Dehydrin-like Proteins in the Necrotrophic Fungus Alternaria brassicicola Have a Role in Plant Pathogenesis and Stress Response

In this study, the roles of fungal dehydrin-like proteins in pathogenicity and protection against environmental stresses were investigated in the necrotrophic seed-borne fungus Alternaria brassicicola. Three proteins (called AbDhn1, AbDhn2 and AbDhn3), harbouring the asparagine-proline-arginine (DPR) signature pattern and sharing the characteristic features of fungal dehydrin-like proteins, were identified in the A. brassicicola genome. The expression of these genes was induced in response to various stresses and found to be regulated by the AbHog1 mitogen-activated protein kinase (MAPK) pathway. A knock-out approach showed that dehydrin-like proteins have an impact mainly on oxidative stress tolerance and on conidial survival upon exposure to high and freezing temperatures. The subcellular localization revealed that AbDhn1 and AbDhn2 were associated with peroxisomes, which is consistent with a possible perturbation of protective mechanisms to counteract oxidative stress and maintain the redox balance in AbDhn mutants. Finally, we show that the double deletion mutant ΔΔabdhn1-abdhn2 was highly compromised in its pathogenicity. By comparison to the wild-type, this mutant exhibited lower aggressiveness on B. oleracea leaves and a reduced capacity to be transmitted to Arabidopsis seeds via siliques. The double mutant was also affected with respect to conidiation, another crucial step in the epidemiology of the disease.


Introduction
Dehydrins belong to the 2a and 2b groups of the large late embryogenesis-abundant (LEA) protein family [1,2], which were initially described as accumulating late in plant seed development. They were also found in vegetative plant tissues following environmental stresses and are believed to play a role in the protection against cold-and dehydration-related stresses [3,4,5]. The dehydrin family seems to be widely distributed as homologous sequences were found in invertebrate and microorganism genomic sequences [6]. Besides the sequence similarities, dehydrins share typical physicochemical features. Their amino acid composition is thus characterized by high percentages of glycine, threonine and serine, and low levels of cysteine and tryptophan residues. Moreover, they exhibit high hydrophilicity, an absence of secondary structures, a high proportion of disordered amino acids and can thus be considered as intrinsically unstructured proteins (IUPs). Disordered regions in dehydrins may constitute flexible linkers or spacers that have a role in forming macromolecular assemblies. In virtue of this structural plasticity, IUPs may serve as potent chaperones [7]. Moreover, a distinctive feature of plant group 2 LEA proteins is a conserved Lys-rich motif named Ksegment [8] usually found in the N terminus of the protein. The K-segment motifs are predicted to form amphipathic a-helical structures similar to the lipid binding class A2 amphipathic ahelical region found in apolipoproteins associated with membranes [9]. This observation raised the hypothesis that one of the roles of the plant dehydrins may be related to membrane stabilization during stress through interaction with hydrophobic surfaces [10].
In fungi, dehydrin-encoding genes have been identified in Tuber borchii and Aspergillus fumigatus during searches for genes controlling fruiting body maturation or conidial dormancy [6,11]. Recently, Wartenberg et al [12] also identified a farnesol-induced dehydrinlike protein (DlpA) in A. nidulans. Homologs were found in ascomycetous fungi but not in other fungal lineages. Their amino acid sequences harbor a repeated and conserved asparagineproline-arginine (DPR) motif, corresponding to a fungal dehydrin signature pattern [6], while a relatively poor sequence homology is generally found outside of DPR motifs. The corresponding transcripts generally accumulate in the same conditions as those affecting plant DHNs transcription, e.g. in response to hyperosmosis, low temperature and salinity stresses. The three dehydrinlike proteins (DprA, DprB and DprC) identified in A. fumigatus and DlpA in A. nidulans were also up-regulated upon addition of dithiothreitol (DTT), an inducer of protein misfolding, and were described as stress protective molecules [11,12,13]. They notably contribute to protection against high pH, freezing, osmotic and oxidative stress. According to these features, they were presumed to function as molecular chaperones or membrane stabilizers. Different locations have been assigned to them. While DprC was associated with the vacuoles, DprA and DprB accumulated in the cytoplasm and peroxisomes. The authors hypothesized a peripheral association of DprA and DprB proteins with the peroxisome due to the absence of peroxisomal-targeting signal. Loss of DprA and DprB was not found to be associated with attenuated virulence in mice even though the conidia of mutants were hypersensitive to killing by lung phagocytes.
The necrotrophic fungus Alternaria brassicicola causes black spot disease and is an economically important seed-borne pathogen of Brassicaceae species. During host infection, A. brassicicola is exposed to high levels of defence compounds, such as phytoalexins and glucosinolate breakdown products, and the ability to overcome these antimicrobial metabolites is a key factor in determining fungal virulence. Seed transmission is also a major component of pathogen fitness, as A. brassicicola strongly depends on this process for its dispersal and long-term survival [14]. In this study, we identified three dehydrin-encoding genes (called AbDhn1, AbDhn2 and AbDhn3) in the A. brassicicola genome via their sequence similarities and physicochemical characteristics. Expression analyses indicated that dehydrin gene transcription was dependent on the AbHog1 MAP kinase and that two AbDHN2 isoforms could be produced through alternative splicing of mRNA. We then investigated their role in protection against dehydrationrelated stresses and, for the first time, we analysed their impact on the pathogenicity during host vegetative tissue infection or the seed transmission process.

AbDhn1, AbDhn2 and AbDhn3 Encode Three Dehydrinlike Proteins
The camalexin-induced sequence P3E9 (GenBank accession No DY543081) was previously shown to encode a 157-residue protein that exhibits similarities to dehydrin-like proteins from Tuber borchii [15]. This protein, here referred to as AbDhn1, indeed contained two amino acid repeats ( Fig. 1) that corresponded to the signature pattern of fungal dehydrins [6] recently named DPR domains [11]. A BLAST search was conducted in the A. brassicicola automatically annotated genome database (http://genome.jgi-psf. org/Altbr1) to check for proteins containing similar motifs. Besides a protein referred to as AB02513.1 corresponding to AbDhn1, two other hits (AB08993.1 and AB05365.1) were found and named AbDhn2 and AbDhn3 (GenBank accession No JX891381 and JX891382), respectively. The nucleotide sequence encoding the latter protein was located at the end of a contig and 39 RACE was performed to obtain the C-terminal encoding sequence. The complete AbDhn3 protein contains 964 amino acids and three DPR motifs (Fig. 1). The automatic annotation at the locus encoding AB08993.1 predicted four introns. Amplification of firststrand cDNA with primers spanning the whole coding sequence generated two products differing of ca 150 bp in size ( Fig. 2A). Analysis of the nucleotide sequences of the two amplicons and comparison with the corresponding genomic sequence suggested that the AbDhn2 gene contains two conventional 54 bp-and 57 bp-introns near the 59 and 39 ends respectively, both being absent in the amplified cDNA sequences, and a third longer 159 bp-intron only present in the larger amplification product (Fig. 2B). Blast search in the A. brassicicola genome assembly database and Southern hybridization (Fig. S1) did not reveal any additional AbDhn2 coding sequence, strongly suggesting that the two transcripts were produced by alternative splicing of a AbDhn2 pre-mRNA. The two deduced AbDhn2 protein isoforms contain 453 amino acids (a isoform) and 400 amino acids (b isoform) and five DPR motifs (Fig. 1). A stretch of eleven amino acids repeated seven times was also observed at the N-terminal part of their sequence (Fig. S2).
The alignment of the DPR domains from the three A. brassicicola dehydrin-like proteins is shown in Fig. 3. Apart from the conserved domains, sequence comparison of these proteins revealed very poor homology, and sequences producing significant alignments (.50% identity and coverage) were only detected in genomes of closely related Pleosporales species (A. arborescens, Pyrenophora tritici-repentis, P. teres, Phaeosphaeria nodorum). However they all shared typical features of dehydrins such as high glycine, threonine and serine content, low cysteine and tryptophan content, high hydrophilicity, i.e. negative GRAVY values ( Table 1). Moreover they lack well-defined tertiary structure as revealed by their high percentage of amino acids predicted as being disordered and as potential binding regions (Fig. 1). In AbDhn1, one of these potential binding regions contained a Wxxx[YF] motif found in proteins involved in peroxisomal matrix import into peroxisomes.

Stress-regulated Expression of AbDhn Genes
Previous experiments showed that the AbDhn1 gene was upregulated during exposure of A. brassicicola to the indolic phytoalexin camalexin [16]. We thus investigated the expression of the three dehydrin-like genes in germinated conidia exposed to camalexin but also to other brassicaceous defense metabolites, i.e. the indolic phytoalexin brassinin, and allyl-isothiocyanate (Al-ITC), a breakdown product of the aliphatic glucosinate sinigrin. As shown in Fig. 4, AbDhn1 and AbDhn2 transcription levels increased right after 0.5 h of exposure to camalexin and brassinin and the highest transcripts levels were observed after 2 h of treatment with the two phytoalexins. Increased AbDhn1 and AbDhn2 expressions were also observed in germinated conidia following 2-4 h of exposure to Al-ITC.
Transcript levels were also measured for the three dehydrin-like genes in conditions previously reported to induce fungal DHN transcription, i.e. low temperature-, salinity-, osmotic-and oxidative-stress [6,11,12,13]. AbDhn1 and AbDhn2 transcript levels dramatically increased immediately following transfer on NaCl containing medium for 0.5 h and up to 2 h, while this upregulation was no longer observed after prolonged (4 h) treatment. A similar strong upregulation response was observed for AbDhn2 after 2-4 h of exposure to cold temperature. Increased expression was also observed for AbDhn1 in the presence of H 2 O 2 and for AbDhn2 and AbDhn3 in the presence of sorbitol, but at much lower levels.
In A. brassicicola the Hog1 MAPK cascade plays an important role in the response to camalexin exposure, oxidative stress [16] and salt stress (unpublished results). To check whether dehydrinlike genes constitute potential targets for this signalling cascade, their expression was assessed in a Dabhog1 mutant. As shown in Fig. 5A, significantly decreased accumulation of the AbDhn transcripts was observed in the MAPK-deficient strain compared to the wild-type under inducing conditions (NaCl stress). As camalexin has also been shown to activate the unfolded protein response (UPR) pathway in A. brassicicola [17], and due to the potential molecular chaperone function of intrinsically unstructured proteins (IUPs), expression of the three dehydrin-like genes was investigated, either in the wild-type strain exposed to the chemical UPR-inducer dithiothreitol (DTT), or in the UPRdefective DabhacA mutant without exogenous stress application. Upon treatment with DTT for 0.5 h, significant upregulation of AbDhn1 and AbDhn2 expression was observed in the wild-type strain (Fig. 4). These inductions were not observed for longer DTT exposure times. In the absence of exogenous stress, a dramatic increase in AbDhn1 transcripts was observed in the DabhacA mutant, deficient for the transcription factor AbHacA, as compared to the wild-type (Fig. 5B).

Dehydrin-like Proteins AbDhn1 and AbDhn2 Accumulate under Salt Stress in Peroxisomes
To check whether increased levels of dehydrin transcripts were paralleled by the accumulation of dehydrin-like proteins, strains expressing AbDhn proteins, under the control of their own promoters and fused at their carboxy-terminal end to sGFP, were engineered. As a control, a strain expressing only sGFP under control of the ToxA promoter [18] was also constructed. Proteins extracted from each strain grown in liquid cultures in the presence or absence of NaCl were analysed by Western blot using anti-GFP antibodies. As shown in Fig. 6, under basal growth conditions no signal was observed except with proteins extracted from the control strain constitutively expressing GFP. After salt-stress exposure, GFP antibodies recognized one protein at ca 40 kDa in extract from the strain expressing AbDhn1-GFP and two proteins at ca 65-70 kDa, i.e. that could correspond to the a and b isoforms of AbDhn2, in extracts from the strain expressing AbDhn2-GFP. No signal was observed with extracts from the strain expressing AbDhn3-GFP, suggesting that this protein was not expressed or at a very low level at least under the salt stress conditions.
The subcellular localization of AbDHN proteins was checked using the same strains. The GFP-Dhn1 signals formed punctate green dots in conidium cells (Fig. 7A) and young vegetative hyphae (Fig. 7B) that tended to be larger in older hyphae. Similar   expression patterns were observed with the strain expressing AbDhn2-GFP, whereas GFP expression in the control strain resulted in diffuse green fluorescence (Fig. S3). Based on previously published results on dehydrin-like proteins from A. fumigatus, a peroxisomal localization of AbDhn1 and AbDhn2 proteins was hypothesised. To achieve labeling of peroxisomes, the red fluorescent protein DsRed was fused to a serine-lysine-leucine tag (SKL) typical of the type 1 peroxisomal targeting sequence PTS1. Expression of DsRed-SKL in A. brassicicola leads to the import of DsRed into peroxisomes and results in the detection of fluorescence signals after excitation. In the transformants that coexpressed AbDhn1-or AbDhn2-GFP and DsRed-SKL, colocalization of sGFP and DsRed was observed in conidia and hyphae ( Fig. 7), confirming that the two dehydrin-like proteins were associated with peroxisomes. For Dhn3, only faint fluorescent signals were observed at the tip of hyphae but the cellular localization could not be accurately determined (not shown).

Dehydrin-like Deficient Mutants are Susceptible to Oxidative Stress
To investigate the role of dehydrin-like proteins in A. brassicicola, knockout mutants deficient for AbDhn1, AbDhn2 or AbDhn3 were constructed by replacing the respective ORFs with a hygromycin B-resistance cassette (Fig. S1A). The resulting replacement mutants (called Dabdhn1, Dabdhn2, and Dabdhn3) were selected among hygromycin-resistant transformants using a PCR screen (not shown). For each genotype, disruption of Dhn genes was confirmed by DNA gel blot analysis (Fig. S1B) in two to three randomly selected transformants. Besides the expected signals, the DNA gel blot analysis did not detect any additional fragments, implying that there were no structural homologs of AbDhn1, AbDhn2 or AbDhn3 in A. brassicicola. Monitoring growth in solid PDA medium (Fig. S4A) and in liquid PDB medium (Fig. S4B) did not reveal any effect of single mutations compared to the wildtype parental strain on the mycelium growth rate, conidia germination and initial hyphal growth. Analyses of growth curves in liquid medium supplemented with menadione (generation of O 2 2 ), H 2 O 2 , Allyl-isothiocyanate (Al-ITC) and brassinin were used to assess the susceptibility of the dehydrin-like mutants to oxidative stress and plant defense metabolites. For a given genotype, all of the tested transformants had a similar response to the applied stress and mean growth curves were thus constructed. Lag time and maximal growth rate variables were calculated from the growth curves using a calculation method described by Joubert et al [19]. For each parameter, Student's Ttest was used to assess significant difference between the treated and untreated samples or between mutants and parental isolate. As shown in Fig. 8, all the dehydrin-like mutants were characterized by high susceptibility to oxidative stress. Almost complete growth inhibition of the mutant strains was observed in the presence of 10 mM menadione and 5 mM H 2 O 2 while in the same conditions the growth of the wild-type was not significantly affected ( Fig. 8). Similarly, Al-ITC induced delayed entry into the log phase (ca. 7 h) and brassinin caused a significant reduction in the maximum slope (from 23% for Dabdhn3 to 42% for Dabdhn1) for the three dehydrin-like deficient mutants compared to the wildtype (Fig. 8).
Similar experiments were performed in the presence of NaCl and sorbitol to assess the susceptibility of the mutant strains to saltand osmotic-stress, respectively. None of the dehydrin-like deficient mutants showed increased susceptibility to any of these stresses compared to the wild-type (data not shown). As the expression patterns of AbDhn1 and AbDhn2 were quite similar, a possible functional redundancy was suspected and an AbDhn1-AbDhn2 double deletion mutant was constructed by transforming a Dabdhn1 mutant strain with an AbDhn2-replacement cassette containing a nourseothricin-resistance marker (Fig. S1A). The double gene replacement strains (called DDabdhn1-abdhn2) were selected for both nourseothricin and hygromycin resistance. DNA gel blot analysis revealed that these transformants had lost both the AbDhn1 and AbDhn2 coding sequences that were replaced by the two disruption cassettes (Fig. S1B). In both liquid and solid media, growth retardation was observed as compared to other genotypes and the mycelial colony formed by the DDabdhn1-abdhn2 mutant was not darkly melanised (Fig. S4) due to weak conidia production. No other phenotypic difference was recorded between the double mutant and the Dabdhn1 and Dabdhn2 single mutant strains when grown in the presence of salt or sorbitol (data not shown).

Conidia from Dehydrin-like Deficient Mutants Display Altered Survival Rates upon Exposure to High and Freezing Temperatures
Nephelometric recording of growth was used to calculate lag times and assess conidia germination after 10 h of storage in water suspensions at normal (20uC), low (4uC and 220uC) or high (40uC) temperatures. As the lag time was found to be directly proportional to the number of germinating conidia [19], the viability rate was estimated from the ratio between lag times before and after storage. Compared to storage at 20uC, increased lag . Expression levels of AbDhn1, AbDhn2 and AdDhn3 sequences in A. brassicicola exposed to various stresses. Firststrand cDNAs were prepared from RNA samples extracted from germinated conidia either exposed to 125 mM camalexin (CAM), 125 mM brassinin (BRA), 2.5 mM allyl-isothiocyanate (Al-ITC), 5 mM menadione (MEN), 5 mM hydrogen peroxide (H2O2), 1 M sorbitol (SORB), 350 mM sodium chloride (NACL), 20 mM dithiothreitol (DTT) or incubated at 4uC (COLD) for the indicated times and used as template for real-time PCR. For each gene, expression induction is represented as a ratio (fold induction) of its relative expression (studied gene transcript abundance/b-tubulin transcript abundance) in each inductive condition to its relative expression in the corresponding control. Each value is the mean of two independent experiments, each with three replicates. For easier visualization of the results, numerical data were transformed into colour-grid representations using JColorGrid software [45] in which the fold gene expression induction (Log 2 values) is represented by a grey scale (on the right). doi:10.1371/journal.pone.0075143.g004 Role of Dehydrins in Alternaria brassicicola PLOS ONE | www.plosone.org Dabhog1 strains were exposed to 350 mM NaCl for 30 min prior RNA extraction. For the three AbDhn genes, expression induction is represented as a log 2 ratio (fold induction) of their relative expression under stress condition to their relative expression in the control without stress. B: Basal transcript levels of AbDhn genes in the DabhacA mutant relative to their expression levels in the reference wild-type strain Abra 43 (Log 2 values). Each value is the mean of two independent experiments, each with three replicates. Asterisks indicate values that are significantly (P,0.01) different than that of the wild-type. doi:10.1371/journal.pone.0075143.g005 Figure 6. Expression of AbDhn-GFP proteins under salt stress. Germinated conidia (24 h-old) from strains expressing AbDhn-GFP fusions under the control of their own promoters or from a strain constitutively expressing GFP (cGFP) were either exposed to 350 mM NaCl for 2 h or left without stress. Proteins extracts were immunoblotted with HRP-coupled antibodies directed against GFP. Numbers correspond to molecular weights in kDa. The two AbDhn2 isoforms are indicated. doi:10.1371/journal.pone.0075143.g006 times (7-10 h at cold and high temperatures, respectively) were recorded with the wild-type. Similar effects were observed for the Dabdhn1 and Dabdhn2 single mutant strains (Fig. 9). By contrast, storage at low (4uC), freezing (220uC) or high (40uC) temperatures strongly decreased the germination capacity of Dabdhn3 conidia. In the two latter conditions, germination of DDabdhn1-abdhn2 conidia was also affected.

Impact of Dehydrin-like Proteins on A. Brassicicola Virulence
The expression of AbDhn genes was examined during infection of Brassica oleracea leaves inoculated with the wild-type strain (Fig. 10). Up to 3 days post-inoculation (dpi), no significant difference was noted between the AbDhn transcript levels in freeliving mycelium (control) and those in infected leaves, while small  . Conidia were used to inoculate microplate wells containing standard PDB medium that was supplemented with either 10 mM menadione, 5 mM H 2 O 2 , 5 mM Al-ITC or 100 mM brassinin. Error bars indicate standard deviations. Each genotype was analysed in triplicate and the experiments were repeated twice times per growth condition. Lag time and maximal growth rate variables were calculated from the growth curves using a calculation method described by Joubert et al. [19]. For each parameter, Student's T-test was used to assess significant difference between the treated and untreated samples or between mutants and parental isolate. doi:10.1371/journal.pone.0075143.g008 necrotic symptoms were already observed. At 6 dpi, higher relative levels of AbDhn (in particular AbDhn2 and AbDhn3) transcripts were recorded. At this stage, large typical necrotic areas with emerging conidia were apparent.
The pathogenic behaviour of the dehydrin-like deficientmutants was examined at both vegetative (leaves) and reproductive (siliques) plant developmental stages. When inoculated on intact B. oleracea leaves at decreasing inoculum concentrations (from 10 5 to 10 3 conidia per mL), irrespective of the inoculum pressure, no difference was recorded when the single deletion mutants were compared to the wild-type strain concerning the symptom aspect (Fig. 11A) or the rate of successful infection (Fig. 11B), estimated as the lesion size and percentage of typical lesions formed at 5 dpi, respectively. By contrast, the double deletion mutant was highly compromised in its pathogenicity, i.e. only small necrotic lesions with very limited spread around the inoculation sites were observed at 5 dpi on tissues inoculated with this strain.
When inoculated on siliques of A. thaliana Ler ecotype and recording seed transmission at 10 dpi, a slight but significant (p,0.05) decrease in global transmission rates was observed for the single AbDhn deletion mutants compared to the wild-type strain Abra 43 (Fig. 12). Much lower infection probabilities were obtained after inoculation with the DDabdhn1-abdhn2 strain. This difference was statistically significant overall and for each silique (p,0.05). The gradient of fungal incidence from the oldest silique (nu 1) and the youngest one (nu 5) was observed for all tested genotypes.

Discussion
Three proteins harbouring the fungal dehydrin signature pattern were identified from the A. brassicicola genome. In silico analyses confirmed that they shared the characteristic features of dehydrin-like proteins via their physicochemical properties. In this respect, A. brassicicola is similar to A. fumigatus, whose genome also encodes three different dehydrin-like proteins. Conversely, DlpA represents the only gene in the A. nidulans genome with sequence similarities to putative dehydrin proteins [12]. The pairwise identity rates indicated marked divergence between AbDhn1, AbDhn2 and AbDhn3 sequences, with less of 16% identity. They were not close to Aspergillus dehydrin-like proteins, with only a maximum of 27% identity between DlpA/AbDhn2 or DprB/ AbDhn2. AbDhn3 differed substantially from other known dehydrins because of its large size and its specific expression pattern. This protein was indeed only weakly expressed in all conditions tested, except during plant tissue colonization. This surprising result suggested that AbDhn3 was specifically involved Figure 9. Susceptibility of AbDhn-deficient mutants to temperature stress. Calibrated water suspensions of conidia from the wild-type (WT) strain Abra43 and AbDhn-deficient mutants were left for 10 h at various temperatures (220uC, +4uC, +20uC, +40uC). Conidia were then used to inoculate microplate wells and nephelometric growth curves were established over a 33 h period. DLag time was calculated as the difference between the lag time at the tested temperature and the lag time at 20uC and was used as a parameter to estimate the effect of the treatment on spore viability. Error bars indicate standard deviations and asterisks indicate values that are significantly (P,0.01) higher than that of the wild-type. Each genotype was analysed in triplicate and the experiments were repeated twice times per growth condition. doi:10.1371/journal.pone.0075143.g009 Figure 10. Quantitative RT-PCR results for the expression of AbDhn genes during the infection kinetics of A. brassicicola wild-type strain on B. oleracea. For each gene, expression induction is represented as a ratio of its relative expression at 1, 3 and 6 dpi (studied gene transcript abundance/tubulin transcript abundance) in each inoculated sample to its relative expression in free-living fungal control cultures. The experiment was performed twice on biologically independent samples with three technical replicates. Error bars indicate standard deviations and asterisks indicate a relative expression significantly different from 1 (Student test, P,0.01). doi:10.1371/journal.pone.0075143.g010 in a mechanism related to the fungal development in host plant, more particularly during the necrosis formation. Abdhn3 expression might be induced following exposure to specific host metabolites, which differ from those (camalexin, brassinin, AlITC) used for in vitro expression analyses. Two AbDhn2 protein isoforms seemed to be produced by alternative splicing of AbDhn2 pre-mRNA. We were unable to demonstrate particular functional characteristics of each isoform or specific accumulation under the conditions tested (data not shown).
Expression analyses suggested the involvement of AbDhn1/2 in the cellular response to oxidative, osmotic and cold stresses. They were also highly expressed in germinated conidia exposed to the brassicaceous defense metabolites, camalexin, brassinin and ITCs. Since the Hog1 MAPK cascade plays an important role in the adaptive response to phytoalexin exposure in A. brassisicola [16] and to oxidative and osmotic stress in several fungal species [20,21,22], we analysed AbDhn gene regulation in a mutant strain lacking the Hog1 MAP kinase. Q-PCR analysis of AbDhn expression revealed the absence of NaCl-induced AbDhn expression in the hog1 deletion strain, indicating that the Hog1 MAP kinase is involved in the regulation of A. brassicicola dehydrin genes, as previously reported for Aspergillus dehydrins [11,12,13].
Conversely, AbDhn proteins do not act downstream of AbHacA, the major UPR transcription regulator in A. brassicicola [17]. In agreement with our results, previous findings in A. nidulans and A. fumigatus showed that treatment with UPR inducers (dithiothreitol (DTT), tunicamycin, farnesol) triggered significant upregulation of dehydrins, suggesting their potential role as molecular chaperones to prevent the aggregation of proteins partially unfolded due to stress damage [11,12,13]. However, no obvious defect was seen in the growth of AbDhn single mutants upon supplementation with DTT (data not shown). We monitored Figure 11. Effects of AbDhn knockouts on pathogenicity. B. oleracea leaves were inoculated with 5 mL drops of conidia suspension (10 5 , 10 4 or 10 3 conidia mL 21 in water). Mutants were inoculated on the right part of the central vein and compared on the same leaf with the parental strain (inoculated on the left part of the central vein). A: Representative result at 5 dpi. B: Percentage of successful infection at 5 dpi. The experiment was repeated twice and for each experiment each genotype was inoculated on 30 leaves at the three inocula concentrations. Error bars indicate standard deviations and asterisks indicate a significant difference with respect to the wild-type aggressiveness using the Student test (P,0.01). doi:10.1371/journal.pone.0075143.g011 basal expression of the three A. brassicicola dehydrin-like genes in a AbHacA mutant defective for the unfolded protein response. In the absence of exogenous stress, a dramatic increase in AbDhn1 (and, to a lesser extent, of Abdhn2 and AbDhn3) transcripts was observed in the DabhacA strain, compared to the wild-type. It was reported that a basal UPR exists in the absence of stress [23,24]. This process should allow the cells to make minor adjustments necessary to buffer dynamic fluctuations in endoplasmic reticulum stress during hyphal growth and maintain continuous ER homeostasis. In a strain lacking the HacA UPR regulator, the basal maintenance of ER is difficult or impossible, which seems to cause cellular stress quite similar to that generated by chemical agents like DTT and tunicamycin. As HacA-targeted genes could not be induced in a cellular DhacA background, we hypothesized that other stress protective molecules, in particular dehydrins acting as chaperones, could be induced in a HacA-independent manner as a compensatory mechanism.
Although AbDhn gene expression is induced under various stresses, the knock-out approach showed that these dehydrin-like proteins had an impact mainly on oxidative stress tolerance. The corresponding deficient single mutants were indeed characterized by a strong susceptibility towards oxidative stress generated by exposure to menadione, H 2 O 2 or Al-ITC. We previously showed that glucosinolate-derived isothiocyanates induced intracellular ROS accumulation in fungal cells [15]. These results are in good agreement with data from Aspergillus strains, indicating a role of DlpA and DprA in stress resistance against ROS [11,12].
Fluorescent protein fusions suggested that AbDhn1 and AbDhn2 were associated with peroxisomes. These organelles are known to participate not only in ROS generation but also in cell rescue from the damaging effects of such radicals [25]. DprA and DprB were previously found to be important for peroxisome function and DprA mutants had altered catalase activity [11]. Catalases have a protective role against H 2 O 2 and are localized in peroxisomes. The association of AbDhn1 and AbDhn2 with the peroxisomes is consistent with a possible perturbation of protective mechanisms to counteract oxidative stress and maintain the redox balance in AbDhn mutants.
Wong Sak Hoi et al [11] reported that DprB was involved in the response to osmotic stress. Although AbDhn expression seemed to be Hog1-dependent, all of the A. brassicicola dehydrin-like deficient mutants had normal tolerance to osmotic stress. In line with this observation, we previously reported that Dabhog1 mutants, while being hypersensitive to oxidative stress, had normal growth in the presence of high sorbitol concentrations [16], suggesting a minor role of this MAP kinase in the response of A. brassicicola to high osmolarity.
We also showed that storage at freezing (220uC) or high (40uC) temperatures decreased the germination capacity of Dabdhn3 conidia. Similarly, it has previously been reported that DrpC in A. fumigatus and DlpA in A. nidulans confer cold and heat tolerance of conidia, respectively [11,12]. We also observed that high temperatures strongly affected germination of DDabdhn1-abdhn2 conidia suggesting a (partial) functional redundancy associated Figure 12. Transmission capacity of A. brassicicola wild-type (WT) and AbDhn-deficient genotypes to Arabidopsis thaliana seeds (Ler ecotype). The seed transmission capacity according to the silique stage and global seed transmission capacity (strain model) were measured as described by Pochon et al [26]. The five youngest siliques of at least five plants were inoculated with each fungal genotype and the experiment was repeated twice. Contaminated siliques were harvested 10 dpi. After dissection, seeds were incubated separately on PDA medium for 2 days. A seed was considered contaminated when incubation resulted in typical A. brassicicola colony development. For each inoculated fungal genotype, the seed infection probability was evaluated from at least 1000 seeds. Values represent infection probabilities with 95% confidence interval. doi:10.1371/journal.pone.0075143.g012 with the heat tolerance of these two proteins. In fact, our study highlighted other functional redundancies of the two proteins, which had very similar expression patterns, in developmental processes and during pathogenesis.
Concerning development, we observed that DDabdhn1-abdhn2 mutants only weakly sporulated. First dehydrin-encoding genes were identified in T. borchii and A. fumigatus during searches for genes controlling fruiting body maturation or conidial dormancy, respectively [6,11]. In A. fumigatus, dehydrin transcipts were detected in dormant conidia, and then their levels decreased dramatically upon germination suggesting an important role for these proteins during conidial dormancy. This is also supported in A. brassicicola by the observation that conidia producing an AbDhn1-or AbDhn2-GFP fusion protein under control of their native promoters displayed strong fluorescence. However, this inability of dehydrin-deficient fungi to form conidia has not been previously reported in any fungal dehydrin single mutant and we showed for the first time that dehydrins were required for asexual sporulation.
Concerning pathogenesis, virulence of the wild-type and AbDhn mutants were first compared on B. oleracea host plants by leaf inoculation experiments. Secondly, we compared the abilities of mutants and the wild-type to transmit to A. thaliana seeds by using the model pathosystem recently described by Pochon et al [26]. In both cases, the double mutant DDabdhn1-abdhn2 was severely compromised in its pathogenicity. One previously reported function attributed to dehydrin is the protection against oxidative stress, which may be generated by the host plant defense system. Oxidative burst is a general plant defense mechanism that occurs at a very early stage of the interaction and is characterized by rapid accumulation of hydrogen peroxide in the extracellular space of plant tissues exposed to biotic stress [27,28]. This oxidative burst occurs in many plant-pathogen interactions and leads to hypersensitive cell death (HR), which is thought to confine the pathogens to initial infection site. HR is efficient against biotrophic pathogens but it does not generally protect plants against infection by necrotrophic pathogens, such as A. brassicicola, which can utilize dead tissues [29]. However, necrotrophic lifestyle requires the induction of fungal detoxification mechanisms to overcome the toxic effects of reactive oxygen species (ROS). Mannitol was proposed to act as an antioxidant agent and protect A. brassicicola cells by quenching ROS [30]. Similarly, dehydrins may contribute to provide protection against oxidative stress. The decreased aggressiveness on B. oleracea and the lower capacity to be transmitted to Arabidopsis seeds via siliques observed for DDabdhn1-abdhn2 could therefore be related to their increased susceptibility to oxidative burst during the early leaf or silique infection stage. At a later infection stage, i.e. during leaf or silique tissue colonization, A. brassicicola is also exposed to glucosinolate-derived isothiocyanates that induce intracellular ROS accumulation in fungal cells [15]. In planta assays were conducted on leaves of Brassica oleracea var. Bartolo and fruits of A. thaliana ecotype Ler that both accumulated various glucosinolates. In addition to their increased susceptibility to extracellular ROS, the low aggressiveness and seed colonization capacity of DDabdhn1-abdhn2 strain may thus also be related to their failure to overcome the intracellular oxidative stress caused by ITC during leaf or silique colonization. Surprisingly, while AbDhn1 was found to be strongly expressed in A. brassicicola when exposed to various in vitro stresses (Fig. 4), in particular when exposed to host defence metabolites, this gene was only weakly up-regulated during leaf infection. However, its maximal expression could occur at earlier times than those that we analyzed (i.e. before 1 dpi). Indeed, it is likely that oxidative burst and production of phytoalexins occur in first hours of infection.
The fact that the AbDhn2 mutant did not exhibit a lower agressiveness may also seem surprising since its in planta expression is strongly induced. This could be explained by the existence of a functional redundancy between dehydrins, which would be partly suppressed in the double mutant.
In conclusion, these results highlight the importance of dehydrin proteins with respect to the ability of A. brassicicola to efficiently accomplish key steps of its pathogen life cycle. During tissue colonization, they probably participate in fungal protection against oxidative stress and are crucial factors in the vertical transmission mechanism (i.e. seed transmission). Moreover, AbDhn1 and AbDhn2 are required for asexual sporulation, which is necessary for efficient horizontal transmission of the pathogen, which is a key element of the fungal disease spreading process.

Strains and Growth Conditions
The A. brassicicola wild-type strain Abra43 used in this study has previously been described and used in the laboratory to generate various deletion mutants [31]. The Dabhog1 and DabhacA mutant strains deficient for the MAP kinase AbHog1 and the transcription factor AbHacA, respectively, have been formerly described [16,17]. For routine culture, A. brassicicola was grown and maintained on potato dextrose agar (PDA, Biokar) supplemented with hygromycin B for mutants. To study hyphal growth in liquid media, conidial suspensions (10 5 spores.ml 21 , final concentration) were inoculated onto microplate wells containing the appropriate test substances in potato dextrose broth (Difco) in a total volume of 300 ml. Microplates were placed in a laser-based microplate nephelometer (NEPHELOstar, BMG Labtech) and growth was monitored automatically over a 33 h period. Nephelometry, which is based on the measurements of scattered light, was proved to be an accurate indicator of the fungal biomass and can be used as reliable tool for the monitoring of fungal growth [19]. Data were exported from Nephelostar Galaxy software in ASCII format and further processed in Microsoft Excel. Two variables, i.e. lag time and maximal slope, were calculated from the growth curves using the calculation method reported by Joubert et al [19]. At least three replicates per treatment were used. To study the susceptibility of the fungal strains to ITC, allyl-ITC (AlITC), purchased from Aldrich Chemical Co. (Milwaukee, WI), was diluted from stock solutions prepared in methanol at the final desired concentrations. The phytoalexin camalexin was synthesized according to Ayer et al [32] and brassinin according to Kutschy et al [33] and Takasugi et al [34]. Stock solutions were prepared in DMSO and added to the medium at the desired concentrations. Solvent concentrations in controls and assays did not exceed 1% (v/v).

RNA Isolation and Expression Analysis by Real-time Quantitative PCR
Total RNA was prepared according to the TRIzol reagent protocol (Invitrogen). Additional cleanup and DNase treatment were performed using the Nucleospin RNA II kit (Macherey-Nagel) according to the manufacturer's protocol. First-strand complementary DNA was synthesized from 5 mg of total RNA and used for real-time PCR. Amplification experiments were conducted as previously described [17]. The relative quantification analysis was performed using the comparative DDCt method as described by Winer et al [38]. To evaluate the gene expression level, the results were normalized using Ct values obtained from btubulin cDNA amplifications run on the same plate.

DNA Procedure and Southern Hybridization
Genomic DNA was extracted from mycelium according to Moller et al [39]. For Southern analysis, DNA fragments resulting from genomic DNA digestion with PstI or SacI were separated on 1% agarose gels and vacuum transferred to Hybond N membranes (Amersham Biosciences). Blots were then probed with a PCR product that was amplified from A. brassicicola genomic DNA and 32 P labelled using the Random Prime Labelling System Rediprime II (Amersham Biosciences).

Generation of Targeted Gene Knockout Mutants and Fusion Strains
The gene replacement cassettes were generated using the double-joint PCR procedure described by Yu et al [40]. The selectable marker inserted in the PCR constructs corresponded to the Hph gene cassette (1436 bp) from pCB1636 [41] or the Nat gene cassette (2150 bp) from pNR2 [42] conferring resistance to hygromycinB and nourseotricin, respectively. Two sets of primers (Table S1) were used with Phusion Hot Start High-Fidelity DNA Polymerase (Finnzymes, Espoo, Finland) to amplify at least 500 bp from the 59 and 39 flanking regions of each targeted gene. The double-joint final PCR was purified and used to transform A. brassicicola protoplasts as described by Cho et al [43]. A. brassicicola wild-type Abra43 was used to obtain single hygromycin resistant transformant strains Dabdhn1-3. Dabdhn1 genotype was used to obtain DDabdhn1-abdhn2 hygromycin and nourseotricin resistant strains. The hygromycin resistant mutants were selected and prescreened by PCR with relevant primer combinations to confirm integration of the replacement cassette at the targeted locus. The gene replacement mutants were further purified by three rounds of single-spore isolation and then confirmed by Southern blot analysis. The AbDhn C-terminal GFP fusion constructs were generated by fusion PCR as described in Fig.  S5. Using A. brassicicola genomic DNA as template, the respective ORFs and 39 flanking regions were amplified with relevant primer combinations (Table S1). In parallel, fragments containing the sGFP and Hyg B cassettes were amplified from the plasmid pCT74 [18] and pCB1636, respectively. The resulting PCR fragments were mixed and subjected to second fusion PCR. A linker containing three glycine residues was introduced at the 39 end of the respective ORFs to replace the stop codons. The final PCR products were transformed in the A. brassicicola wild-type (as described above) to make AbDhn-GFP fusion mutants. Transformants with expected genetic integration events were identified by PCR. Labelling peroxisomes in strains expressing AbDhn-GFP fusion was performed by transformation with a linear Nat-DsRed-SKL cassette amplified from plasmid pDsRed-SKL [44]. Observations were performed under a Nikon (Nikon Instruments, Melville, NY) A1S1 confocal laser microscope equipped with argon-ion (488 nm) and diode (561 nm) lasers.

Western Blot Analysis
Production of AbDhn proteins in A. brassicicola strains expressing C-terminal GFP fusion was monitored by Western blot using HRP coupled antibodies directed against GFP (Miltenyi Biotec, Germany). Samples were prepared from mycelia obtained by growing conidial suspensions at 25uC for 24 h in PDB (10 5 conidia mL 21 ) and then exposed to 350 mM NaCl final concentration for 2 h. Mycelia were collected by filtration on filter paper, ground with a mortar and pestle to a fine powder under liquid nitrogen and homogeneized in ice-chilled buffer (500 mM Tris-HCl, pH 8.7, 700 mM sucrose, 50 mM EDTA, 100 mM KCl, 2% (v/v) b-mercaptoethanol, 1 mM PMSF, 50 mM NaF, 5 mM Na pyrophosphate, 0.1 mM Na vanadate, 10 mM b-glycerophosphate). Proteins were then extracted with one volume of water-saturated phenol and precipitated from the phenolic phase with five volumes of 100 mM NH 4 + acetate prepared in methanol. The protein concentration in the extracts was calculated using a BCA protein assay reagent (Pierce, Rockford, IL). Equal quantities (10 mg) of protein samples were loaded on 10% polyacrylamide gels and blotted onto nitrocellulose membranes (Schleicher and Schuell, Dassel, Germany). Antibody binding was visualized using an ECL Plus kit Western (Amersham Biosciences, Buckinghamshire, UK).

Pathogenicity Assays
For plant infection assays on Brassica oleracea plants (var. Bartolo), 5 mL drops of A. brassicicola conidia suspension (10 5 ,10 4 or 10 3 conidia mL 21 in water) were inoculated on leaves from 5 week-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). The experiment was repeated twice and for each experiment each genotype was inoculated onto 30 leaves at the three inoculum concentrations.
Seed contamination assessments were estimated as described by Pochon et al [26]. Two 2.5 mL drops of an A. brassicicola conidial suspension (1610 5 conidia mL 21 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-monthold 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 mL 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 days after inoculation. 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 contaminated when incubation resulted in typical A. brassicicola colony development. Seed contamination data were analyzed using logistic (logit) generalized linear models as previously described [26]. Figure S1 Verification of deletion mutants. A : Schematic representation of the AbDhn1, AbDhn2 and AbDhn3 loci (light-gray boxes) with flanking regions (dark-gray boxes) and replacement constructs with the hygromycin B (Hph) and nourseotricin (Nat) resistance genes (white boxes). The positions of SacI and PstI sites, probes and the sizes of expected hydridizing fragments are shown. B: Southern hybridization of genomic DNA from wild-type Abra43 (WT) and transformants. Each DNA was digested with either SacI or PstI and the blot hybridized with the Abdhn1, Abdhn2, or Abdhn3 probes. (TIF) Figure S2 AbDHN2 amino acid sequence. The eleven amino acids repeat at the N-terminal end of the sequence is indicated in red and blue characters. The five conserved DPR motifs are indicated in bold characters. Residues in italic are absent in the ß isoform and the underlined residue corresponds to a potential phosphorylated serine present only in the a isoform. (TIF) Figure S3 Constitutive GFP expression in an A. brassicicola mutant. Observation was performed using confocal microscopy. Bars = 10 mm. (TIF) Figure S4 Comparison of growth rates of the wild-type strain and dehydrin-like deficient mutants in solid and liquid nutritive media. A: In vitro growth tests were carried out on PDA plates. Growth was recorded after 7 days of incubation at 24uC. B: Nephelometric monitoring of the growth of different genotypes. Conidia from the wild-type and mutants were used to inoculate microplate wells containing standard PDB medium. Growth was automatically recorded for 25 h at 25uC using a nephelometric reader. The unit of the Y-axis corresponds to the Relative Nephelometric Unit (RNU). Lag time and maximal growth rate variables were calculated from the growth curves using a calculation method described by Joubert et al [19]. For each parameter, Student's T-test was used to assess significant difference between mutants and parental isolate.

Supporting Information
(TIF) Figure S5 Schematic representation of the construction of AbDhn-Gfp fusion cassette. Targeted ORF and 39 flanking region were amplified from A. brassicicola wild-type genomic DNA with two sets of primers (P1/P2 and P3/P4). Reverse primer P2 was designed so that the original stop codon was replaced by three Glycin codons. The sGfp coding sequence (stripped boxes) and Hph gene cassette were amplified from pCT74 [18] and pCB1636 [41], respectively. The four PCR fragments were fused using the double-joint PCR procedure described by Yu et al [40]. The resulting DNA fragment was introduced into A. brassicicola by protoplasts transformation and double crossing-over events replaced the wild locus by the AbDhn-sGfp-Hph cassette. Table  S1 lists the different primers that were used for each AdDhn gene: