The Aspergillus giganteus antifungal protein AFPNN5353 activates the cell wall integrity pathway and perturbs calcium homeostasis

Background The antifungal protein AFPNN5353 is a defensin-like protein of Aspergillus giganteus. It belongs to a group of secretory proteins with low molecular mass, cationic character and a high content of cysteine residues. The protein inhibits the germination and growth of filamentous ascomycetes, including important human and plant pathogens and the model organsims Aspergillus nidulans and Aspergillus niger. Results We determined an AFPNN5353 hypersensitive phenotype of non-functional A. nidulans mutants in the protein kinase C (Pkc)/mitogen-activated protein kinase (Mpk) signalling pathway and the induction of the α-glucan synthase A (agsA) promoter in a transgenic A. niger strain which point at the activation of the cell wall integrity pathway (CWIP) and the remodelling of the cell wall in response to AFPNN5353. The activation of the CWIP by AFPNN5353, however, operates independently from RhoA which is the central regulator of CWIP signal transduction in fungi. Furthermore, we provide evidence that calcium (Ca2+) signalling plays an important role in the mechanistic function of this antifungal protein. AFPNN5353 increased about 2-fold the cytosolic free Ca2+ ([Ca2+]c) of a transgenic A. niger strain expressing codon optimized aequorin. Supplementation of the growth medium with CaCl2 counteracted AFPNN5353 toxicity, ameliorated the perturbation of the [Ca2+]c resting level and prevented protein uptake into Aspergillus sp. cells. Conclusions The present study contributes new insights into the molecular mechanisms of action of the A. giganteus antifungal protein AFPNN5353. We identified its antifungal activity, initiated the investigation of pathways that determine protein toxicity, namely the CWIP and the Ca2+ signalling cascade, and studied in detail the cellular uptake mechanism in sensitive target fungi. This knowledge contributes to define new potential targets for the development of novel antifungal strategies to prevent and combat infections of filamentous fungi which have severe negative impact in medicine and agriculture.


Background
All organisms have evolved several defence systems in order to protect themselves against bacteria, fungi and viruses. Higher organisms have developed a complex network of humoral and cellular responses, called adaptive immunity. A second defence system, the innate immunity, consists of many components, including small peptides with a broad antimicrobial spectrum [1,2]. The production of such proteins with antimicrobial activity is not limited to higher eukaryotes, but also found in microorganisms, including fungi. The diversity of these proteins is reflected in their mode of action and their species-specificity. Some of them form pores in the membrane, others are known to inhibit cell wall synthesis or interfere with nucleic acids and their synthesis [3,4]. They can be involved in the inhibition of protein synthesis or interfere with cell cycle control [3,4]. A relatively new group of antimicrobial proteins secreted by filamentous ascomycetes includes small, cationic and cysteine-rich proteins. So far, only few antifungal proteins have been characterized, namely AFP from Aspergillus giganteus, ANAFP from Aspergillus niger, PAF from Penicillium chrysogenum and NAF from Penicillium nalgiovense [5][6][7][8].
The mode of action of these proteins is not fully understood. Nevertheless, there is evidence, that their toxicity is mediated by interaction with distinct molecules or receptors at the outer layers of the cell, e.g. cell wall or plasma membrane. Deleterious effects can then be induced either by transmitting signals from the outer layers into the cell, or by internalization of the protein and interaction with internal molecules [9][10][11][12][13][14][15]. Similar to substances that perturb the cell wall, such as caspofungin, congo red or calcofluor white (CFW) [10,16], the A. giganteus antifungal protein AFP was found to modulate the cell wall composition by enhancing the expression of the α-1,3-glucan synthase A gene (agsA), possibly by the activation of the cell wall integrity pathway (CWIP), and inhibiting chitin synthesis in sensitive fungi [10]. This, however, stands in contrast to the mode of action of the P. chrysogenum antifungal protein PAF which fails to activate the CWIP [9]. However, the central players that trigger cell wall remodelling in AFPsensitive fungi have not been investigated so far.
Another mechanistic function of antifungal proteins is the interference with ion, especially Ca 2+ ion homeostasis and signalling [15,17,18]. We could recently show that the P. chrysogenum antifungal protein PAF severely perturbed the Ca 2+ homeostasis of Neurospora crassa by rapidly elevating the cytoplasmic Ca 2+ [Ca 2+ ] c resting level [17]. Numerous reports indicate that the activity of antifungal proteins can be antagonized by the external addition of Ca 2+ ions to the test medium [15,[17][18][19][20][21] pointing towards the induction of adaptive responses which may be triggered by Ca 2+ signalling [15,17].
The aim of this study was to characterize in more detail the mode of action of the A. giganteus AFP variant protein AFP NN5353 and to investigate the pathways that might be affected/modulated by this antifungal protein. Therefore, we focussed our interest on the involvement of the CWIP and the Ca 2+ signalling in the toxicity of AFP NN5353 . To address these questions, we used the highly AFP NN5353 sensitive model organisms A. nidulans and A. niger for which appropriate mutant strains were available.

Results
In silico analysis of AFP NN5353 CLUSTALW amino acid (aa) sequence analysis of AFP NN5353 with other known antifungal proteins revealed that AFP NN5353 from A. giganteus strain A3274 is a protein homologous to AFP from A. giganteus strain MDH 18894 [8,22]. AFP NN5353 exhibits > 90% identity with AFP, but only 42% identity with the P. chrysogenum PAF and 27% identity with the A. niger ANAFP. In fact, the secreted mature form of AFP NN5353 consists of 51 aa and differs only in 5 aa from AFP ( Figure 1). Three aa exchanges belong to structurally related aa, one aa exhibits weak similarity and one aa is different (position 4). These aa exchanges do not influence the theoretical isoelectric point (pI) of AFP NN5353 , which is the same as for AFP (pI 9.3, http://expasy.org/tools/ protparam.html). Most importantly, AFP NN5353 still contains the putative chitin-binding domain CKYKAQ present in AFP but not in PAF or ANAFP and also harbors all conserved cysteine residues important for protein stabilization [10,23].

Antifungal activity of the protein AFP NN5353
To investigate the antifungal specificity of AFP NN5353 , fifteen filamentous fungi were tested for their susceptibility to the protein. Since antifungal proteins might be useful for biotechnological applications, filamentous human and plant pathogenic fungi were selected as test organisms (e.g. Fusarium oxysporum, Botrytis cinerea, Mucor sp. and A. fumigatus) in addition to the model organisms A. nidulans and A. niger. As shown in Table  1, thirteen out of fifteen tested moulds were found to be sensitive against AFP NN5353 . A. nidulans wild type, N. crassa wild type and A. niger wild type were the most sensitive strains to AFP NN5353 . The minimal inhibitory concentration (MIC) of AFP (the concentration that completely inhibited conidial germination in liquid growth assays) was 0.2 μg/ml for A. nidulans, 0.5 μg/ml for N. crassa and 1 μg/ml for A. niger. Two strains were unaffected at the protein concentrations tested: M. circenelloides and M. genevensis were insensitive against AFP NN5353 when concentrations up to 500 μg/ml were used.

AFP NN5353 interferes with the cell wall integrity of A. nidulans
It is known that antifungal compounds such as congo red, caffeine, CFW or caspofungin interfere with cell wall biosynthesis and weaken the cell wall in fungi (reviewed by [24]). The remodeling of the cell wall by these antifungal compounds is mediated by the activation of the CWIP. In fungi, extracellular signals are transmitted via the membrane bound small GTPase RhoA to the central regulators Pkc and Mpk, which are regulated by phosphorylation/dephosphorylation. The signal transduction cascade eventually enforces transcription of cell wall synthesis genes, partly via the transcription factor RlmA [16,25]. Respective loss-offunction or conditional mutants show hypersensitive phenotypes in the presence of cell wall perturbing agents [9,[24][25][26]. Similar to substances that weaken the cell wall, the A. giganteus antifungal protein AFP modulates the cell wall composition by inhibiting chitin synthesis in sensitive fungi (e.g. A. niger, A. oryzae) and inducing the expression of agsA most likely by the activation of the CWIP [10].
To study the involvement of the CWIP in AFP NN5353 toxicity, we first tested whether the osmotic stabilizer sorbitol counteracts the toxicity of AFP NN5353 . In the absence of AFP NN5353 A. nidulans proliferated less well in the presence of 1 M sorbitol and reached only 30% growth compared to the growth in standard medium (100%). Nevertheless, the addition of 1 M sorbitol to the growth medium strongly reduced the activity of AFP NN5353 on A. nidulans wild type. The osmotic stabilizer ameliorated growth in the presence of 0.05 μg/ml AFP NN5353 by 80% compared to a 10% growth rate in the absence of sorbitol (Table 2). This was even more accentuated when 0.1 and 0.2 μg/ml AFP NN5353 were applied, suggesting that AFP NN5353 indeed weakens the cell wall of A. nidulans.
To investigate whether AFP NN5353 induces agsA gene transcription similar to AFP via the Pkc/Mpk signalling pathway, we tested the effect of the antifungal protein on the transgenic A. niger strain RD6.47 which expresses a nuclear-targeted GFP protein fused to the A. niger agsA promoter. RD6.47 germlings were treated with AFP NN5353 (conc. 10 to 100 μg/ml) for 2 h and analyzed microscopically. As shown in Additional file 1, a nuclear signal was clearly detectable in germlings of RD6.47 treated with ≥ 50 μg/ml AFP NN5353 , similar to that when exposed to 10 μg/ml caspofungin. In untreated germlings, however, no signal could be observed. These observations perfectly match with the data obtained for AFP [10]. It has to be noted here that antifungal protein concentrations higher than the MIC determined for conidia (> 10-50 fold) are needed to inhibit the growth of germlings or hyphae of sensitive fungi [10,27] (data not shown).
Next, we tested several A. nidulans mutant strains affected in central players of the CWIP for their susceptibility to AFP NN5353 by determining their radial growth in the presence or absence of the antifungal protein. Since RhoA is an essential protein in A. nidulans, two strains with ectopic copies of the constitutively active rhoA G14V allele and the dominant rhoA E40I allele [28] were tested in comparison to the wild type strain (GR5). The rhoA G14V mutation prevents the hydrolysis of GTP and therefore renders RhoA constantly active [28]. Similarly, the GTP hydrolysis is inhibited in the RhoA E40I strain, but this mutation also perturbs the binding of the GTPase activating protein (GAP) to RhoA and possibly disturbs downstream effectors of RhoA-GAP [28]. The constitutively active RhoA G14V and the dominant RhoA E40I strain exhibited the same sensitivity towards AFP NN5353 as the wild type strain at low protein concentrations (≤ 0.2 μg/ml) ( Figure 2A). Interestingly, the dominant RhoA E40I strain was more Figure 1 Clustalw sequence alignment http://www.ebi.ac.uk/Tools/msa/clustalw2/ of the antifungal proteins AFP NN5353 and AFP from A. giganteus, ANAFP from A. niger and PAF from P. chrysogenum. Identical amino acids (aa) are marked with (*), aa with strong similarity are indicated with (:) and aa with weak similarity are marked with (.).  resistant to AFP NN5353 than the wild type strain or the RhoA G14V strain at higher protein concentrations (1 μg/ ml) ( Figure 2A). Therefore, we suggest that the toxicity of AFP NN5353 is transmitted by RhoA-GAP targets and not by RhoA itself. These mutants performed similarly when exposed to the orthologous P. chrysogenum antifungal protein PAF [9].
In addition, mutants defective in PkcA and MpkA activity were tested for their AFP NN5353 susceptibility. As pkcA is an essential gene in A. nidulans, a conditional alcA-PKC mutant strain was used, where the pkcA gene was put under the control of the alcA promoter, which is repressed by glucose but derepressed by glycerol [26]. Both the conditional alcA-PKC mutant (cultivated under repressive conditions) and a ΔmpkA mutant were hypersensitive to AFP NN5353 compared to their recipient strains R153 and GR5, respectively, indicating that the activity of PkcA and MpkA confers a certain resistance to AFP NN5353 (Figure 2A). The hypersensitive phenotype of the ΔmpkA mutant was also confirmed by liquid growth inhibitory assays. In unchallenged liquid condition, the GR5 and the ΔmpkA mutant showed a comparable proliferation rate ( Figure  2B). In the presence of 0.05 μg/ml AFP NN5353 , however, Figure 2 AFP NN5353 susceptibility of A. nidulans mutants RhoA G14V , RhoA E40I , alcA-PkcA and ΔmpkA compared to the respective recipient strains GR5 and R153. (A) A total of 2 × 10 3 conidia were point inoculated on agar plates (CM for GR5, RhoA G14V , RhoA E40I and ΔmpkA, repressive MM containing 1% glucose according to [26] for R135 and alcA-PkcA) containing the appropriate supplements and 0, 0.2 and 1 μg/ml AFP NN5353 for GR5, RhoA G14V , RhoA E40I , R135 and alcA-PkcA. The ΔmpkA mutant and its reference strain GR5 were exposed to 0, 0.5 and 1 μg/ml AFP NN5353 . The plates were incubated at 37°C for 48 h. (B) 1 × 10 4 conidia/ml of the ΔmpkA mutant and GR5 were treated with 0.05 μg/ml AFP NN5353 or without protein (controls) in a total volume of 200 μl of appropriately supplemented CM in 96-well plates.
the mpkA deletion strain did not germinate whereas the GR5 strain still exhibited 11% growth. Note that growth inhibition in liquid conditions requires less antifungal protein to monitor its toxicity than on solid media probably due to less diffusion in the latter case (data not shown).
From these data we conclude that AFP NN5353 interferes with the cell wall homeostasis of A. nidulans and that this interaction is mediated by PkcA/MpkA signalling, although independently from RhoA.

AFP NN5353 disrupts calcium homeostasis in A. niger
Supplements other than osmotic stabilizers can also antagonize the activity of antifungal proteins from plants and ascomycetes. For example, the addition of cations such as Ca 2+ ions to the growth medium reversed the antifungal activity of the P. chrysogenum PAF [17], the A. giganteus AFP [15,21] and of plant defensins [29,30] which are usually positively charged due to their high pI. A cation-sensitive antifungal mode of action can for example be associated with the perturbation of the intracellular Ca 2+ homeostasis by antifungal peptides [17,18] but might also result from the interference of cations with antifungal-target interaction(s).
Therefore, we tested to which extend these effects also account for the antifungal activity of AFP NN5353 . To this end, we selected A. niger as model organism because this mould was highly sensitive to AFP NN5353 and a transgenic strain was available that expressed the recombinant codon optimized Ca 2+ -sensitive photoprotein aequorin for measuring the [Ca 2+ ] c resting level in response to AFP NN5353 [31]. First, we tested the activity of AFP NN5353 in Vogels* medium supplemented with 5-20 mM CaCl 2 or without CaCl 2 as a control (data not shown). Addition of CaCl 2 did not influence the growth of A. niger up to a concentration of 20 mM. The growth of A. niger exposed to AFP NN5353 , however, ameliorated in the presence of increasing concentrations of CaCl 2 . 20 mM CaCl 2 neutralized the toxicity of 0.5-1.0 μg/ml AFP NN5353 and the treated samples resumed growth to 100% (Table 3).
Next, we determined the influence of AFP NN5353 on the intracellular Ca 2+ signature. Before AFP NN5353 addition, the resting level of the intracellular Ca 2+ was 0.08 μM. We could show, however, that the [Ca 2+ ] c resting level was significantly increased in twelve h old A. niger cultures that were treated with 20 μg/ml AFP NN5353 . The [Ca 2+ ] c resting level rose to a maximum of 0.19 μM within the first 8 min and stayed elevated throughout the time of measurement (60 min), whereas the Ca 2+ level of the untreated control remained at 0.08 μM (Figure 3). This indicated that AFP NN5353 indeed disrupts Ca 2+ homeostasis in A. niger.
To exclude the possibility that the AFP NN5353 induced rise in the [Ca 2+ ] c resting level is due to membrane permeabilization and/or pore formation, we studied the effects of AFP NN5353 on germlings in the presence of CMFDA, a membrane permeant dye that is metabolized by viable cells, and the membrane impermeant dye propidium iodide (PI). Additional file 2 shows that samples treated with 20 μg/ml AFP NN5353 for 10 min metabolized CMFDA but did not take up PI, resulting in green but no red fluorescence, similar to untreated controls. This indicated that the plasma membrane was still intact after 10 min of protein treatment. Samples exposed to ethanol did not metabolize CMFDA but appeared bright red due to PI internalization, indicating that here the membrane was permeabilized. We therefore conclude that the rapid increase in [Ca 2+ ] c within the first 10 min of protein treatment is not the result of uncontrolled Ca 2+ influx due to plasma membrane permeabilization.
The calcium chelator BAPTA abrogates the AFP NN5353induced calcium signature The increased [Ca 2+ ] c in response to AFP NN5353 treatment could originate from extracellular and/or from intracellular Ca 2+ stores, such as mitochondria, vacuoles, Table 3 The effect of 20 mM external CaCl 2 (in Vogels* medium) on the growth inhibitory activity of AFP NN5353 on A. niger strain A533. AFP NN5353 (μg/ml) Vogels* Vogels* + 20 mM Ca 2+ 0 100 ( SD ± 10) 100 ( SD ± 8) 0. 5 12 ( SD ± 3) 101 ( SD ± 9) 1.0 no growth 105 ( SD ± 6) OD 620 was measured after 24 h of incubation. The growth of untreated controls was normalized to 100% to evaluate the percent growth of samples in the presence of AFP NN5353 . Vogels* medium without CaCl 2 supplementation contains 0.7 mM Ca 2+ . Results are expressed as mean ± SD (n = 3).   (Figure 4). However, a pretreatment of the samples with 10 mM BAPTA before the addition of AFP NN5353 inhibited the protein-specific increase in [Ca 2+ ] c resting level (Figure 4). Interestingly, the elevated [Ca 2+ ] c in response to a 40 min AFP NN5353treatment dropped to the resting level immediately after the addition of 10 mM BAPTA (Figure 4) [31,32]. One of these physiological stimuli is mechanical perturbation, which is achieved by the rapid injection of isotonic medium into the test system. This stimulus results in a unique Ca 2+ signature, likely involving different components of the Ca 2+ -signalling and Ca 2+ homeostatic machinery. Changes in this specific Ca 2+ signature in the presence of compounds, such as AFP NN5353 , can give insights into the mode of action of these compounds. In our study, twelve h old cultures of A. niger were pre-incubated with AFP NN5353 for 60 min and thereafter subjected to mechanical perturbation (rapid injection of 100 μl Vogels medium). The resulting Ca 2+ signature, including [Ca 2+ ] c resting level, kinetics and amplitude, were determined and compared with controls that were not exposed to the protein but also subjected to mechanical perturbation. As shown in Figure 5, AFP NN5353 provoked a less pronounced [Ca 2+ ] c amplitude; however, the [Ca 2+ ] c level remained elevated even after the stimulus specific response had stopped.
AFP NN5353 binding and uptake are essential for protein toxicity in A. nidulans To understand the function of antifungal proteins, the identification of the site of action in target organisms is crucial. So far, controversial reports exist of the  localization of the homologous A. giganteus AFP protein. AFP has been detected to bind to outer layers, e.g. the cell wall or the plasma membrane of sensitive fungi [20,21] and a time-and concentration-dependent intracellular localization was reported [20]. In another study, Alexa-labelled AFP was shown to be internalized by the fungal cell and to localize to the nucleus [33].
To dissect the uptake and localization of AFP NN5353 , we performed indirect immunofluorescence staining with A. nidulans wild type exposed to a sublethal concentration of AFP NN5353 (0.2 μg/ml). We applied a protein amount below the toxic concentration for hyphae to maintain the cellular structure and to avoid apoptotic cell disruption [34]. Our study revealed that the protein was internalized after 90 min of incubation, mostly in hyphal tips, but also within hyphal segments ( Figure 6A, B). The protein seemed not to localize to cell compartments, but was distributed in the cytoplasm. Similar results were obtained with A. niger wild type (data not shown). Control experiments proved the specificity of the intracellular immunofluorescent signals: no intracellular fluorescent signals were detected in samples where either AFP NN5353 (Figure 6C, D) or the primary antibody or the secondary antibody was omitted (data not shown).
To analyse the AFP NN5353 localization in more detail, A. nidulans was incubated with AFP NN5353 in the presence of latrunculin B, a potent inhibitor of actin polymerization and endocytosis [35][36][37]. At low latrunculin B concentrations (5 μg/ml), protein uptake was severely reduced compared to the positive control without latrunculin B (data not shown), whereas 20 μg latrunculin B/ml completely inhibited the uptake of 0.2 μg/ml AFP NN5353 . The solvent of latrunculin B, DMSO, had no adverse effect on protein uptake (data not shown). This indicates that AFP NN5353 enters the A. nidulans cells by an endocytotic mechanism ( Figure 6E, F).
Based on our observation that Ca 2+ ions antagonize the growth inhibitory activity of AFP NN5353 , we questioned whether Ca 2+ prevents actin-mediated internalisation of the antifungal protein. Indeed, the presence of 10 mM CaCl 2 inhibited protein uptake ( Figure 6G, H). Most interestingly, no specific fluorescent signals were detectable in M. circinelloides when treated with up to 500 μg/ml of antifungal protein (data not shown), indicating that AFP NN5353 does not bind to insensitive strains.

Discussion
In this study we provide important insights into the mechanistic basis of AFP NN5353 , a AFP homologous protein.
Species specificity tests revealed that AFP NN5353 is active against a broad range of filamentous fungi, including human and plant pathogens. Although the proteins AFP NN5353 and AFP are almost identical and show a similar toxicity, MICs for AFP NN5353 differed slightly from those reported for AFP [21]. We attribute this discrepancy to differences in the experimental setups, e.g. fungal strains, medium composition, conidial inoculum, incubation times, cultivation temperature etc., rather than to the differences in the primary sequence of both proteins.
It has been reported that the closely related AFP protein interfered with cell wall synthesis [10] and our finding that the osmotic stabilizer sorbitol neutralized AFP NN5353 toxicity further corroborated this assumption. Two A. nidulans mutants, the conditional alcA-PkcA and the mpkA deletion mutant showed a hypersensitive phenotype when exposed to AFP NN5353 . This is in agreement to the reported function of cell wall stressing agents, such as CFW or caffeine in S. cerevisiae and A. nidulans [9,16,24,26,38,39] and to the Penicillium antifungal protein PAF [9]. Importantly, Mpk function is essential for CWIP activation in both, unicellular and filamentous fungi [10,16,40] and triggers the activation of the transcription factors Rlm1p and SBF which regulate the expression of cell cycle regulated genes and genes involved in the synthesis and remodelling of the fungal cell wall in S. cerevisiae [41,42]. Similarly, RlmA dependent induction of the expression of the ags gene was also reported for aspergilli [25]. Importantly, the activation of the CWIP can occur in a RhoA-dependent, e.g. with CFW [9,43], or RhoA-independent way, the latter proved for PAF and caffeine [9,16] and for AFP NN5353 (this study). As proposed by [28] the dominant rhoA E40I allele suffers from a perturbation of its GAP binding domain and downstream effectors of Rho-GAP might be disturbed. Therefore, we hypothesize that Rho-GAP targets might be involved in the toxicity of AFP NN5353 similarly to the mode of action of the P. chrysogenum PAF [9]. Our assumption of the activation of the CWIP by AFP NN5353 was further strengthened by the fact, that AFP NN5353 treatment induced agsA expression in the A. niger reporter strain. This result was consistent with the activity of AFP and caspofungin [10], but differed to the function of PAF, where no CWIP activation and no induction of cell wall biosynthesis genes occurred [9].
Therefore, we conclude that AFP NN5353 triggers cell wall remodeling via Pkc/Mpk signalling. We further deduce from our data that similarities and differences exist in the molecular targets and the mode of action of antifungal proteins from filamentous fungi, e.g. AFP NN5353 and PAF -despite their homology. This phenomenon was also reported for other closely related antifungal proteins, such as the plant defensins MsDef1 and MtDef4 from Medicago spp. [44].
Apart from the activation of the CWIP, the perturbation of the Ca 2+ homeostasis represents a major mechanistic function of antifungal proteins in sensitive fungi [17,18]. The intracellular Ca 2+ response to AFP NN5353 in A. niger reflected that of the Penicillium antifungal protein PAF in N. crassa [17]. The rapid and sustained increase of the [Ca 2+ ] c resting level depended on a sustained influx of Ca 2+ ions from the external medium. Moreover, the AFP NN5353 induced changes in the Ca 2+ signature of mechanically perturbed A. niger cells further underlines the disruption of the Ca 2+ response and homeostasis by AFP NN5353 . The addition of CaCl 2 to the growth medium reduced the susceptibility of A. niger towards the antifungal protein and decreased the AFP NN5353 specific rise in the [Ca 2+ ] c resting level. Both observations point towards an adaptive response which is mediated most probably via Ca 2+ signalling. First, high extracellular Ca 2+ concentrations trigger chitin synthesis in A. niger and thereby confer increased protection against antifungal proteins as shown for AFP [15]. Second, it primes the Ca 2+ homeostatic machinery to better maintain a low [Ca 2+ ] c resting level when challenged with the antifungal protein, e.g. by (i) the increase of the activity of existing Ca 2+ pumps/transporters to counteract the AFP NN5353 -specific intracellular Ca 2+ perturbation, or (ii) the modulation of the expression of Ca 2+ channels/pumps/exchangers [17]. The former hypothesis (i) might be supported by the observation that the addition of CaCl 2 only 10 min before A. niger was challenged with AFP NN5353 restored the low [Ca 2+ ] c resting level. However, the perturbation of the Ca 2+ homeostasis by a sustained elevation of the [Ca 2+ ] c resting level indicates that A. niger is not able to restore the low [Ca 2+ ] c resting level after exposure to AFP NN5353 and this might trigger programmed cell death (PCD) on the long term as it was shown to occur in A. nidulans in response to the P. chrysogenum PAF [34].
Since AFP was shown to cause membrane permeabilization [21], the influx of Ca 2+ might be due to changes in membrane permeability for this ion, if not the formation of pores. However, our staining experiments with CMFDA and PI exclude this possibility at least in the first 10 min of exposure to AFP NN5353 when the [Ca 2+ ] c resting level reaches its maximum. This result is further corroborated by the fact that higher external concentrations of Ca 2+ reduced the AFP NN5353 specific rise in [Ca 2+ ] c resting level which -in our opinion -would not occur with leaky membranes. However, we do not exclude changes in membrane permeability at longer exposure times to this antifungal protein and more studies are needed to answer this question.
Finally, we observed that the internalization of AFP NN5353 is characteristic for sensitive but not resistant moulds. A lack of binding of AFP NN5353 to insensitive fungi might point towards the absence or inaccessibility of a putative interacting molecule at the cell surface. AFP NN5353 localized to the cytoplasm of target fungi only when actin filaments were formed. This is in agreement with the endocytotic uptake and intracellular localization of the P. chrysogenum antifungal protein PAF in sensitive filamentous fungi [14,45]. Importantly, we observed that AFP NN5353 was internalized by hyphae even under subinhibitory concentrations (0.2 μg/ml for A. nidulans) which suggests that a threshold concentration is required to cause severe growth defects in target fungi.
The presence of high concentrations of extracellular Ca 2 + counteracted AFP NN5353 uptake. This finding parallels well with the report of [20] that the presence of cations, such as Ca 2+ , interfered with the binding of AFP to the surface of F. oxysporum and with our observations made with the Penicillium PAF (unpublished data). One possible explanation might be that extracellular Ca 2+ ions compete with AFP NN5353 for the same molecular target on the fungal surface which might represent a first binding receptor or even a "gate" for protein uptake [20,21] or, alternatively, that the interacting target is repressed under these conditions [17]. An additional explanation might be that the primary cell-surface localized AFP NN5353 target might be masked due to a Ca 2+ -dependent stimulation of chitin synthesis and cell wall remodeling as recently observed for AFP in A. niger [15]. This further suggests that the activation of the CWIP and the agsA induction does not mediate sufficient resistance to survive the toxic effects of AFP NN5353 . Instead, according to the "damage-response framework of AFP-fungal interactions" [15], the chitin response might represent the better strategy for fungi to survive the antifungal attack.

Conclusions
Based on the growth inhibitory activity, antifungal proteins like AFP NN5353 can be well considered as promising candidates for future antimycotic drug developments. However, for biotechnological exploitation, the detailed knowledge on the mode of action is demanded. Our study shows that the detrimental effects caused by the A. giganteus antifungal protein AFP NN5353 in sensitive target aspergilli are based on the interaction of this protein with more than one signalling pathway. In Figure 7, we present a tentative working model. The toxicity of AFP NN5353 is mediated via PkcA/MpkA signalling which occurs independently from RhoA. Instead, so far unidentified RhoA-GAP effector molecules might contribute to AFP NN5353 toxicity. The activation of the CWIP by AFP NN5353 induces the agsA gene expression which is, however, insufficient to counteract toxicity of the protein. Furthermore, AFP NN5353 leads to an immediate and significant increase of the [Ca 2+ ] c resting level in the cell. This sustained perturbation of the Ca 2+ homeostasis could lead to PCD [17,34]. The presence of extracellular Ca 2+ neutralizes the toxic effects of AFP NN5353 and improves the resistance of the target organism possibly by decreasing the elevated [Ca 2+ ] c resting level and stimulating the fortification of the cell wall by the induction of chsD expression as shown for AFP [15]. Further investigations are in progress to clarify how these pathways are interconnected and interfere with each other on the molecular level.

Strains, Media and Chemicals
Fungal strains used in this study are listed in Table 5. All strains were obtained from the culture collections FGSC, ATCC, CBS, from the Institute of Microbiology, Division of Systematics, Taxonomy and Evolutionary Biology at the Leopold Franzens University of Innsbruck, or the strain collection of the Department of Biotechnology, National Institute of Chemistry, Ljubljana, Slovenia. Unless otherwise stated, all fungi were grown in complete medium (CM) [19] with the respective supplements [28,38]. R153 and alcA-PkcA were grown in defined minimal medium (MM) according to [26]. Ca 2+ response experiments were performed in Vogels medium [46]. For experiments with CaCl 2 supplementation, the KH 2 PO 4 concentration of the culture media was reduced from 37 mM to 10 mM to avoid precipitation of supplemental Ca 2+ and these media were called CM* and Vogels*. Chemicals were purchased from Sigma. AFP NN5353 and polyconal rabbit anti-AFP NN5353 antibody were generous gifts from Mogens T. Hansen, Novozymes, Denmark. The antifungal protein was isolated from A. giganteus strain A3274 (CBS 526.65), purified and analyzed by HPLC as described in the patent application WO94/01459 [47].

Growth inhibition assays
Antifungal activity assays were performed in 96-well plates in CM or Vogels medium inoculated with 1 × 10 4 conidia/ ml and supplemented with various concentrations of AFP NN5353 or with equivalent amounts of buffer (untreated controls). Fungal growth was monitored microscopically with an Olympus CK40 microscope equipped with a Zeiss MRc digital camera and the growth rates were determined spectrophotometrically as described previously [19]. Alternatively, 2 × 10 3 conidia were spotted in Figure 7 Tentative model of the mechanistic function of the A. giganteus antifungal protein AFP NN5353 on Aspergillus sp. The response against AFP NN5353 attack is mediated via PkcA/MpkA signalling and results in increased agsA transcription. However, the activity of the CWIP occurs independently from RhoA and so far unidentified RhoA-GAP effector molecules might contribute to the AFP NN5353 toxicity. Furthermore, AFP NN5353 leads to an immediate and significant increase of the [Ca 2+ ] c resting level in the cell. The sustained perturbation of the Ca 2+ homeostasis could lead to PCD [17,34]. The presence of elevated concentrations of extracellular Ca 2+ counteracts the toxic effects of AFP NN5353 and improves the resistance of the target organism by decreasing the elevated [Ca 2+ ] c resting level. Whereas cell wall remodelling via CWIP seems to be insufficient to counteract AFP NN5353 activity, the fortification of the cell wall by the induction of chsD expression might represent an adequate response to increase resistance [15]. 5 μl aliquots on appropriately supplemented agar plates. The plates were then incubated at 37°C for up to 72 h. Every 24 h, the plates were photographed and the colony diameters were determined. All assays were performed as technical triplicates and biological duplicates.
Analysis of the induction of the agsA expression by a GFP-based reporter system The A. niger reporter strain RD6.47 carries the agsA promoter fused to a nucleus-targeted GFP (H2B::eGFP) [27]. Activation of the CWIP can be monitored by the increase in nuclear fluorescence. Analysis of the activation of the agsA promoter by 10-100 μg/ml AFP NN5353 was performed as described in [10]. As a positive control, caspofungin at a concentration of 10 μg/ml was used. Fluorescence images were taken from coverslips observed with an Axioplan 2 microscope (Zeiss) equipped with a Sony DKC-5000 digital camera.

Fluorescence staining Indirect immunofluorescence staining
A. nidulans was grown over night on glass cover slips at 30°C in CM. They were further incubated for 90 min in the presence or absence (controls) of 0.2 μg/ml AFP NN5353 . The samples were stained as described previously [14] and incubated with rabbit-anti-AFP NN5353 antibody (1:2.500, Novozymes, Denmark) for at least 60 min. Immunocomplexes were detected with FITC-conjugated swine-anti-rabbit IgG (1:40, DAKO, Germany). All samples were embedded in Vectashield mounting medium (Vector Laboratories, Burlingame, USA). Microscopy was done with a Zeiss Axioplan fluorescence microscope or a Zeiss confocal laser scanning microscope as described in [14].
For incubation with latrunculin B (Sigma, Austria), samples were treated with 0.2 μg/ml AFP NN5353 and 10 μg/ml latrunculin B for 80 min. As a control, samples were treated with DMSO to exclude artifacts evoked by the dissolvent of latrunculin B.
For detection of AFP NN5353 in the presence of elevated concentrations of CaCl 2 , fungi were grown in CM* medium and then treated with 0.2 μg/ml AFP NN5353 in the presence of 10 mM CaCl 2 for 90 min.

Analysis of membrane permeabilization and cell viability
To determine if AFP NN5353 permeabilized the plasma membrane of A. niger germlings, we used a combination of propidium iodide (PI) and fluorescein diacetate (cell tracker, CMFDA green, Invitrogen) according to [48]. Twelve h old A. niger germlings were grown in Vogels medium and pretreated with the two dyes (final conc. 5 μg/ml each) for 15 min before AFP NN5353 was added to a final concentration of 20 μg/ml. Samples without AFP NN5353 served as controls for positive CMFDA staining, while ethanol (70%) was used to permeabilize the membrane for positive PI staining.
Analysis of the calcium response to AFP NN5353 application 10 5 conidia/ml of the A. niger strain A533 expressing codon optimized aequorin were grown in Vogels* medium containing 10 μM coelenterazine (Biosynth, Switzerland) at 30°C for twelve h in the dark. The [Ca 2+ ] c resting level and mechanical perturbation experiments and the calibration of [Ca 2+ ] c were performed as described in [17]. Additional file 2: Viability staining of A. niger germlings after AFP NN5353 exposure. Twelve h old A. niger germlings were stained with fluorescein diacetate (CMFDA, middle pannels) and propidium iodide (right pannels). The left panels show the respective light micrographs. All samples were pretreated with the dyes for 15 min before 20 μg/ml AFP NN5353 was added (B). Controls remained untreated (A) or were exposed to 70% ethanol (C). Scale bar, 50 μm.