Cultivation and transformation of Botrytis

B. cinerea strains were routinely cultured on agar containing malt extract (ME) medium [3]. For growth tests, agar plates containing Gamborg minimal medium (GB5) containing 25 mM glucose were used. For sporulation tests, 10 μl droplets containing 10⁵ conidia ml^-1 were inoculated onto ME agar plates, and incubated at 20–22°C for 7 days. Conidia were scraped off the plates with 10 ml water using a glass spatula, filtered through glass wool, and counted using a hemocytometer. For induction of sclerotia, ME agar plates were inoculated in the same way, but incubated in darkness for 14 days.

B. cinerea transformation using knockout constructs with resistance markers was performed as described [3, 32]. Protoplasts transformed with CypR cassettes (containing a 344 bp promoter fragment of Aspergillus nidulans trpC (chrom.4:2565112–2565127), a 1380 bp fragment containing Bcpos5 (chrom.10:1077653–1079032) and a 146 bp terminator fragment of B. cinerea niaD (chrom.7:456591–456736). were selected in SH agar containing 0.3 μg ml^-1 Cyp (Syngenta, Chorus fungicide formulation). Non-transformed colonies appeared with a frequency of <10⁻⁷ per transformed protoplast. Single spores of the transformants were transferred after three days to plates containing GB5 agar with 25 mM glucose and 0.3 μg ml^-1 Cyp, for further cultivation and verification of the transformants.

For generation of marker-free multiple knockout mutants, our published transformation protocol [32] was modified as following: To 2x 10⁷ B. cinerea protoplasts suspended in 100 μl TMSC buffer, 10 μg pTEL-Fen and up to four RNPs, each consisting of pre-complexed 6 μg Cas9-Stu^2x and 2 μg sgRNA (two RNPs per gene) were added. The transformed protoplasts were mixed with liquified 200 ml SH agar adjusted to 39.5°C, and poured into ten 90 mm petri dishes. After three days of incubation at 20–22°C, small agarose pieces containing individual transformants were cut out with a pointed scalpel and transferred to 5 cm plates containing selection-free 4x ME agar (4x ME: 4% malt extract, 1.6% glucose, 1.6% yeast extract, 1.5% agar, pH 5.5), to accelerate growth and sporulation of freshly generated transformants, and rapid loss of pTEL-Fen. After two days, plates with no or very little growth were discarded. Hyphal tips of fast growing colonies, all considered as original transformants, were transferred to new 4x ME plates and allowed to grow for 5 days until sporulation. Conidia or sporulating mycelium were used for DNA isolation and PCR analysis to detect the desired editing events in the gene or genes targeted by the RNP pair, and the absence of WT DNA in the deleted region. Rapid loss of pTEL in the absence of FenR selection was confirmed for about 90% of the transformants; transformants which remained FenR (probably due to chromosomal integration of pTEL-Fen) were discarded. With FenS transformants containing the expected deletion(s), a single round of single spore isolation was performed. From the purified transformants, DNA was again isolated, to verify by PCR the generation of a homokaryotic deletion, i.e. the complete loss of WT DNA in the deleted region by primers that amplified an internal sequence of the deleted region(s). Verified transformants were used for phenotypic characterization. Depending on the efficiency of transformation and editing, between two and five genetically identical transformants were used for subsequent growth and infection tests. Transformants showing normal growth and differentiation in vitro (which was observed for most but not all transformants) were selected for the next round of transformation.

DNA manipulations

To generate a cyprodinil resistance (CypR) cassette, the Bcpos5 (Bcin10g02880) coding sequence including two introns was amplified by PCR from genomic DNA of B. cinerea B05.10, by using primers CypR_ol_Ptrp_FW CypR_ol_TniaD_RV. The 3’-terminal primer CypR_ol_TniaD_RV was changed in sequence to generate the cypR-associated L412F substitution [33]. The resulting fragment was flanked with fragments containing the Aspergillus nidulans trpC promoter (PtrpC) generated with primers Gib_pTEL_S_EcorV_PtrpC & PtrpC_ol_CypR_RV, and the niaD terminator of B. cinerea (TniaD) generated with primers TniaD_ol_CypR_FW & Gib_pTEL_S_EcorV_TniaD, both amplified from plasmid pTEL-Fen [32]. The CypR cassette was integrated into pTEL-Start linearized with EcoRV. To test its functionality as a resistance marker, the CypR cassette was attached to the flanking regions of several target genes, using a modular cloning approach.

Deletion constructs were generated with resistance cassettes for nourseothricin (natR/^N) and cyprodinil (cypR/^C) for spl1^C, xyn11A^N, nep2^N, ieb1 ^C, xyg1^C, xyn11A^C, and nep2^C. For this, 0.5–1 kb genomic regions flanking the coding sequences were amplified, spliced together with a resistance cassette (trpC promoter–resistance gene–niaD terminator) into a pBS-KS vector by Gibson assembly, and transformed into E. coli. Before transformation into B. cinerea, deletion constructs were released from the plasmids by restriction digestion.

For generation of a quadruple mutant with resistance markers, a xyn11A^N mutant was transformed with a spl1^C k.o. cassette to generate a xyn11A^N spl1^C double mutant. This mutant was cotransformed with two Cas9-sgRNA complexes targeting nep1 and nep2, and nep1^H nep2^F k.o. cassettes with 60 bp homology flanks (amplified using pTEL-Fen or pTEL-Hyg as template) as repair templates. Transformants with resistance to fenhexamid (1 μg ml^-1) and hygromycin (17.5 μg ml^-1) were tested for the knockout of nep1 and nep2. The double mutant pg1pg2^R, kindly provided by Jan van Kan (Wageningen University), was constructed transforming a hygR pg1 mutant [11] with a pg2^N knockout construct, with selection for nourseothricin (35 μg ml^-1). The knockouts of pg1 and pg2 were confirmed by PCR analysis. Primers used for synthesis of sgRNAs, construction of knockout constructs, and screening of transformants for correct knockouts and homokaryosis are shown in S1 Table.

Genome sequencing

For full genome sequencing of B. cinerea, gDNA was normalized to 15 ng μl^-1. Sequencing libraries were prepared with 30 ng of normalized DNA using the NEBNext Ultra II FS DNA Library Prep Kit for Illumina (New England BioLabs, Ipswich, Massachusetts, USA). Custom 8 bp barcodes were added to each library during the preparation process, and sample then pooled together. The pool was cleaned with magnetic beads included in the library preparation kit, and run on a lane of the HiSeqX instrument (Illumina, San Diego, CA, USA) in a 150-cycle paired end run, resulting in sequence yield of 969 Gb for 153 samples. This resulted in a range from 4.9 to 7.9 Gb for Botrytis samples. The B05.10 reference genome and annotation gtf file were created by combining the nuclear genome (ASM83294v1) and the mitochondrial genome (GenBank: KC832409.1). Reads were quality checked using FastQC version 0.11.4 [34] and multiqc [35] and trimmed using Trimommatic version 0.39 [36] to remove adapters and low quality bases. Trimmed reads were aligned to the reference genome using bwa mem version 0.7.17 [37]. SAM and BAM files were manipulated using Samtools version 1.7 [38]. Duplicated reads were marked using Picard Tools version 2.20.2 [39]. Variants were called using Samtools mpileup and bcftools version 1.9 and filtered for a read depth of 10 and quality score of >20. Variants were annotated using SnpEff version 4.1g [40]. Chimeric and discordant reads were identified using Samtools. To identify larger deletions such as those introduced by CRISPR/Cas9, IMSindel was used using default parameters (https://www.nature.com/articles/s41598-018-23978-z). Sequencing of the three B. cinerea strains resulted in genome coverages of 149x (WT), 122x (6x mutant), and 199x (12xbb mutant). The fraction of ≥Q30 bases was between 91.07 and 91.29.

Infection tests and secretome analyses

Infection tests were performed with attached leaves of Phaseolus vulgaris (genotype N9059) and Nicotiana benthamiana, detached leaves of tomato (Solanum lycopersicum, cv. Marmande) and maize (Zea mays, cv. Golden Bantam), and apple fruit (Malus domestica, cv. Golden Delicious). Leaf inoculations were performed as previously described [41], using 20 μl droplets with 10⁵ conidia ml^-1 in GB5 minimal medium (GB5: 3.05 g l^-1 GB5, 10 mM KH₂PO₄, pH 5.5) with 25 mM glucose. For inoculation of tobacco leaves, conidia suspensions did not work well due to their hydrophilic surfaces. Therefore, before inoculation, 10 μl of conidia were inoculated onto agar discs (5 mm ∅, 2 mm thickness; containing GB5 medium with glucose) and incubated for 24 h. The discs were placed with the germinated conidia downwards onto the leaves. To achieve maximal accuracy and comparability, up to three droplets each of WT and mutant conidial suspensions or agar discs were applied on both sides of the midrib of one leaf or leaflet. They were considered as technical replicates to determine lesion sizes per strain and leaf. Control experiments confirmed that neighbouring inoculations did not affect each other. Apple fruit was inoculated with conidia suspensions after wounding with a cork borer of 5 mm diameter above the equatorial line. Lesions were photographed after 48 to 96 h, and lesion areas determined by image analysis using ImageJ software, after subtraction of the inoculation areas.On planta secretomes were obtained from detached tomato leaves densely inoculated with 25 μl droplets containing 10⁵ conidia ml^-1 in GB5 medium with 25 mM glucose, and incubated at 20–22°C and 100% humidity in flat glass trays covered with saran wrap. After 48 h, droplets were collected, frozen at -80°C, thawed, centrifuged at 4°C for 60 min at 4000 g, sterile filtered and again frozen in aliquots at -80°C until further analysis. MS/MS-based proteomic analysis for confirmation of loss of CDIPs in the deletion mutants was performed as described [3]. To determine CDI activity of WT and mutant secretomes, the secretomes (containing ca. 5–10 μg protein ml^-1) were diluted with GB5 medium to concentrations of 1 or 2 μg protein ml^-1, and ca. 20–50 μl each of the solutions were infiltrated into attached leaves of Nicotiana benthamiana or Vicia faba (cv. Fuego). The size of necrotic lesions in the infiltrated leaf area was recorded after two days. Heat treatment of the secretomes was performed by incubation for 20 min at 95°C in a heating block. Size fractionation of the secretomes was done using ultrafiltration cartridges (Amicon Ultra-4, Merck Millipore Ltd, Tullagreen, Co., Cork, Ireland) with 10 kDa molecular weight cutoff. Before infiltration, the small molecular weight fraction was concentrated two-fold in a speed vacuum concentrator.

Generation and infection of tobacco plants silenced for sobir1 and bak1

Silencing of bak1 in N. benthamiana plants was performed using the tobacco rattle virus (TRV) based plasmid construct pTRV2:NbBAK1 with Agrobacterium tumefaciens GV301 [42], as described in Seifbarghi et al. (2020) [43]. Three weeks after virus infection, attached leaves showing typical symptoms of Bak1 silencing (stunting and epinasty of leaves; [44]) were inoculated with agar plugs containing germlings of B. cinerea as described above, and kept in a humid chamber at ambient light for 48 h.

Microscopical analysis of infection

To compare the early infection process of B. cinerea WT and multi-k.o. mutants microscopically, inoculations of detached Phaseolus vulgaris leaves were performed. To investigate fungal appressorium formation and the ability to penetrate the plant surface in the absence of plant defence responses, dead onion epidermal cell layers were used. For this, epidermal layers were removed from the concave side of onion, fixed with tape onto glass coverslips and killed by incubation at 65°C for 30 min in a water bath. After thorough washing with water, samples were air-dried and inoculated with 20 μl droplets containing 5*10⁴ conidia ml^-1 in 1 mM fructose, and incubated in a humid chamber for 24 h. Staining of superficial fungal structures was done with 0.03% trypane blue in lactophenol for 20 min, followed by replacement with water and microscopic analysis as previously described [41]. For confocal microscopy, Phaseolus vulgaris leaves were inoculated with 1 μl of 10⁵ conidia ml^-1 in GB5 medium with 25 mM glucose. After 24 h, developing lesions were stained with 10 μg ml^-1 calcofluor white (fluorescence brightener 28, Sigma) for 5 min and thoroughly washed with water subsequently. Images were acquired using a Zeiss LSM880 AxioObserver confocal laser scanning microscope equipped with a Zeiss EC Plan-Neofluar 5x/0.16 M27 objective (DFG, INST 248/254-1). Fluorescent signals of calcofluor white (excitation/emission 405 nm/410-523-571 nm) and chlorophyll (excitation/emission 633 nm/638-721 nm) were processed using the Zeiss software ZEN 2.3 or ImageJ (https://imagej.nih.gov/ij/).

Isolation of B. cinerea metabolites

B. cinerea B05.10 (WT), 10x and 12xbb mutants were grown on malt agar medium at 26°C with 24 h of daylight and used to inoculate roux bottles (1000 ml) containing 150 ml of modified Czapek-Dox medium (50 g glucose, 1 g yeast extract, 1 g K₂HPO₄, 2.5 g NaNO₃, 0.5 g MgSO₄·7H₂O, 0.01 g FeSO₄·7H₂O, pH 7, in 1 l of water). Bottles were inoculated with mycelium grown on six small slices of agar (0.8 mm) and incubated for 6 and 12 days at 26°C under daylight and static conditions.

After fermentation of the strains in 4 l culture medium for 6 and 12 days, media were filtered and the aqueous phases were extracted three times with ethyl acetate. The obtained extracts were dried with anhydrous sodium sulfate. The solvents were evaporated under vacuum to yield crude yellow oily extracts. These extracts were fractionated by column chromatography eluting with hexane:ethyl acetate mixtures containing increasing percentages of ethyl acetate (20−100%). Spectroscopic analysis by ¹H- and ¹³C-NMR was used to detect the presence of the toxins in each fraction. Interesting fractions were purified by HPLC with an increasing gradient of ethyl acetate. The metabolites structures were analysed and characterised by spectroscopic methods and direct comparison with authentic samples, previously isolated from strains of B. cinerea [45].

¹H- and ¹³C-NMR spectra were obtained on Agilent 400 and 500 MHz spectrometers with SiMe₄ as the internal reference. HPLC was performed with a Hitachi/Merck L-6270 apparatus equipped with a UV–vis detector (L 6200) and a differential refractometer detector (RI-71).

Statistical analyses

Statistical analyses were carried out with the GraphPad Prism software. For comparative infection assays, two or three pairwise inoculations of WT and mutant were performed on opposite sites of leaves or fruits. Relative necrotic areas of the mutants (% of WT) were calculated for each leaf/fruit, based on two or three technical replicates of WT vs. mutant pairs. Values from at least three inoculation dates with at least three leaves or fruits each were analyzed by one-sample t test between WT and each of the analysed mutants. Comparison of necrotic areas between different deletion mutants was carried out by one-way ANOVA followed by Tukey’s multiple comparison test and compact letter display. To evaluate the significance of differences between WT and mutants, one-way ANOVA followed by Dunnett’s multiple comparison post-hoc test was performed.

In the scatter plots, single data points and mean values ± standard errors (SEM) are indicated. Limits of box plots represent 25th percentile and 75th percentile, horizontal line represents median, and whiskers display minimum to maximum values. Evaluation of MS/MS proteomics data for significant deviations in the abundance of detected proteins in WT and 10x, 11x, 12xbb and 12xpg mutants was performed using the Perseus bioinformatics platform [46].

Generation of B. cinerea CDIP single and multiple mutants

Before the CRISPR/Cas9 protocol for marker-free mutagenesis was available [32], different selection markers were used to generate single and up to quadruple mutants by standard mutagenesis, using knockout constructs integrated into the target genes by homologous recombination. In addition to the established markers conferring resistance to hygromycin (HygR), nourseothricin (NatR) and fenhexamid (FenR), the anilinopyrimidine fungicide cyprodinil (Cyp)[47] was developed as a new marker for selection in B. cinerea. The mode of Cyp resistance (CypR) has been uncovered by a functional genomics approach [33]. Many CypR B. cinerea field strains contain a mutation leading to an L412F exchange of a mitochondrial NADPH kinase encoded by Bcpos5. A CypR selection marker was generated by integrating Bcpos5^L412F into a constitutive expression cassette (Fig 1A). Functionality of the CypR marker was confirmed by targeted mutagenesis of several genes, which yielded robust numbers of transformants. However, several of the apparant transformants turned out as spontaneous CypR mutants, as revealed via PCR by the absence of the CypR cassette (Fig 1B and 1C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Establishment of cyprodinil resistance (CypR) as a novel selection marker for B. cinerea.

A: Schematic picture of a CypR cassette containing a resistant variant of Bcpos5 gene due to the L412F amino acid exchange. PtrpC: Aspergillus nidulans trpC promoter fragment. TniaD: B. cinerea niaD terminator fragment. B: Transformation efficiency with the CypR cassette and characterization of the CypR transformants. in loco: Correct knockout mutants (replacement of WT DNA by CypR cassette). Ectopic: Detection of CypR cassette outside of target gene. Spontaneous: CypR ‘transformants’ lacking the CypR cassette. C: Growth of a CypR resistant transformant (spl1^C), and the sensitive B. cinerea WT strain on agar plates with GB5 medium with 25 mM containing different Cyp concentrations.

https://doi.org/10.1371/journal.ppat.1010367.g001

The genes analysed by mutagenesis include most of the currently known B. cinerea CDIPs and phytotoxins (Table 1). Single knockout mutants were generated for the previously characterized genes xyn11A, spl1, xyg1, ieb1, and nep2. Furthermore, we constructed mutants of gs1 encoding a putative glucoamylase previously reported as a CDIP [18], and of plp1 (PAMP like protein) encoding the homologue of a CDIP of the apple pathogen Valsa mali named VmE02 [48]. All mutants showed normal vegetative growth and differentiation in vitro. When infection experiments were performed with tomato and Phaseolus bean leaves and apple fruits, no significant differences in virulence compared to WT were observed (S1 Fig). Therefore, none of the deleted genes alone plays a major role for infection on any of the tested plants. These results are inconsistent with previous studies, which reported reduced virulence of B. cinerea mutants in xyn11A, spl1 and xyl1 [16, 21, 49]. Next, a quadruple mutant (4x^R: xyn11A spl1 nep1 nep2) was generated by using four different resistance markers, including the newly established CypR marker. As described below, the 4x^R mutant was weakly impaired in virulence, besides a slight growth retardation.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Cell death-inducing metabolites and proteins of Botrytis cinerea.

https://doi.org/10.1371/journal.ppat.1010367.t001

Development of a marker-free CRISPR/Cas9 method for rapid generation of homokaryotic, multiple knockout mutants

We have recently described a powerful CRISPR/Cas9-based method for B. cinerea gene editing without introducing resistance markers into the transformants. It is based on simultaneous introduction of an unstable telomere vector and Cas9-sgRNA ribonucleoprotein complexes (RNPs) into protoplasts, which allows transient selection of transformants containing the desired editing events [32]. For improved serial mutagenesis, the protocol was modified to generate deletions by two RNPs targeting one gene, without addition of a repair template, which results in excision of the sequence between the cleavage sites by non-homologous end joining (NHEJ). When the protocol was tested for knockout of spl1, several thousand FenR transformants were obtained, and 68% of them verified by PCR analysis as being edited with the correct deletion. For serial mutagenesis, two genes were targeted simultaneously, using four RNPs in each transformation. These experiments usually yielded high editing rates, resulting in the isolation of mutants containing the expected single and double deletions. In some cases, however. e.g. ieb1 and pg1, gene knockouts occurred with lower effiency, possibly due to the chosen sgRNAs or an inherently lower accessibility of these genes for editing (Table 2). Unexpectedly, a large fraction of the primary transformants appeared to be homokaryotic when tested by PCR of their genomic DNA, and complete loss of the deleted DNA in these transformants was confirmed after a single spore isolation step. This represents a great advancement over traditional transformation methods for B. cinerea, which usually resulted in heterokaryotic transformants that had to be purified by several rounds of single spore isolation before homokaryosis was achieved [52].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Serial deletion of B. cinerea genes encoding CDIPs and phytotoxic metabolites.

https://doi.org/10.1371/journal.ppat.1010367.t002

The improved transformation protocol was used for serial inactivation of up to 12 genes encoding CDIPs and key enzymes for biosynthesis of the phytotoxins, botrydial and botcinins. Using the 10x mutant as recipient, two 12x mutants were generated, one impaired in biosynthesis in botrydial and botcinins (12xbb) and the other mutant lacking PG1 and PG2 (12xpg). This was done based on previous data which showed that these two pairs of phytotoxic molecules play significant roles for infection [9, 12]. The expected gene deletions, and the absence of remaining WT DNA in the deleted regions were verified by PCR after purification of homokaryotic transformants (S2 Fig). Sequencing revealed in most cases precise (±2 bp) excisions as predicted from the RNP-directed DNA cleavage sites, except for the 12xpg mutant in which a ca. 3 kb larger deletion than expected had occurred in pg2 (S2 Table).

Genome sequencing reveals only few off-target mutations in multiple mutants

In a previous study we have shown that editing of B. cinerea using Cas9-RNPs leads to only few off-target mutations [32]. To investigate the mutations that were introduced by the repeated mutagenesis procedure in this study, Illumina-based genome sequencing was performed with B. cinerea WT, the 6x mutant, and the 12xbb mutant. All targeted deletions were confirmed (S2 Table). Similar to our previous results, few off-target mutations were observed in the 6x and 12xbb mutants (S3 Table). Among a total of nine clear-cut off-target changes, eight were point mutations, and one was a single nucleotide insertion within an oligo-(T) stretch. In no case, the mutations were found to be due to off-target Cas9 activities caused by the sgRNA(s) used. Remarkably, eight of the nine changes were found in both mutants, and only one mutation was restricted to the 12xbb mutant. Thus, eight mutations had occurred during the first four rounds of transformation leading to the 6x mutant, and just one more mutation in the three following rounds leading to 12xbb mutant. Four changes were located in genes, and resulted in alterations of their coding sequences. Potentially the most severe change was a nonsense mutation in Bcin12g06810, a gene located next to a telomere of chromosome 12, leading to the premature stop of translation. The predicted protein lacks any known domains, and belongs to a family of more than 10 paralogs restricted to close relatives of the genus Botrytis within the Helotiales. Three changes were missense mutations leading to amino acid exchanges in a protein with unknown function (Bcin03g08690), a protein similar to phosphoinositide-specific phospholipases (Bcin16g02140), and a secreted protein (Bcin06g00550) with unknown function. In addition to the nine clear-cut mutations, two ambiguous changes were found which occurred in the WT in a ‚heterozygous‘ state, suggesting preexisting heterogeneous mutations across nuclei in WT strain and the selection of nuclei carrying these mutations in the KO mutants.

Phenotypes of mutants deficient in multiple cell-death inducing proteins and metabolites

All mutants displayed a growth rate similar to WT, except for a slighty reduced growth of the marker-containing 4x^R mutant, which was not observed in the marker-free 4x mutant with the same knockouts (Fig 2A). Conidia formation, which is stimulated in the light, and sclerotia formation, which is induced by cultivation in complete darkness, were also unaffected in all tested mutants (Fig 2B and 2C). When incubated on glass surfaces for two days, WT germlings form large aggregates of appressoria-like structures, so-called infection cushions. They are believed to represent alternative infection structures besides simple appressoria [53]. Infection cushions with similar morphology were formed by the WT and the 12xpg and 12xbb mutants (Fig 2D). These data confirmed that none of the secreted proteins are significantly involved in vegetative growth, reproduction and pathogenic differentiation on artificial surfaces.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Growth and in vitro differentiation of B. cinerea B05.10 (WT) and mutants generated in this study.

A: Relative radial growth on GB5 minimal agar medium with 25 mM glucose (3 days). One way ANOVA and Dunnett’s post-hoc test (control: WT) are displayed; ***: p < 0.001. B: Conidia formation on ME plates incubated for 10 days under permanent light to induce conidia formation. C: Sclerotia formation on ME plates incubated for 14 days in darkness to induce sclerotia formation. Mutants marked with superscript letters were generated with resistance markers. For B and C, a one-sample t test against a hypothetical value of 100% (mean of WT) did not show any significant differences. For A-C, the means of three experiments, with one to three replicates each, are shown. D: Infection cushions formed on glass slides after 48 h. Upper scale bars: 50 μm; lower scale bars: 150 μm.

https://doi.org/10.1371/journal.ppat.1010367.g002

To test the effects of multiple knockouts on infection, leaves of Phaseolus bean, tomato and maize, and apple fruits were inoculated, and necrosis formation quantified after 48 to 96 h (Fig 3). On all tested tissues, the same trend was observed that infection efficiency of the mutants generally decreased with increasing number of deleted genes. Whereas a marker-free 4x mutant and a marker-containing 4x^R mutant were not or only weakly affected in virulence, mutants with more deletions showed reduced infection on all tissues. However, differences were observed on different tissues: On bean leaves and apple fruit, the effects of multiple knockouts were stronger than on tomato and maize leaves. For example, the 10x mutant formed lesions which were only ca. 60% in size compared to WT lesions on bean leaves and apple, but similar or only slightly smaller lesions on tomato and maize leaves. Compared to the 10x mutant, 11x, 12xpg and 12xbb mutants showed further reductions of lesion sizes, except on maize leaves. These comparisons allowed to assign clear contributions to virulence for pg1 (10x vs. 11x), pg1 plus pg2 (10x vs. 12xpg, but not 11x vs. 12xpg), encoding endopolygalacturonases, and bot2 plus boa6 encoding key biosynthesis enzymes for botrydial and botcinins (Fig 3). In contrast, only minor effects on virulence were observed for the other eight CDIPs, and their contributions appeared to differ between the tissues.The combined effects of pg1 and pg2 were confirmed by the reduced virulence of a pg1 pg2 double mutant, generated by classical mutagenesis with selection markers, on all tested tissues (S3 Fig), in accordance to previous results [11, 12]. When WT, pg1 pg2 mutant and 12xpg mutant were inoculated next to each other on the same leaf or fruit, the differences in lesion formation between each of the three strains were evident (S3 Fig). The differences in virulence between pg1 pg2 and 12xpg confirms the relatively small effects on virulence of the other ten CDIPs that are deleted in the 12xpg mutant. Similarly, the effects of the combined bot2 and boa6 knockouts which were evident by the low virulence of the 12xbb mutant compared to the 10x mutant were confirmed by the infection phenotype of a bot2 boa6 double mutant generated previously with resistance markers [32]. The bot2 boa6 mutant showed reduced lesion formation in all tissues, in particular apples compared to WT, consistent with published data [8] but it was still more virulent than the 12xbb mutant when the three strains were inoculated on the same tissue (S4 Fig). Despite their reduced virulence, the 12xpg and 12xbb mutants were eventually able to sporulate on infected bean and tomato leaves, similar to WT (S5 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Infection tests of multiple CDIP/ phytotoxin mutants on different host tissues.

A: Attached Phaseolus bean leaves (48 h.p.i.). B: Detached tomato leaves (48 h.p.i.). C: Detached maize leaves (72 h.p.i.). D: Apple fruits (96 h.p.i.). The p values by one-sample t test to a hypothetical value of 100% (WT) are shown, indicating significant reductions of the mutants compared to WT lesions. * p < 0.05; ** p < 0.01; *** p < 0.001. For comparison of the mutants, results of a one-way ANOVA and Tukey’s post hoc test are displayed with compact letter display. The pictures show lesions caused by 12xbb and 12xpg mutants in comparison to WT.

https://doi.org/10.1371/journal.ppat.1010367.g003

Microscopical analysis of infection

The infection process of B. cinerea can be divided into penetration, primary lesion formation, lesion expansion, and sporulation. Inoculations of killed onion epidermal cells revealed that in the absence of induced plant defence responses, germination and penetration of the 12x mutants proceeded largely normally. The germ tubes of the 12xpg and 12xbb mutants were slightly longer than WT germ tubes, indicating a small delay in penetration of the mutants (Fig 4A and 4B). To analyse the early stages of infection on living plant tissue, microscopic studies were performed. On detached Phaseolus leaves, plant cell necrosis after 24 h by up to 6x mutants was similar to WT, whereas significantly reduced necrosis was observed with 8x, 10x, and 12x mutants (Fig 4C and 4D). These data indicate that plp1 and ieb1, alone or in combination with six other CDIPs, are necessary for tissue invasion and host cell killing.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Microscopic analysis of penetration and host necrosis induction by B. cinerea WT and mutants.

A: Penetration of onion epidermis cells by germinated conidia of WT, 12xpg and 12xbb mutants. Superfical structures, namely conidium (c), germ tube (gt) and appressorium (ap) are stained with trypan blue, whereas intracellular hyphae (ih) remain unstained. Scale bars: 10 μm. B: Quantification of germ tube length of germinated and penetrated conidia. One-way ANOVA and Dunnett´s post hoc test (control: WT). *** p value < 0.001. C: Induction of host cell necrosis by B. cinerea on detached Phaseolus leaves (24 h.p.i.). Fungal hyphae were stained with calcofluor white, host cell death is visible by loss of red autofluorescence. Scale bars: 500 μm. D: Mean extent of host necrosis under the inoculation site. One-way ANOVA and Dunnett´s post hoc test (control: WT). ** p < 0.01; *** p < 0.001.

https://doi.org/10.1371/journal.ppat.1010367.g004

Analysis of the secretomes of B. cinerea WT and multiple mutants

To verify the loss of proteins encoded by the deleted genes in the mutants, a proteome analysis of on planta produced secretomes was performed as described previously [3]. Out of the 12 CDIPs analysed, nine could be detected by MS/MS in the WT secretome, consistent with previous studies [2, 3], whereas Nep1, Plp1 and Xyl1 were never detected in our or other published secretome studies. In all mutants, proteins were missing when the respective gene had been deleted (S4 Table). Since most of the deleted CDIPs are highly expressed in the WT, we checked whether their loss was compensated by overexpression of other proteins in the secretomes of the 10x, 11x, 12xpg and 12xbb mutants. However, because of the high variability of intensities of the detected secreted proteins between independent samples, compensatory changes were difficult to determine. Analysis of the proteome data using the Perseus bioinformatic platform [46] did not reveal evidences for differential protein abundance in WT and mutant secretomes.

The on planta secretomes of B. cinerea are highly phytotoxic when infiltrated into leaves [2, 22]. Similar toxicity was observed for the secretomes of WT and up to 8x mutants. The secretomes of 10x, 12xpg and 12xbb mutants showed reduced cell death inducing activity. Compared to 10x, the secretomes of 12xpg and 12xbb showed decreased activity on V. faba and N. benthamiana, respectively, indicating differential effects of the loss of PG1/PG2 (in 12xpg) and botrydial/botcinin (in 12xbb) on different plant species (Fig 5A and 5B). As the 10x and the 12xbb mutant differed in their ability to synthesize botrydial and botcinins, we investigated the contribution of the two phytotoxins. Secretomes collected from the two mutants were fractionated by ultrafiltration through a membrane with 10 kDa molecular weight cutoff. No protein could be detected in the filtrates, and only the filtrate of the 10x but not the 12x mutant caused necrosis, which could be attributed to the presence and absence of botrydial and botcinins, respectively (Fig 5C). Heating of the 10x mutant filtrate to 95°C for 20 min did not signficantly reduce its phytotoxic activity (Fig 5D).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Phytotoxic activity of B. cinerea secretomes after leaf infiltration.

A-B: Necrotic lesions caused by WT and mutant secretomes (2 μg ml^-1) in infiltrated tobacco (A) and faba bean (B) leaves after three days. Values are the means of at least three experiments and two or three leaves per experiment. One-way ANOVA and Dunnett´s post hoc test (control: WT); **p<0.01; ***p<0.001; n≥15 (tobacco) and n≥8 (faba bean). C: Cell death inducing activity on faba bean leaves of non-proteinaceous (<10 kDa) secretome fractions of a 10x mutant and a 12xbb mutant unable to synthesize botrydial and botcinins (n = 6). D: Effects of heating (95°C for 20 min) on CDI activity of the non-proteinaceous fraction of the 10x mutant (n = 6). C, D: Two-tailed t test, **p<0.01.

https://doi.org/10.1371/journal.ppat.1010367.g005

Production of botrydial and botcinins by B. cinerea multiple mutants

To confirm that the differences in the secretomes of the 10x and the 12xbb mutants were due to the presence or absence, respectively, of botrydial and botcinins, the production of these metabolites by the WT and both mutants was investigated. Botryanes (including botrydial) and botcinins are known to reach their maximum amount after five and twelve days, respectively, under the fermentation conditions described in Materials and Methods, therefore these two time points were selected to investigate their production. Fermentation broths were subjected to extraction using ethyl acetate as solvent. Evaporation of solvents from 6 days old cultures yielded oily extracts (WT: 570 mg; 10x: 220 mg; 12xbb: 496 mg), and from 12 day old cultures dense yellow oily extracts (WT: 785 mg; 10x: 276 mg; 12xbb: 650 mg). Toxins were isolated by column chromatography and analysed by HPLC, as well as by ¹H- and ¹³C-NMR. As shown in Fig 6, the extracts of the WT and the 10x mutant yielded a variety of botryanes and botcinins in different amounts as described previously [45], whereas no derivatives with botryane and botcinin skeletons were detected in the 12xbb mutant.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Production of botryanes (including botrydial) and botcinins by B. cinerea and multiple knockout mutants.

A: Structures of botryanes and botcinins detected. B: Metabolite yields (in mg) obtained after 6 d and 12 d of fermentation by B. cinerea WT and 10x mutant, but not by the 12xbb mutant. -: Not detected.

https://doi.org/10.1371/journal.ppat.1010367.g006

Susceptibility of Nicotiana benthamiana sobir1 mutants and plants silenced for bak1 to infection by B. cinerea

For several B. cinerea CDIPs, evidence has been obtained that they are recognized by plant immune receptors that are dependent on BRASSINOSTEROID INSENSITIVE1-ASSOCIATED KINASE1 (BAK1) and/or and the receptor-like kinase SUPPRESSOR OF BIR1 (SOBIR1) for PTI-mediated defence activation. BAK1 has been described as a central regulator of innate plant immunity and is necessary for the function of most known immune receptors [44]. SOBIR1 has been reported to be required for recognition of pathogen-derived secreted proteins, also referred to as secreted invasion patterns [54]. To evaluate the role of BAK1 and SOBIR1, B. cinerea WT and the 12xpg mutant were inoculated on Nicotiana benthamiana WT and sobir1 knockout mutants that have been generated recently by CRISPR/Cas9 [55], and on plants in which expression of bak1 was suppressed by tobacco rattle virus 2 (TRV2)-induced gene silencing (VIGS). As observed on other tissues, lesion formation by the 12xpg mutant was significantly reduced compared to WT on all plants. Furthermore, N. benthamiana WT and sobir1 plants were equally susceptible to infection by B. cinerea (Fig 7A and 7B). In the same line, TRV2-bak1 silenced plants showed similar susceptibility as TRV2-GUS control plants to both B. cinerea WT and 12xpg mutant (Fig 7C and 7D). Unexpectedly, these data do not support a major role of activation of PTI-mediated plant defence for B. cinerea infection.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Infection of Nicotiana benthamiana WT and sobir1 mutants, and plants silenced for bak1, by B. cinerea WT and 12xpg mutant.

A: Lesion formation on N. benthamiana WT and sobir1 leaves (48 h.p.i.). B: Necrosis induction of 12xpg mutant relative to WT. B. cinerea. WT-induced lesions (on N.b. WT: 143.8±9.2; on N.b. sobir1#1: 129.5±10.9; on N.b. sobir1#2: 145.1±17.2 mm²) and relative lesion sizes of 12xpg were not significantly different (n.s.) between N.b. WT and sobir1 mutants (one way ANOVA and Dunnet’s post-hoc test). Data are from three independent experiments, with two plants and two or three leaves per plant each. C: Lesion formation by B. cinerea WT (upper left sides of leaves) and 12xpg mutant (upper right sides) on plants subjected to VIGS (48 h.p.i.). Sites of TRV2 inoculation for VIGS induction are indicated by red arrows. Note the stunted growth and wrinkled leaves of the plant silenced with TRV2-bak1. D: Necrosis induction of 12xpg mutant relative to WT. WT-induced lesions (on N.b. TRV2-GUS: 191,5±9.2; on N.b. TRV2-bak1: 178.4±9.2 mm²) and relative lesion sizes of 12xpg were not significantly different (n.s.) on these plants (one way ANOVA and Tukey’s post-hoc test). Data are from four independent experiments with two batches of VIGS-silenced leaves, with two plants per experiment and two or three leaves per plant.

https://doi.org/10.1371/journal.ppat.1010367.g007