The A. brassicicola wild-type strain Abra43 used in this study has previously been described (Dongo et al., 2009; Joubert et al., 2011a). For routine culture, A. brassicicola was grown and maintained on potato dextrose agar (PDA) or on Vogel minimal medium (Vogel, 1956). For osmotic stress experiments mycelia were grown on PEG-infused agar plates (Verslues et al., 2006). Colony diameters were measured daily and used for calculation of radial growth. The method based on microscale liquid cultivation (from conidial suspensions) and automated nephelometric recording of growth, followed by extraction of relevant variables (lag time and growth rate), was described by Joubert et al. (2010). To study the susceptibility of fungal strains to ITC, allyl-ITC (AlITC), benzyl-ITC (BzITC) or phenetyl-ITC (PhITC), were diluted from stock solutions prepared in acetone at the final desired concentrations (2.5 and 5 mM). ITC were purchased from Aldrich Chemical Co. (Milwaukee, WI). To study the effects of plant extracts on mannitol accumulation, plant extracts were prepared from primary leaves of tomato or radish as described by Ehrenshaft and Upchurch (1993) and sterilized by filtration through a 0.2-μm nitrocellulose filter. Potato dextrose Broth (PDB) containing either 10% (v/v) aqueous plant-leaf extract or an equal volume of sterile distilled water were inoculated with conidia (10⁵ conidia/mL final concentration). Cultures were grown at 24°C with gentle agitation (150 rpm) for 7 days.

Propidium iodide (PI) was used as a cell viability marker. Viable cells with intact membranes exclude PI, whereas the membranes of dead and damaged cells are permeable to PI. Fungal suspensions were prepared on PDB with conidia for 10⁵ conidia/mL (final concentration). Non-germinated conidia and germinated conidia after 15 h of incubation (150 rpm, 24°C) were treated with H₂O₂ (8 mM) or Al-ITC (5 mM). After 30 min of exposure, cells were washed twice with cold phosphate-buffered saline (137 mM NaCl, 2.7 mM KCl, 4.3 mM Na₂HPO₄.7H₂O, and 1.4 mM KH₂PO₄, pH 7.4) and then stained with PI 2 μg/mL (Sigma-Aldrich).

The gene replacement cassettes were generated using the double-joint PCR procedure (Yu et al., 2004). The selectable marker inserted in the PCR constructs corresponded to the Hph gene cassette (1436 bp) from pCB1636 (Sweigard et al., 1995) or the Nat gene cassette (2150 bp) from pNR (Malonek et al., 2004) conferring resistance to hygromycin B and nourseotricin, respectively. The sets of primers used to amplify the 5′ and 3′ flanking regions of each targeted gene are presented in the Table 1. The double-joint final PCR products were used to transform A. brassicicola protoplasts as described by Cho et al. (2006). The A. brassicicola wild-type Abra43 was used to obtain single hygromycin resistant transformant strains Δabmpd and Δabmdh. The Δabmpd genotype was used to obtain ΔΔabmpd-abmdh hygromycin and nourseotricin resistant strains. Potential transformants were prescreened by PCR with relevant primer combinations (Table 1) to confirm integration of the replacement cassette at the targeted locus. Two putative gene replacement mutants for each construct were further purified by three rounds of single-spore isolation and then confirmed by Southern blot analysis.

The Abmdh or Abmpd C-terminal GFP fusion constructs were generated by fusion PCR (Figure 2). Using A. brassicicola genomic DNA as a template, the respective ORFs and 3′ flanking regions were amplified with relevant primer combinations (Table 1). In parallel, a fragment containing the GFP cassettes and Hyg B cassettes were amplified from the plasmid pCT74 (Lorang et al., 2001). The resulting PCR fragments were mixed and subjected to second fusion PCR. A linker containing 3 glycine residues was introduced at the 3′ end of the respective ORFs to replace the stop codons. The final PCR products were transformed in the A. brassicicola wild-type to make AbMpd- and AbMdh-GFP fusion proteins. The transformants with expected genetic integration events were identified by PCR and Southern blot analyses (data not shown).

Genomic DNA extraction and Southern blot analysis were conducted as previously described by Joubert et al. (2011a). Total RNA extractions and amplification experiments were conducted as previously described (Joubert et al., 2011b) using specific primers for AbMdh and AbMpd genes (Table 1).

For plant infection assays on Brassica oleracea plants (var. Bartolo), 5 μL drops of A. brassicicola conidia suspension (10⁵, 10⁴ or 10³ conidia/mL in water) were inoculated on leaves from 5 weeks-old plants. Inocula were symmetrically deposited on the left and right sides from the central vein. The plants were then maintained under saturating humidity (100% relative humidity). Symptoms were observed and samples collected at 2, 4, 6 days post-inoculation (dpi) for the determination of major soluble carbohydrates contents and AbMpd and AbMdh expression analysis. For in planta sporulation assays, symptomatic tissues were sampled and vortexed for 30 s in water containing Tween 20 (0.02%, v/v). The concentration of the resulting conidia suspensions was estimated microscopically using a haemocytometer. For the microscopic analyses, B. oleracea leaf fragments were discolored, cleared and fungal structures were stained with solophenyl flavine 7GFE 500 (Ciba Specialty Chemicals, North Carolina, USA) as described by Hoch et al. (2005). Specimens were observed under a Leica fluorescent microscope (using 480 nm excitation and 527 nm emission).

Seed contamination assessments were estimated as described by Pochon et al. (2012). Two 2.5 μL drops of an A. brassicicola conidial suspension (1 × 10⁵ conidia mL⁻¹ in water) supplemented with 0.01% (v/v) Tween 20 were placed on the five youngest siliques (one drop at the silique base and one in the middle) from 1-month-old A. thaliana (Ler) plants. At least five plants per fungal genotype were inoculated and the experiment was repeated twice. As a control for all experiments, two 2.5 μL drops of a 0.01% (v/v) Tween 20 solution were placed on five siliques of one plant. The plants were then maintained under saturating humidity for 2 days in the dark. Contaminated siliques were harvested 10 dpi. Inoculated or control siliques were dissected with sterile forceps and seeds were carefully harvested to avoid contact with the fungus potentially present on the outer surface of siliques. Seeds were incubated separately on PDA medium for 2 days. A seed was considered as contaminated when incubation resulted in typical A. brassicicola colony development.

B. oleracea seeds were artificially inoculated by incubation (1 h) in a conidia suspension (5 mL at 10⁵ conidia/mL). After removing the solution, the seeds were air-dried for 2 h and separated into two batches. The initial contamination rate was determined on one seed batch before storage. One seed per microplate well was placed in 300 μl of PDB and fungal growth was recorded in a laser-based nephelometer. The mean lag time was calculated and representative of the initial seed infection rate. The second seed batch was stored in a dry dark place at 24°C for 6 months and processed as above. As the lag time was found to be directly proportional to the number of germinating conidia, the viability rate was estimated from the ratio between lag times before and after storage. This experiment was repeated twice for each fungal genotype.

Three-day-old cultures grown in PDB medium were harvested, submerged in liquid nitrogen for 5 min and stored at −80°C until use. Extraction procedures for cell-free extracts were carried out at 4°C and MPD and MDH enzymatic activities were measured as described by Velez et al. (2007). For both enzymes, specific activities were defined as the μmole of NADP(H) or NAD(H) oxidized per minute per mg of protein. Three independent experiments were done for each sample.

Ethanolic extractions of cells from each sample used for hexose sugar and polyol measurements were performed as described by Stoop and Pharr (1993) with minor modifications. Dry powdered samples were suspended in 80% ethanol solution and incubated at 80°C for 5 min. After 5 min of centrifugation at 1000 g, the supernatants were recovered and pellets were re-extracted twice. Pooled ethanolic solutions were evaporated using a vacuum concentrator (speedVac UNIEQUIP), and residues were dissolved in sterile water or D₂O for analysis. High performance liquid chromatography (HPLC) was performed on a Carbopac PA-1 column (Dionex Corp., Sunnyvale, CA, USA) as described by Rosnoblet et al. (2007). For each sample, three independent experiments were done from separate cultures.

Brassicicolin A was extracted as described by (Gloer et al., 1988) from cultures (minimal medium plus thiamine, Pedras et al., 1997) of Abra43 and ΔΔabmpd-abmdh strains. Each filtered culture broth (1 L) was extracted with ethyl acetate (3 × 300 ml), and the organic phase was dried over MgSO₄ and evaporated to generate 25 mg and 13.5 mg of crude extracts from Abra43 and ΔΔabmpd-abmdh strains, respectively. Liquid chromatography-mass spectrometry (LC/MS) was performed on each extract using a Bruker Esquire 3000 Plus electrospray ionization-ion trap mass spectrometer coupled with a Waters 2790 high performance liquid chromatography (HPLC-ESI-MSⁿ). Elution was carried out on an Hypersil RP18 column (250 × 4.6 mm, 5 μm, Termo) using the following gradient: initial mobile phase AcN/H₂O 0.01% formic acid 15/85 reaching 60/40 in 35 min and maintained for 10 min before reaching 100/0 (v/v) until 46 min, with a flow rate of 1 mL/min. Only HPLC grade solvents were used. All samples were diluted in a solution of acetonitrile and filtered (UptiDisc™ PVDF 0.22 μm) prior to HPLC injection. They were analysed at 10 mg/ml concentration. The ESI parameters were as follows: solvent split ratio 1:9; nebulizer: 30 psi; dry gas (N₂): 7 L/min; dry temperature: 340°C; skim: 40 V; trap drive: from 90 to 178, octopole RF amplitude: from 144 to 210 Vpp; capillary exit: from −156 to −240, capillary voltage 4500 V. The ion trap mass spectrometer was run in negative ion scanning mode for m/z ranging from 80 to 1500. MSⁿ was performed at a fragmentation amplitude ranging from 0.8 to 2.0 V depending on the samples. Preparative thin-layer chromatography (PTLC) was carried out on silica gel 60F254 (0.25 mm, Merck) using MeoH/CHCL₃ (5/95) as eluent. This experiment was done twice from separate cultures of each fungal genotype.

For nuclear magnetic resonance (NMR) metabolite analysis, 500 μl of the D₂O-samples were transferred to a 5 mm NMR tube, then analyzed on a BRUKER Advance DRX 500 MHz spectrometer equipped with a multinuclear QNP probe (Bruker, Wissembourg, France). Proton-decoupled ¹³C NMR spectra (sweep width = 31 450 Hz) were recorded at 125 MHz excitation frequency, 30-degrees pulse angle (6.5 μs pulse duration) at 2 s intervals. The free induction decays were collected as 32 K data points and processed with a 1–2 Hz exponential line broadening for ¹³C NMR. Maleic acid (δ_CH 130.4 ppm) was the external reference for chemical shifts. Identifications were made by comparison with spectra of pure known standards. For brassicicolin A, ¹H (500 MHz) and 2D NMR spectra (HMQC and COSY) were recorded in CDCl₃ in a capillary probe (Bruker TXI 1.7 mm) with chloroform resonances (δ_H 7.28, δ_C 77.0 ppm) as internal references. For each sample, NMR analysis was done twice.

The presumed Mpd and Mdh loci were identified by a homology search against the A. brassicicola genome assembly (http://genome.jgi-psf.org/Altbr1/Altbr1.home.html) with genes previously described in A. alternata (Velez et al., 2007). AbMdh and AbMpd sequences (GenBank accession No JX403801 and JX403800, respectively) consisted of 851 and 1173 nucleotides, respectively. Blast search on the whole genome sequence and Southern analyses suggested the presence of only one copy of each gene (Figure 3B). Among the putative regulatory elements identified on sequences upstream of the ATG, consensus sequences for the binding of transcription factors involved in response to thermal, osmotic and oxidative stresses (Msn2p/Msn4p, HsF2) were found on the two genes.

The resulting AbMdh protein belongs to the short-chain group of the dehydrogenase/reductase superfamily (Jornvall et al., 1995) and has 95, 88, and 75% identity to the corresponding proteins described in A. alternata, S. nodorum, and B. cinerea, respectively. The AbMpd amino acid sequence shares 92, 82, and 57% identity with that of A. alternata, S. nodorum, and B. cinerea, respectively, and contains both NAD-interacting domains and a specific mannitol-1-phosphate dehydrogenase motif.

For each targeted gene, 2 replacement mutants (called Δabmdh1-2, Δabmpd1-2) were generated by replacing the targeted ORF with a hygromycin B resistance cassette. Two ΔΔabmpd-abmdh1-2 double deletion mutants were then constructed by transforming the Δabmpd genotype with an AbMdh-replacement cassette containing a nourseothricin-resistance marker (Figure 3). In all further experiments, the phenotypic characters for transformants of the same genotype were not found to be significantly different.

To confirm that gene inactivation impacted the enzyme activity, the transformants were grown in liquid culture and analysed for their ability to reduce fructose using NADPH as cofactor, or for fructose-6-phosphate conversion to mannitol-1-phosphate. Enzymatic assays (Table 2) confirmed that AbMdh and AbMpd deletions abolished mannitol dehydrogenase and mannitol-1-phosphate dehydrogenase activities, respectively.

The effects of the Δabmdh, Δabmpd, and ΔΔabmpd-abmdh mutations on the accumulation of sugars and sugar alcohols were estimated by ¹³C NMR. Ethanolic extracts of sporulating mycelium grown in synthetic Vogel medium with glucose (2%) were obtained and analysed (Figure 4A). The wild-type and Δabmpd extracts exhibited similar sugar profiles. However, lower amounts of mannitol and higher amounts of trehalose were found in the latter genotype. Conversely, the ¹³C-NMR spectra of Δabmdh mutants were dominated by mannitol resonances. Mannitol was absent in extracts from ΔΔabmpd-abmdh mutants in which trehalose and glycerol appeared to be the major compounds. Quantitative estimations of the mannitol content in the different genotypes during in vitro development were obtained by HPLC analysis of extracts from mature conidia and young non-sporulating mycelia (Figure 4B). While the wild-type accumulated nearly the same amount of mannitol in conidia and mycelia, this polyol was exclusively detected in conidia of the Δabmpd mutants. By contrast, the Δabmdh mutants preferentially accumulated mannitol in mycelia. No mannitol was detected in either conidia or young mycelia (30 h post germination) from the ΔΔabmpd-abmdh mutants. However, traces of mannitol were detected from the double deletion strains in 1-week-old cultures (data not shown).

Mannitol has been proposed to act as a potent protective metabolite against oxidative stress. As the mannitol contents of conidia and mycelia from the different mutant genotypes were significantly different from that of the wild-type, the effects of AbMpd and AbMdh inactivation in A. brassicicola on conidia germination and initial mycelium growth in the presence of oxidative stressors were examined. Analysis of nephelometric growth curves revealed that under non-stress conditions (PDB medium), no significant phenotypic differences in conidia germination (based on the lag time parameter) or mycelium growth (based on the maximum growth rate parameter) were detected in any of the tested mutants as compared to the wild-type (data not shown). By contrast, Δabmdh and Δabmpd mutants, and to a lesser extent ΔΔabmpd-abmdh mutants, were far more susceptible than the wild-type to 1 mM H₂O₂ treatments (Figure 5A). The Δabmpd and Δabmpd-abmdh mutants were also more susceptible to exposure to the superoxide-generating compound menadione than the wild-type and Δabmdh genotypes (Figure 5B). As some brassicaceous defense metabolites have antifungal properties that might be at least partially linked to their capacity to generate oxidative stress (Sellam et al., 2007a), the susceptibility to allyl-ITC (Al-ITC), benzyl-ITC (Bz-ITC), and phenetyl-ITC (Ph-ITC) and brassinin were investigated (Figures 5C–F). A genotype susceptibility pattern similar to that obtained with menadione was observed in the presence of 5 mM Al-ITC, 5 mM Bz-ITC, 10 mM Ph-ITC and 200 μm brassinin. Note that after 24 h of exposure to ITC, the mannitol content of wild-type mycelia was 1.4-fold higher than in control and a 3-fold increased expression of AbMpd was recorded in cultures after 3 h of ITC treatment, while the AbMdh expression level remained unchanged (data not shown).

Based on these susceptibility patterns, comparing the response of the four genotypes to H₂O₂ on one hand and to menadione or ITC on the other suggested different modes of action for these oxidants. This was further supported by comparing the effects on the cell viability of non-germinated and germinated conidia of low-time exposure to H₂O₂ and Al-ITC at their respective IC₅₀ (Figure 6). On germinated conidia, treatment with Al-ITC resulted in PI-positive hyphae suggesting a loss of cell viability. No PI staining was observed in non-germinated conidia exposed to Al-ITC. Conversely, H₂O₂ caused cell death in non-germinated or germinated conidia.

Besides its potential role as antioxidant molecule, mannitol has been proposed to provide protection against drought stress (Dijksterhuis and de Vries, 2006). Media with different water potentials (from −0.25 MPa to −1.7 MPa) were prepared using a PEG-infused plate protocol (Verslues et al., 2006), inoculated with the different A. brassicicola genotypes and radial growth was recorded after 15 days of incubation. A shown in Figure 7, strains lacking functional MPD (Δabmpd and ΔΔabmpd-abmdh) were much more susceptible to low water potential treatments than the wild-type. Conversely, Δabmdh mutants were more tolerant than the wild-type.

The HPLC profiles of the major soluble carbohydrates present over the course of A. brassicicola infection on B. oleracea leaves were established (Figure 8A). At 2 dpi, the wild-type strain produced characteristic appressoria-like structures at the tips of germ tubes in contact with the leaf epidermis (Figure 8B). Small necrotic symptoms were observed at 4 dpi, and they continued to expand into large typical necrotic areas surrounded by chlorotic halos at 6 dpi. At this infection stage, necrotic spots exhibited dense conidia formation on the surface. The HPLC analysis showed that, while the mannitol level remained below the detection limit in control plants, mannitol accumulated throughout infection and revealed a twenty-fold increase from 2 dpi to 4 dpi. A small decrease in mannitol was then observed at 6 dpi when sporulation was abundant. However, the mannitol level at this stage was still significantly higher than that detected at 2 dpi. Trehalose was only detected at low levels at 4 dpi and 6 dpi. Sucrose, which was the major carbohydrate in control non-inoculated plant samples, quickly decreased to undetectable levels at 4 dpi.

The expression of AbMdh and AbMpd during infection was also examined (Figure 8C). At 2 dpi, the expression level of AbMpd increased and remained higher than during in vitro growth over the time course of the experiment. Increased AbMdh expression was observed at a later stage and reached 3.5-fold its basal expression level at 6 dpi. These results are consistent with the metabolic profiling, which indicated a prevalence of mannitol production during plant infection. In order to follow the in vitro and in planta spatial expression patterns of both genes, strains expressing AbMpd and AbMdh, under the control of their own promoter and fused at their carboxy-terminal end to SGFP, were engineered. None of the transformants showed visible phenotypic changes discernible from the wild-type except for expression of green fluorescence (data not shown). In both strains, the fluorescence signal was detected in mature conidia and in the young germ tubes (Figure 9). In the in vitro-produced mature hyphae, the signal was still detectable but was much weaker in the AbMpd-GFP strain than in the AbMdh-GFP strain. Consistent with the in planta expression patterns, GFP fluorescence appeared to be stronger in the mutant during host plant infection. In AbMpd- and AbMdh–GFP fusion mutants, fluorescence intensity increased dramatically during the in vitro or in planta conidiation process, reaching a maximum in young conidia (Figure 9).

To test the hypothesis that host plants would elicit changes in fungal mannitol production, the A. brassicicola wild-type strain was cultured for 7 days in the presence and absence of leaf extracts from host (B. oleracea) or non-host (Solanum lycopersicum) plants. The amount of mannitol in the fungal mycelia was then determined by HPLC (Figure 10). Fungal growth was essentially unaffected by plant extracts (data not shown). A. brassicicola responded to the presence of host plant extracts by accumulating significantly higher levels of mannitol as compared to the amount detected in control culture or in the presence of non-host extract.

The accumulation of mannitol throughout infection of B. oleracea by A. brassicicola and the increased susceptibility of mannitol biosynthesis mutants to oxidative stress prompted us to comparatively evaluate the pathogenicity of the different fungal genotypes. The wild-type and Δabmdh, Δabmpd, and ΔΔabmpd-abmdh mutants were all able to produce typical symptoms (Figure 11A). However, as judged from the lesion sizes at low inoculum charge, significant decreases in aggressiveness (up to 85% that of the wild-type at 10³ conidia per ml) were recorded for the ΔΔabmpd-abmdh mutants and to lesser extent for the Δabmdh and Δabmpd mutants. Closer inspection of symptoms suggested that weak in planta sporulation occurred on necrosis obtained after inoculation with AbMdh deficient mutants (Figure 11B). This was confirmed by measuring the quantity of conidia produced per mm² of necrotic tissue. All genotypes produced significantly fewer conidia in planta than the wild-type, with a 90% reduction for Δabmdh mutants.

To check whether the lower aggressiveness of mutants could be correlated with in planta spore germination and/or plant tissue penetration defects, observation of solophenyl stained samples at 1 dpi was performed to quantify germinated conidia and appressoria-like structures at the plant surface. Although nearly 95% of conidia from all genotypes were germinated at this time, only 5 and 20% had differentiated infection structures in samples inoculated with the ΔΔabmpd-abmdh and the Δabmpd mutants, respectively, vs. 40% for samples inoculated with the wild-type or Δabmdh mutants. Following host penetration, the fungus has to produce necrotic factors to progress within infected tissues. Brassicicolin A, a fungal metabolite considered as being a mannitol derivative, represents a potent necrotic toxin produced by A. brassicicola (Pedras et al., 2009). We hypothesized that the weak virulence in mutants lacking mannitol may also be explained by the absence of brassicicolin A synthesis. To test this hypothesis, ethyl acetate extracts of culture filtrates from submerged cultures of A. brassicicola wild-type and ΔΔabmpd-abmdh strains were analysed by HPLC-UV-MS. The resulting total ion chromatograms (TIC) of both extracts revealed a major metabolite at R_t = 39 min (Figures 12A,B) exhibiting the quasimolecular ion [M-H]⁻ (m/z 683) of brassicicolin A (C₃₂H₄₈N₂O₁₄), whereas typical MS/MS fragments (1 V) were recorded at m/z, 565 (loss of two acetyl units) and 473 (loss of one α-hydroxyisovaleryl unit together with one α-isocyanoisovaleryl unit) (data not shown). The corresponding compound, which seems to accumulate in the same amounts in both strains, was purified through preparative TLC from the wild-type extract and analyzed (¹H NMR, COSY and HMQC) in a capillary NMR probe (500 MHz). The resulting data (data not shown) were also in full agreement with former measurements obtained by Gloer et al. (1988) for this mixture of stereoisomers and confirmed that the isolated compound was brassicicolin A. We concluded that brassicicolin A was present in organic extracts from the culture broths of both the wild-type strain and the ΔΔabmpd-abmdh mutant and that the attenuated virulence of mannitol-deficient mutants could not be linked to the loss of brassicicolin A production.

A. brassicicola is a seed-borne pathogen and has potential for long-term survival on dry seeds. As the differential abilities of the mutants to overcome a water potential stress have been observed, their capacity to survive on artificially contaminated seeds during storage was examined. After 6 months of storage, the percentage of viability was estimated using laser nephelometry growth curves on the basis of Δlag time values. As shown in Figure 13A, ΔΔabmpd-abmdh mutant survival was significantly lower than that of other tested genotypes. Similarly, the ability of A. brassicicola to efficiently infect seeds has been correlated with their capacity to cope with severe stress conditions consecutive to gradual dehydration of maturating reproductive organs (Iacomi-Vasilescu et al., 2008). Using the model pathosystem recently described for investigating seed transmission in Arabidopsis plants (Pochon et al., 2012), the abilities of the mutants to transmit to seeds were compared with that of the wild-type. As shown in Figure 13B, the transmission capacity to A. thaliana seeds was significantly affected in Δabmpd and Δabmdh mutants and almost completely abolished in the ΔΔabmpd-abmdh mutants.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.