•Proteins belonging to the cerato-platanin family are small proteins with phytotoxic activity. A member of this family, BcSpl1, is one of the most abundant proteins in the Botrytis cinerea secretome.
•Expression analysis of the bcspl1 gene revealed that the transcript is present in every condition studied, showing the highest level in planta at the late stages of infection. Expression of a second cerato-platanin gene found in the B. cinerea genome, bcspl2, was not detected in any condition.
•Two bcspl1 knock-out mutants were generated and both showed reduced virulence in a variety of hosts.
•bcspl1 was expressed in Pichia pastoris and the recombinant protein was able to cause a fast and strong necrosis when infiltrated in tomato, tobacco and Arabidopsis leaves, in a dose-dependent manner. The BcSpl1-treated plant tissues showed symptoms of the hypersensitive response such as induction of reactive oxygen species, electrolyte leakage, cytoplasm shrinkage, and cell autofluorescence, as well as the induction of defense genes considered to be markers of the hypersensitive response. The Arabidopsis bak1 mutation partially prevented the induction of necrosis in this plant by BcSpl1. Two different BcSpl1-derived 40-amino acids peptides were also active in inducing necrosis.
The fungal plant pathogen Botrytis cinerea has a necrotrophic life style and a wide host range, being able to infect a large variety of plant species, including many important crops (Elad et al., 2004). Its invasion strategy has been proposed (Govrin & Levine, 2000; van Kan, 2006; Choquer et al., 2007; Williamson et al., 2007) to include the active induction of the plant defense response known as the hypersensitive response (HR), a form of plant cell death (Mur et al., 2008), in the tissues surrounding the lesion. The hypersensitive response is an effective defense against biotrophic plant pathogens, restricting access to water and nutrients, but can be exploited by necrotrophic pathogens such as B. cinerea to generate dead tissue around the infected area (Govrin & Levine, 2000).
A number of metabolites and proteins secreted by B. cinerea have been shown to cause cell death when applied to plant tissues and some of them have also been shown to induce symptoms of HR, such as oxalic acid (Kim et al., 2008), NLPs (Schouten et al., 2008) and Xyn11A (Noda et al., 2010). A contribution of this necrotizing activity to the infection process has only been shown for Xyn11A, which has differentiated xylan-degrading and necrotizing activities, and contributes to virulence with the latter.
We have recently identified 105 proteins secreted by B. cinerea early after germination, in conditions that resemble plant infection (Espino et al., 2010). Among them, one of the most abundant proteins was BcSpl1, a protein of the cerato-platanin family that has also been shown to be induced by ethylene (Chaguéet al., 2006). BcSpl1 was present in every condition studied for the early secretome, and it was quite abundant especially under induction with tomato extract, being the sixth most abundant protein. In two previous works studying the proteins secreted by fully mature mycelium of B. cinerea (Shah et al., 2009a,b), BcSpl1 was also found in every culture condition, and it was especially abundant in the secretome induced with plant extracts such as tomato and strawberries.
The cerato-platanin family of proteins (Pfam family PF07249, http://pfam.janelia.org/family/PF07249), sometimes referred to as snod-like proteins, is composed of small proteins of c. 120 amino acids that are distantly related to hydrophobins. The ‘founder member’ of this family is the cerato-platanin produced by Ceratocystis fimbriata (Pazzagli et al., 1999), an ascomycete that is pathogenic for the European plane tree (Platanus acerifolia). We therefore use here the term cerato-platanin as a generic name for the members of this protein family.
Cerato-platanins are all extracellular proteins (Pazzagli et al., 1999; Espino et al., 2010), although the C. fimbriata cerato-platanin has also been detected in the fungal cell wall by immunological methods (Boddi et al., 2004). They are all moderately hydrophobic, with four conserved cysteine residues that have been experimentally proven to form two disulfide bridges (Pazzagli et al., 1999). Cerato-platanins from two Ceratocystis species (Pazzagli et al., 1999; Comparini et al., 2009) lack any posttranslational modification, apart from the removal of the signal sequence and the formation of the disulfide bridge, as judged from the agreement between the predicted and empirically calculated molecular weights. By contrast, cerato-platanin from Trichoderma virens is glycosylated (Vargas et al., 2008).
Cerato-platanins have been shown to induce plant defenses and, on occasions, localized necrosis in the treated area. The C. fimbriata protein induces autofluorescence and cell death in tobacco leaves, in a dose dependent manner (Pazzagli et al., 1999), as well as transcriptional changes in 78 plane tree genes which include genes involved in defense and stress responses (Fontana et al., 2008). Treatment with cerato-platanin induces the ability to inhibit germination of C. fimbriata conidia in plane tree leaves (Fontana et al., 2008). The protein from T. virens is also able to induce plant defense genes and the production of hydrogen peroxide (Djonovic et al., 2006; Buensanteai et al., 2010), but no necrosis. The cerato-platanin Sm1 from Leptosphaeria maculans induces autofluorescence in Brassica napus leaves, but no necrotizing activity was reported (Wilson et al., 2002). MSP1, a cerato-platanin from Magnaporthe grisea, was initially reported not to be able to induce any response in plants (Jeong et al., 2007), although later studies showed that transgenic Arabidopsis thaliana plants expressing MSP1 become necrotic when the expression of the transgene was turned on (Yang et al., 2009). As far as we know, the mechanism of phytotoxicity is completely unknown for any member of the cerato-platanin family.
Genes coding for cerato-platanin have been knocked-out in two fungi. In the case of L. maculans, the mutants were as virulent as the wild-type strain (Wilson et al., 2002). The deletion of the M. grisea msp1 gene, however, caused a significant reduction of the virulence (Jeong et al., 2007), which was attributed to a lower capacity to extend the infection as the ability to penetrate the plant tissue was not perturbed.
Here we show that the cerato-platanin BcSpl1 is required for virulence in Botrytis cinerea and that its contribution to the infection process may be related with its ability to induce the hypersensitive cell death in the host.
Materials and Methods
Organisms, strains and culture conditions
Botrytis cinerea strain B05.10 (Büttner et al., 1994) was used as wild-type and control strain. Fungal strains were kept as conidia in 15% glycerol at −80°C for long time storage, or in silica gel at 4°C for routine use (Delcan et al., 2002). The silica stock was used to inoculate tomato-agar plates (25% tomato fruit extract, 2% agar, pH 5.5) to obtain conidia as described by Benito et al. (1998). Tomato (Lycopersicon esculentum cv Moneymaker), tobacco (Nicotiana tabacum cv Havana) or Arabidopsis plants were maintained at controlled temperature, humidity, and photoperiod in a phytotron. The following Arabidopsis seeds were obtained from the Nottingham Arabidopsis Stock Centre: Columbia ecotype Col-O (stock id. N1092), used as wild type, and the two bak1 mutants bak1-1 (stock id. N6125) and bak1-3 (stock id. N662069). The last of these is a homozygous line (SALK_034523C) derived from the SALK T-DNA insertion line SALK_034523 (Alonso et al., 2003). Arabidopsis seeds were germinated in solid 0.5× Murashige and Skoog (MS) basal salt medium (Duchefa, Haarlem, the Netherlands).
Standard molecular techniques
Recombinant DNA methods were performed as described by Sambrook & Russell (2001). Genomic DNA from B. cinerea was extracted using a method previously developed in our laboratory (González et al., 2008). Polymerase chain reaction amplifications were made using Phusion DNA polymerase (Finnzymes, Keilaranta, Finland) when the product was to be cloned. All other PCRs were made with Taq DNA polymerase (GenScript, Picataway, NJ, USA). Oligonucleotides were from Invitrogen (Paisley, Scotland) and their nucleotide sequences are indicated in Table 1. Peptides were from GenScript.
Table 1. Oligonucleotide primers used in this study
aOligonucleotides used for quantitative real-time (Q-RT-PCR) of the following Botrytis cinerea genes: actA (ACTAFW and ACTARV), bcspl1 (RNACP5Pri and RNACP3Pri), and bcspl2 (HOMOCPT-FW and HOMOCPT-RV).
bOligonucleotides used for Q-RT-PCR of the corresponding Nicotiana tabacum genes: Tac9 (actin), HIN1, HSR203J, PR1A, and PR5.
Botrytis cinerea transformations were carried out according to the protocol of Hamada et al. (1994), modified by van Kan et al. (1997) and transformants were then purified by single-conidia isolation on hygromycin-containing plates to ensure homokaryosis. Southern blots were carried out with digoxigenin labeled probes using the DIG DNA Labeling and Detection kit (Roche, Basel, Switzerland). two-dimensional (2D) electrophoresis and identification of the protein spots were carried out as explained previously (Espino et al., 2010). The statistical software package SPSS 17.0 (IBM, Armonk, NY, USA) was used for statistical analysis. Image analysis was done with imagej (National Institutes of Health, Bethesda, MD, USA).
Generation of bcspl1 mutants
To obtain mutants in the bcspl1 gene, a plasmid was constructed carrying the hygromycin resistance cassette flanked by the 5′ and 3′ regions of bcspl1 (see the Supporting Information, Fig. S1). Oligonucleotides PROPLAOUT and PROPLAIN were used to amplify a 519-bp 5′ region of the gene from B. cinerea B05.10 genomic DNA and the product was cloned in the EcoRV and HindIII sites of plasmid pLOB1 downstream of the resistance cassette. pLOB1 contains a hygromycin resistance cassette (GenBank accession no. AJ439603) in a pUC18 background. Oligonucleotides TERPLAIN and TERPLAOUT were used to amplify a 547-bp 3′ region of bcspl1 and the product was cloned in the SmaI and EcoRI sites, upstream of the resistance cassette. The resulting plasmid was transformed into B. cinerea B05.10 protoplasts. Twelve of the transformants obtained were purified by the isolation of single germinating conidia in order to isolate homokaryons. To ensure the site-directed integration of transforming DNA, a PCR reaction was performed on genomic DNA from all of them using primers CHECK RV, which binds to the hygromycin resistance cassette, and PROCEPLAT, which binds to a genomic region in the vicinity of bcspl1 not included in the transforming DNA. A second PCR reaction was performed to ensure homokaryosis with primers INIPROT and FINPROT, which bind to bcspl1 at both sides of the integration site for the hygromycin resistance cassette. Two bona fide mutants, Δbcspl1.1 and Δbcspl1.2, showed only the expected bands (Fig. S1). The disruption of bcspl1 in these two strains was also confirmed by Southern blot analysis, and no additional ectopic integrations were detected by hybridization with a probe-specific for the bcspl1 gene (Fig. S1).
The bcspl1 mutation was rescued by transformation with a plasmid carrying the whole bcspl1 open reading frame plus 1.2-kb promoter and 0.65-kb terminator regions. To construct the transforming plasmid, bcspl1 was amplified from B. cinerea genomic DNA with oligonucleotides ProCP-NotI and TerCP-XbaI, and then cloned at the indicated sites of plasmid pNR2 (Kars et al., 2005) upstream of the nourseothricin resistance cassette. The resulting plasmid was transformed into the two B. cinerea bcspl1 mutants and the transformants were analysed for the presence of the whole bcspl1 by PCR, using primers ProCP-NotI and TerCP-XbaI. One rescued strain for each bcspl1 mutant, Δbcspl1.1-Res and Δbcspl1.2-Res, were identified (not shown) and used in subsequent work.
Expression of bcspl1 in Pichia pastoris
The EasySelect Pichia Expression Kit (Invitrogen, Carlsbad, CA, USA) was used to express the protein in the yeast Pichia pastoris, according to the manufacturer’s instructions. A 357-bp fragment from bcspl1, carrying almost the whole cerato-platanin open reading frame, from the end of the predicted signal peptide to the stop codon, was amplified by PCR from B. cinerea B05.10 genomic DNA with oligonucleotides INIPROT and FINPROT, and cloned in the EcoRI and XbaI sites of plasmid pPICZαA. The recombinant protein coded by this construction contains the Pichiaα-factor signal sequence at the N-terminus of the cerato-platanin sequence and the c-myc and 6xHis epitopes at the C-terminus. The resulting plasmid, pCP, was linearized with SacI and then transformed into P. pastoris KM71H, and one of the transformants expressing bcspl1 was chosen for all subsequent work. The supernatant of a culture of this strain, induced for 2 d with 0.5% methanol in BMMH medium, was the starting material in the purification of the recombinant protein.
Purification of BcSpl1 took advantage of the 6xHis epitope, and was carried out with HisTrap FF prepacked minicolums (GE Healthcare Life Sciences, Uppsala, Sweden), following manufacturer’s recommendations. The yeast culture supernatant was adjusted to pH 8 by the addition of 1/3 vol. of 4× equilibration buffer (200 mM NaH2PO4 pH 8, 1.2 M NaCl) and run through the column at 1 ml min−1. After washing with washing buffer (50 mM NaH2PO4 pH 8.0, 300 mM NaCl, 5 mM imidazole), the protein was eluted by the addition of water (adjusted at pH 4.5) and 1 ml fractions were collected. Sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) was used to test for the presence of the protein in the chromatographic fractions and those containing BcSpl1 were pooled and frozen at −70°C in small aliquots until use. Typically, c. 7 mg of protein were obtained from 200 ml of culture supernatant. Molar concentrations for the protein were calculated assuming a molecular mass for the recombinant protein of 14 739 Da, the theoretical mass of the polypeptide resulting after removing the 89 amino acids α-factor signal sequence. Protein concentrations were determined as described by Bradford (1976), using BSA as a standard.
Quantitative real-time PCR (Q-RT-PCR)
Mycelia for RNA extraction were prepared as explained elsewhere (ten Have et al., 2010). For the expression of tobacco defense genes upon infiltration with BcSpl1, the infiltrated area of the leaf was excised at the indicated times. Total RNA from B. cinerea, B. cinerea-infected tomato plants or BcSpl1-treated tobacco plants was isolated with the RNeasy plant mini kit (Qiagen). Absence of contaminant genomic DNA was assayed by PCR with primers PROPLAOUT and TERPLAOUT, and, when necessary, it was eliminated by treatment with RNase-free DNaseI. Amplification was carried out in an iCycler iQ Real-Time PCR system (Bio-Rad), with the Bio-Rad iQ SYBR Green supermix and the primers listed in Table 1. The B. cinerea and tobacco actin genes, actA and Tac9, respectively, were used to correct for sample-to-sample variation in the amount of RNA. The amplification of a single fragment was verified for every PCR reaction by running the final product on a 12% polyacrylamide gel electrophoresis. The relative mRNA amounts were calculated by the ΔΔCt method from the mean of three independent determinations of the threshold cycle (Applied Biosystems, 2004; Schmittgen & Livak, 2008). Deviation from the mean was calculated from the standard deviation (SD) in the ΔΔCt value, using the expression 2−(ΔΔCt ± SD).
Assays were carried out with detached tomato (cv Moneymaker) or tobacco (cv Havana) leaves, gerbera petals, tomato fruits (injured with a needle), grape berries (inoculated at the wounds left after detaching the pedicels) and slices of apple, kiwi or squash. Inoculations were made with 5-μl droplets of a conidial suspension (2 × 105 conidia ml−1 in 60 mM KH2PO4, 10 mM glycine, 0.01% Tween 20, 0.1 M glucose). The infected plant material was incubated at 22°C under conditions of high humidity on water-soaked filter paper in closed containers, and lesions at different time-points were photographed. To evaluate the virulence on tomato and tobacco leaves, the shape of the lesions was approximated to an ellipse for which the two radii were measured. Lesion sizes were calculated as the geometrical mean of these two radii, that is, as the radius of a hypothetical circle with the same area as the ellipse. Quantitative results are presented as the rate of progression of lesion size (radius), calculated for each individual infection from two measures taken at different days.
Reactive oxygen species (ROS) and electrolyte leakage assays
The generation of H2O2 and O2− was assayed in BcSpl1-infiltrated leaves with 3,3′-diaminobenzidine (DAB) and nitroblue tetrazolium (NBT), respectively. The protein solution was first infiltrated into the leaves with a 1-ml syringe without needle, through stomata in the underside. Four hours later the leaves were cut and treated with either DAB or NBT. Treatment of tomato and Arabidopsis leaves with DAB was done by immersing the petioles overnight at 22°C in 1 mg ml−1 DAB, pH 3.8. In the case of tobacco, leaf discs were vacuum infiltrated with the same solution for 1 h. Treatment with NBT of tomato and tobacco leaf discs, as well as whole Arabidopsis leaves, was done by vacuum infiltration with 0.5 mg ml−1 NBT for 1 h. To visualize DAB and NBT deposits, treated tissues were incubated for 12 h in ethanol to eliminate chlorophyll, and photographed.
To assay electrolyte leakage, tobacco leaves were first infiltrated with the protein solutions (BcSpl1 or BSA) on the plant and then, after 4 h, two leaf discs of 1 cm diameter were cut and submerged in 1.5 ml water at 4°C with shaking at 200 rpm. At the indicated times the conductivity was measured with a Basic 30 ion conductivity meter (Crison, Barcelona, Spain).
Cell death and autofluorescence assays
Two microliters of 34 μM BcSpl1 were applied to the inner surface of freshly prepared onion epidermis and incubated for 24 h at 20°C and high humidity. To visualize autofluorescence, the epidermis was examined with an Olympus BX-50 fluorescence microscope equipped with a U-MWIB filter to detect green fluorescence. To assay cell death, the same sample was then incubated with 0.4% Trypan blue at 50°C for 10 min and rinsed with water to check for the presence of blue-stained dead cells under a light microscope.
Cerato-platanin BcSpl1 is a conserved protein
In a previous work (Espino et al., 2010), we identified BcSpl1 in the early secretome of B. cinerea B05.10 as a quite abundant protein. A gene coding for this protein was found for the two B. cinerea strains with sequenced genomes, B05.10 (gene ID BC1G_02163.1) and T4 (gene ID BofuT4_P011930.1), although the protein is not identical for both strains. Two single amino acid substitutions, Glu64 vs Gln64 and Ala110 vs Thr110, differentiate, respectively, the T4 and the B05.10 proteins. Hereafter we will refer only to the protein from the strain B05.10 because that is the one we have analysed. A signal peptide is predicted for BcSpl1 in the first 18 amino acids (Emanuelsson et al., 2007) and the putative mature protein is highly similar to other members of the cerato-platanin family. An alignment of BcSpl1 with the closest homologues in Sclerotinia sclerotiorum, Trichoderma atroviride, Neurospora crassa, Gibberella zeae and Magnaporthe oryzae (not shown) resulted in c. 39% identical residues for all these proteins, and in 72% similar residues. The four cysteine residues characteristic of the family, involved in disulfide bridge formation (Pazzagli et al., 1999), are perfectly conserved. BcSpl1 has a predicted molecular mass, 12.0 kDa, similar to those proteins of the family previously characterized (Pazzagli et al., 1999; Djonovic et al., 2006). As for all members of the family, the hydrophobic amino acids content is relatively high, 37%, but the overall hydropathicity (GRAVY) determined by Protparam (Gasteiger et al., 2005) resulted 0.030, indicating a soluble protein. A Trp residue that has been implicated in the dimerization of cerato-platanins from Trichoderma species (Vargas et al., 2008), which renders the protein unable to cause necrosis in plants, as well as a glycosylation site interfering with dimerization, are both absent in the B. cinerea protein.
The bcspl1 gene shows variable levels of expression
The fact that BcSpl1 had been found in all conditions studied in three reports on the B. cinerea secretome (Shah et al., 2009a,b; Espino et al., 2010) prompted us to study in more detail the expression of the corresponding gene by Q-RT-PCR. In axenic cultures, bcspl1 mRNA levels were found to increase with respect to ungerminated conidia in every condition studied (Fig. 1a). At 12 h postinoculation this increase ranged from fivefold for media supplemented with strawberry to 300-fold with tomato. Times courses were also different for the different axenic cultures. In glucose, bcspl1 mRNA levels reached a maximum at 12 h after inoculation and decreased thereafter, while in strawberry extract bcspl1 expression increased steadily for the first 48 h.
In B. cinerea-infected tomato leaves (Fig. 1b), the amounts of bcspl1 mRNA increased following a two-phase curve reminiscent of the infection process: a first peak during the germination of conidia and formation of the initial black infection spot, and a second peak when the lesions are expanding. The highest value found in planta, late in the infection, was c. 2.5 times higher than the maximum value found in axenic cultures, so we can conclude that bcspl1 is expressed in every condition studied, but preferentially in planta during the late stages of infection.
The B. cinerea genome contains a second cerato-platanin gene
A search of the genome databases for the B. cinerea strains T4 (http://urgi.versailles.inra.fr) and B05.10 (http://www.broadinstitute.org), using the blast algorithm (Altschul et al., 1990), revealed an additional cerato-platanin gene (gene IDs. BofuT4_P000440.1 and BC1G_08735.1, respectively) coding for a protein highly similar to BcSpl1 (Fig. S2). The corresponding gene was named bcspl2 and its expression was studied by Q-RT-PCR with primers HOMOCTP-FW and HOMOCTP-RV (Table 1). However, no PCR product could be obtained in any case (Fig. S2). Moreover, sequences corresponding to this gene did not appear in any of the expressed sequences tag (EST) libraries generated during the genome sequencing (http://urgi.versailles.inra.fr), suggesting that either this gene is not expressed at all (i.e. it is a pseudogene) or it is expressed only under very particular conditions.
BcSpl1 contributes to virulence
In order to determine whether BcSpl1 plays any role in the infection process, two bcspl1 mutants (Δbcspl1.1 and Δbcspl1.2) were obtained (Fig. S1). Genomic DNA from both mutant strains was analysed by Southern blot and PCR amplification and both of them showed the expected band pattern in agreement with a single integration of the foreign DNA at the bcspl1 locus and with the absence of untransformed nuclei (Fig. S1). Moreover, Q-RT-PCR using cDNA prepared from these two mutants showed no expression of bcspl1 (not shown).
The 2D electrophoresis of the secretomes obtained from 3-d-old cultures of Δbcspl1.1 and Δbcspl1.2, showed two clear differences in the spot pattern, compared with the wild type, in the gel region expected for BcSpl1 (Fig. S1). Two protein spots of similar molecular mass to BcSpl1, but with different isoelectric points, were absent in both mutants. The two spots were identified as BcSpl1 by peptide mass fingerprinting. Among the peptides identified, it was found the N-terminal end of the mature protein, experimentally confirming the predicted 18-aa signal peptide. Some other, less-clear differences between the secretomes of the bcspl1 mutants and that of the wild type could be observed and are now under investigation.
Virulence of the two bcspl1 mutant strains on tomato and tobacco leaves was studied by measuring the increase in lesion size per day. Both mutants showed a low but statistically significant (P <0.05 by Student’s t test) decrease in virulence, of c. 20% in tomato and 40% in tobacco, with respect to the wild-type strain (Fig. 2). Visual inspection of infections in different fruits and vegetables such as tomatoes, apples, squash, grapes and gerbera petals showed also a general trend of decreased virulence for the two bcspl1 mutants, in comparison with the wild type (Fig. S3).
Reintroduction of the wild-type bcspl1 gene in the bcspl1 mutant strains resulted in restoration of the virulence to the levels of the wild-type strain B05.10 (Fig. 2), unequivocally assigning the decrease in virulence to the mutation of bcspl1.
No additional phenotypic differences were found between the two bcspl1 mutant strains and the wild type (not shown). Growth rates on solid medium (tested in malt extract agar, Gamborg’s B5 agar with 1% glucose, tomato-agar and potato dextrose agar) and conidia production was similar to the wild type. Wettability of the fungal colony surface, which was tested because cerato-platanins and hydrophobins share considerable sequence similarity and hydrophobin mutants usually show easier access of water to the interior of the colony (Linder et al., 2005), also showed no differences between the wild type and the two mutants.
BcSpl1 causes a rapid necrosis of plant tissues
BcSpl1 was expressed in P. pastoris using the vector pPICZαA (Invitrogen Easyselect Pichia expression kit) and the protein was purified from the culture supernatant. Infiltration of tomato, tobacco, and Arabidopsis leaves with a 34 μM solution of BcSpl1 caused a rapid appearance of a necrotic zone in the infiltration area (Fig. 3a). In tomato and tobacco, necrotic symptoms were evident within 1 h after infiltration (Fig. S4), while first symptoms in Arabidopsis appeared c. 8 h after treatment (not shown). Control infiltrations with water or BSA at the same molar concentration did not produce any effect in the leaves. As a positive control, we also infiltrated the well-known fungal elicitor ethylene-inducing xylanase (EIX) (Ron & Avni, 2004) and found that necrotic symptoms generated by infiltration with 34 μM BcSpl1 appeared faster and were more intense that those generated with 314 μM EIX (Fig. 3a).
Infiltration with different BcSpl1 concentrations revealed a dose–response curve (Fig. 3b) reminiscent of saturation. The minimum dose to produce appreciable necrosis 4 h after infiltration was 17 μM, and concentrations equal or higher than 68 μM all produced the same effect. Concentrations below the minimum value, although not able to generate any effect at 4 h after infiltration, were, however, able to produce at least some chlorosis at 4 d (Fig. 3b).
To rule out a general toxic effect of cerato-platanin on any cell type, we treated mammalian (HeLa) cells with the protein. A BcSpl1 concentration of 34 μM in the culture medium, for up to 16 h, did not produce cell death (assayed with Trypan blue) while lowering the pH of the medium to 4.0 made the cells to change their typical nonspherical morphology to a more rounded one, and to take the blue dye inside the cells – both clear symptoms of cell death (not shown).
The production of hydrogen peroxide and superoxide anion were assayed with DAB and NBT, respectively. In both instances, tomato, tobacco, or Arabidopsis leaves were first infiltrated with BcSpl1 and 4 h later assayed for the production of ROS. In all plants assayed an increase in brown DAB precipitate could be observed in the areas where BcSpl1 had been infiltrated, as well as a blue precipitate indicative of superoxide anion production (Fig. 4).
Infiltration of tobacco leaves with 34 μM BcSpl1 also caused electrolyte leakage from the treated tissue. When leaf discs cut from the infiltrated area were incubated in water, an increase in conductivity over time was observed that was significantly higher than that obtained for the control infiltrations with BSA at the same molar concentration (Fig. 5a). When this assay was done with different BcSpl1 concentrations, a dose–response curve was obtained (Fig. 5b) that resembled the dependency of the necrotizing activity on the BcSpl1 concentration (Fig. 3b). The concentration producing half the maximum effect on electrolyte leakage, c. 25 μM, corresponds to those producing an intermediate level of necrosis in leaves.
Autofluorescence and cytoplasm shrinkage are also symptoms of HR, and we have indeed observed that infection with B. cinerea induces both symptoms in onion epidermis (Fig. 6a). When 2-μl drops of 34 μM BcSpl1 were applied to the inner surface of onion epidermis, the cells under the drop developed clear autofluorescence and their cytoplasmic membrane separated from the cell wall resulting in a reduction of cell volume (Fig. 6b). Staining the treated onion epidermis with Trypan blue revealed that the fluorescent cells were indeed dead. Control incubation with BSA at the same molar concentration did not produce any of these effects.
BcSpl1 induces markers of HR and PR proteins
The induction of four pathogenesis-related genes was studied upon infiltration with BcSpl1 (Fig. 7): HSR203J and HIN1, which are considered markers of HR in tobacco (Pontier 1994, 2001), and PR-1 and PR-5, which are pathogenesis-related genes under the control of the transcription coactivator NPR1, the master regulator of systemic acquired resistance (SAR) (Spoel et al., 2009). The four genes were activated soon after infiltration of tobacco leaves with BcSpl1, especially HSR203J and PR-1, whose transcript levels relative to actin mRNA raised > 1000-fold above control (untreated leaves).
BcSpl1-derived peptides are able to reproduce the whole protein activity
Three peptides comprising each about one-third of the mature protein were assayed for necrosis-inducing activity in tobacco leaves. Peptides 1 (residues 19–58 of the immature protein) and 2 (residues 59–98), at a concentration of 34 μM, were able to produce chlorosis in the infiltration area after 4 d (Fig. 8a), while peptide 3 (residues 99–137) did not produce any effect. Concentrations of the peptides higher than 34 μM, up to 240 μM, did not produce a stronger effect in any case, and lower concentrations, 17 or 8 μM, did not result in any visible effect (not shown). These results seem to indicate that the protein region with necrosis-inducing activity comprises peptides 1 and 2, but not peptide 3. This idea is further reinforced by the fact that infiltration with a mixture of peptides 1 and 2, as well as a mixture of peptides 1, 2 and 3, produced a level of necrosis in tobacco leaves similar to that of the whole protein (Fig. 8b), while infiltration with the other two possible combinations of the three peptides did not.
The Arabidopsis BAK1 signaling protein is required for full BcSpl1 activity
BAK1 (also called SERK3) is an Arabidopsis regulatory protein that has been shown to be required for the perception of the archetypal plant elicitors flagellin and EF-Tu (Zipfel & Robatzek, 2010), as well as several other elicitors (Segonzac & Zipfel, 2011). BAK1 has been shown to form a complex with the flagellin receptor FLS2 upon ligand binding, initiating the signaling cascade (Chinchilla et al., 2007; Heese et al., 2007). Two independently isolated bak1 mutants were assayed for their responsiveness to BcSpl1, bak1-1 (Li et al., 2002) and bak1-3 (SALK line SALK_034523C). Both mutants showed a reduced sensitivity to BcSpl1 when the protein was included in the germination medium (Fig. 9a), and when the roots of 5-d-old seedlings were submerged in medium containing the cerato-platanin (Fig. 9b), indicating that BAK1 also plays a role in the perception of BcSpl1.
Among the vast array of proteins secreted by B. cinerea, one of the most abundant is the cerato-platanin BcSpl1 (Shah et al., 2009a,b; Espino et al., 2010), and we have shown here that it is required for full virulence and that induces necrosis in several hosts with symptoms of HR.
BcSpl1 is a well conserved member of the cerato-platanin family, with all the structural features that are typical of the family such as low molecular weight, four conserved cysteine residues, and a relatively high proportion in hydrophobic amino acid residues. The 18-amino acids signal peptide predicted for the protein was experimentally corroborated because the N-terminal peptide of the mature protein, with the expected molecular mass, was obtained in the identification of the protein by peptide mass fingerprinting. The resulting mature protein sequence has no predicted N- or O- glycosylation sites.
As we were able to express in yeast and purify BcSpl1, we could study the effect of the recombinant protein on plants. Similarly to what has been observed for cerato-platanins from C. fimbriata (Pazzagli et al., 1999) and M. grisea (Yang et al., 2009), BcSpl1 was able to induce necrosis in plant leaves. The effect seemed more rapid and efficient for the B. cinerea cerato-platanin than for the two other proteins. Pazzagli et al. (1999) reported a minimum dose of 80 μM for the C. fimbriata protein to produce observable necrosis 24 h after infiltration of tobacco leaves, while we were able to observe clear necrosis 4 h after infiltration, in the same plant species, with protein concentrations down to 17 μM (Fig. 3b). Even lower protein concentrations, down to 2 μM, produced some chlorosis days after infiltration (Fig. 3b). The necrosis-inducing activity of BcSpl1 proved also to be higher than other well-studied fungal protein elicitors such as xylanases (Bailey et al., 1990; Enkerli et al., 1999; Noda et al., 2010). Side by side comparison of BcSpl1 and xylanase EIX from Trichoderma viride (Figs 3a and S4) showed a much stronger necrotizing activity for BcSpl1. Even the infiltration with an EIX concentration of 314 μM, about ninefold higher than the one used for BcSpl1, did not induced a response as intense or fast as BcSpl1.
Not all cerato-platanins reported so far have been shown to induce necrosis in treated plants. This is the case, for example, for the M. grisea protein MSP1, where application on wounded wheat leaves did not produce any effect (Jeong et al., 2007). However, expression of the corresponding M. grisea gene in Arabidopsis under an inducible promoter led to plants that died rapidly upon induction of the transgene (Yang et al., 2009), indicating that the exogenous application of MSP1 did not reach the necessary concentration and/or the location of the target in the plant tissue. To our knowledge, at least two other cerato-platanins have been tested for their ability to produce necrosis with negative results: Sm1 from Trichoderma virens (Djonovic et al., 2006) and SP1 from L. maculans (Wilson et al., 2002). Apart from the intrinsic differences in the proteins, or the different plant species used in the assays, the lack of response maybe also related to the effective concentration of protein reached inside the leaf tissues, as in both cases the application of the protein was done by placing drops of protein solution on leaves that had been previously punctured. However, both Sm1 and SP1 proteins, were able to induce the production of ROS, cell autofluorescence and the expression of defense genes.
It is interesting, therefore, that all cerato-platanins studied to date, including B. cinerea BcSpl1, were able to induce plant defense responses related to HR. One of these responses is the induction of autofluorescence (Heath, 1998), which has been observed for cerato-platanins from C. fimbriata (Pazzagli et al., 1999), T. virens (Djonovic et al., 2006), L. maculans (Wilson et al., 2002) and B. cinerea (Fig. 6). Another HR symptom, the induction of ROS (Mur et al., 2008) has also been observed in BcSpl1 (Fig. 4), similar to what has been seen for the T. virens protein (Djonovic et al., 2006). Other HR symptoms have only been observed, to our knowledge, for the B. cinerea protein, such as electrolyte leakage (Fig. 5) and cytoplasm shrinkage (Fig. 6). All these observations seem to imply that, instead of a direct toxic effect of cerato-platanin on plant cells, these proteins are being recognized by the plant immune system and that this recognition culminates, at least in some cases, with the programmed death of the affected cells. This idea is further reinforced by the fact that BcSpl1 induces defense genes in planta (Fig. 7), including markers of HR, as has been observed for cerato-platanins from T. virens (Djonovic et al., 2006; Buensanteai et al., 2010) and C. fimbriata (Fontana et al., 2008). Substances perceived by the plant immune system, usually referred to as pathogen/microbe-associated molecular patterns (PAMPs or MAMPs) usually interact with their respective plant receptors via a specific, highly conserved region of the protein (Zipfel & Robatzek, 2010). This could also be the case for BcSpl1 as two adjacent 40-amino acids BcSpl1 fragments are able to reproduce, at least partially, the effect of the whole protein (Fig. 8). Interestingly, an equimolar mixture of these two peptides generates a response comparable to that produced by the intact BcSpl1, suggesting that the putative necrosis-inducing motif of BcSpl1 spans the junction between these two protein fragments. In fact, the best conserved part of cerato-platanins comprises these two fragments (see alignment of Pfam family PF07249). Considering the hypothesis of a plant receptor for BcSpl1, the two peptides with the partial motif could be binding the receptor with lower affinity, generating a partial response, while the mixture of the two, as well as the whole BcSpl1, would bind with a higher affinity and would generate a stronger response. These possible explanations are now under investigation. Regarding the hypothesis that BcSpl1 is perceived as a PAMP by the plant immune system, it is interesting that the Arabidopsis BAK1 protein, a component of the PAMP signaling cascade (Chinchilla et al., 2007; Heese et al., 2007; Schulze et al., 2010), also seems to play a role in the induction of necrosis by BcSpl1. Two null mutants in the Arabidopsis bak1 gene showed a moderately diminished sensitivity to BcSpl1, similar to what has been observed for several other PAMPs such as flagellin and EF-Tu (Chinchilla et al., 2007). Although the absence of BAK1 does not completely abolish the action of BcSpl1, the lower sensitivity observed for the mutants is a good indication that the necrosis-inducing activity of BcSpl1 may be carried out via activation of the plant immune system. A plant membrane receptor for BcSpl1 should exist if this hypothesis is true, and it would not be surprising if this yet-to-discover receptor belongs to the LRR-RK family of proteins as the two proteins known to interact with BAK1 are members of this family, and hundreds of LRR-RKs with unknown ligands are coded by the Arabidopsis genome.
Along with B. cinerea bcspl1, three cerato-platanin genes have been knocked out. In two out of these three cases, B. cinerea BcSpl1 and M. grisea MSP1, cerato-platanins were shown to contribute to virulence. Conversely, mutation of the cerato-platanin gene sp1 from L. maculans did not cause any effect on virulence (Wilson et al., 2002). It is tempting to speculate that the contribution of BcSpl1 to the virulence of B. cinerea could reside on its ability to induce HR, as it has been proposed that this fungus (as well as other necrotrophs) can make use of HR for its own benefit (Govrin & Levine, 2000; van Kan, 2006; Choquer et al., 2007; Williamson et al., 2007). The proposal is that these organisms would actively stimulate HR in plants with the purpose of generating dead tissue to grow in. However, the fact that cerato-platanin is also required for full virulence in M. oryzae does not seem to corroborate this hypothesis, as this fungus is considered to be, at least until late in the infection process, a biotrophic fungus for which induction of the plant HR would be deleterious for a successful infection (Valent & Khang, 2010).
Although the only known activity of cerato-platanins is their effect on plants, the induction of tissue necrosis and plant defenses, it is difficult to imagine that these proteins have no other function. A protein with only this activity may confer an evolutionary advantage to necrotrophs, but not to biotrophs or to fungal organisms not interacting with plants. However, cerato-platanin proteins available in Pfam belong to a wide and diverse group of fungi, strongly suggesting that this protein may have an additional function. Unfortunately, all our fitness tests carried out with the two bcspl1 mutants in B. cinerea failed to identify any other phenotypic difference from the wild type other than a reduced virulence on plants, so this putative main function of cerato-platanin still awaits discovery. Our present working hypothesis is that cerato-platanins, being a common fungal secreted protein of unknown function, can be detected by the plant immune system as a PAMP, and that the intensity of the defense response mounted afterwards can lead, with the right elicitor at the right concentration, to HR and the accompanying cell death.
Support for this research was provided by grants from the Ministerio de Educación y Ciencia (AGL2006-09300) and Gobierno de Canarias (PI2007/009). MF was supported by Gobierno de Canarias. We thank J. Ávila and P. Martín for providing access to the fluorescence microscope.