The novel Cladosporium fulvum lysin motif effector Ecp6 is a virulence factor with orthologues in other fungal species

Authors

  • Melvin D. Bolton,

    1. Laboratory of Phytopathology, Wageningen University, Binnenhaven 5, 6709 PD Wageningen, the Netherlands.
    2. Department of Plant Pathology, North Dakota State University, Fargo, ND 58105-5012, USA.
    Search for more papers by this author
    • These authors contributed equally.

    • Present address: USDA – ARS, Northern Crop Science Laboratory, Fargo, ND 58105-5677, USA.

  • H. Peter Van Esse,

    1. Laboratory of Phytopathology, Wageningen University, Binnenhaven 5, 6709 PD Wageningen, the Netherlands.
    2. Centre for BioSystems Genomics (CBSG), PO Box 98, 6700 AB Wageningen, the Netherlands.
    Search for more papers by this author
    • These authors contributed equally.

  • Jack H. Vossen,

    1. Laboratory of Phytopathology, Wageningen University, Binnenhaven 5, 6709 PD Wageningen, the Netherlands.
    2. Swammerdam Institute for Life Sciences, Mass Spectrometry, Nieuwe Achtergracht 166, 1018 WV Amsterdam, the Netherlands.
    Search for more papers by this author
    • These authors contributed equally.

    • §

      Present address: Plant Research International BV, PO Box 16, 6700 AA Wageningen, the Netherlands.

  • Ronnie De Jonge,

    1. Laboratory of Phytopathology, Wageningen University, Binnenhaven 5, 6709 PD Wageningen, the Netherlands.
    Search for more papers by this author
  • Ioannis Stergiopoulos,

    1. Laboratory of Phytopathology, Wageningen University, Binnenhaven 5, 6709 PD Wageningen, the Netherlands.
    Search for more papers by this author
  • Iris J. E. Stulemeijer,

    1. Laboratory of Phytopathology, Wageningen University, Binnenhaven 5, 6709 PD Wageningen, the Netherlands.
    Search for more papers by this author
  • Grardy C. M. Van Den Berg,

    1. Laboratory of Phytopathology, Wageningen University, Binnenhaven 5, 6709 PD Wageningen, the Netherlands.
    Search for more papers by this author
  • Orlando Borrás-Hidalgo,

    1. Laboratory of Phytopathology, Wageningen University, Binnenhaven 5, 6709 PD Wageningen, the Netherlands.
    2. Centre for Genetic Engineering and Biotechnology, PO Box 6162, Havana, 10600, Cuba.
    Search for more papers by this author
  • Henk L. Dekker,

    1. Swammerdam Institute for Life Sciences, Mass Spectrometry, Nieuwe Achtergracht 166, 1018 WV Amsterdam, the Netherlands.
    Search for more papers by this author
  • Chris G. De Koster,

    1. Swammerdam Institute for Life Sciences, Mass Spectrometry, Nieuwe Achtergracht 166, 1018 WV Amsterdam, the Netherlands.
    Search for more papers by this author
  • Pierre J. G. M. De Wit,

    1. Laboratory of Phytopathology, Wageningen University, Binnenhaven 5, 6709 PD Wageningen, the Netherlands.
    2. Centre for BioSystems Genomics (CBSG), PO Box 98, 6700 AB Wageningen, the Netherlands.
    Search for more papers by this author
  • Matthieu H. A. J. Joosten,

    1. Laboratory of Phytopathology, Wageningen University, Binnenhaven 5, 6709 PD Wageningen, the Netherlands.
    Search for more papers by this author
  • Bart P. H. J. Thomma

    Corresponding author
    1. Laboratory of Phytopathology, Wageningen University, Binnenhaven 5, 6709 PD Wageningen, the Netherlands.
    2. Centre for BioSystems Genomics (CBSG), PO Box 98, 6700 AB Wageningen, the Netherlands.
      *E-mail bart.thomma@wur.nl; Tel. (+31) 317 484536; Fax (+31) 317 483412.
    Search for more papers by this author

*E-mail bart.thomma@wur.nl; Tel. (+31) 317 484536; Fax (+31) 317 483412.

Summary

During tomato leaf colonization, the biotrophic fungus Cladosporium fulvum secretes several effector proteins into the apoplast. Eight effectors have previously been characterized and show no significant homology to each other or to other fungal genes. To discover novel C. fulvum effectors that might play a role in virulence, we utilized two-dimensional polyacrylamide gel electrophoresis (2D-PAGE) to visualize proteins secreted during C. fulvum–tomato interactions. Three novel C. fulvum proteins were identified: CfPhiA, Ecp6 and Ecp7. CfPhiA shows homology to proteins found on fungal sporogenous cells called phialides. Ecp6 contains lysin motifs (LysM domains) that are recognized as carbohydrate-binding modules. Ecp7 encodes a small, cysteine-rich protein with no homology to known proteins. Heterologous expression of Ecp6 significantly increased the virulence of the vascular pathogen Fusarium oxysporum on tomato. Furthermore, by RNA interference (RNAi)-mediated gene silencing we demonstrate that Ecp6 is instrumental for C. fulvum virulence on tomato. Hardly any allelic variation was observed in the Ecp6 coding region of a worldwide collection of C. fulvum strains. Although none of the C. fulvum effectors identified so far have obvious orthologues in other organisms, conserved Ecp6 orthologues were identified in various fungal species. Homology-based modelling suggests that the LysM domains of C. fulvum Ecp6 may be involved in chitin binding.

Introduction

Cladosporium fulvum (syn. Passalora fulva) is a biotrophic pathogen that causes leaf mold of tomato (Lycopersicon esculentum Mill. syn. Solanum esculentum) (Thomma et al., 2005). After germination of conidia, the fungus produces runner hyphae that penetrate stomata predominantly on the lower side of the leaf. Once inside the apoplast, C. fulvum does not penetrate host cells or develop haustoria but remains confined to the intercellular space between plant mesophyll cells (de Wit, 1977). Despite much research on the C. fulvum–tomato interaction, the molecular components that C. fulvum utilizes for infection and colonization are largely unknown (Thomma et al., 2005).

Plant pathogens secrete molecules called effectors that contribute to the establishment of disease to their hosts. As the complete set of effectors of a potential pathogen determines the outcome of the interaction with a possible host, it is important to make an inventory of this effector catalogue. Many plant pathogenic bacteria inject effector proteins into the cytoplasm of host cells by means of the type III secretion system (TTSS) to subvert host cellular physiology to the bacterium's advantage (Grant et al., 2006; Tang et al., 2006). This process is orchestrated by specific cis-elements in the promoters of genes encoding type III effector proteins, a feature that has been exploited to identify such effectors in genome-wide functional screens (Guttman et al., 2002; Chang et al., 2005). In a similar way, the discovery that several oomycete effector molecules enter the host cytoplasm through a specific host targeting RXLR-DEER motif (Whisson et al., 2007) has been exploited to identify oomycete effector catalogues. It is currently predicted that the genomes of oomycete plant pathogens contain hundreds of such effectors (Whisson et al., 2007; Jiang et al., 2008).

The effectors of extracellularly growing plant pathogenic fungi are usually very rich in cysteine residues involved in disulphide bridges, thereby protecting them against proteinases that occur frequently in apoplastic spaces of their host plants (Joosten and de Wit, 1999; Rep, 2005; Thomma et al., 2005; Kamoun, 2006). At present, relatively few whole-genome sequences of plant pathogenic fungi are available when compared with bacteria (Xu et al., 2006). As most effector proteins from extracellular pathogenic fungi are secreted, apoplastic extract from colonized plants is an important resource for the discovery of molecular factors important in several plant diseases (Joosten and de Wit, 1999; Rep, 2005; Thomma et al., 2005; Kamoun, 2006).

As C. fulvum is restricted to the tomato apoplast during colonization, all communication and exchange of molecular components between C. fulvum and its host occurs in the apoplastic space. So far, analysis of the protein composition of the apoplastic space of C. fulvum-infected tomato leaves has mainly been focused on identification of race-specific avirulence proteins (Avrs) that are secreted by the fungus during infection and invoke a resistance response in tomato genotypes carrying cognate C. fulvum resistance (Cf) genes (van Kan et al., 1991; Joosten et al., 1994; Luderer et al., 2002; Westerink et al., 2004). In addition, a number of extracellular proteins (Ecps) secreted during infection by all strains of C. fulvum have been identified (Joosten and de Wit, 1988; Wubben et al., 1994; Laugéet al., 1998; 2000; Haanstra et al., 1999; 2000). Like Avrs, Ecps induce a resistance response in tomato accessions carrying as yet unidentified Cf-Ecp resistance genes. Collectively, the Avrs and Ecps are the secreted effecor proteins. In total, eight C. fulvum secreted effector proteins have been characterized in detail and their corresponding genes have been cloned (van Kan et al., 1991; van den Ackerveken et al., 1993; Joosten et al., 1994; Laugéet al., 2000; Luderer et al., 2002; Westerink et al., 2004; Thomma et al., 2005). All these secreted effector proteins are relatively small (ranging from 3 to 15 kDa) and contain a high and even number of cysteine residues that appear to be involved in disulphide bridge formation (Kooman-Gersmann et al., 1997; van den Burg et al., 2003). These bridges provide a compact tertiary structure that contributes to stability and activity of the secreted effector proteins in the protease-rich tomato apoplast (Joosten et al., 1997; Tornero et al., 1997; Jorda et al., 1999; Krüger et al., 2002; van Esse et al., 2006). All of these effector proteins elicit a defence response in plants carrying the cognate Cf genes in a ‘gene-for-gene’ manner (Kruijt et al., 2005). The observation that these proteins are maintained within the population together with their abundance and specific accumulation during pathogenesis suggests that these proteins play an important role in fungal virulence (Thomma et al., 2005). Indeed, transformants containing gene knockouts of either Ecp1 or Ecp2 were shown to have impaired aggressiveness in mature tomato plants (Laugéet al., 1997). Recent data show that also Avr2 is a genuine virulence factor of C. fulvum (H.P. van Esse et al., submitted). It has previously been shown that Avr2 interacts with, and inhibits, the tomato cysteine protease Rcr3 which, in compliance with the guard hypothesis, is required for Cf-2-mediated immunity (Rooney et al., 2005). In compatible interactions, however, Avr2 inhibits several additional extracellular host cysteine proteases that are required for host basal defence (H.P. van Esse et al., submitted). Protection of chitin, a major constituent of fungal cell walls, against plant chitinases by the chitin-binding Avr4 effector protein (van den Burg et al., 2006) was recently shown to contribute to C. fulvum virulence (van Esse et al., 2007).

In addition to the secreted effectors, Nrf1 and Aox have been identified as virulence factors of C. fulvum (Segers et al., 2001; Thomma et al., 2006). The nitrogen response regulator Nrf1 was found to control expression of Avr9 but no other known Avr or Ecp genes in planta (Pérez-García et al., 2001; Thomma et al., 2006). Interestingly, disruption of the Nrf1 gene reduces C. fulvum virulence significantly (Thomma et al., 2006). Similarly, targeted disruption of Aox1, a starvation-induced acetaldehyde dehydrogenase, caused decreased colonization of the host plant (Segers et al., 2001).

To visualize extracellular proteins present in compatible and incompatible C. fulvum–tomato interactions, the apoplastic proteome of C. fulvum-infected tomato was analysed using two-dimensional polyacrylamide gel electrophoresis (2D-PAGE). Several proteins that are produced during infection were identified by mass spectrometry (MS) and coding sequences for three novel C. fulvum proteins were obtained by reverse genetics employing PCR with degenerate primers based on MS/MS sequence tags and N-terminal sequencing. We used RNA interference (RNAi) for functional analysis in C. fulvum and demonstrate that one of the identified secreted effectors is crucial for C. fulvum virulence.

Results

Quantification of C. fulvum biomass in infected tomato leaves

In a compatible interaction involving the susceptible MoneyMaker Cf-0 (MM-Cf-0) tomato cultivar which lacks resistance genes against this pathogen, the fungus colonizes the apoplast around leaf mesophyll cells. Conidiophores emerge from stomata 7 days post inoculation (dpi) to produce conidia (Fig. 1A). Using real-time PCR to quantify fungal biomass in the plant tissue it is evident that fungal biomass gradually increases until the fungus is extensively sporulating (Fig. 1B). In the incompatible interaction, such as with resistant MoneyMaker Cf-4 (MM-Cf-4) tomato plants that recognize C. fulvum strains expressing wild-type Avr4 (Joosten et al., 1994), no disease symptoms are visible (not shown). Real-time PCR confirms that in such an incompatible interaction no significant increase in fungal biomass occurs when compared with the compatible interaction (Fig. 1B).

Figure 1.

Disease progression of Cladosporium fulvum on tomato.
A. Typical symptoms caused by C. fulvum on susceptible MM-Cf-0 tomato plants at 3, 6, 9, 13 and 16 days post inoculation (dpi). The fungus is not visible at early stages of infection (3 dpi) but develops white patches of conidiophores (6 dpi) that expand and cover almost the whole leaf (9 dpi). Subsequently, the conidiophores start to produce conidia (13 dpi) which give the leaf a green-brownish velvet-like appearance (16 dpi).
B. Quantitative real-time reverse transcription PCR to measure C. fulvum growth on resistant MM-Cf-4 tomato plants (white) and on susceptible MM-Cf-0 tomato plants (grey) at 3, 6, 9, 13 and 16 dpi. The extent of colonization is determined by the relative quantification (RQ) of transcript levels of the constitutively expressed C. fulvum actin gene (measure for fungal biomass) to the constitutively expressed tomato glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene (measure for plant biomass) shown on a logarithmic scale. Bars represent mean values and standard errors of three leaflets taken from two plants at each time point analysed. The experiment was repeated twice with similar results.

Characterization of the C. fulvum-infected tomato apoplast proteome

In previous analyses, the protein composition of the apoplastic space of C. fulvum-infected tomato leaves has mainly focused on identification of effectors that are secreted by the fungus during infection and that invoke a resistance response in tomato. For an inventory of the apoplast proteome of C. fulvum-infected tomato and to identify secreted fungal proteins that might play a role in virulence, 2D-PAGE was utilized that allowed the comparison of MM-Cf-0 and MM-Cf-4 plants infected by a race 5 C. fulvum strain (compatible and incompatible interaction respectively). At 2 weeks post inoculation of susceptible MM-Cf-0 plants, the fungus has generated considerable biomass and has extensively colonized the host tissue (Fig. 1), likely resulting in a large quantity of fungal proteins in the apoplast as compared with resistant MM-Cf-4 plants. Therefore, this time point was chosen for detailed analysis of fungal proteins (Fig. 2). Proteins present in 2 ml of apoplastic fluid isolated from the two different interactions were analysed with 2D-PAGE. Separation of the proteins in the first dimension was carried out on Immobiline DryStrips (pH 4–7) and for the second dimension 12.5% polyacrylamide gels were used. After Coomassie brilliant blue staining, 16 protein spots specific for, or highly induced during, the compatible interaction were excised from the gel (Fig. 2). Subsequently, the proteins were digested with trypsin and the generated peptides were analysed with matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) MS and peptide fragment spectra were obtained with liquid chromatography (LC) MS/MS. Peptide mass fingerprints and peptide sequence information were used to search for protein identity in databases. This resulted in the identification of a tomato endochitinase and the C. fulvum proteins Ecp1, Ecp2 and Ecp5 (Table 1). Proteins present in the other spots could not be identified solely based on the data obtained in the MS analysis.

Figure 2.

The apoplast proteome of Cladosporium fulvum-infected tomato analysed with two-dimensional polyacrylamide gel electrophoresis (2D-PAGE). Coomassie brilliant blue-stained 2D-PAGE gels obtained after electrophoresis of soluble proteins present in apoplastic fluid collected from a compatible (A; race 5 C. fulvum strain inoculated onto MM-Cf-0 plants) and an incompatible (B; race 5 C. fulvum strain inoculated onto MM-Cf-4 plants) interaction at 14 days post inoculation. The proteins were focused over a non-linear gradient of pH 4–7. Molecular weight markers for the second dimension are indicated on the left. The part of the gel containing the C. fulvum-derived differentially accumulated proteins is shown. Protein spots for which identification was pursued are numbered (see also Table 1).

Table 1.  Apoplast proteins identified with mass spectrometry.
ProteinSpot numbersPeptides confirmed with MS/MSN-terminal amino acids (aa)Reference
  • a. 

    Determined with MALDI-TOF-generated peptide mass fingerprints (PMF).

Ecp11, 25Joosten and de Wit (1988)
Ecp237Wubben et al. (1994)
Ecp544Laugéet al. (2000)
CfPhiA5–10646 aaThis study
Ecp611–13426 aaThis study
Ecp714525 aaThis study
Endochitinase16, 17PMFaJoosten and de Wit (1989)

Six of these non-identified protein spots (5–10; Fig. 2) resulted in a comparable peptide mass fingerprint and are therefore likely to be derived from the same protein. One of the protein spots (5; Fig. 2) was subsequently subjected to N-terminal sequencing, resulting in a 46-amino-acid sequence that was found to harbour the previously identified MS/MS tags. As the obtained sequence showed homology to a structural Aspergillus nidulans phialide protein, this protein was designated CfPhiA (Table 1). This is a protein that typically occurs on phialides, which are sporogenous cells that release conidia from their apex by budding (Melin et al., 2003).

Three other protein spots (11–13; Fig. 2) also generated a comparable mass fingerprint, implicating that also these spots may be derived from the same protein. N-terminal sequencing of spot 12 resulted in a 26-amino-acid sequence harbouring the identified MS/MS sequence tags and the corresponding protein was designated Ecp6 (Table 1).

The remaining protein spots (14, 15; Fig. 2), of which we obtained peptide mass fingerprints as well as peptide fragment spectra, were also subjected to N-terminal sequencing. For protein spot 14, the 25-amino-acid sequence that was obtained matched the corresponding MS/MS sequence tags and the protein was designated Ecp7 (Table 1). Although sequence information based on MS/MS was available for protein spot 15, this protein was not considered for further study because N-terminal sequence failed repeatedly.

Cloning of extracellular protein genes

Degenerate primers were designed based on the N-terminal protein sequences of CfPhiA and Ecp6 and were used in combination with an oligo-(dT) primer to amplify the coding regions of the corresponding genes using a cDNA library from C. fulvum-infected tomato leaves as template. For Ecp7, a degenerate primer based on an MS sequence tag was used because the N-terminal sequence was not yet available when the cloning was initiated. In all cases, a cDNA sequence was successfully amplified which corresponded to MS/MS and N-terminal peptide sequences. For CfPhiA, a 720 bp fragment encoding the mature protein and part of the 3′UTR was cloned (Fig. S1). The predicted mature CfPhiA protein contains 175 amino acids and has a predicted molecular mass of about 19 kDa and an isoelectric point (pI) of 5.0. blastp analysis (Altschul et al., 1997; Schäffer et al., 2001) of the amino acid sequence showed that this protein shares similarity to putative proteins of several fungal species including A. nidulans, A. fumigatus and Neurospora crassa. Of these orthologues, the PhiA protein from A. nidulans has been functionally characterized (Melin et al., 2003), and was found to be essential for growth and sporulation of the fungus as phiA mutants were found to be impaired in phialide development. Therefore, it is likely that the C. fulvum putative orthologue CfPhiA has a similar function.

A 742 bp fragment with the coding region for the mature Ecp6 protein and the 3′UTR was cloned (Fig. S1). Ecp6 encodes a mature protein of 199 amino acids, including eight cysteines, and has a predicted molecular mass of 21 kDa and a pI of 4.6. Furthermore, Ecp6 contains five predicted N-glycosylation sites, explaining the location of the Ecp6 protein spots on the 2D-gel. Based on blastp analysis, Ecp6 was found to share significant homology to the glycoprotein CIH1 identified in the plant pathogenic fungus Colletotrichum lindemuthianum (Perfect et al., 1998). Although the contribution of CIH1 to pathogenicity is unknown, it has been shown to accumulate during infection on bean in the walls of intracellular hyphae and the interfacial matrix which separates the hyphae from the invaginated host plasma membrane (Perfect et al., 1998).

For Ecp7, a 464 bp cDNA fragment was cloned containing the coding region for 84 amino acids of the mature Ecp7 protein. N-terminal sequencing of Ecp7 revealed that a stretch of 16 amino acids precedes the peptide that was identified as an MS tag, and based on which the degenerate primer for cloning the cDNA was designed (Fig. S1). Therefore it should be concluded that Ecp7 encodes a mature protein of 100 amino acids which includes six cysteines and has a predicted molecular mass of 11 kDa and a pI of 6.0. blastp analysis of the amino acid sequence revealed no significant homology of Ecp7 to other protein sequences deposited in public databases.

CfPhiA, Ecp6 and Ecp7 are expressed during infection

With real-time PCR assays using genomic DNA from C. fulvum as a template and Avr2 as a single-copy reference gene (Luderer et al., 2002), it was determined that the C. fulvum genome contains only one copy of the CfPhiA, Ecp6 and Ecp7 genes (results not shown). Furthermore, real-time PCR analysis of CfPhiA, Ecp6 and Ecp7 transcripts, using the constitutively expressed C. fulvum actin gene as an endogenous control, revealed that all genes are expressed in both compatible and incompatible interactions (Fig. 3). CfPhiA expression is induced already early in the compatible interaction, at 6 dpi, and maintains this level of expression for all time points analysed. In the incompatible interaction, CfPhiA is also induced, although its expression level is approximately half of that found in the compatible interaction (Fig. 3). Both Ecp6 and Ecp7 show a low but steady level of expression in the incompatible interaction when compared with that of the C. fulvum actin gene, while the genes are clearly induced in the compatible interaction. While Ecp7 peaks at 9 dpi (Fig. 3), Ecp6 is maximally expressed at 13 dpi (Fig. 3). In contrast to the expression pattern of the CfPhiA gene, the patterns of Ecp6 and Ecp7 typically resemble those of other genes encoding secreted C. fulvum effectors. For example, C. fulvum Avr9 is highly expressed throughout the compatible interaction, with maximum expression at 9 dpi, whereas its expression in the incompatible interaction remains low (Fig. 3). Nevertheless, the expression level of the Avr9 gene is much higher than those of Ecp6 and Ecp7 (Fig. 3).

Figure 3.

Expression analysis of the newly identified Cladosporium fulvum extracellular proteins. The expression of CfPhiA, Ecp6, Ecp7 and Avr9 genes was monitored during the interaction of C. fulvum with MM-Cf-4 tomato (incompatible; white bars) and MM-Cf-0 tomato (compatible; grey bars) at 3, 6, 9, 13 and 16 days post inoculation. Real-time reverse transcription PCR was used for the relative quantification (RQ) of transcript levels of the C. fulvum CfPhiA, Ecp6 and Ecp7 genes relative to the constitutively expressed C. fulvum actin gene as an endogenous control. The RQ of the Avr9 gene is shown as an example of the expression profile of a typical C. fulvum effector gene. The mean and standard error of the results obtained from three leaflets taken from two plants at each time point assayed are shown. The experiment was repeated twice with similar results.

Heterologous expression of Ecp6 in Fusarium oxysporum f. sp. lycopersici enhances virulence on tomato

In contrast to C. fulvum, Fusarium oxysporum may easily be transformed using Agrobacterium-mediated transformation, generally resulting in large numbers of transformants (Mullins et al., 2001). To investigate whether C. fulvum Ecp6 and Ecp7 may act as fungal virulence factors, we overexpressed these Ecps in F. oxysporum f. sp. lycopersici. To this end, the sequences encoding the mature proteins were fused in frame with the sequence encoding the C. fulvum Avr4 signal peptide for extracellular targeting (Joosten et al., 1997) into a binary vector under control of the fungal constitutive ToxA promoter (Ciuffetti et al., 1997). Using Agrobacterium-mediated transformation a large number of transformants were obtained, and presence of the transgene was confirmed by PCR (data not shown). Four transformants were randomly picked for each of the C. fulvum Ecps and tested in an inoculation assay on tomato. Upon inoculation of tomato plants with transformants that overexpress Ecp7, disease development was indistinguishable from disease caused by the non-transformed progenitor strain (data not shown). In contrast, on tomato plants that were inoculated with each of the four transformants that overexpress Ecp6, disease symptoms developed earlier and were more severe compared with the inoculation with the non-transformed progenitor F. oxysporum f. sp. lycopersici strain (Fig. 4A and B) or the transformants that overexpress Ecp7 (data not shown). With reverse transcription PCR it was confirmed that in each of the transformants, but not in the progenitor F. oxysporum f. sp. lycopersici strain, Ecp6 was expressed (Fig. 4C).

Figure 4.

Symptoms caused by wild-type Fusarium oxysporum f. sp. lycopersici and heterologous Ecp6 overexpression transformants on susceptible tomato.
A and B. Side view (A) and top view (B) of the disease phenotype caused by F. oxysporum f. sp. lycopersici wild-type (WT) and four independent heterologous Ecp6 overexpression transformants (Ecp6-1 to Ecp6-4) on susceptible tomato MoneyMaker plants when compared with mock-inoculated tomato (mock) at 14 days post inoculation.
C. Reverse transcription PCR to detect in planta transcription of heterologously expressed C. fulvum Ecp6 in F. oxysporum f. sp. lycopersici wild-type and four independent heterologous Ecp6 overexpression transformants (Ecp6-1 to Ecp6-4) on susceptible tomato MoneyMaker plants when compared with mock-inoculated tomato (mock) at 14 days post inoculation.

RNAi-mediated silencing of Ecp6 compromises C. fulvum virulence on tomato.

RNAi has been successfully employed for gene functional analysis in filamentous fungi (Nakayashiki et al., 2005). This is particularly relevant for fungi like C. fulvum for which homologous recombination is not straightforward. Recent evidence has shown that PEG-mediated transformation may generate somaclonal variation that may be circumvented by Agrobacterium-mediated transformation which is, however, significantly less efficient (van Esse et al., 2007). Therefore, RNAi was recently successfully implemented to silence the expression of C. fulvum effector genes (van Esse et al., 2007).

Based on the results obtained with heterologous expression of C. fulvum Ecp6 in F. oxysporum f. sp. lycopersici, we applied RNAi-mediated silencing for functional analysis of the C. fulvum Ecp6 gene using Agrobacterium-mediated transformation with constructs aimed at generating double-stranded RNA that targets these genes (RNAi). A pGREEN-based binary vector, carrying transfer DNA (T-DNA) that contains either a nourseothricin resistance cassette or a hygromycin resistance cassette, and an inverted-repeat fragment of the target gene under control of the fungal constitutive ToxA promoter (Ciuffetti et al., 1997), was used to provoke RNAi-mediated gene silencing. To target the expression of the Ecp6 gene, two RNAi constructs were generated based on different sections of the Ecp6 coding region. Agrobacterium-mediated transformation of the RNAi constructs generated several antibiotic-resistant transformants for each construct. Analysis of the transformants indicated that their growth in vitro was indistinguishable from that of the progenitor race 5 isolate (data not shown). As C. fulvum effector genes show variable expression when cultured in vitro (Thomma et al., 2006), 4-week-old MM-Cf-0 tomato plants were inoculated with three transgenic C. fulvum strains to determine whether the introduction of the inverted-repeat construct resulted in Ecp6 silencing. Utilizing real-time PCR, a strong reduction in transcription of the target gene was found when compared with the progenitor isolate in several transformed isolates using expression of the C. fulvum actin gene as a reference (Fig. 5A). At 10 dpi, transformants Ecp6i-1 and Ecp6i-4 of the first construct, and Ecp6i2-1 of the second construct, showed a reduction to 36%, 27% and 48% of the wild-type Ecp6 expression level respectively (Fig. 5A). At later time points, the level of Ecp6 reduction increased for the Ecp6i2-1 transformant, while the reduction in the Ecp6i-1 and Ecp6i-4 remained rather consistent, which may possibly be attributed to different regions of the transcript that are targeted for gene silencing (data not shown).

Figure 5.

Expression analysis and quantification of growth of Cladosporium fulvum RNAi transformants silenced for Ecp6.
A. The expression of Ecp6 is monitored during a compatible interaction between C. fulvum and MM-Cf-0 tomato involving the wild-type (WT) C. fulvum and RNAi transformants at 10 days post inoculation. Real-time PCR was used to measure the relative quantification (RQ) of transcript levels of the Ecp6 genes, as compared with the constitutively expressed C. fulvum actin gene as an endogenous control. Bars represent mean values and standard error of the results obtained from three leaflets taken from two infected plants.
B. Growth of WT C. fulvum and RNAi transformants was quantified on MM-Cf-0 tomato plants. The transcript levels of the constitutively expressed C. fulvum actin gene (measure for fungal biomass) relative to the levels of the constitutively expressed tomato glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene (measure for plant biomass) are shown to determine the degree of fungal colonization of the MM-Cf-0 tomato leaves. Bars represent mean values and standard error of the results obtained from three infected leaflets taken from two plants.

Visual inspection of the inoculated MM-Cf-0 tomato plants showed a clearly delayed progression of disease for the Ecp6 RNAi transformants (Fig. 6). While conidiophores were emerging from the stomata on the lower surface of tomato leaves inoculated with the wild-type progenitor strain at 10 dpi, the leaves inoculated with transformant Ecp6i-4 were devoid of these structures (Fig. 6). Although leaves inoculated with transformants Ecp6i-1 (Fig. 6) and Ecp6i2-1 (data not shown) showed some fungal growth, the extent of leaf colonization was significantly less than that observed for the wild-type strain. To measure the extent of fungal growth of RNAi transformants compared with the parental wild-type strain, the constitutively expressed C. fulvum actin gene was used as a marker in real-time PCR analyses (Fig. 5B). The constitutively expressed tomato chloroplast glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene was used as a reference for the ratio of fungal biomass to plant biomass to determine the degree of colonization. After inoculation of MM-Cf-0 tomato lines, all Ecp6 RNAi transformants showed significant reduction in growth compared with the parental race 5 isolate (Fig. 5B).

Figure 6.

Typical symptoms caused by C. fulvum wild-type (WT) and RNAi transformants silenced for Ecp6 at 10 days post inoculation onto susceptible tomato plants (MM-Cf-0).

Ecp6 sequence analysis from a worldwide collection of C. fulvum strains

As our results showed that Ecp6 is a virulence factor of C. fulvum, we assessed sequence variation of Ecp6 in a worldwide collection of strains (Stergiopoulos et al., 2007a,b). We first obtained 691 bp of genomic sequence upstream of the region that encodes the mature Ecp6 protein by gene walking. Sequence analysis using the gene prediction algorithm fgenesh (Salamov and Solovyev, 2000) identified a putative start codon and predicted intron/exon boundaries using the genetic codes of several fungi present as models in the database. These were confirmed by cloning the Ecp6 cDNA from infected plant material, showing that the Ecp6 ORF is 669 bp, interrupted by two introns of 68 and 111 bp, respectively, and encodes a protein of 222 amino acids (Fig. 7).

Figure 7.

Allelic variation of the Cladosporium fulvum Ecp6 gene. Open reading frames are shown as light grey boxes and introns as black boxes. The predicted signal peptide is indicated as dark grey box. The white flag indicates a single-nucleotide polymorphism (SNP) that leads to an amino acid substitution in the Ecp6 protein. Silent mutations are indicated by a T. The figure is drawn at scale.

The full-length sequence of Ecp6 was obtained from a collection of 50 C. fulvum strains (Table S1). Analysis of the sequence 62 bp upstream of the start codon to 91 bp downstream of the stop codon revealed that variation within Ecp6 was very limited, resulting in a total of five single-nucleotide polymorphisms (SNPs) within these strains (Fig. 7). One SNP (G > A at 494 bp downstream of the putative start codon) occurred inside the second intron of Ecp6, and was only detected in one Canadian strain (#34; Table S1). The other four SNPs all occurred in seven strains originating from North America (#31, #34, #40, #41; Table S1), and Japan (#67, #71, #74; Table S1). While one SNP (G > A at 128 bp) occurred in the first intron, two other SNPs are silent mutations (C > T at 335 bp and G > A at 662 bp). Only one SNP (C > A at 142 bp) is predicted to result in an amino acid substitution (Thr25 > Asn; Fig. 7).

Orthologues of Ecp6 are found in several fungal species

Interrogation of the C. fulvum Ecp6 protein sequence using blastp (Altschul et al., 1997; Schäffer et al., 2001) and Pfam analysis (Finn et al., 2008) indicates that the Ecp6 protein contains three lysin motif (LysM) domains. These domains are widespread protein modules of approximately 40 amino acids, originally identified in a bacterial autolysin that degrades bacterial cell walls (Joris et al., 1992). LysM domains are also found in eukaryotic proteins, and presently LysM domains are implicated in binding of diverse carbohydrates that occur in bacterial peptidoglycan, fungal chitin and Nod factor signals that are produced by Rhizobium bacteria during the initiation of root nodules on legumes (Butler et al., 1991; Amon et al., 1998; Ponting et al., 1999; Bateman and Bycroft, 2000). We queried all available fungal genome sequences and expressed sequence tag (EST) libraries (Table 2) for the presence of Ecp6-like proteins using blastp or tblastn respectively. The retrieved sequences were subsequently analysed for predicted protein domains using HMMER (http://hmmer.janelia.org/) loaded with the current Pfam HMM library (http://pfam.sanger.ac.uk). Prediction of significant LysM domains (E-value cut-off 0.001) was used as a selection criterion for further analysis. Subsequently, all sequences containing predicted LysM domains were aligned, permitting for the selection of fungal proteins with high overall similarity to C. fulvum Ecp6. In this way, a list of 16 putative C. fulvum Ecp6-like proteins was generated, containing five Aspergillus niger proteins, two Magnaporthe grisea proteins, and one from each Mycosphaerella fijiensis, M. graminicola, Botrytis cinerea, Sclerotinia sclerotiorum, A. nidulans, A. oryzae, A. flavus, C. lindemuthianum and Leptosphaeria maculans. For these 17 proteins, using clustalw (Chenna et al., 2003) a multiple sequence alignment analysis was performed (Fig. S2). In addition to the LysM domains, the positions of the cysteine residues that flank the LysM domains, and the high abundance of proline, serine and threonine residues in the LysM linker regions appear to be conserved (Fig. S2). Subsequently a neighbour-joining tree (Saitou and Nei, 1987) was constructed to reveal evolutionary relationships (Fig. 8). Based on this tree, the 16 Ecp6-like proteins can be divided into three groups. C. fulvum Ecp6 clusters with three Ecp6-like proteins of M. graminicola, M. fijiensis and L. maculans that all contain three LysM domains (Group 1, Fig. 8). The second group of Ecp6-like proteins encompasses the two M. grisea Ecp6-like proteins and CIH1 from C. lindemuthianum that are shorter than other Ecp6-like proteins and have only two LysM domains (Group 2, Fig. 8). The largest group of Ecp6-like proteins, encompassing the five A. niger proteins in addition to those of A. nidulans, A. oryzae, S. sclerotiorum and B. cinerea, contain two LysM domains and a weak, but not significant, signature of a third LysM domain (Group 3, Fig. 8).

Table 2.  Fungal whole genome and EST sequence libraries screened for Ecp6-like sequences.
SpeciesaNo. of hitsbLysMc
  • a. 

    LysM-containing Ecp6-like proteins of the species indicated in bold are included in the alignment shown in Fig. 8. The asterisks indicate species for which a LysM-containing Ecp6-like protein is identified, but as the overall homology of these proteins to C. fulvum Ecp6 and C. lindemuthianum CIH1 is low they are not included in the alignment shown in Fig. 8.

  • b. 

    blastp and tblastn searches (E-value < 0.001) were performed using Cladosporium fulvum Ecp6 and Colletotrichum lindemuthianum CIH1 as queries.

  • c. 

    Sequences were analysed using HMMER (http://hmmer.janelia.org) loaded with the current Pfam HMM library (http://pfam.sanger.ac.uk) for the presence of LysM domains (E-value < 0.001). Numbers between brackets indicate how many of the hits contain a predicted LysM domain. N.a. is not applicable.

Cladosporium fulvum (Ecp6)1Yes
Colletotrichum lindemuthianum (CIH1)1Yes
EST sequences
 Alternaria brassicicola0N.a.
 Blumeria graminis0N.a.
 Colletotrichum gloeosporioides f. sp. aeschynomene0N.a.
 Colletotrichum trifolii0N.a.
 Fusarium sporotrichioides0N.a.
 Leptosphaeria maculans1Yes
 Ophiostoma novo-ulmi0N.a.
 Phycomyces blakesleeanus0N.a.
Whole genome sequences
 Aspergillus flavus2Yes (1)
 Aspergillus nidulans1Yes
 Aspergillus niger5Yes (5)
 Aspergillus oryzae1Yes
 Batrachochytrium dendrobatidis*1Yes
 Botrytis cinerea1Yes
 Candida sp.0N.a.
 Chaetomium globosum*3Yes
 Cryphonectria parasitica*1Yes
 Cryptococcus neoformans*1Yes
 Fusarium graminearum0N.a.
 Fusarium oxysporum1No
 Fusarium verticilliodes0N.a.
 Histoplasma capsulatum*1Yes
 Laccaria bicolor0N.a.
 Lodderomyces elongisporus0N.a.
 Magnaporthe grisea2Yes (2)
 Mycosphaerella fijensis1Yes
 Mycosphaerella graminicola1Yes
 Nectria haematococca0N.a.
 Neurospora crassa1No
 Phanerochaete chrysosporium0N.a.
 Pichia stipitis0N.a.
 Podospora anserina1No
 Postia placenta0N.a.
 Sclerotinia sclerotiorum1Yes
 Stagonospora nodorum*1Yes
 Sporobolomyces roseus0N.a.
 Ustilago maydis1No
Figure 8.

Homologues of Ecp6 in other fungal species. Neighbour-joining tree of 17 Ecp6-like sequences from different fungal species. The evolutionary history of Ecp6-like protein sequences was inferred by neighbour-joining analysis (Saitou and Nei, 1987) and bootstrap values (%) are indicated at the nodes. The tree is drawn to scale, with branch lengths representing evolutionary distances. The positions containing alignment gaps were eliminated in pair-wise sequence comparisons. A total of 220 positions were calculated in the final data set.

Homology modelling of Ecp6 LysM domains

Although LysM domains have been identified in over 1500 proteins, the three-dimensional (3D) structure of only three LysM domains has been reported. Two of these are of bacterial origin, the 3D structure of a LysM domain of the Escherichia coli membrane-bound lytic murein transglycosylase D (MltD; PDB code: 1EOG; Bateman and Bycroft, 2000) and the LysM domain of the Bacillus subtilis spore protein ykuD of unknown function (PDB code: 1Y7M; Bielnicki et al., 2006). Recently, the 3D structure of the LysM domain of the human hypothetical protein SB145 was determined using nuclear magnetic resonance (NMR) imaging (PDB code: 2DJP). The structural organization of the three LysM domains from these different proteins is highly similar, and characterized by a βααβ fold, with the two helices stacking on one side of the plate generated by a double-stranded antiparallel β-sheet.

Recently, the first characterization of an interaction of a LysM domain with its ligand was reported (Ohnuma et al., 2008). Binding of oligomers of N-acetylglucosamine [(GlcNAc)n], a monosaccharide derivative of glucose that is a building block for bacterial peptidoglycan and fungal chitin, to the LysM domains of a chitinase from Pteris ryukyuensis was monitored with NMR spectroscopy. The stoichiometry of (GlcNAc)n/LysM binding was found to occur in a 1:1 ratio. Furthermore, using (GlcNAc)5 it was shown that binding of this oligomer to the LysM domain occurs at a shallow groove formed by the N-terminal part of helix 1, the loop between strand 1 and helix 1, the C-terminal part of helix 2, and the loop between helix 2 and strand 2.

To predict the ligand binding site with corresponding binding specificities of the C. fulvum Ecp6 LysM domains, homology-based modelling based on the 3D structure of the LysM domain of the MltD structure was performed. The MltD and Ecp6 LysM domains show moderate but significant overall sequence similarity (53%, 47% and 33%, respectively, for LysM domains 1, 2 and 3; Fig. 9A). Moreover, by assessing local Kyte–Doolittle (KD) hydrophobicity values (Kyte and Doolittle, 1982), the conserved secondary structure could be predicted, which was subsequently used to predict the 3D structure. The predicted 3D structure of the three individual Ecp6 LysM domains is highly similar, with small changes in the position of the second loop of the third LysM domain (Fig. 9B). Moreover, due to sufficient similarity (52%) of LysM domain 1 of C. fulvum Ecp6 to LysM domains 1 and 2 of P. ryukyuensis PrChi-A (Fig. 9A), ligand binding can be modelled according to the interaction between chitin oligomers and PrChi-A LysM domains. The molecular surface of the first LysM domain of Ecp6 (Fig. 9B, panel 1) was computed and is shown in panel 4 of Fig. 9B. In the surface of this LysM domain, a cavity is observed that fulfils the requirements to act as binding site of chitin oligomers, based on the structural homology with PrChi-A.

Figure 9.

Homology models for the LysM domains of Cladosporium fulvum Ecp6.
A. Alignment of the individual LysM domains of C. fulvum Ecp6 (this study), Escherichia coli MltD (Bateman and Bycroft, 2000) and Pteris ryukyuensis PrChi-A (Ohnuma et al., 2008). Identical amino acid residues are shaded in black and similar residues (75% threshold according to Blosum62 score) are shaded in grey.
B. LysM domains modelled based on the MltD LysM solution structure (Bateman and Bycroft, 2000). Panels 1, 2 and 3 display the three-dimensional ribbon structures of the Ecp6 LysM domains 1, 2 and 3 respectively. Panel 4 shows the computed molecular surface of Ecp6 LysM domain 1. The arrow indicated in panel 1 indicates the direction of looking to obtain the view in panel 4. The arrow in panel 4 indicates the shallow groove described as the site of interaction of PrChi-A with chitin oligomers (Ohnuma et al., 2008).

Discussion

In this study, we employed a combined 2D-PAGE and proteomics approach to identify C. fulvum proteins produced and accumulating in compatible as compared with incompatible interactions. It was anticipated that proteins that accumulated exclusively in the compatible interaction were predominantly of pathogen origin. Indeed in this way, three novel extracellular C. fulvum proteins (CfPhiA, Ecp6 and Ecp7) could be identified in addition to the previously described effector proteins Ecp1, Ecp2 and Ecp5 (Table 1). CfPhiA was found to have homology to the PhiA protein from A. nidulans which is important for phialide and conidium development (Melin et al., 2003). Several attempts to generate RNAi transformants for CfPhiA have failed (data not shown), which suggests that silencing of CfPhiA might be detrimental or even lethal. Furthermore, the expression pattern of CfPhiA compared with that of Avr9 suggests that CfPhiA is likely not a genuine effector of the fungus.

Of the newly identified fungal extracellular proteins, Ecp7 especially resembles the previously identified Avrs and Ecps. It is relatively small (the mature protein contains 100 amino acid residues of which six are cysteines) with a calculated molecular mass of approximately 11 kDa. The even number of cysteine residues suggests their involvement in disulphide bridges that aid in their stability and activity in the harsh protease-rich apoplast (Joosten et al., 1997; Kooman-Gersmann et al., 1997; Thomma et al., 2005). In addition, the Ecp7 expression profile during infection of tomato resembles that of other effector genes (Fig. 5). However, like for most of the previously identified Avrs and Ecps, the Ecp7 amino acid sequence did not show significant homology to sequences present in public databases. Despite the use of multiple transformants generated with two different RNAi constructs to target Ecp7 expression, we have not been able to obtain unambiguous evidence showing that Ecp7 is a virulence factor of C. fulvum (data not shown). This is in contrast to the findings for C. fulvum Ecp6.

The mature Ecp6 protein contains 199 amino acids and has an estimated molecular mass of 21 kDa, making it the largest of the abundantly secreted effector proteins of C. fulvum identified so far. Previous studies on the genes encoding secreted C. fulvum effectors have shown that Avr genes accumulated considerably more polymorphisms than Ecp genes (Stergiopoulos et al., 2007a). This was suggested to be due to the lack of selection pressure imposed on the pathogen to overcome resistance mediated by R genes that recognize Ecps, as these have not been deployed yet in commercial tomato lines (Stergiopoulos et al., 2007a). In line with these findings, polymorphisms in Ecp6 were only rarely observed. Of the 50 C. fulvum strains, only seven strains contained allelic variants of Ecp6. All seven of these strains, which have previously been shown to be related (Stergiopoulos et al., 2007b), contained the same four SNPs, while one strain contained an additional fifth SNP. Of these five SNPs, only one resulted in an amino acid change, while the four others concerned silent or intron mutations. The occurrence of mostly synonymous modifications in Ecp genes was hypothesized to imply selective constraints for maintaining Ecp protein sequences or, alternatively, a recent common ancestor gene (Stergiopoulos et al., 2007a). However, our finding that Ecp6 markedly contributes to C. fulvum virulence, and that Ecp6 has orthologues in other fungal species, favours the second hypothesis.

The Ecp6 protein contains three lysin motifs (LysM domains) that were originally found in a variety of enzymes that bind to and hydrolyse peptidoglycans present in bacterial cell walls, of which lysozyme is the best-known example (Joris et al., 1992; Kariyama and Shockman, 1992; Ruhland et al., 1993; Birkeland, 1994; Longchamp et al., 1994). More recently, LysM motifs have been found to occur in plant plasma membrane receptors (Zhang et al., 2007), where they have so far been implicated in two different types of interactions with microbes (Knogge and Scheel, 2006). LysM receptor kinases are involved in the perception of oligosaccharide nodulation (Nod) factors secreted by Rhizobium bacteria to establish a symbiosis with their legume hosts (Limpens et al., 2003; Madsen et al., 2003; Radutoiu et al., 2003; 2007; Arrighi et al., 2006; Smit et al., 2007). LysM receptors also function in chitin signalling in plant innate immune responses against fungal pathogens (Kaku et al., 2006; Miya et al., 2007; Wan et al., 2008). For example, an insertion in the LysM-containing receptor-like kinase gene CERK1 (also known as LysM RLK1) resulted in loss of ability to respond to the chitin elicitor (GlcNAc)8 or crab shell chitin, as measured by production of reactive oxygen species, MAP kinase signalling and induction of chitooligosaccharide-responsive genes. Moreover, enhanced susceptibility towards the fungal pathogens Alternaria brassicicola and Erysiphe cichoracearum was observed for these mutants, showing that this LysM-containing receptor-like kinase is required for chitin signalling in plant innate immune responses (Miya et al., 2007; Wan et al., 2008).

LysM domains are also found in different chitinases from various organisms (Amon et al., 1998; Ponting et al., 1999). The involvement of LysM proteins in perception of chitin (β-1,4-linked poly N-acetyl-d-glucosamine), peptidoglycan (a heteropolymer with alternating units of N-acetyl-d-glucosamine and N-acetyl-muramic acids) and the N-acetyl-d-glucosamine backbone of Nod factors supports a role for LysM domains in binding of the N-acetyl-d-glucosamine oligosaccharide. Using domain swaps between Nod factor receptors, it was demonstrated that these receptors mediate specific perception of Nod factors from different Rhizobium bacteria and that this recognition depends on the structure of the Nod factor (Radutoiu et al., 2007). Moreover, a single-amino-acid change in one of the LysM domains resulted in altered Nod factor recognition, strongly suggesting that the LysM domains constitute the binding domains for the lipochitinoligosaccharide Nod factors (Radutoiu et al., 2007). A high-affinity chitin-binding protein was isolated from the plasma membrane of suspension-cultured rice cells. This extracellular membrane-anchored protein, CEBiP, contains two LysM domains. Knockdown of CEBiP expression diminished the elicitor-induced oxidative burst as well as expression of chitin-induced genes. Moreover, binding assays as well as affinity labelling showed that the plasma membrane of knockdown lines for CEBiP carried less elicitor binding sites (Kaku et al., 2006).

Chitin binding has also been demonstrated for the C. fulvum effector protein Avr4 which contains an invertebrate chitin-binding domain (van den Burg et al., 2003). Through this chitin-binding activity, Avr4 was found to protect C. fulvum hyphae from hydrolysis by plant chitinases (van den Burg et al., 2006; van Esse et al., 2007). It is tempting to speculate that C. fulvum Ecp6 is a chitin-binding protein too. To that end it is interesting to note that the Ecp6 homologue CIH1 of the plant pathogenic fungus C. lindemuthianum is found to be present at the surface of intracellularly growing fungal structures present in infected plant tissue (Perfect et al., 1998). Ecp6 may potentially act as a functional homologue of Avr4 through the ability to bind to chitin. Such functional redundancy might explain why the C. fulvum strain Can38, which harbours a frameshift mutation in the Avr4 gene and as a consequence does not produce Avr4, is still able to infect tomato (Joosten et al., 1997). Alternatively, Ecp6 may act as a ‘stealth factor’ by shielding fungal hyphae in a similar fashion as has been suggested for hydrophobins (Whiteford and Spanu, 2002). Furthermore, the fungus may avoid recognition by the plant by sequestering chitin mono- or oligomers that act as elicitors of defence responses once they are released by the activity of plant chitinases.

Using homology modelling, docking sites for the interaction between LysM domains and their ligands have been predicted (Mulder et al., 2006; Radutoiu et al., 2007). We have used homology modelling to show that the Ecp6 LysM domains are likely to structurally resemble previously characterized LysM domains, and that based on structural calculations GlcNAc oligomers may indeed dock to the LysM domains of Ecp6 in a similar fashion as to the LysM domains of PrChi-A (Ohnuma et al., 2008). Future experiments will reveal whether the LysM domains of Ecp6 are able to bind chitin and, moreover, how Ecp6 contributes to fungal virulence.

Experimental procedures

Fungal and plant materials, and infection assays

The wild-type race 5 strain of C. fulvum was stored in 50% glycerol at −80°C until revitalized on potato dextrose agar (PDA; Oxoid, Hampshire, England) and was grown at room temperature in the dark. Two-week-old C. fulvum PDA plate cultures were used to harvest conidia by adding sterile water to the plates and rubbing the surface with a sterile glass rod to release the conidia. Conidial suspensions were filtered through Miracloth (Calbiochem-Behring, La Jolla, CA), centrifuged at 4000 r.p.m. and washed twice with sterile water after which the conidial concentration was determined. Subsequently, the conidia were used for plant inoculations or Agrobacterium tumefaciens-mediated transformation.

All tomato plants were grown under standard greenhouse conditions: 21°C during the 16 h day period, 19°C at night, 70% relative humidity (RH) and 100 Watt m−2 supplemental light when the sunlight influx intensity was below 150 Watt m−2. The tomato (S. esculentum) cultivar MoneyMaker, containing no resistance genes against C. fulvum (MM-Cf-0), and a MoneyMaker near isogenic line containing the Cf-4 locus (MM-Cf-4) were used for all inoculations. C. fulvum was inoculated as described previously (de Wit, 1977). Per 5-week-old tomato plant, 5 ml of conidial suspension (1 × 106 conidia per ml) was used for spray inoculation on the lower surface of the leaves until drop-off. Plants were kept at 100% RH under a plastic cover for 48 h after inoculation. All experiments, starting from plant inoculations, were repeated at least twice.

Preparation of protein samples and 2D-PAGE

Leaves were harvested from C. fulvum-infected MM-Cf-0 and MM-Cf-4 lines at 14 dpi and apoplastic fluid (AF) was isolated by vacuum infiltration (van Esse et al., 2006) using de-mineralized water followed by centrifugation for 5 min and stored at −20°C until further analysis. AF from both interactions was freeze-dried and the residue was re-suspended in 3.5 ml of water. After centrifugation (10 min at 4000 g) samples were desalted using a PD-10 desalting column (GE Healthcare, UK), freeze-dried again and stored at −20°C. Freeze-dried protein samples were dissolved in 340 μl of Rehydration Buffer [7 M urea, 2 M thiourea, 4% CHAPS, 60 mM DTT, 0.002% (w/v) bromophenol blue] along with 3.4 μl of IPG buffer pH 4–7 (GE Healthcare). The samples were vortexed briefly and centrifuged (10 min at 4000 g). The protein samples were applied to Immobiline DryStrips of 18 cm with a non-linear pH 4–7 gradient (GE Healthcare), covered with paraffin oil and allowed to re-hydrate overnight at room temperature. Isoelectric focusing was performed using the Ettan IPGphor electrophoresis apparatus (GE Healthcare) at 20°C maintaining 50 μA per strip. A total focusing of 70 k Vh was achieved by following a running protocol using a step-n-hold gradient (1.5 h 0–3500 V, 6 h 3500 V). After first dimensional isoelectric focusing, the strips were stored at −20°C.

Subsequently, strips were placed in equilibration buffer [EB; 50 mM Tris, pH 8.8, 6 M urea, 30% (v/v) glycerol and 2% (w/v) SDS] supplemented with 65 mM DTT. After 15 min, the buffer was replaced by EB supplemented with 135 mM iodoacetamide, and the strips were incubated for another 15 min. The proteins were subsequently separated on 12.5% polyacrylamide gels; the gels were run at 70 V for the first 30 min and subsequently at 200 V until the bromophenol blue reached the bottom of the gels. Gels were stained with Coomassie brilliant blue overnight and de-stained with 10% ethanol and 7.5% HAc in water.

Mass spectrometry

Protein spots were excised from the gel and digested with trypsin with an in-gel method (Shevchenko et al., 1996). The collected extracts of the resulting tryptic peptides were freeze-dried and stored at −20°C. The peptides were re-dissolved in 8 μl of 50% acetonitrile, 5% formic acid. MS and MS/MS information was acquired with a Q-Tof1 (Waters, Manchester, UK) coupled with a nano-LC Ultimate system (LC Packings Dionex, Sunnyvale, CA). After the dilution of 1–2 μl of sample 12 times with water, peptides were separated on a nano-analytical column (75 μm inside diameter × 15 cm C18 PepMap, LC Packings, Dionex) using a gradient of 2–50% acetonitrile, 0.1% formic acid in 20 min. The flow of 300 nl min−1 was directly infused into the Q-Tof1, operating in data-dependent MS and MS/MS modes. The resulting MS/MS spectra were processed with Masslynx software (Waters, Manchester, UK) and used to search in MASCOT using the MSDB database. As sequence data of both C. fulvum and tomato are far from complete, MS/MS data from un-assigned spectra were analysed by using the Masslynx Pepseq software for de novo sequence information. Both blast (http://www.expasy.org/tools/blast) and msblast were used to search for possible homologous proteins with the generated sequence information. For MALDI-TOF analysis, a 1 μl volume was spotted on a target plate after mixing the samples 1:1 (v/v) with a solution of 10 mg ml−1α-Cyano-4-hydroxycinnamic acid in 50% ethanol/50% acetonitrile/0.1% TFA. Reflectron MALDI-TOF spectra were acquired on a TofSpec 2E (Waters, Manchester, UK). For peptide mass fingerprinting the resulting peptide mass lists were used to search in MASCOT using the same MSDB database.

Cloning of CfPhiA, Ecp6 and Ecp7

Based on the N-terminal CfPhiA sequence MDPIDVVWK, the forward degenerate primer Deg-PhiA along with an oligo-(dT) primer (Table S2) was used to isolate the CfPhiA coding sequence. Likewise, degenerate forward primers (Table S2) were designed matching the ETKATDCG and QITTQDFG sequences from the N-terminal sequences of Ecp6 and Ecp7 respectively. Using the degenerate primers and a poly T primer PCR products were amplified from a cDNA library derived from a compatible interaction between C. fulvum and tomato using the high fidelity polymerase ExTaq (Takara, Shiga, Japan). Products were cloned into the pGEM-T Easy vector (Promega, Madison, WI) and sequenced.

Construction of plasmids for RNAi in C. fulvum

Two constructs for overexpression of inverted-repeat constructs for RNAi based on two different parts of the Ecp6 coding sequence were generated. For the first RNAi construct targeting the 3′ end of Ecp6, 218 bp of Ecp6 was PCR-amplified from cDNA using a forward primer that added an NcoI restriction site to the 5′ end (Ecp6i-F) and a reverse primer that added EcoRI and NotI restriction sites to the 3′ end (Ecp6i-R; Table S2). PCR reactions were carried out under the following conditions: an initial denaturation step for 2 min followed by denaturation for 15 s at 94°C, annealing for 30 s at 55°C and extension for 1 min at 72°C for 30 cycles, followed by a final elongation step at 72°C for 5 min. PCR products were separated on 1% agarose gels and were purified using the DNeasy kit (Qiagen, Valencia, CA). Subsequently, PCR products were cloned into the pGemT-Easy vector. Vectors were digested with NcoI and NotI or with NcoI and EcoRI. Both digested inserts were cleaned from gel using the QIAquick gel extraction kit (Qiagen) and subsequently ligated with a NotI- and EcoRI-digested 129 bp spacer segment from the Pichia pastoris Aox-1 gene into the NcoI-digested plasmid pFBB302 (Dr Brandwagt, Wageningen University). The plasmid pFBB302 is constructed in the backbone of the pGreen II binary vector (Hellens et al., 2000) and contains a nourseothricin resistance cassette (Malonek et al., 2004) to select for fungal transformants, and the UidA reporter gene flanked by the constitutive ToxA fungal promoter (Ciuffetti et al., 1997) and trpC terminator (Punt et al., 1987). Digestion with NcoI releases the UidA coding sequence and allows ligation of the inverted-repeat RNAi sequence.

For the second RNAi construct targeting the 5′ end of Ecp6, two Ecp6 PCR products were generated of 250 and 318 bp with the same forward primer that added an EcoRI restriction site to the 5′ end (Ecp6i2-F) and two different reverse primers that added a NotI restriction sites to the 3′ end (Ecp6i2 k-R and Ecp6i2 l-R respectively; Table S2). PCR reactions and gel cleaning were performed similar as for the first RNAi construct. Subsequently, PCR products were cloned into the pGemT-Easy vector, digested with NotI and EcoRI, cleaned from gel and ligated into the EcoRI-digested plasmid pFBT004. The plasmid pFBT004 is a modified version of pFBB302, in which the nourseothricin resistance cassette is replaced by a hygromycin resistance cassette (Punt et al., 1987).

A. tumefaciens-mediated transformation of C. fulvum

RNAi plasmids were transformed into A. tumefaciens strain LBA1100 [containing the binary vector pSoup (Hellens et al., 2000)] by electroporation. A 3 ml culture of A. tumefaciens was grown overnight in 1× YT (Sambrook and Russell, 2001) supplemented with kanamycin (25 μg ml−1). The following day, the culture was centrifuged and re-suspended in 50 ml of fresh minimal medium (MM) (Hooykaas et al., 1979) supplemented with kanamycin (25 μg ml−1) and grown overnight. The following day, the culture was centrifuged and re-suspended in 10 ml of fresh MM. One millilitre of re-suspended bacteria was used to inoculate 50 ml of induction medium [IM; MM salts plus 40 mM 2-(N-morpholino)ethanesulphonic acid (MES), pH 5.3, 10 mM glucose and 0.5% (w/v) glycerol] supplemented with 200 μM acetosyringone (AS) and was grown for an additional 4–5 h until the culture reached an optical density (OD600) of 0.25. At that point, the A. tumefaciens culture was centrifuged and re-suspended in 10 ml of sterile water. In addition, while A. tumefaciens cultures were growing in IM+AS medium, C. fulvum conidia were harvested and subsequently suspended in 50 ml of B5 medium (Duchefa Biochemie BV, Haarlem, the Netherlands) at a concentration of approximately 1 × 106 conidia ml−1 and placed in a rotary shaker (125 r.p.m.) at room temperature to induce germination of conidia. After 4–5 h, germinated conidia were centrifuged twice at 4000 r.p.m. and re-suspended in sterile water to a final volume of 1 × 107 conidia ml−1.

Five hundred microlitres from the induced A. tumefaciens cell suspension was mixed with 10 ml of germinated conidia and plated (200 μl per plate) on a 0.45-μm-pore, 45-mm-diameter nitrocellulose filter (Whatman, Hillsboro, OR) and placed on co-cultivation medium (IM + 200 μM AS and 5 mM glucose and 1.5% technical agar). The co-cultivation mixture was incubated at 22°C for 2 days. Following incubation, the filter was transferred to PDA supplemented with 50 μg ml−1 nourseothricin (Werner BioAgents, Jena, Germany) or with 100 μg ml−1 hygromycin B (Duchefa Biochemie BV, Haarlem, the Netherlands) as a selection agent for transformants and 200 μg ml−1 cefotaxime (Duchefa Biochemie BV, Haarlem, the Netherlands) to kill A. tumefaciens cells. Individual transformants were transferred to new selection plates and incubated until conidiogenesis under normal growth conditions. Conidia from these plates were stored in 50% glycerol at −80°C until further analysis.

Real-time PCR analyses

Three leaflets were harvested from inoculated MM-Cf-0 and MM-Cf-4 plants at 3, 6, 9, 13 and 16 dpi. Leaf samples were composed of three leaflets from the second, third and fourth tomato leaves of two tomato plants taken at each time point, immediately frozen in liquid nitrogen and stored at −80°C until used for RNA analysis. A similar procedure was used for RNAi transformant analysis. Ecp6 RNAi transformants Ecp6i-1 and Ecp6i-4 along with Ecp7 RNAi transformants Ecp7i-1, Ecp7i-3 and Ecp7i-7 were randomly chosen for inoculation and analysis with the progenitor race 5 wild-type strain inoculated on MM-Cf-0 plants. Leaf samples were taken at 10 dpi, immediately frozen in liquid nitrogen and stored at −80°C until used for RNA analysis.

Total RNA was isolated from infected leaf material using the RNeasy kit (Qiagen, Valencia, CA), including an in-column DNase treatment (Qiagen) according to manufacturer's instruction. Total RNA was used for cDNA synthesis using an oligo-(dT) primer and the SuperScript II reverse transcriptase kit (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. Quantitative real-time PCR was conducted using an ABI7300 PCR machine (Applied Biosystems, Foster City, CA) with the GoldStar SYBR green PCR kit (Eurogentec, Seraing, Belgium). All primer sequences are shown in Table S2. Expression primers were designed so that the reverse primer was not included in the RNAi construct to prevent detection of the constitutively expressed RNAi construct. For the first RNAi construct, primer pair Ecp6-RNAi-RQ-F and Ecp6-RNAi-RQ-R was used, and for the second RNAi construct primer pair Ecp6-RNAi2-RQ-F and Ecp6-RNAi2-RQ-R. Real-time PCR conditions were as follows: an initial 95°C denaturation step for 10 min followed by denaturation for 15 s at 95°C, annealing for 30 s at 60°C and extension for 30 s at 72°C for 40 cycles, and analysed on the 7300 System SDS software (Applied Biosystems, Foster City, CA). To ensure no genomic DNA contaminated RNA samples, real-time PCR was also carried out on RNA without the addition of reverse transcriptase. All experiments, including leaf inoculations, were repeated twice.

Heterologous expression of C. fulvum Ecp6 in F. oxysporum f. sp. lycopersici

For C. fulvum Ecp6, the cDNA corresponding to the mature protein was amplified using primer Ecp6OE-F that also contained the sequence encoding the C. fulvum Avr4 signal peptide for extracellular targeting (Table S2). For C. fulvum Ecp7, the cDNA corresponding to the mature protein was amplified in two steps. As the 5′ coding sequence was lacking from our cDNA clone, a primer was designed to add a 5′ codon-optimized sequence stretch based on the N-terminal protein sequence (Ecp7NtermF) and used in combination with the reverse primer Ecp7OE-R (Table S2). The resulting PCR product was used as template for a second PCR with primer Ecp7OE-F that also contained the sequence encoding the C. fulvum Avr4 signal peptide for extracellular targeting and a HindIII restriction site in combination with the reverse primer Ecp7OE-R that contained a XmaI restriction site (Table S2). All PCR reactions were carried out under the following conditions: an initial denaturation step for 2 min followed by denaturation for 15 s at 94°C, annealing for 30 s at 56°C and extension for 1 min at 72°C for 30 cycles, followed by a final elongation step at 72°C for 5 min. PCR products were separated on 1% agarose gels and purified using the DNeasy kit (Qiagen, Valencia, CA). Subsequently, PCR products were cloned into the pGemT-Easy vector and sequenced. A correct clone was digested with EcoRI (for Ecp6) or HindIII and XmaI (for Ecp7), cleaned from gel, and ligated into the EcoRI- (for Ecp6) or HindIII- and XmaI- (for Ecp7) digested plasmid pFBT004. The constructs were transformed into A. tumefaciens strain LBA1100 [containing the binary vector pSoup (Hellens et al., 2000)] by electroporation essentially as described by Mersereau et al. (1990). Agrobacterium-mediated transformation of F. oxysporum f. sp. lycopersici was performed as described (Mullins et al., 2001).

Ecp6 gene walking

Three primers designed on the region encoding the mature Ecp6 protein (TSP1, TSP2 and TSP3; Table S2) were used to amplify the genomic DNA sequence upstream of the region that encodes the mature Ecp6 protein using the DNA Walking SpeedUpTM Premix Kit (Seegene, Rockville, MD) according to the manufacturer's instructions. Amplified products were cloned in the pGEM-T Easy vector (Promega, Madison, WI) and sequenced. Putative open reading frames (ORFs) were predicted using the fgenesh program (Salamov and Solovyev, 2000) of the MOLQUEST software package (available at http://softberry.com/berry.phtml; Softberry, NY, USA) using the genetic codes of several fungi present in the database as models. ORFs were verified by cloning Ecp6 cDNA. For this purpose, total RNA was isolated from leaves of MM-Cf-0 plants inoculated with a race 5 strain of C. fulvum at 11 dpi and used for cDNA synthesis using an oligo-(dT) primer (Table S2) and the SuperScript II reverse transcriptase kit (Invitrogen, Carsbad, CA) as described previously (van Esse et al., 2007). The generated cDNA was used as template for the primers Ecp6_ChrWal_F1 and Ecp6_R (Table S2) to amplify the predicted Ecp6 ORF. The primers Ecp6_F3, Ecp6_F2, Ecp6_R3, Ecp6_R2 (Table S2) that hybridized outside the predicted Ecp6 ORF were used as negative controls. The 50 μl PCR reaction mixes contained 5.0 μl of 10× SuperTaq PCR reaction buffer, 10 mM of each dNTP (Promega Benelux bv, Leiden, the Netherlands), 20 μM of each primer, 1 unit of SuperTaq DNA polymerase (HT Biotechnology, Cambridge, UK) and approximately 100 ng of cDNA as template. The PCR programme consisted of an initial 5 min denaturation step at 94°C, followed by 35 cycles of denaturation at 94°C (30 s), annealing at 55°C (30 s) and extension at 72°C (60 s). A final extension step at 72°C (7 min) concluded the reaction. Amplified products were cloned in the pGEM-T Easy vector (Promega, Madison, WI) and sequenced.

Ecp6 allelic variation

Allelic variation in Ecp6 was determined for 50 C. fulvum strains (Table S1) that are part of a previously described collection (Stergiopoulos et al., 2007a,b). Strains were cultured on half-strength PDA (Oxoid, Hampshire, England) at 22°C. Conidia were harvested from 15-day-old cultures and freeze-dried prior to DNA extraction. Genomic DNA isolations were performed using the DNeasy Plant Mini Kit (Qiagen, Valencia, CA) according to the manufacturer's instructions. The forward primer Ecp6_F3, located 424 bp upstream of the Ecp6 translation start codon, and the reverse primer Ecp6_R3, located 99 bp downstream of the Ecp6 stop codon, were used to amplify Ecp6 (Table S2). The 50 μl PCR reaction mixes contained 5.0 μl of 10× SuperTaq PCR reaction buffer, 10 mM of each dNTP (Promega Benelux bv, Leiden, the Netherlands), 20 μM of each primer, 1 unit of SuperTaq DNA polymerase (HT Biotechnology, Cambridge, UK) and approximately 100 ng of genomic DNA as template. The PCR programme consisted of an initial 5 min denaturation step at 94°C, followed by 35 cycles of denaturation at 94°C (30 s), annealing at 55°C (30 s) and extension at 72°C (60 s). A final extension step at 72°C (7 min) concluded the reaction. Amplified PCR products were excised from 0.8% agarose gels, purified using the GFXTM PCR DNA and Gel Band Purification Kit (Amersham Biosciences UK limited, Buckinghamshire, England), and sequenced using the forward primers Ecp6_F2 and Ecp6_F in combination with the reverse primer Ecp6_R3 (Table S2).

Bioinformatical analysis of Ecp6-like proteins

EST sequences from various fungal pathogens were downloaded from the COGEME Phytopathogenic Fungi and Oomycete EST Database version 1.6 (http://cogeme.ex.ac.uk)(Soanes and Talbot, 2006). The genome sequences of various fungi listed in Table 2 were consulted at the website of Fungal Genome Initiative of the Broad Institute of MIT and Harvard (http://www.broad.mit.edu/annotation/fgi/) or at the website of the USA Department of Energy Joint Genome Institute (http://genome.jgi-psf.org). The mining of Ecp6-like proteins was performed using NCBI blast, and the Standalone-blast version 2.2.3 (Altschul et al., 1997; Schäffer et al., 2001). Hmmpfam analysis of each identified candidate was performed by running a customized Perl script for Pfam HMM detection, available at ftp://ftp.sanger.ac.uk/pub/databases/Pfam, using Bioperl version 1.4 (http://bioperl.org) and HMMER version 2.3.2 (http://hmmer.janelia.org), which was loaded with the current Pfam ls and fs models (02.10.2007), for whole domain and fragment models respectively. An E-value of 0.001 was used as cut-off. The retained sequences were analysed in BioEdit version 7.0.5.3 (http://www.mbio.ncsu.edu/BioEdit/bioedit.html). Multiple sequence alignment was performed by clustalw version 1.83 and for phylogenetic tree construction Molecular Evolutionary Genetic Analysis 4.0 (mega) was used (Kumar et al., 2001; Tamura et al., 2007). Phylogeny construction of fungal Ecp6-like proteins was performed by neighbour-joining analysis. We used p-distance as the distance parameter as specified in the program mega. The inferred phylogeny was tested by 500 bootstrap replicates (Felsenstein, 1985).

Three-dimensional modelling was performed using the Protein Homology/analogY Recognition Engine (Phyre), a Protein fold recognition server (http://www.sbg.bio.ic.ac.uk/~phyre/; Kelley et al., 2000; Bennett-Lovsey et al., 2008). Estimated precision of generated models was used as an indication of significance. Subsequent analyses, visualization and preparation of 3D figures were performed in the Swiss-PdbViewer version 3.7 (http://www.expasy.org/spdbv).

Acknowledgements

The authors thank Bert Essenstam and Henk Smid at Unifarm for excellent plant care. We acknowledge Bas Brandwagt and John van't Klooster for providing materials and assistance and three anonymous reviewers for useful suggestions. This project was carried out within the research programme of the Centre of BioSystems Genomics (CBSG) which is part of the Netherlands Genomics Initiative/Netherlands Organization for Scientific Research. B.P.H.J.T. is supported by a Vidi grant of the Research Council for Earth and Life sciences (ALW) of the Netherlands Organization for Scientific Research (NWO). I.S. is supported by ERA-PG project ERA-PG 31855.00030.

Ancillary