of various strains of C. fulvum, sampled throughout the world, was isolated according to the procedure described by Cenis (1992), after growth of the fungus for 5 days in liquid B5 medium. After DNA isolation, the Avr4E ORF was PCR amplified using forward primer 4E-F (5′-ggacgagtcttcgaagga-3′), annealing 94 bases upstream of the ATG start codon and reverse primer 4E-R3 (5′-ctgagattagaaggtagttag-3′), annealing 74 bases downstream of the TAG stop codon, resulting in a product of 532 bp. Sequence analysis of the Avr4E ORF of the various strains was also performed using the 4E-F and 4E-R3 primers.
RNA and DNA gel blot analysis
Total RNA from C. fulvum-infected tomato leaves was isolated 21 days after inoculation using a hot phenol-extraction procedure, according to the manufacturer's instructions (CLONTECH Laboratories). Glyoxal-denatured RNA (15 µg) was separated on 0.01 M sodium phosphate, pH 7.0 agarose gels and transferred to Hybond N+ membrane (Amersham-Pharmacia) by capillary blotting with 0.025 M sodium phosphate buffer, pH 7.0 (Sambrook et al., 1989). Genomic DNA gel blots were made as described by Van den Ackerveken et al. (1992). Both RNA and DNA gel blots were treated by UV cross-linking and prehybridized at 65°C for 60 min in modified Church and Gilbert buffer (0.5 M phosphate buffer, pH 7.2, 7% SDS and 1 mM EDTA) (Church and Gilbert, 1984). Subsequently, the blots were hybridized overnight at 65°C in the same buffer containing radiolabelled DNA fragments of the ORF of the Avr4, Avr4E or actin gene of C. fulvum. The blots were washed at 65°C with 0.5× SSC (75 mM NaCl, 7.5 mM sodium citrate), containing 0.1% SDS and Kodak X-OMAT films were subsequently exposed to the blots.
Construction of PVX derivatives and PVX inoculation procedure
The recombinant constructs PVX::Avr4E and PVX::avr4ELT were obtained by PCR using genomic DNA isolated from C. fulvum race 5 and race 4, respectively, as templates. Constructs that carry the endogenous signal sequence of Avr4E for extracellular targeting were obtained by PCR with forward primer PVX4E-N (5′-ccatcgatgcagttttccaacccctca-3′) and reverse primer PVX4E-R (5′-ccatcgatctatctgtttgccatcctctc-3′) (ClaI sites underlined). PVX constructs carrying the PR-1a signal sequence (Hammond-Kosack et al., 1995), upstream of the sequence encoding mature Avr4E (mutant) protein, were obtained by PCR-mediated overlap extension. First, to obtain the PR-1a signal sequence with an Avr4E overhang at the 3′ end, PCR was performed using forward primer OX10 (5′-caatcacagtgttggcttgc-3′) and reverse primer PR4E-R (5′-gcgcgagaaatcggcacggcaagagtggg-3′) on a plasmid containing PVX::Avr4 (Joosten et al., 1997) as a template. The DNA fragments encoding mature Avr4E and avr4ELT were obtained by PCR using forward primer 4E-F (5′-tcttgccgtgccg atttctcgcgcgattgc-3′) and PVX4E-R, together with genomic DNA of race 5 and race 4, respectively, as a template. The two sets of PCR products were subsequently used as templates in an overlap extension PCR with primers OX10 and PVX4E-R. The PCR fragments were digested with ClaI and ligated into the ClaI-digested PVX-expression vector pTXΔGC3A (Hammond-Kosack et al., 1995), downstream of the PVX coat protein promoter. In vitro transcription, amplification of the virus particles on N. clevelandii and PVX inoculations on (transgenic) tomato plants were performed as described by Hammond-Kosack et al. (1995).
Construction of binary plasmids and transformation of Agrobacterium tumefaciens
Binary plasmids carrying Avr4E or Avr4 and their derivatives were constructed as follows: PCR fragments containing the ORF of Avr4E and avr4ELT were generated using AT4E-N (5′-tagctcgagcgatttctcgcgcgattgcc-3′, Xhol site underlined) and AT4E-B (5′-cgcggatccctatctgtttgccatcctctc-3′, BamHl site underlined) with PVX::Avr4E and PVX::avr4ELT vectors as a template respectively. Fragments containing the sequences that encode mature avr4EL and avr4ET were obtained by overlap extension PCR with PVX::Avr4E as a template. For avr4EL, the following sets of primers were used: AT4E-N with F82L-R (5′-cacaagaacagctctctc-3′) and F82L-F (5′-gagagagct gttcttgtg-3′) with AT4E-B. For avr4ET, AT4E-N with M93T-R (5′-cgactccgacgtccgcgc-3′) and M93T-F (5′-gcgcggacgtcg gagtcg-3′) with AT4E-B were used. PCR fragments containing the ORFs encoding Avr4, avr4 (C35Y), avr4 (Y38H), avr4 (C41Y) and avr4 (C80Y) (Joosten et al., 1997) were generated using primer A4-NX (5′-tagctcgagcaaggccccaaaactcaacc-3′, XhoI site underlined) and A4-BC (5′-cgcggatccctattgcg gcgtctttaccg-3′, BamHI site underlined) with genomic DNA of C. fulvum race 5, race 126.96.36.199 (C35Y), race 4 (Y38H), race 4 (C41Y) and race 188.8.131.52.11 (C80Y) as templates respectively. The amplified fragments were digested with XhoI/BamHI, cloned into XhoI/BamHI-digested pBluescript SK+ (Amersham-Pharmacia) and sequenced. When the correct sequence was obtained, the plasmid was subsequently digested with XhoI and BamHI, and the fragments containing the Avr ORFs were subsequently cloned into XhoI/BamHI-digested binary plasmid pNW30, a derivative of pRH271 (R.A.L. van der Hoorn, unpubl.). This binary plasmid pNW30 carries the Cauliflower Mosaic Virus (CaMV) 35S-promoter and the PI-II terminator (An et al., 1989), both of which flank the PR-1a signal sequence (Hammond-Kosack et al., 1995) and the sequence encoding a double c-Myc-tag.
A binary plasmid carrying Hcr9-4E was constructed by PCR with primers Cf 4E-F (5′-agctccatgggttgtgtaaaact tatatttttcatgc-3′) and Cf 4E-R (5′-agctctgcagctaatatatcttttcttgt gcttttttcattctcg-3′), using a genomic clone containing Hcr9-4E (Takken et al., 1998) as a template, after the procedure as described for binary plasmid carrying Hcr9-4D (pCf-4) constructed by Van der Hoorn et al. (2000).
The binary plasmids were transferred to Agrobacterium tumefaciens strain MOG101 (Hood et al., 1993) by electroporation.
Agroinfiltration and analysis of apoplastic washing fluids
Agrobacterium-mediated transient expression was performed essentially as described by Van der Hoorn et al. (2000). Cultures containing recombinant Agrobacterium carrying the different binary plasmids were resuspended to a final OD600 of 2 and infiltrated into tobacco leaves in the presence of 200 µM acetosyringone. As tobacco leaves at different developmental stages might exhibit different transient expression levels, the stability of the Avr (mutant) proteins was determined within one tobacco leaf. Three-to-four days after co-infiltration with Agrobacterium carrying Hcr9-4D or Hcr9-4E, the NIA of the Avr4 and Avr4E (mutant) proteins was scored. AF was isolated 3 days after infiltration of similar leaves with cultures that carry Avr-encoding genes, according to the method described by De Wit and Spikman (1982). Ten times concentrated AF (7 µl) was supplied with 2 µl of 5× SDS loading buffer [50 mM Tris-HCl pH 6.8, 5% (w/v) SDS, 10% (v/v) glycerol, 5% 2-mercaptoethanol and 0.0025% bromophenolblue] and incubated at 95°C for 5 min. Proteins were separated by SDS-PAGE on gels containing 15% (w/v) acrylamide (Laemmli, 1970) and transferred to nitrocellulose membranes (Schleicher and Schuell) by blotting for 2 h at 250 mA. Detection of c-Myc-tagged proteins was performed by incubation of the filters in blocking buffer (BB) 1× phosphate-buffered saline (PBS) (Oxoid) and 5% skimmed milk powder (ELK; Campina), followed by overnight incubation in BB supplied with 1:1000 diluted antibodies raised against c-Myc (rabbit polyclonal IgG, Santa Cruz Biotechnology). After 3 × 15 min washes in wash buffer (WB) 1× PBS, 0.05% Tween-20, filters were incubated for 2 h in BB supplied with a 1:1000 diluted secondary antibody [Anti-Rabbit Ig-horseradish peroxidase (Amersham-Pharmacia)]. After 3 × 15 min washes in WB, c-Myc-tagged proteins were detected using Super Signal Chemiluminescent Substrate (Pierce) and Kodak X-OMAT films.
We thank Guy Bauw (Department of Genetics, University of Gent, Belgium) for determination of amino acid sequences of the purified Avr4E protein and Richard Oliver (School of Biology, University of Perth, Australia) for providing the genomic library of C. fulvum race 4. The Hcr9-4D- and Hcr9-4E-transgenic plants were a kind gift of Jonathan Jones (Sainsbury Laboratory). We also thank Renier van der Hoorn (Laboratory of Phytopathology) for providing vector pRH271 that was used for agroinfiltration and for critically reading the manuscript. N. Westerink and B.F. Brandwagt were supported by research grants from Wageningen University and NWO-STW (Netherlands Organization for Scientific Research, Technology Foundation) respectively. M.H.A.J. Joosten was supported by a VIDI Grant from the Netherlands Organization for Scientific Research (NWO).