Identification of the GATA-type transcription factor BcLTF1

To identify genes in B. cinerea that are associated with the infection process, 2,367 hygromycin-resistant mutants obtained through A. tumefaciens-mediated transformation (ATMT) were screened for virulence on detached tomato leaves. The assay revealed 560 mutants that were either avirulent or less virulent than the parental wild-type strain B05.10 [39]. So far, phenotypes have been confirmed for 193 mutants on primary leaves of living bean plants (Phaseolus vulgaris), including the mutants PA31, PA2411, and PA3417 (Fig. 1A). These mutants are impaired in their ability to penetrate (PA3417) or to colonize the host tissue (PA31, PA2411), and are furthermore affected in light-dependent differentiation (data not shown). TAIL-(thermal asymmetric interlaced) PCR analyses revealed that the T-DNA integrations in the genomes of the mutants have occurred in the same genomic region (Fig. 1B). Identical integration events occurred for the RB flanks of the T-DNAs in all mutants i.e. 1.608 kb upstream of the annotated open reading frame (ORF) B0510_3555 (Broad Institute; for details see Materials and Methods), while different situations were found for the LB-flanking regions. In PA3417, the T-DNA insertion caused the deletion of four base pairs, in the other two mutants TAIL-PCR failed because no amplicons were obtained (PA31), or the remaining vector sequence was amplified (PA2411).

Download:

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

Figure 1. Identification of bcltf1 as putative virulence factor by an ATMT approach.

(A) Virulence phenotypes of the recipient strain B05.10 and the ATMT mutants PA31, PA2411, and PA3411. Primary leaves of Phaseolus vulgaris were inoculated with conidial suspensions and incubated for 5 d. (B) T-DNA insertion sites in the ATMT mutants were identified by TAIL-PCR analyses. In all three mutants, the insertion of the T-DNA (purple arrow) took place 1.608 kb upstream of an ORF hereinafter referred to as bcltf1 (blue arrows). The genomic region in strain B05.10 (B. cinerea Database, Broad Institute) with annotated genes and matching EST sequences (green bars) is shown. (C) Sequence alignment of BcLTF1 with homologous proteins from A. nidulans and N. crassa. The alignment was generated using ClustalW (http://genius.embnet.dkfz-heidelberg.de/) and the following protein sequences: B. cinerea LTF1 (481 aa; B0510_3555), A. nidulans NsdD (461 aa; AAB16914), and N. crassa SUB-1 (466 aa; NCU01154.5). Amino acids that are identical in all sequences are shaded black, amino acids that are shared by B. cinerea and N. crassa or A. nidulans are shaded in gray. The nuclear localization signal (NLS) is highlighted in blue, the GATA zinc finger motif (PF00320) with the four conserved cysteine residues (indicated by asterisks) is highlighted in pink.

https://doi.org/10.1371/journal.pgen.1004040.g001

BlastP analyses using the sequence of B0510_3555 as query revealed the GATA-type transcription factors NsdD and SUB-1 from A. nidulans (e-value 6e-46) and N. crassa (e-value 9e-40), respectively, as putative orthologues. The alignment of protein sequences of B0510_3555, NsdD and SUB-1 showed low levels of sequence similarity (48% and 35%, respectively), except for the highly conserved GATA zinc finger domains in the C-terminal regions (Fig. 1C). Due to its expression profile and functions in B. cinerea - as described in the following - the protein is referred to as BcLTF1 (B. cinerea light-responsive transcription factor 1).

To confirm the linkage between the identified T-DNA integrations upstream of bcltf1 and the observed phenotypes of PA31, PA2411, and PA3417, bcltf1 was deleted and over-expressed in the B05.10 genomic background. Two different replacement mutants exhibiting identical phenotypes were generated by deleting either the ORF only (Δbcltf1-A6) or the ORF including 1.7-kb of the 5′-noncoding region (Δbcltf1-B1). Expression of bcltf1 under control of a constitutive promoter (PoliC) in the wild-type background resulted in two over-expressing mutants (OE::bcltf1-T6 and -T7). Complementation of deletion mutants was done by re-introduction of bcltf1 into the native gene locus (bcltf1^COM) (for details, see Materials and Methods, Table 1, Fig. S1, Table S1).

Download:

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

Table 1. B. cinerea strains used in this study.

https://doi.org/10.1371/journal.pgen.1004040.t001

Bcltf1 expression is strongly induced by light

Given that BcLTF1 is implicated in regulation of light-dependent differentiation as suggested by the ATMT phenotypes, we studied its expression pattern during conditions inducing either conidiation (incubation in continuous light (LL) or in 12 h light/12 h dark rhythm (LD)) or sclerotia formation (incubation in continuous darkness (DD)) in the wild-type strain B05.10. Northern blot analyses revealed that bcltf1 was highly expressed during incubation in the light while only very weak expression occurred during sclerotial development in the dark. For comparison, expression of bcssp1 (homologue of S. sclerotiorum “sclerotia-specific protein 1” [40]) is restricted to developing sclerotia while expression of bcpks13 (polyketide synthase (PKS) putatively involved in melanin biosynthesis [41]) is restricted to conidiation in LL (Fig. 2A). To further analyze the effect of light on the expression of bcltf1, we exposed wild-type cultures that were grown for 2 d in DD for different periods to white light (5 to 300 min). High expression levels of bcltf1 became visible from 15 min light pulse (LP) while only low expression occurred in cultures that were kept in DD. In contrast to other light-induced genes such as bop1 (opsin-like protein [30]) and bcccg1 (homologue of N. crassa “clock-controlled gene-1” [42]) whose expression levels decreased again after prolonged exposure to light (from exposure times of 120 min and longer), expression levels of bcltf1 remained stable (Fig. 2B). Light induction of bcltf1, bop1, and bcccg1 was also observed during infection of P. vulgaris leaves (in stage of lesion spreading at 3 dpi), even though expression levels of bcltf1 were very low in comparison to those found in cultures grown on solid PDAB medium made from mashed bean leaves (Fig. 2C). Finally, dark-grown submerged cultures of B05.10 were exposed to ROS under dark conditions. Exposure to light, but neither the addition of hydrogen peroxide (H₂O₂), nor the presence of singlet oxygen (¹O₂), superoxide anions (O₂−•) and hydroxyl radicals (OH•) affected the expression of bcltf1. Expression levels of bcccg1 increased in response to all treatments in contrast to bcgpx1 (encoding a glutathione peroxidase) that is specifically induced by H₂O₂ (Fig. 2D).

Download:

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

Figure 2. Expression of bcltf1 depends on light.

Gene expression was studied in wild-type strain B05.10 by northern blot analyses. rRNA or bcactA are shown as loading controls. For more details see Materials & Methods. (A) Bcltf1 is expressed during conidiation but not during sclerotial development. B05.10 was incubated on solid CM for the indicated time periods in continuous light (LL) and light-dark (LD) for induction of conidiation, and in continuous darkness (DD) for induction of sclerotium formation (sclerotia initials at 6 dpi). Genes exhibiting stage-specific expression profiles are shown: expression of bcpks13 is restricted to premature conidiation in LL, bcssp1 is exclusively expressed in developing sclerotia. (B) Expression of bcltf1 is induced by light. Wild type was incubated on solid CM for 52 h in DD or LL. DD-grown cultures were then exposed to white light for the indicated periods (5 to 300 min). (C) Bcltf1 is weakly expressed during infection of P. vulgaris. Primary leaves of living plants were inoculated with conidial suspensions of the wild type (four droplets per leaf) and incubated for 3 d in DD. Then plants were exposed to light for the indicated periods (30 to 300 min) or kept in DD. For comparison, light responses of B05.10 grown on solid PDAB (containing mashed bean leaves) are shown. (D) Expression of bcltf1 does not respond to oxidative stress. Liquid CM was inoculated with non-sporulating mycelia, and incubated for 3 d in DD at 170 rpm. Then, cultures were harvested (DD, controls), exposed to light (30 to 180 min), or exposed in darkness to oxidative stress induced by H₂O₂ (5 to 25 mM) or singlet oxygen (generated by simultaneous addition of 10 mM NaClO and 10 mM H₂O₂ (N+H)). All cultures that were exposed to oxidative stress were harvested after 30 min. Bcgpx1 is a known H₂O₂-induced gene.

https://doi.org/10.1371/journal.pgen.1004040.g002

Taken together, among the conditions tested bcltf1 is exclusively highly expressed in response to light, and represents therefore an eligible candidate for controlling differentiation processes (conidiation vs. sclerotial development) and virulence in a light-dependent manner.

BcLTF1 is localized in the nuclei

To localize BcLTF1 in living hyphae, the coding sequence of bcltf1 was fused to the codon-optimized gene encoding the green fluorescent protein (GFP) [43] and expressed under control of a constitutive promoter in the bcltf1 deletion mutant (Δbcltf1+PoliC::bcltf1-gfp) (for details, see Materials and methods, Table 1, Fig. S1E). Merged images of GFP fluorescence and nucleic acid-specific Hoechst staining patterns showed that BcLTF1-GFP localizes to the nuclei in germinated conidia irrespective of the applied light conditions (Fig. 3A). Nuclear localization of BcLTF1-GFP was also observed in infectious hyphae during infection of onion epidermal cells (Fig. 3B). Similar results were obtained for BcLTF1-GFP that was expressed from its native promoter (Δbcltf1+Pbcltf1::bcltf1-gfp). GFP fluorescence, although weaker when compared to that of the constitutively expressed fusion protein, was detected in the nuclei independently of the illumination treatment (data not shown). Expression of BcLTF1-GFP under control of PoliC or its native promoter in the Δbcltf1 genomic background indicated the functionality of the fusion protein, as the native expression rescued all phenotypes of the deletion mutant while PoliC-driven expression resulted in phenotypes that are presumably due to the inappropriate expression of BcLFT1 especially in dark conditions (Fig. S2, Fig. S1F). Accordingly, mutants expressing PoliC::bcltf1-gfp in the wild-type background were generated as BcLTF1-overexpressing strains hereafter referred to OE::bcltf1 (T6, T7) (Fig. S1F, Table 1).

Download:

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

Figure 3. BcLTF1-GFP fusion proteins are localized in the nuclei.

(A) Subcellular localization of BcLTF1-GFP does not depend on illumination conditions. The fusion protein was expressed from the constitutive A. nidulans oliC promoter in the Δbcltf1 background. Conidia were incubated on microscope slides for 18 h in DD (dark), or 12 h in DD followed by 6 h in LL (light). Nuclei were visualized using the fluorescent dye Hoechst 33342. Scale bars, 5 µm. (B) BcLTF1-GFP localizes to the numerous nuclei in infectious hyphae. Onion epidermis strips were inoculated with conidial suspensions and incubated for 24 h in DD. Conidia and hyphae on the surface were stained with the fluorescent dye calcofluor white (CFW). Sites of penetration are indicated by black arrows. Scale bars, 20 µm.

https://doi.org/10.1371/journal.pgen.1004040.g003

Bcltf1 mutants are affected in light-dependent differentiation programs

The deletion as well as the over-expression of bcltf1 cause severe growth phenotypes that are related to the light treatment. Deletion mutants (Δbcltf1-A6, -B1) failed to grow on minimal medium (CD) in any light condition (Fig. 4A), and radial growth rates on synthetic complete medium (CM) correlated with the exposure times to light: while growth rates nearly equivalent to the wild type were observed in DD (92% of WT), incubation in LD and LL conditions significantly decreased growth rates to 43% and 27% of DD, respectively (Fig. 4B). The inhibitory effect of white light could be assigned to the blue light fraction, as growth rates were similarly decreased under incubation in continuous white and blue light, while growth rates in green, yellow and red light were comparable to those of Δbcltf1 cultures grown in DD (Fig. S3). Although Δbcltf1 mutants were severely impaired in growth in LD, they colonized the whole Petri dish after two weeks of incubation accompanied by reduced aerial hyphae formation and precocious and enhanced conidiation (155% of conidia produced by WT; data not shown). In contrast, the overexpressing mutants produced more aerial mycelia and reduced numbers of conidia giving the colonies a fluffy appearance. Light of different wavelengths (in LD rhythm) failed to alter the phenotypes of Δbcltf1 and OE::bcltf1 mutants. In contrast, the wild type formed more aerial hyphae and fewer conidia and sclerotia during incubation in blue and yellow/red light, respectively (data not shown). In general, phenotypes of the dark-grown cultures were more pronounced than those incubated in LD as both Δbcltf1 and OE::bcltf1 mutants failed to produce any sclerotia. Instead, the deletion mutants produced conidia and the overexpression mutants accumulated cotton ball-like mycelial aggregates in the dark (Fig. 4C).

Download:

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

Figure 4. Mutations of bcltf1 alter light-dependent growth and differentiation programs.

(A) BcLTF1 is essential for allowing growth on minimal medium. Indicated strains were incubated for 2 d on solid CD. Scale bar, 1 cm. (B) Exposure to light negatively affects radial growth rates of Δbcltf1 mutants on synthetic complete medium. Strains were incubated on solid CM in LL (white bars), LD (gray bars), and DD (black bars). Colony diameters (growth in cm) were determined after 3 d of incubation. Mean values and standard deviations were calculated from five colonies per strain and light condition. Asterisks indicate significant differences compared to WT:B05.10 in each condition (p<0.001). (C) Both overexpression and deletion of BcLTF1 modulate the differentiation program in DD. Strains were incubated for 14 d on solid CM. Close-up views of the cultures incubated in DD are shown in the lower panel. Scale bar, 5 mm.

https://doi.org/10.1371/journal.pgen.1004040.g004

Evidently, light and BcLTF1 have an impact on colony morphology and the mode of reproduction, however, other differentiation programs in B. cinerea are not affected by light and mutations of bclft1. Conidial germination is induced by nutrients or hydrophobic surfaces in a similar fashion in wild type, OE::bcltf1 and Δbcltf1 mutants in LL and DD, and germling fusions via conidial anastomosis tubes (CATs) were observed for all strains (Fig. S4A; data not shown). Nevertheless, the forced expression of bcltf1 that is marked by bright nuclear BcLTF1-GFP signals, results in malformed germ tubes (Fig. S4B). The lack of the GFP signal accompanied by a wild-type-like morphology in a percentage of germ tubes suggests that the overexpression construct might become lost during conidiogenesis, so the outcome of studies involving OE::bcltf1 conidia needs to be carefully considered.

Moderate expression levels of bcltf1 are crucial for virulence

Though all T-DNA insertions upstream of bcltf1 impair virulence, differences exist between mutant PA3417, that is not able to penetrate resulting in the accumulation of mycelia on the top of the leaf, and mutants PA31 and PA2411 that are impaired in their ability to colonize the host tissue (Fig. 1A). However, the penetration defect of PA3417 is restricted to the conidia as infections from mycelia have been observed (data not shown). To examine how BcLTF1 is implicated in the infection process, the defined mutants (Δbcltf1, OE::bcltf1) were tested for their capabilities to penetrate and to proliferate inside a living host. To monitor the penetration events, onion epidermal strips were inoculated with conidial suspensions or agar plugs with non-sporulating mycelia (Fig. 5A). At 24 hpi, all strains had formed branched infection structures (infection cushions) originated from the already established mycelia on the agar plugs. Presumably, OE::bcltf1 mutants produced them more frequently than the wild type due to the significantly increased proliferation of aerial hyphae. In contrast, infection cushions were less often observed for Δbcltf1 mutants known to form fewer aerial hyphae (data not shown). Conidia derived from wild-type and Δbcltf1 strains formed short germ tubes that immediately penetrated the host cells indicating that the transcription factor is dispensable for entering the host. However, the majority of conidia with elevated bcltf1 expression levels produced elongated and branched germ tubes that failed to penetrate.

Download:

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

Figure 5. BcLTF1 is required for full virulence.

(A) Overexpression of bcltf1 affects the formation of infection structures. Onion epidermal strips were inoculated with conidial suspensions (on the left) or non-sporulating mycelia (on the right), and incubated for 24 h in DD. Conidia and hyphae on the surface were stained with lactophenol blue whereas invasively growing hyphae remain colorless. Scale bars, 50 µm. (B) Deletion of bcltf1 impairs the capability to colonize primary leaves of P. vulgaris. Leaves of living plants were inoculated with conidial suspensions and incubated in LD. White scale bars, 1 cm; black scale bars, 2 mm. (C) Spreading lesions provoked by Δbcltf1 mutants are characterized by increased H₂O₂ accumulation. Primary leaves were inoculated with conidial suspensions, detached 1 to 4 dpi, and incubated for 2 h in DAB solution. Plant tissues were decolorized prior to microscopy. (D) Addition of ascorbic acid restores virulence of Δbcltf1 mutants. Conidia were suspended in GB5 (gray) or in GB5 supplemented with 5 g/l ascorbic acid (black). Mean values and standard deviations of lesion diameters were calculated from eight lesions per strain and condition at 3 dpi. Asterisks indicate significant differences compared to WT:B05.10 in each condition (p<0.001). Pictures of lesions derived from ascorbic acid-supplemented conidial suspensions were taken 3 dpi.

https://doi.org/10.1371/journal.pgen.1004040.g005

Primary leaves of living bean plants (P. vulgaris) were inoculated with conidial suspensions or mycelia-containing agar plugs derived from wild-type and mutant strains, and incubated in LD and DD to study the impact of light on infection. As can be seen from the time course experiments, Δbcltf1 mutants were significantly impaired in the infection process irrespective of the inoculation method and incubation conditions (in DD or LD) because in all cases distinct primary lesions emerged one day later than for the wild type and the OE::bcltf1 mutants (Fig. 5B, Fig. S5A,B). The reduced proliferation of Δbcltf1 hyphae inside the host became visible by 1 dpi in onion cells, and was likewise visualized in living bean tissues by trypan blue staining at 2 and 3 dpi (Fig. S5C). However, infection by the deletion mutant progressed until the whole leaf was colonized (9 dpi) and conidia were formed. Although germ tubes of OE::bcltf1 were severely impaired in their capacity to penetrate onion epidermal cells, infections of P. vulgaris derived from conidia and mycelia were comparable to those caused by the wild type with regard to lesion expansion. In contrast, colonization of the host tissue by OE::bcltf1 was accompanied by increased accumulation of aerial hyphae and reduced conidiation (Fig. 5B, Fig. S5A). In summary, light or its absence did not affect the infection process by wild type and bcltf1 mutant strains, and elevated and reduced expression levels of bcltf1 resulted in reduced penetration rates via germ tubes (OE::bcltf1) and retarded infection (Δbcltf1), respectively.

Increased ROS accumulation accounts for reduced virulence of Δbcltf1 mutants

B. cinerea induces an oxidative burst and hypersensitive cell death in the host, and it was shown that the degree of virulence on Arabidopsis correlates with the levels of superoxide (O₂−) and hydrogen peroxide (H₂O₂) [44]. To monitor the accumulation of H₂O₂ in wild type- and Δbcltf1-infected plant tissues, inoculated bean leaves were sampled at 1 to 4 dpi and treated with 3,3′-diaminobenzidine (DAB); a compound that is oxidized by H₂O₂ in the presence of peroxidases yielding a dark-brown product. Microscopic observations revealed H₂O₂ accumulation in anticlinal walls of epidermal cells that were in close contact with wild-type germlings at 1 dpi. In contrast, no H₂O₂ accumulation was observed beneath the Δbcltf1 infection droplets at this time point even though conidia had germinated and entered the plant tissue (data not shown). However, after a further two days, significantly increased H₂O₂ accumulation in spreading lesions of the bcltf1 deletion mutant was observed (Fig. 5C). Therefore, in this particular case, increased ROS accumulation does not positively correlate with virulence. To explore the possibility that the elevated ROS levels may negatively affect the ability of the Δbcltf1 mutant to colonize the host tissue, we added ascorbic acid as an antioxidant to the conidial suspensions of wild type and bcltf1 mutants prior to inoculation. Whereas no differences were observed for infections by wild type and OE::bcltf1 strains, the addition of the antioxidant restored virulence of Δbcltf1 mutants as deduced from lesion sizes at 3 dpi (Fig. 5D). Consequently, the abnormal ROS accumulation may account for the virulence defect of the Δbcltf1 mutants.

BcLTF1 is required for maintaining ROS homoeostasis

For reasons thus far un-established, BcLTF1 is required to tolerate light during saprophytic growth on solid media, but not during plant colonization. To unravel the mechanism of light sensitivity, we modified the standard growth medium with different supplements and monitored colony growth and morphology during incubation in LL, LD and DD. Osmotic stabilization of the medium by adding 0.7 M sorbitol elevated growth rates of Δbcltf1 mutants in LL and LD conditions almost to those found in DD (66% and 94% of DD) (Fig. 6A) suggesting that the mutants have difficulty to cope with hypo-osmotic conditions, especially during illumination. On the other hand, the addition of 0.7 M NaCl appeared to cause ionic stress and prevented growth of Δbcltf1 mutants also during incubation in DD (25% of DD). These findings may imply that cell wall integrity and/or the osmotic stress response is impaired when bcltf1 is absent.

Download:

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

Figure 6. BcLTF1 is needed to cope with oxidative stress and regulates the generation of ROS.

(A) Radial growth rates of Δbcltf1 mutants are affected by supplements. Strains were incubated in LL (white), LD (gray), or DD (black) on solid CM (control), and CM with 0.7 M NaCl, 0.7 M sorbitol, 300 µM menadione, 7.5 mM H₂O₂, 800 µM DTT (dithiothreitol) or 28 mM AA (ascorbic acid). Mean values and standard deviations were calculated from five colonies per strain and condition. Asterisks indicate significant differences compared to WT:B05.10 in each condition (p<0.001). (B) Salt and oxidative stress prevent growth of Δbcltf1 mutants even in the absence of light. Pictures were taken after 2 d in DD. (C) Ascorbic acid and DTT restore light tolerance but not the proliferation of aerial hyphae. Pictures were taken after 2 d in LL. Hardly visible Δbcltf1 colonies are highlighted by dashed lines. (D) Bcltf1 mutants are affected in generation of H₂O₂. Strains were incubated for 2 d in DD on CM or CM supplemented with 750 µM DTT. Then, colonies were flooded with DAB solution. DTT prevents oxidation of DAB in the range of wild-type colonies but not in the range of Δbcltf1 colonies (on the left). Identical quantities of fresh mycelia of B05.10, Δbcltf1 (A6, B1) and OE::bcltf1 (T6, T7) were incubated in DAB solution (on the right). (E) The cytosolic glutathione redox potential is not significantly affected in the absence of bcltf1. Reactions of redox-sensitive reporter roGFP2 expressed in Δbcltf1 and wild type were compared. Reduction of roGFP2 is displayed after an oxidation event provoked by addition of 10 mM H₂O₂ after 2 min. Indicated values are the means of triplicates; standard deviations are shown. For details, see Materials and Methods.

https://doi.org/10.1371/journal.pgen.1004040.g006

Considering the finding of the altered ROS accumulation in Δbcltf1-infected plant tissues and the compensating effect of ascorbic acid in virulence, we pursued two strategies to study whether ROS play a role in the light-dependent growth defect of the deletion mutants. Firstly, the effect of oxidative stress caused by addition of 300 µM menadione (artificial source of superoxide radicals (O⁻₂•)) and 7.5 µM H₂O₂ (a source of highly reactive hydroxyl radicals (OH•) upon reaction with metal ions), respectively, was monitored revealing that the deletion mutants were not able to cope with these conditions irrespective of the light conditions (Fig. 6A,B). Secondly, the use of antioxidants was intended to protect fungal strains against ROS. Both the addition of 800 µM dithiothreitol (DTT) and 28 mM ascorbic acid restored radial growth rates of Δbcltf1 mutants in LL and LD, but not the proliferation of aerial hyphae (Fig. 6C). To study whether the growth defect of the Δbcltf1 mutants on minimal medium (CD) can be rescued by reducing oxidative stress, we supplemented CD medium containing either nitrate or ammonium a single nitrogen source with ascorbic acid. In fact, the supplementation increased the growth rates of the mutants to certain extents in LD and DD, but failed to restore wild-type growth rates (Fig. S6). The fact that Δbcltf1 mutants are hypersensitive to exogenously applied oxidative stress even in the absence of light, and that on the contrary, the supplementation with antioxidants facilitated growth during incubation in LL indicate that the deletion mutants are unable to cope with oxidative stress that arises during the exposure to light.

To detect ROS production by the different strains, colonies of wild type and bclft1 mutants grown on CM and DTT-supplemented CM were stained with DAB solution. The intensive brown coloration of mutant colonies, even in presence of DTT that creates a reducing environment, indicated an increased accumulation of H₂O₂ in the range of the Δbcltf1 colonies in comparison to the wild type (Fig. 6D). In a second approach, the H₂O₂ production by equal amounts of biomasses of wild type, bcltf1 deletion and overexpressing mutants was studied, confirming the increased H₂O₂ production by the deletion mutants and revealing furthermore the decreased production by mutants overexpressing BcLTF1 (Fig. 6D). Formation of O⁻₂• was monitored by treating wild type and mutant mycelia with nitro blue tetrazolium (NBT). Blue precipitates were found in hyphal tip segments of all strains (data not shown) demonstrating that mutations of bcltf1 do not alter the net accumulation of O₂⁻. The fact that Δbcltf1 mutants accumulate higher amounts of H₂O₂ under all light conditions, either caused by its inappropriate production and/or detoxification, may explain the general hypersensitivity of the mutants to (additional) exogenously applied oxidative stresses such as the exposure to menadione, H₂O₂ or light.

To see whether the cellular redox status (glutathione pool) is altered in Δbcltf1 mutants, we expressed the redox-sensitive green fluorescent protein (roGFP2) in the mutant background and compared the ratio of fluorescence intensities (395 nm = oxidized state/488 nm = reduced state) with those of the wild type expressing roGFP2 [45]. Similar initial redox statuses were found in Δbcltf1 and wild-type hyphae, and roGFP2 was oxidized and subsequently reduced in a similar fashion in both strains after addition of 10 mM H₂O₂ (Fig. 6E). Thus, the altered ROS homoeostasis of the mutants that is characterized by the overproduction of H₂O₂ and the hypersensitivity to oxidative stress and light is not accompanied by an altered cytosolic redox status.

BcLTF1 modulates the transcriptional responses to light

To gain broader insight into the functions of BcLTF1 especially in those that are related to illumination, a genome-wide approach was initiated to compare gene expression profiles in the wild type and the Δbcltf1 mutant in response to a light stimulus. For these experiments, strains were cultivated in DD, and then exposed to white light for 60 min (WT:B05.10+LP; Δbcltf1+LP) or kept in darkness for additional 60 min (WT:B05.10-D; Δbcltf1-D) (for details, see Material and Methods). RNA from four biological replicates was extracted, labeled and hybridized to NimbleGen microarrays containing oligonucleotides representing all predicted B. cinerea genes as well as non-matching expressed sequence tags (ESTs) [2].

Statistical tests were performed to detect differentially expressed genes: light responses in wild type (+LP/D in WT) and mutant (+LP/D in Δbcltf1), and Δbcltf1 effects in darkness (Δbcltf1/WT in D) and light (Δbcltf1/WT in LP). In total, 2,156 differentially expressed spotted genes corresponding to 2,074 distinct genes were identified and could be assigned to 11 expression profiles (Fig. 7, Table S2). 293 genes showed a light-responsive pattern in the wild-type background: the expression of 244 genes was induced by light (referred to as light-induced genes; bclig1-244), and the expression of 49 genes was repressed by light (referred to as light-repressed genes; bclrg1-49). The expression levels of the majority of light-responsive genes were altered in the absence of BcLTF1, i.e. light induction/repression did not happen (genes in profiles 5 and 6) or just disappeared as genes are already de-regulated in Δbcltf1 in the absence of light (genes in profiles 7 and 8). Only 34% of light-responsive genes (in profiles 3 and 4) exhibited similar light responses in both genomic backgrounds, demonstrating that the absence of bcltf1, which itself is responsive to light, severely affects the expression pattern of other light-responsive genes. Interestingly, some genes were found whose expression responded to light exclusively in the absence of BcLTF1; they were either induced (12 genes in profile 9) or repressed by light (44 genes in profile 10). Nevertheless, the loss of the BcLTF1 had a more pronounced effect on genome-wide gene expression than the light stimulus. In total, the expression levels of 1,972 spotted genes corresponding to 1,905 distinct genes were changed in the Δbcltf1 mutant and most of them (1,539 genes; 81%) were not related to a light response: 748 genes were overexpressed (profile 1) and 791 genes were underexpressed (profile 2) irrespective of the illumination condition (Fig. 7, Table S2).

Download:

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

Figure 7. BcLTF1 modulates the transcriptional responses to light.

Wild type B05.10 and the Δbcltf1 mutant were grown on solid CM for 52 h in DD; then, cultures were exposed for 60 min to light (+LP) or kept for additional 60 min in darkness (D). Four biological replicates were performed. RNA samples were labeled and hybridized to microarrays with all predicted genes of B. cinerea. Statistical analysis revealed 2,074 differentially expressed genes. Relative expressions of the genes (i.e. log2-normalized intensities scaled by gene) in the four conditions (16 hybridizations) were clustered, and 11 of the 14 theoretically possible expression profiles were found (for more details, see Materials and Methods). This figure represents the centroid views of the 11 expression profiles (x = hybridization and y = log2-normalized intensity scaled by gene). The percentages among the 2,074 genes are indicated for each profile. Genes whose expression is only affected by the deletion of bcltf1 belong to profiles 1 (748 genes) and 2 (791 genes). Profiles 3 (94 genes) and 4 (5 genes) comprise genes whose expression is affected by light in a similar manner in both genomic backgrounds (WT, Δbcltf1). Genes (436) belonging to the other profiles are affected by light treatment and the deletion of bcltf1.

https://doi.org/10.1371/journal.pgen.1004040.g007

Functional enrichment analyses using the GSEA (Gene Set Enrichment Analyses [46]–[47]) method and defined sets of genes were performed to reveal biological processes that are affected by light and/or in the absence of BcLTF1. Hence, the group of light-induced genes in wild type is enriched for genes implicated in the oxidative stress response (OSR) or light perception, while amino acid-transporter-encoding genes are overrepresented in the group of light-repressed genes. Even though a majority of bcltf1-dependent genes display similar expression profiles in darkness and light, the group of overexpressed genes in the Δbcltf1 mutant in darkness is significantly enriched in genes involved in light perception and respiration, and the group of overexpressed genes in the light is enriched in secondary metabolism-related genes. In both light conditions, the group of underexpressed genes in Δbcltf1 is enriched in transporter-encoding genes (Table 2).

Download:

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

Table 2. Functional categories that are enriched due to light treatment or deletion of BcLTF1.

https://doi.org/10.1371/journal.pgen.1004040.t002

Identification of putative regulators of light-dependent differentiation

Currently, it is not known how light and its absence, trigger differentiation in B. cinerea. Because young undifferentiated mycelia of the wild type, that are able to form conidia or sclerotia dependent on the illumination conditions, were subjected to the transcriptomic analysis, two sets of “developmental genes” among the light-responsive genes can be expected, those implicated in induction of conidiation or in suppression of sclerotial development.

Light responses are based on the perception of light as a signal and the induction of subsequent signaling events. Though it is not clear why photoreceptors should be regulated on the level of gene expression, their light-responsive expression is a known phenomenon in N. crassa [37], [48]. Accordingly, expression of most of the predicted photoreceptor-encoding genes of B. cinerea was induced by light (Fig. 8A, Table S3). Bcvvd1 encoding a putative blue light receptor was the strongest up-regulated gene with a 37-fold increase. Further light-induced genes are those encoding cryptochromes (bccry1, bccry2), opsins (bop1, bop2) and one of three phytochromes (bcphy2). Whereas similar intensities were found for bcvvd1 in the wild type and the Δbcltf1 mutant, amplitudes of light induction of the other photoreceptor-encoding genes were much lower in the mutant than in the wild-type background as the genes in Δbcltf1 were already higher expressed in the dark (bccry1, bccry2, bop2, bcphy2) or less induced by light treatment (bop1, bop2) than in the wild type. Bcwcl1 encoding a putative blue light-sensing transcription factor was not found among the light-induced genes in this microarray experiment. However, further analysis showed that bcwcl1 is expressed higher during incubation in LL than in DD (data not shown), indicating that light also affects the expression of this gene. Among the light-induced genes bclov3 (PAS/LOV protein 3) might be related to light perception as the deduced protein sequence shares similarities with N-terminal parts of plant phototropins. These proteins are autophosphorylating protein kinases harboring N-terminal blue light-sensory (LOV) and C-terminal kinase domains [49].

Download:

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

Figure 8. Differentially expressed genes may encode regulators of light-dependent differentiation.

Relative expressions of the genes (i.e. log2-normalized intensities scaled by gene) in the four conditions (16 hybridizations) were clustered and depicted by color scale, where shades of green and red represent under- and overexpressed genes, respectively. Homologues in N. crassa were identified by a bidirectional best-hit (BDBH) approach. Light-induced genes in B. cinerea and N. crassa are marked with orange asterisks. For more details see Table S3. (A) Expression of most photoreceptor-encoding genes is similarly induced by light in wild type and Δbcltf1 strains. All predicted photoreceptors were clustered. (B) Expression levels of 45 TF-encoding genes are altered in response to light or deletion of bcltf1. TFs that were differentially expressed in at least one condition were subjected to clustering.

https://doi.org/10.1371/journal.pgen.1004040.g008

Regulators that function downstream of signaling cascades are transcription factors (TFs) whose activities might be transcriptionally and/or post-transcriptionally modified. The genome of B. cinerea contains 406 genes encoding TFs of different families [2], [50], 45 TF-encoding genes were differentially expressed in at least one of the four conditions, i.e. displaying a light response in wild type or Δbcltf1 mutant, or a Δbcltf1 effect in darkness or in response to light (Fig. 8B, Table S3). Hierarchical clustering of the selected TFs across the 16 hybridizations highlights that the differential expression is mainly due to the bcltf1 mutation (43 TF-encoding genes) and to a smaller extent to the light condition (seven TF-encoding genes). In addition to bcltf1, the expression of five further TF-encoding genes was induced by light and the genes were accordingly named bclft2-bclft6. BcLTF2 and BcLTF3 are C2H2-TFs homologous to the light-responsive TFs SAH-1 and CSP-1 of N. crassa [37], [48]. BcLTF4 to BcLTF6 belong to the family of Zn2Cys6-TFs, and only a homologue of BcLTF6 is present in N. crassa. While the expression of bcltf3, bcltf4 and bcltf5 was similarly induced by light in both genomic backgrounds, the light-dependent expression pattern of the other two TF-encoding genes are affected in different ways in the absence of BcLTF1. Both bcltf2 and bcltf6 were overexpressed in the Δbcltf1 mutant in darkness, but only the expression level of bcltf6 was further increased in response to the light stimulus. Therefore, BcLTF1 is involved in repressing the transcription of bcltf2 and bcltf6 in the dark. Moreover, it can be assumed that BcLTF1 represses the expression of 16 other TF-encoding genes in a light-independent manner, including bcboa13 that belongs to the botcinic acid (BOA) gene cluster (see next section). The expression of one TF-encoding gene (BofuT4_P161860) was repressed by light in the wild type and the effect was found to be absent in the Δbcltf1 background, as the gene was no longer expressed (even in darkness). The gene encodes a member of the Zn2Cys6-TF family and is physically linked with genes of unknown functions displaying a similar expression pattern (Bclft1g1499-1505; Table S2).

Absence of BcLTF1 affects expression of secondary metabolism-related genes

Fungal secondary metabolite (SM)-biosynthetic genes are typically located adjacent to each other and exhibit similar patterns of expression. These clusters usually include a gene encoding the key enzyme responsible for synthesis of the raw product, genes encoding enzymes for modifications and the transport of the compound, and genes whose products regulate cluster gene expression [51]. The genome of B. cinerea comprises 43 genes encoding SM key enzymes including 21 polyketide synthases (PKSs), nine non-ribosomal peptide synthetases (NRPSs), six sesquiterpene cyclases (STCs), three diterpene cyclases (DTCs) and one dimethylallyl tryptophan synthase (DMATS) [2]. In our transcriptomic approach, 38 of the 43 genes were significantly expressed in at least one condition (see Materials and Methods). The expression patterns of 17 of these genes (45%) are significantly altered in the absence of BcLTF1, including 9 PKS-, 3 NRPS-, 4 STC- and the single DMATS-encoding gene (Fig. S7A). Remarkably, the majority (15) of these BcLTF1-dependent genes are not responsive to light. This group includes bcbot2 (bcstc1) that is part of the BOT-biosynthetic cluster (bcbot1-5) as well as bcboa6 and bcboa9 (PKSs) that belong to the two gene clusters (bcboa1-6, bcboa7-17) required for BOA biosynthesis. The other cluster genes were overexpressed as the key enzymes (Fig. S7B), suggesting that the mutants may produce more toxins under the tested conditions. Additionally, the two PKS-encoding genes bcpks12 and bcpks13 were under- and overexpressed in the Δbcltf1 mutant, respectively, and are considered to be associated with dihydroxynaphthalene (DHN)-melanin biosynthesis [41]. Beside the PKS yielding the polyketidic precursor, probably three other enzymes encoded by bcbrn1, bcbrn2 and bcsd1 are involved in melanin biosynthesis (Fig. S8). Increased expression levels of bcpks13, bcbrn1, bcbrn2, and bcscd1 in Δbcltf1 mutants were associated with the dark pigmentation of the culture medium, while expression of bcpks12 was exclusively detected in developing sclerotia of the wild type. Therefore, bcpks13 and bcpks12 are differentially expressed in Δbcltf1 mutants as well as during the life cycle of the wild type.

The expression patterns of only three key enzyme-encoding genes (bcnrps2, bcphs1, bcstc5) were significantly affected by light. Notably, one of them corresponds to a cluster of light-induced genes (bcphs1, bcphd1, bccao1) that encode homologues of enzymes involved in the biosynthesis of retinal, the chromophore for opsin, in Fusarium fujikuroi (phytoene synthase, phytoene dehydrogenase, and carotenoid oxygenase). As in F. fujikuroi, the cluster contains a forth co-regulated gene encoding an opsin (bop2) [52]–[54] (Fig. S9). Though genes were still induced by light in the absence of BcLTF1, the absolute expression levels were much lower than those found in the wild type in response to light, suggesting decreased retinal production and an altered function of BOP2 in the absence of BcLTF1.

Light and BcLTF1 affect the expression of ROS-related genes

Given the finding that mutations of bcltf1 affect the accumulation of H₂O₂, we inspected the transcriptomic data for genes involved in ROS metabolism to gain insight whether the net over-accumulation is due to an inappropriate detoxification and/or generation of ROS. Oxidative stress response (OSR) systems include superoxide dismutases (SODs) that convert O⁻₂• into the less toxic H₂O₂, and catalases (CATs), peroxidases (PRDs), and peroxiredoxins (PRXs) that detoxify H₂O₂ and thereby prevent the formation of OH• via the Fenton reaction. Important non-enzymatic mechanisms include the oxidation of compounds such as glutathione, ascorbate and carotenoids. Reduced glutathione (GSH) provided by glutathione reductases (GLRs) is used by glutathione peroxidase (GPX), glutaredoxins (GRXs) as well as by glutathione S-transferases (GSTs) that conjugate GSH with various substrates such as peroxides [55]. In B. cinerea, several enzyme activities are represented by multiple genes, thus, four SODs, eight CATs, 16 PRDs, nine PRXs, one TRX, five GRXs and 26 GSTs have been identified (Table S4). The set of light-induced genes in the wild type is significantly enriched for OSR-related genes (Table 2), among them are two CATs (bccat3, bccat4), two PRDs (bcprd1, bcprd2) and three GSTs (bcgst3, bcgst8, bcgst15) (Table S4). These genes are not induced by light in the absence of bcltf1 suggesting that their products may contribute to coping with light-related oxidative stress. Nevertheless, twenty genes among them SOD-, CAT-, PRD-, PRX- and GST-encoding genes exhibited increased or decreased expression levels in the mutant background, reflecting a general deregulation of OSR-related genes.

Several enzymes of B. cinerea may generate ROS either as secondary messenger (e.g. formation of O⁻₂• by the NADPH oxidase (NOX)) or as side products of metabolic processes (e.g. oxidation of carbohydrates, fatty acids and amino acids yielding H₂O₂). Forty-eight putative oxidase-encoding genes were identified in the genome, and expression levels of twenty-five genes were affected by either the light stimulus and/or the deletion of bcltf1. Four oxidase-encoding genes were induced by light in both genomic backgrounds, and a further 11 oxidase-encoding genes exhibited increased expression levels in the mutant background. Seven out of the 11 identified DAO (D-amino acid oxidase)-encoding genes are less expressed in the bclft1 mutant than in the wild type, irrespective of the illumination condition, a fact that might be related with the weak expression of amino acid transporter-encoding genes (Table 2). Additionally, two out of the four identified putative NADH:flavin oxidoreductase/NADH oxidase-encoding genes showed increased expression levels in the absence of BcLTF1 while the genes encoding the catalytic subunits of the NOX complex (bcnoxA and bcnoxB) were slightly underexpressed (Table S4).

Absence of BcLTF1 results in increased alternative respiration

Beside the mentioned ROS-generating enzymes, the mitochondrial electron transport chain (ETC) massively contributes to intracellular ROS levels. Complexes I and III are forming O₂⁻ that is in turn dismutated to H₂O₂ to prevent the formation of OH• [56]. Though expression levels of nuclear-encoded subunits of the different protein complexes of the respiratory chain were not affected in the Δbctlf1 mutant (data not shown), strongly increased expression levels were observed for the gene encoding the alternative oxidase (BcAOX1) (Table S4, Fig. S10). This enzyme accepts electrons from the ubiquinone pool and reduces oxygen directly, hence bypassing the electron flux through complexes III and IV (cytochrome c oxidase, COX), and related ROS generation [57]. In addition to complex I (NADH:ubiquinone oxidoreductase), B. cinerea possesses five putative NADH dehydrogenases that may deliver electrons to the ubiquinone pool thereby bypassing complex I. In fact, bcnde2 and bcndi1 encoding external and internal NADH dehydrogenases, respectively, were overexpressed in the deletion mutants. Additionally, genes encoding cytochrome c (bccyc1), a cytochrome c peroxidase (bcccp2), an uncoupling protein (bcucp1) and several other mitochondrial carrier proteins were more highly expressed in the absence of BcLTF1, further supporting the deregulation of the ETC in the mutant background. CCP2 may accept electrons from complex III/cytochrome c for reduction of H₂O₂, hence avoiding the electron flux through COX [58], and uncoupling proteins are considered to decrease ROS production by facilitating the transport of ions across the membrane [56], [59]. Taken together, the observed transcriptional changes in the Δbcltf1 mutant suggest that alternative enzymes compete with the ETC complexes for electrons to decrease the generation of mitochondrial ROS.

B. cinerea strains and growth conditions

Strain B05.10 of B. cinerea Pers. Fr. [B. fuckeliana (de Bary) Whetzel] is an isolate from Vitis (Table 1) and is used as the recipient strain for genetic modifications. Genome sequences of B05.10 and T4 were recently published [2], [91]. Strains were cultivated on plates containing solid synthetic complete medium (CM) [92], modified Czapek-Dox (CD) as minimal medium (2% sucrose, 0.1% KH₂PO₄, 0.3% NaNO₃, 0.05% KCl, 0.05% MgSO₄×7 H₂O, 0.002% FeSO₄×7 H₂O, pH 5.0), or solid potato dextrose medium (Sigma-Aldrich, Germany) supplemented with 10% homogenized bean leaves. Acidification of the culture medium (solid CM pH 8) was monitored using the pH indicator bromothymol blue (0.1%) (Sigma-Aldrich, Germany). Cultures were incubated at 20°C under white light (12 h light/12 h darkness (LD) or continuous light (LL)) for conidiation, and in continuous darkness for induction of sclerotial formation. White light (9000 lx at culture level) was generated by using a set of Sylvania Standard F18W/29-530 “warm white” and F36W/33-640 “cool white” fluorescent bulbs. Wavelengths of transmitted light were controlled using Petri dish chambers covered with Roscolux polyester filters (#381 Baldassari Blue, #389 Chroma Green, #312 Canary, #27 Med Red; http://www.rosco.com/filters/roscolux.cfm); transmission spectra of the filters are shown in Fig. S3. Crossings using SAS405 (MAT1-2) as reference strain were performed as described previously [31]. For DNA and RNA isolation, mycelia were grown on solid CM medium with cellophane overlays.

Standard molecular methods

Fungal genomic DNA was prepared according to Cenis [93]. For Southern blot analysis, fungal genomic DNA was digested with restriction enzymes (Fermentas, Germany), separated on 1% (w/v) agarose gels and transferred to Amersham Hybond-N+ filters (GE Healthcare Limited, UK). Blot hybridizations with random-primed α-³²P-dCTP-labelled probes were performed as described previously [94]. PCR reactions were performed using the high-fidelity DNA polymerase Phusion (Finnzymes, Finland) for cloning purposes and the BioTherm Taq DNA Polymerase (GeneCraft, Germany) for diagnostic applications. Replacement fragments and expression vectors were assembled in Saccharomyces cerevisiae by exploiting its homologous recombination machinery [36], [66]. Sequencing of DNA fragments was performed with the Big Dye Terminator v3.1 sequencing kit (Applied Biosystems, USA) in an ABI Prism capillary sequencer (model 3730; Applied Biosystems). For sequence analysis, the program package DNAStar (Madison, USA) was used. Protocols for protoplast formation and transformation of B. cinerea were described by Schumacher [66]. Regenerated protoplasts were overlaid with SH agar containing 70 µg/ml hygromycin B (Invitrogen, The Netherlands) and 140 µg/ml nourseothricin (Werner-Bioagents, Germany), respectively. Resistant colonies were transferred to agar plates containing CM medium supplemented with the respective selection agent in a concentration of 70 µg/ml. Single conidial isolates were obtained by spreading conidial suspensions on selective media and transferring single colonies to new plates.

Northern blot analyses

Total RNA was isolated making use of the TRIzol reagent (Invitrogen, The Netherlands). Samples (25 µg) of total RNA were transferred to Hybond-N+ membranes after electrophoresis on a 1% (w/v) agarose gel containing formaldehyde, according to the method of Sambrook et al. [95]. Blot hybridizations were carried out by use of random-primed α-³²P-dCTP-labelled probes [94]. Analyzed genes are (Fig. 2): Bcssp1 (BC1G_03185) encoding the homologue of S. sclerotiorum ssp1 [40]; bcpks12 (AY495617) and bcpks13 (AY495618) encoding polyketide synthases putatively involved in melanin biosynthesis [41]; bop1 (BC1G_02456) encoding an opsin-like protein, expression is induced upon H₂O₂ [30] and light treatment (this study); bcccg1 (BC1G_11685) representing the homologue of N. crassa ccg-1 [42]), expression is induced upon H₂O₂ and light treatment (this study); bcactA (AJ000335) encoding actin is used in this study to detect fungal RNA in mixed (in planta) samples; bcgpx1 (BC1G_02031) encoding the glutathione peroxidase, expression is induced upon H₂O₂ treatment [96].

Identification of bcltf1 as virulence-associated gene

Strains PA31, PA2411 and PA3417 belong to an insertional mutant library of B. cinerea B05.10 that was generated by A. tumefaciens-mediated transformation (ATMT). Southern blot analyses demonstrating the existence of single T-DNA copies in the three mutants (data not shown) were performed according to Giesbert et al. [39]. The T-DNA-flanking regions were identified by TAIL-(thermal asymmetric interlaced) PCR as described previously [39]. BlastN analyses using the amplified T-DNA flanking regions revealed discontinuous sequences at the respective locus in both genome sequences of T4 (1-kb gap upstream of BofuT4_P09344; 480 aa) and B05.10 (2-kb gap upstream of BC1G_10441; 133 aa) (B. cinerea genome project, URGI; http://urgi.versailles.inra.fr/Species/Botrytis/) [2]. Re-sequencing of both strains yielded continuous sequences (B. cinerea Database, Broad Institute; http://www.broadinstitute.org/annotation/genome/botrytis_cinerea/ [91], hence, the T-DNA insertion occurred 1.608 kb upstream of the ORF hereafter called bcltf1 (Fig. 1B). Bcltf1 of B05.10 encodes a protein of 481 aa (BcLTF1^B05.10: B0510_3555), that of T4 a protein of 480 aa (BcLTF1^T4: BofuT4_P09344, BcT4_9306). Putative transcriptional start (TSS) and PolyA signals (Fig. 1B) were predicted using the FGENESH tool (http://linux1.softberry.com/). The nuclear localization signal (NLS) and the GATA zinc finger motif were predicted by WoLF PSORT (http://wolfpsort.org/) and Pfam Search (http://pfam.sanger.ac.uk), respectively (Fig. 1C). No nuclear export signals (NES) were identified by using NetNES 1.1 (http://www.cbs.dtu.dk/services/NetNES/). Putative phosphorylation sites for protein kinase A (PKA: S⁷⁸, S⁹⁵, S¹⁸⁴, S¹⁸⁵) and protein kinase C (PKC: T⁹¹, S¹⁰⁵, S¹²⁴, S¹⁹⁴, T⁴⁵³, T⁴⁵⁷, S⁴⁶⁷, S⁴⁶⁸) in BcLTF1^B05.10 were identified by running NetPhosK (http://www.cbs.dtu.dk/services/NetPhosK/).

Construction of bcltf1 mutants

For generation of fragments mediating the replacement of bcltf1 (Δbcltf1-A) and bcltf1+1.7 kb of the 5′-noncoding region (Δbcltf1-B), respectively, different 5′ flanks (A, B) were assembled with the same 3′ flank and a hygromycin resistance cassette by yeast recombinational cloning [36] (Table S1, Fig. S1A). Replacement fragments were transformed into B. cinerea B05.10. Homologous integration events in hygromycin-resistant transformants were detected by diagnostic PCR using the primers pCSN44-trpC-T and pCSN44-trpC-P, binding in the hygromycin resistance cassette and the primers bcltf1-A-hi5F/B-hi5F and bcltf1-hi3R, binding upstream and downstream of the bcltf1-flanking regions, respectively (Table S1, Fig. S1B). Single spore isolates were screened for the absence of bcltf1 alleles using primers bcltf1-WT-F and bcltf1-WT-R matching the substituted coding region. For Southern blot analysis, genomic DNA of the mutants and the recipient strain B05.10 was digested with PstI, and the blot was hybridized with the 3′ flank of the replacement fragment. The hygromycin resistance cassette contains an additional PstI restriction site resulting in smaller hybridizing fragments in the replacement mutants (2.3 kb) than in the wild type (7.2 kb) (Fig. S1C). Strain Δbcltf1-A1 was discarded because a second ectopic integration event occurred. Taken together, two independent mutants were generated lacking either only bcltf1 (Δbcltf1-A6) or bcltf1 and its 5′-noncoding region (Δbcltf1-B1). The mutant Δbcltf1-A6 was complemented by targeted integration of bcltf1 or bcltf1-gfp at the native gene locus thereby replacing the hygromycin resistance cassette. For vector construction, all fragments indicated in Fig. S1A were amplified by PCR using oligonucleotides listed in Table S1, and assembled in one step in S. cerevisiae yielding plasmids pbcltf1-COM and pbcltf1-gfp^{in loco}. Replacement constructs were transformed into Δbcltf1-A6, and diagnostic PCR confirmed the targeted integration in two transformants per construct (bcltf1^COM:T2, T3; and Δbcltf1+Pbcltf1::bcltf1-gfp: T2, T5) (Fig. S1D). For over-expressing the BcLTF1-GFP fusion protein, the ORF was introduced in vector pNAN-OGG comprising gene flanks for targeted integration at bcniiA (locus of nitrite reductase), a nourseothricin resistance cassette and the expression cassette with gfp under control of the constitutive oliC promoter [66] (Fig. S1E). The construct was transformed into B05.10 yielding transformants WT+PoliC::bcltf1-gfp (T6, T7), and into Δbcltf1-B1 yielding transformants Δbcltf1+PoliC::bcltf1-gfp (T3, T4). Targeted integration at bcniiA was verified by diagnostic PCR as described previously [66] (data not shown). Northern blot analyses revealed the constitutive expression of bcltf1-gfp in the mutant background and the overexpression of bcltf1-(gfp) in the wild-type background, respectively (Fig. S1F).

Germination and germling fusion assays

Conidial germination was monitored according to Doehlemann et al. [97]. For testing nutrient- and hydrophobicity-induced germination, cleaned conidia were incubated in Gamborg B5+10 mM glucose on glass surfaces and in double distilled water on polypropylene foil, respectively, in LL or DD conditions. Germination rates were determined in triplicates after 4 h and 8 h of incubation for glucose-induced and after 24 h for polypropylene-induced germination. Germling fusions via conidial anastomosis tubes were monitored according to Roca et al. [98]. Briefly, conidia were plated on solid Vogel's medium and incubated for 17 h at 21°C in the dark. Samples were analyzed by using light microcopy.

Virulence assays

For penetration assays, onion epidermal layers were prepared as described [31]. Conidial suspensions or non-sporulating mycelia from 3-d-old cultures were used for inoculation. The staining with lactophenol aniline blue (Sigma-Aldrich, Germany) allowed the identification of the penetrations sites after 24 h of incubation. For infection assays on French bean (Phaseolus vulgaris cv. 90598), primary leaves of living plants were inoculated with 7.5-µl droplets of conidial suspensions (2×10⁵ conidia/ml Gamborg B5+2% glucose) or non-sporulating mycelia [31]. Infected plants were incubated in plastic boxes at 22°C under natural illumination conditions (LD) and continuous darkness (DD), respectively. Trypan blue and DAB (3,3′-diaminobenzidine) staining of lesions were performed as described previously [31]. Bright field images were taken using the Zeiss SteREO Discovery V.20 stereomicroscope or the AxioScope.A1 microscope equipped with an AxioCamMRc camera and the AxiovisionRel 4.8 software package (Zeiss, Germany).

Detection of reactive oxygen species (ROS)

To detect accumulating H₂O₂, strains were grown on solid CM covered with cellophane for 3 d in DD. Fresh mycelia (25 mg) were weighed, placed in wells of a 24-well plate and flooded with 1 ml DAB solution (0.5 mg/ml DAB in 100 mM of citric acid buffer, pH 3.7). Samples were incubated for 2 h in darkness at room temperature. The colorless DAB is oxidized by H₂O₂ in the presence of fungal peroxidases and horseradish peroxidase (positive control) resulting in a brown coloration of the staining solution. To detect accumulating O₂.⁻, germinated conidia were incubated for 2 h in NBT staining solution (0.5 mg/ml nitrotetrazolium blue chloride in 50 mM sodium phosphate buffer, pH 7.5). Samples were analyzed by using light microcopy. The colorless substrate NBT is converted to a blue precipitate in the presence of O₂⁻.

Measurement of the cellular glutathione redox potential

Vector pNAN-GRX1-roGFP2 carrying the reduction-oxidation sensitive green fluorescent protein (roGFP) and a nourseothricin resistance cassette [45], was transformed into the mutant Δbcltf1-A6 (Table 1). The homologous integration of the construct at bcniiA in two independent transformants (T2, T3) was verified by diagnostic PCR as described previously [66]. Reactions of roGFP2 in wild type B05.10 (WT+grx-rogfp2) and bcltf1 deletion mutants (Δbcltf1+grx-rogfp2) were monitored using a fluorometer as described by Heller et al. [45]. Briefly, germinated conidia (conidia cultivated for 12 h in liquid CM) were washed twice and transferred to 96-well plates. Fluorescence was measured at the bottom with 3×3 reads per well and an excitation wavelength of 395±5 nm for the oxidized state and 488±5 nm for the reduced state of roGFP2 using a Tecan Safire fluorometer. Relative fluorescence units (RFU) were recorded to calculate the Em₃₉₅/Em₄₈₈ ratio.

Fluorescence microscopy

For microscopy, conidia were suspended in Gamborg B5 solution supplemented with 2% glucose and 0.02% ammonium phosphate, and incubated overnight under humid conditions on microscope slides or on onion epidermal strips. Nuclei were stained using the fluorescent dye Hoechst 33342 [66], and fungal cell walls using calcofluor white (CFW) as described previously [60]. Fluorescence and light microscopy was performed with a Zeiss AxioImager.M1 microscope equipped with the ApoTome.2 technology for optical sectioning with structured illumination. Differential interference microscopy (DIC) was used for bright field images. Specimens stained by Hoechst 33342 and CFW were examined using the filter set 49 DAPI shift free (excitation G 365, beam splitter FT 395, emission BP 445/50), GFP fluorescence using filter set 38 (excitation BP 470/40, beam splitter FT 495, emission BP 525/50). Images (optical sections and Z-stacks) were captured with an AxioCam MRm camera and analyzed using the Axiovision Rel 4.8 software package (Zeiss, Germany).

Microarrays analyses

Wild type B05.10 and the Δbcltf1 mutant were grown for 3 d on solid CM in LD before non-sporulating mycelia were transferred to solid CM covered with cellophane overlays. Cultures were incubated for 52 h in DD, and then half of the cultures were transferred to white light (+LP). After 60 min all cultures (WT-D; WT+LP; Δbcltf1-D; Δbcltf1+LP) were harvested. Material derived from four independent experiments was used for RNA isolation (Trizol procedure). Total RNA was treated with the DNA-free kit to remove any trace of DNA (Ambion - Applied Biosystems, France). Synthesis of double-stranded cDNA, Cy3-labeling and hybridization on microarrays were done by PartnerChip (http://www.partnerchip.fr/) using the procedures established by NimbleGen (Roche) and the reagents from Invitrogen (Life technologies, France). To study the complete transcriptome of B. cinerea, NimbleGen 4-plex arrays with 62,478 60-mer specific probes covering all the 20,885 predicted gene models and non-mapping ESTs of B. cinerea [2] were used. The arrays also include 9,559 random probes as negative controls. Data processing, quality controls and differential expression analysis were performed using ANAIS methods [99]. Probe hybridization signals were subjected to RMA-background correction, quantile normalization, and gene summarization [100]–[101]. Genes with a normalized intensity over the threshold (1.5×95^th percentile of random probes) in at least one hybridization were considered as expressed and kept for further analyses. Differentially expressed genes (light response in wild type and mutant, Δbcltf1 effect in darkness and light) were then identified using a one-way ANOVA test. To deal with multiple testings, the ANOVA p-values were then submitted to a False Discovery Rate correction. Genes with a corrected p-value <0.05, and more than a 2-fold change in transcript level were considered as significantly differentially expressed. BlastN analyses for the 20,885 spotted genes [2; URGI database] were performed to identify the matching gene models derived from the re-sequencing approaches by Staats & van Kan [91; Broad Institute]. By this, the number of differentially expressed genes was reduced because incorrect gene models derived from discontinuous sequences were revised and furthermore previously non-matching ESTs could be mapped. The four gene lists were then pooled and cluster analyses were performed to highlight the different expression profiles encountered. For this purpose, the log2-normalized intensities scaled by gene across the 16 hybridizations were clustered using Genesis tools [102]. Gene enrichment analyses were further performed with GSEA toolkit [46]–[47] to highlight significantly enriched functions compared to the complete list of functionally annotated B. cinerea genes [2]. Details on the experiments, raw and normalized values are available at NCBI GEO (http://www.ncbi.nlm.nih.gov/geo/) (Accession: GPL17773). Furthermore, lists of differentially expressed genes are available at http://urgi.versailles.inra.fr/Data/Transcriptome.