Strains and growth conditions

Fungal strains used throughout this study are listed in S1 Table. All P. chrysogenum shaking cultures were inoculated with 10⁸-10⁹ conidia in 200 mL defined minimal medium (MM) and grown for 72 h at 25°C as described previously [12]. Protein isotopic ¹⁵N-labelling for NMR analysis was performed by replacing the nitrogen source by 0.3% Na¹⁵NO₃ (Eurisotop) in MM [7]. N. crassa was used as PAF-sensitive model organism and cultivated in 5-fold diluted Vogel's medium (0.2 x Vogel's) [13] at 25°C for growth inhibition assays, fluorescence staining experiments and measurements of intracellular Ca²⁺ fluxes. N. crassa conidia were generated from surface cultures cultivated on Vogel's agar at 37°C for 24 h under continuous light.

High-yield expression of PAF and PAF^D19S

An approx. 2080 bp PstI restriction fragment from pSK275paf4400 [14] comprising the paf gene (420 bp) and approx. 1280 bp of the 5'-UTR and 370 bp of the 3'-UTR was ligated into the PstI site of an empty pSK275 resulting in plasmid pSK275paf [14]. For site-directed mutagenesis the preferential codon usage of P. chrysogenum was taken into account to design two inverse and overlapping oligonucleotides that carried a mismatch sequence coding for the new amino acid replacing the original one (S2 Table). For PCR ligation two overlapping PCR products were amplified, containing the desired mutation (PCR 1: mismatch primer forward and primer M13; PCR 2: mismatch primer reverse and opaf12) and combined in a third PCR reaction using primers T7var and opaf11 (Q5^® High-Fidelity DNA Polymerase, NEB). The final PCR product was digested with NheI/NotI and cloned into the NheI/NotI digested pSK275paf, replacing the original paf sequence. The expression of the mutated paf gene was still under the control of the strong paf promoter and the expression plasmid was named pSK257paf^D19S. The correct mutation of the paf nucleotide sequence was verified using Sanger sequencing (Eurofins/MWG Operon).

In all transformation experiments the paf deletion mutant P. chrysogenum Δpaf [14] was used as recipient strain for pSK275paf and pSK257paf^D19S for the production of PAF and PAF^D19S, respectively. The recipient strain was grown in MM for 48 h at the growth conditions described above and protoplastation and transformation were carried out as described [15,16]. Transformants were selected on MM agar plates supplemented with 0.3-0.6 mg mL^-1 pyrithiamine hydrobromide (Sigma-Aldrich).

Protein purification and MS analysis

PAF and PAF^D19S were purified from 1 L supernatant of 72 h shaking cultures in MM as described previously [12] with the following changes: After loading of the cleared and ultra-filtered supernatant onto a CM-sepharose column, the proteins were eluted with 0.1-0.3 M NaCl and protein containing fractions were pooled, dialyzed against ultra-pure ddH₂O, concentrated and filter sterilized. Protein concentrations were determined spectrophotometrically and the purity checked by SDS-PAGE. Proteins were stored at -20°C (in aqueous solution or in lyophilized form for NMR analyses).

The identity of the purified PAF and PAF^D19S was proved by the determination of the molecular mass by electro-spray ionization mass spectrometry (ESI-MS) at the Protein Micro-Analysis Facility (Medical University of Innsbruck). In brief, protein samples were dissolved in 50% aqueous methanol containing 0.1% formic acid and injected into a CESI 8000 (SCIEX, USA) coupled to a Q Exactive (Thermo Scientific, 180 nL min^-1 flow rate). Protein mass determination was performed by deconvolution using the integrated Xcalibur Xtract software (Thermo Scientific).

NMR measurements, signal assignment and structure calculations

¹⁵N-PAF^D19S NMR sample was prepared by dissolving lyophilized protein in phosphate buffer (10 mM Na₃PO₄, 40 mM NaCl, 0.04% NaN₃, pH = 6.0) to the final concentration of 1.6 mM. NMR experiments for structure determination were carried out at 298 K using AVANCE-II 700 and 500 MHz spectrometers (Bruker). For proton chemical shift referencing DSS (2,2-dimethyl-2-sila-pentane-5-sulfonic acid) was used as external reference and heteronuclear shifts were referenced indirectly from the gyromagnetic ratios for ¹⁵N and ¹³C. Spectra were processed with TopSpin 3.1 (Bruker) and analysed with CARA 1.8.4 [17]. Sequence specific resonance assignment was obtained from 2D ¹H–¹H NOESY (130 ms mixing time), 3D ¹⁵N HSQC-TOCSY (60 ms) and 3D ¹⁵N HSQC-NOESY (130 ms) spectra at 298 K. ¹³C-¹H HSQC spectra were acquired using the natural abundance ¹³C isotopes (80 transients x 640 experiments in the indirect dimension). Sequential resonance assignments were carried out with the identification of NH(i)-Hα(i-1) distance proximities through the backbone using the combination of 3D TOCSY-HSQC and NOESY-HSQC spectra. Chemical shifts of PAF (BMRB: 19657) were used to support the resonance assignment and structure refinement of PAF^D19S. For the calculation of PAF^D19S structure, data were collected from 2D ¹H-¹H NOESY only. After initial calculations with NOE data, structure refinement was performed using C_α and C_β chemical shift data, and disulphide restraints. Backbone torsion angles were calculated with TALOS+ [18,19]. This procedure had no effect on the overall fold of conformational ensemble, but backbone RMSD values were lower than without torsion angle restraints. Cyana 2.1 algorithm was used in combination with Atnos/Candid procedure for structure calculations [20] and NOE assignment, respectively. Disulphide pattern of PAF^D19S was assumed to be identical with that of the wild-type PAF [9], and was given explicitly as covalent bond restraint input to Cyana. Ensembles of 100 structures were calculated and from those, 20 calculated structures were selected according to lowest energy. Structure visualization and analysis was done using MOLMOL [21] and in-house scripts. ¹⁵N-CEST NMR experiments and temperature dependent experiments were performed as described before [8]. In order to obtain comparable datasets, the same instrumentation was used in the same temperature range (265-343K). Modelling of two-state thermal unfolding was carried out using in-house written MATLAB^® scripts as described before [8]. CLEANEX NMR data were fitted against the theoretical function using an in-house written MATLAB^® script to yield NH-H₂O exchange rates as described [22].

Antifungal activity assays

The growth inhibition assays were carried out in 96-well plates (Nunclon^®D, Thermo Scientific) as described [7,23]. Briefly, 10³-10⁴ N. crassa conidia were incubated with increasing concentrations of PAF and PAF^D19S in liquid medium in a total volume of 200 μL per well. Where appropriate, 0-10 mM CaCl₂, MgCl₂ or NaCl were added. The fungal growth was monitored microscopically and by measuring the optical density (OD_620nm) after 24-48 h of incubation (25°C) with a GENios Plus Microplate Reader equipped with Magellan software (Tecan). The minimal effective concentration (MEC) was defined as concentration that reduced growth by ≥ 90%. The germination efficiency and germ tube length of N. crassa was determined by incubating 5 x 10⁴ conidia mL^-1 in liquid medium with 0-32 μM antifungal proteins or 50% ethanol (control) at 25°C for 6 h under continuous stirring. Approx. 100 conidia were analysed for the presence of germ tubes and the tube length was measured by using AxioVision software (Zeiss). Fungistatic and fungicidal effects of PAF and PAF^D19S were determined by the method of Muñoz et al. [24]. N. crassa conidia (10⁴ mL^-1) were incubated in distilled water with 0-32 μM PAF, PAF^D19S or 50% ethanol (control) at 25°C for up to 24 h under continuous stirring. Samples were taken at different time points, diluted and plated on Vogel's agar containing 0.002‰ (w/v) dichloran to slow down colony expansion [25]. The plates were prepared in duplicates and incubated at 37°C for 24-48 h to determine colony numbers. To visualize cell death by pore formation in the cell membrane propidium iodide (PI) staining was performed on 2.5 x 10⁵ conidia mL^-1 in liquid medium or on conidia germinated on cover slides for 6-30 h at 25°C. The conidia/germlings were exposed to 32 μM PAF and PAF^D19S for various times at 25°C and then stained with 0.5 μg mL^-1 PI for 5 min. All experiments were prepared at least in duplicates and repeated twice and statistical calculations were done with Microsoft Excel. Statistical significance was evaluated using Student’s two-tailed t-test.

Labelling of PAF and PAF^D19S with the fluorophore BODIPY

The proteins were labelled with the green fluorophore BODIPY FL EDA (Life Technologies) as described with minor changes [26]. In brief, protein samples (0.4 mM) were dissolved in 0.1 M MES buffer pH = 4.5. BODIPY was added to a final concentration of 10 mM and subsequently EDAC (Life Technologies) and Sulfo-NHS (Life Technologies) were added to a final concentration of 10 mM and 5 mM, respectively. The reaction mixture was stirred in darkness for 3 h at 25°C, followed by dialysis against ddH₂O. Protein concentration and degree of labelling were determined spectrophotometrically.

Microscopy

Microscopic images were generated with a CK40 microscope (Olympus) equipped with an Axiocam MRm camera (Zeiss). PI staining experiments, the uptake and localization studies of fluorescence labelled PAF proteins were analysed with a Zeiss Axioplan fluorescence microscope, equipped with an AxioCam MRc camera (Zeiss, excitation/emission filters 365/420 nm for blue fluorescence, 500/535 nm for green fluorescence, 546/590 or 565/620 nm for red fluorescence). For imaging, special care was taken to apply the same exposure times in each experiment to allow direct comparison of the fluorescence signal intensities between samples. Image editing was done with GIMP (GNU Image Manipulation Program, version 2.8.10).

Aequorin-based measurement of intracellular Ca²⁺ fluxes in response to PAF and PAF^D19S

The measurement assays of [Ca²⁺]_c concentration were performed as described previously [23]. In brief, conidia of the N. crassa strain expressing AeqS [27] were suspended in liquid 0.2 x Vogel's media containing 2.5 μM of the co-enzyme coelenterazine (Biosynth AG) and incubated at 25°C for 6 h in the dark. After adding the antifungal proteins, a multimode plate reader (TriStar LB 941, Berthold Technologies) was used to measure bioluminescence at 25°C, over a time course of 60 min, taking measurements of the relative light units (RLUs) per well for 95 cycles (a cycle being the time it takes to measure all wells in the experiment). To convert relative light units (RLUs) into dynamic measurements of micromolar [Ca²⁺]_c in populations of conidial germlings during an experiment, an empirically derived equation was used as previously described [28,29].

The production of PAF^D19S

PAF^D19S was generated applying PCR-based mutagenesis of the paf gene. We took advantage of the P. chrysogenum paf gene deletion mutant (strain Δpaf) [14] to insert a plasmid carrying the mutated paf gene. After three rounds of single spore isolation, positive transformed P. chrysogenum PAF^D19S clones were tested for best protein production over a time course of 96 hours before one positive clone was selected for the highest secretion of PAF^D19S into the supernatant (data not shown). PAF^D19S was purified in a single-step chromatography and eluted from the cation-exchange column with 0.3 M NaCl.

The purity of the protein preparation and the identity of PAF^D19S were verified by ESI-MS (S1 Fig). A single peak corresponding to the average mass of 6.215 kDa for PAF^D19S was detected. These data correlated with the calculated theoretical mass (web.expasy.org/protparam) of the oxidized protein form indicating the presence of three intra-molecular disulphide bonds and the absence of post-translational modifications, except for cleavage of the pre-pro sequence, as reported also for wild-type PAF (Fig 1A and 1B) [7]. The overall net charge of PAF at pH 7 changed from 4.7 to 5.7 in PAF^D19S (web.expasy.org/protparam).

PAF and PAF^D19S exhibit similar solution structures, but differ in their surface charge distribution, dynamics and thermal unfolding properties

To determine the effect of the amino acid exchange on the protein solution structure the protein variant PAF^D19S was ¹⁵N-labelled and analysed by NMR spectroscopy. The chemical shift dispersions of the protein variant compared to the chemical shifts of PAF pointed towards a very similar folded overall structure of both proteins. This was further supported by comparable ¹³C chemical shifts (from natural abundance ¹³C isotopes), which are strong indicators for backbone conformation (S3 Table).

Using conventional NOE distance constraints, the 3D solution structure of PAF^D19S was determined (PDB ID: 2nb0, BMRB ID: 25957). With the exception of the exchanged Ser19 residue and its neighbour Ala20, the maximum ¹³C chemical shift difference was lower than 0.2 ppm compared to the PAF resonances (S3 Table). This indicated highly similar structures with only a minor change in backbone conformation around the mutated residue. At Ser19, PAF^D19S lost its negative charge and this electrostatic change was further relayed to distant regions of the protein, most remarkable reflected in alterations in the local conformation of the loop between residues 32–41 (Fig 1C and S2 Fig), demonstrating a serious consequence of the mutation.

Analysis of the lysine residues in terms of their backbone and side-chain conformation as well as their closest negatively charged spatial neighbour reveals a remarkable redistribution of the positively charged side-chains at the surface relative to wild-type PAF, also supported by differences in the NOE pattern between the two molecules (S3 Fig). Importantly, the loss of a large negatively charged surface patch is the consequence of the C-terminal Asp55 getting buried by lysine side chains in the D19S mutant.

CLEANEX NMR exchange experiments [22] showed that many of the residues in loop regions gave k_ex,_NH-H2O values in the range of 0.1–50 s^-1. Ser19, as a part of short loop 2, remained solvent accessible (k_ex,NH-H2O = 19 ± 3 s^-1, pH = 6) like Asp19 in PAF (k_ex,NH-H2O = 28 ± 3 s^-1, pH = 6) (Fig 2 and S4 Fig). However, the exchange rates and therefore the solvent accessibility generally decreased at the solvent exposed loop regions of PAF^D19S and at Asn side-chain carboxamides (Fig 2 and S4 Fig). The decrease of the exchange rates in PAF^D19S may be a consequence of the increase of the protein net charge (6.06 vs. 5.07) that could potentially strengthen the first hydration shell, diminishing the access to bulk water. We want to note at this point that the very slow exchange rates from residues located in β-strands are not available by this technique.

Download:

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

Fig 2. Comparison of NH-water exchange rates measured in CLEANEX NMR experiments.

The NH-water exchange rates (s^-1) are shown for PAF (green) and PAF^D19S (blue) at 298 K and pH = 6.0. Note that the very slow exchange rates from residues located in β-strands are not available by this technique, while measurement of fast exchange is inaccurate as shown by the error bars. Some Asn side-chain carboxamides are separately shown (to the right from the dashed line from residue 55: N18, N33, N40, N49, N50).

https://doi.org/10.1371/journal.pone.0169920.g002

According to ¹⁵N-relaxation data at 298 K and pH = 6, the dynamics was slightly enhanced around residue Ser19. The S² order parameter showed a decrease of 0.1 units at Ser19, in contrast to the smooth S² function around Asp19 in PAF (S5 Fig) [7]. Similar to PAF [8], weak but unambiguous ¹⁵N CEST effects were observed around the termini of PAF^D19S, at residues Tyr3, Thr47, Ala51 and Asp53, displaying the presence of distinct, low populated conformers. These minor conformers (< 1%) exchange slowly (~150/s) with the major conformer and exhibit extreme ¹⁵N^H chemical shift for example in case of Asp53, which was shifted by approx. -11 ppm with respect to the major conformer (S6 Fig). Three more residues in PAF^D19S have nearly identical CEST shifts as in PAF (Tyr3: -5 ppm, Thr47: -6 ppm, Ala51: +4.5 ppm), suggesting that the presence of the same sporadic conformer is not influenced by the D19S mutation. Thermal unfolding experiments of PAF^D19S showed surprisingly different parameters compared to PAF. In PAF most of the residues could not be fitted using a two-state model (below arbitrary 6% fitting error) with the exception of some non-conserved residues in loop regions, namely Lys2, Ser10, Lys11, Asp19, Asp32, Asn33 and Tyr48 as described before [8]. In the pertinent fitting protocol used for all experiments the fit errors for residues K30 and F31 in PAF were slightly above the arbitrary 6% limit (6.8% and 8.8%, respectively) and as a result, even more residues turned from three-state to two-state folder if based on these formal criteria. In contrast, the fit errors were significantly lower in PAF^D19S and the range of "two-state folder" residues increased considerably when applying the same criteria (S7 Fig). All the "two-state folder residues" in PAF remained two-state folders in PAF^D19S. However, in addition to these regions residues 4, 6, 12, 20, 21, 23, 25–28, 30–33, 47, 48, 51–53 and 55 could be fit by the two-state thermal unfolding model in this PAF mutant (Fig 3). This observation is consistent with a putative scenario in which hidden intermediate states are important contributors to PAF function [8].

Download:

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

Fig 3. Modelling of thermal unfolding of PAF and PAF^D19S.

(A-C) Fitting of a representative residue Ala20 by the two-state (A for PAF, C for PAF^D19S), and the three-state unfolding model (B) for PAF. Both folded (red ○) and unfolded (blue □) fractions are shown on A and C. While Ala20 could not be fit below 6% fit error (A, 10.01%) in case of PAF, fitting by the three-state model was successfully applied (B, fit error = 1.96%, folded: green ●, unfolded 1: orange ●, unfolded 2: blue □, sum of unfolded 1 and 2: red ○). In case of PAF^D19S Ala20 is rather a two-state folder (C, fit error = 3.37%). The estimated relative error of the experimental 2D peak volumes in the ¹⁵N-¹H HSQC spectra must be within 5%, according to repeated experiments. (D) Ribbon representation of PAF (left) and PAF^D19S (right), displaying the error limits of the thermal unfolding experiment fits. Red: residues that could be fit by the two-state model below 6% error limit. Green: residues that could not be fit below the error limit by the two-state model. Grey: not detected residues (signals could not be integrated because of overlaps). "Sticks" represent the site of the mutation (Asp/Ser19).

https://doi.org/10.1371/journal.pone.0169920.g003

The exchange of aspartic acid 19 to serine severely reduces the antifungal activity of PAF

We next analysed the impact of the amino acid exchange on the toxicity of PAF. The PAF-sensitive test organism N. crassa was exposed to 2-fold increasing protein concentrations (0-128 μM) and the growth rates of the treated samples were compared with the untreated control. Conidia of N. crassa have been previously shown to germinate in the presence of inhibitory concentrations of PAF, but their subsequent growth is inhibited [23]. The growth of N. crassa germlings was reduced by 90% after 30 h of incubation in the presence of 0.06 μM PAF and 32 μM PAF^D19S. The MEC of the protein variant was significantly higher than the MEC of PAF indicating that PAF^D19S severely lost antifungal activity.

Compared to untreated controls no difference in germ tube lengths could be detected in 6 h-old N. crassa germlings exposed to the MEC of PAF (0.13 μM: 40.5 ± 10.5 μm versus 47.0 ± 3.7 μm in the untreated control). However, when applying 32 μM PAF (250-fold higher than its MEC) the germ tube length reached only 50% of the control (22.6 ± 3.6 μm; p = 0.04). In contrast, PAF^D19S reduced the germ tube length by 50% of the control at a concentration corresponding to its MEC (32 μM: 25.3 ± 4.1 μm; p = 0.06). No germination was observed in ethanol-treated control conidia.

Our results showed that none of the tested proteins inhibited the spore germination in N. crassa but exclusively reduced germ tube length in a concentration dependent manner. The differences in antifungal activities suggested an altered mechanism of action of PAF^D19S compared to that of PAF.

The antifungal activity is specifically Ca²⁺ ion sensitive

In previous studies we reported that the extracellular ion concentration influences the antifungal activity of PAF [12,23]. To compare the ion sensitivity of PAF and PAF^D19S, we supplemented the growth medium with increasing concentrations of monovalent (NaCl) and divalent ions (CaCl₂, MgCl₂). The test medium (0.2 x Vogel's) per se contained 0.03 mM CaCl₂, 0.16 mM MgCl₂ and 2.16 mM Na₃-citrate. No considerable effect on the activity of PAF and PAF^D19S could be observed with the addition of 0–10 mM NaCl (Table 1). Instead, the divalent cations CaCl₂ and MgCl₂ reduced the toxicity of PAF and PAF^D19S in a concentration dependent manner, whereby the PAF-neutralizing effect of Ca²⁺ ions was more prominent than that of Mg²⁺ ions (Table 1). The little growth of N. crassa at 0.06 μM PAF (1.7 ± 0.5% growth) could be increased to 92.6 ± 3.9% by adding 10 mM CaCl₂ to the culture medium. The addition of 10 mM MgCl₂ was less effective than CaCl₂ and ameliorated growth under PAF challenge only to 54.0 ± 2.0%. The addition of 0.3 mM divalent ions had no growth promoting effect. In contrast, PAF^D19S was significantly more ion sensitive than PAF. The addition of only 0.3 mM CaCl₂ to the culture medium ameliorated the low fungal growth in the presence of 32 μM PAF^D19S (2.6 ± 0.6% growth) to 52.1 ± 4.6% growth and 10 mM of this ion further improved growth to reach 94.6 ± 12.1%. The growth of PAF^D19S-exposed N. crassa cells could be significantly ameliorated with the addition of 10 mM MgCl₂ (79.5 ± 2.1%).

Download:

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

Table 1. The effect of ion supplementation of the culture medium on the growth-inhibitory activity of PAF and PAF^D19S in N. crassa.

https://doi.org/10.1371/journal.pone.0169920.t001

Our results demonstrated that the antifungal activity of PAF was most sensitive to Ca²⁺ ions and this sensitivity was most pronounced in PAF^D19S. This pointed towards an important role of Asp19 in the Ca²⁺-dependent antifungal mode of action of PAF.

PAF^D19S-uptake in N. crassa is impaired

We could show previously that the toxic activity of PAF is closely linked with its binding to and active internalization by PAF-sensitive fungi [30]. We therefore investigated and compared the uptake and localization of PAF^D19S with that of PAF. To this end we labelled both proteins with the green fluorescent dye BODIPY and verified that they had comparable labelling efficiencies and a similar antifungal activity to the unlabelled proteins (data not shown). To visualize the fluorescence signals protein conjugate concentrations of 0.8-32 μM were applied.

BODIPY-PAF accumulated in the outer regions (cell envelope) of N. crassa conidia within the first minutes after addition (S8 Fig), and with longer incubation time localized in vacuoles and the cytoplasm (Fig 4 and S9 Fig). Uptake of BODIPY-PAF resulted in the formation of large vacuoles in conidia as well as in hyphae (Fig 4). BODIPY-PAF did not interact with the PAF-insensitive fungus A. terreus [12] even after 1 h of exposure (S9 Fig), which proved staining specificity and correlated with previous findings [30].

Download:

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

Fig 4. The localization of BODIPY-labelled PAF and PAF^D19S in 6 h-old N. crassa germlings.

After the treatment for 1 h with 32 μM BODIPY-PAF the conidial cell wall, vacuoles and the cytoplasm were stained. PAF treated hyphae showed large vacuoles (asterisks, left panels). Only weak signals at the outer layers and in the cells could be observed with BODIPY-PAF^D19S. Scale bar = 20 μm.

https://doi.org/10.1371/journal.pone.0169920.g004

We next investigated the localization of BODIPY-PAF^D19S. The germination of N. crassa conidia in the presence of 4 μM BODIPY-PAF^D19S resulted in very weak fluorescent staining, and this was primarily associated with conidia (Fig 5). When BODIPY-PAF^D19S was applied at its MEC (32 μM) on germlings, the fluorescent staining slightly increased and was also more visible in hyphae (Fig 4).

Download:

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

Fig 5. The effect of the addition of ions to the growth medium on binding and uptake of BODIPY-labelled PAF and PAF^D19S.

Conidia of N. crassa were germinated in the presence of 4 μM BODIPY-PAF or BODIPY-PAF^D19S for 6 h in liquid medium supplemented with 1 mM NaCl, MgCl₂ and CaCl₂, respectively. In the control the 0.2 x Vogel’s growth medium was without ion supplementation. Scale bar = 30 μm.

https://doi.org/10.1371/journal.pone.0169920.g005

The impaired antifungal activity of PAF^D19S was reflected by reduced protein binding and internalization, two events that are indispensable for the full antifungal activity of PAF [30]. Therefore, it is plausible that this weak protein–target interaction may be readily disturbed by the addition of low Ca²⁺ concentrations as observed in growth inhibition assays. To address this latter hypothesis, we performed staining experiments in the presence of increasing ion concentrations in the culture broth. The very low fluorescence signals of BODIPY-PAF^D19S in N. crassa conidia and germlings were slightly diminished with the addition of 1 mM MgCl₂ and readily disappeared with 1 mM CaCl₂ supplementation in the medium (Fig 5). At the same Ca²⁺ ion supplementation of the medium (1 mM) the intracellular signals of BODIPY-PAF completely disappeared in the germ tubes but not in the conidia (Fig 5). No effect in the staining intensity and pattern of BODIPY-PAF and BODIPY-PAF^D19S had the supplementation with NaCl (Fig 5).

The greater fluorescent labelling of the conidial cell envelope (comprised of the cell wall and plasma membrane) with PAF and PAF^D19S suggested that this cell structure exhibits a greater binding affinity to both proteins. Furthermore, our results indicate that Ca²⁺ ions (directly or indirectly) most effectively disturb the interaction of PAF and PAF^D19S with the target fungal cells and prevent protein uptake.

PAF^D19S fails to disrupt intracellular Ca²⁺ homeostasis

Another indication for the fungal cell killing activity of PAF is the induction of the perturbation of the intracellular [Ca²⁺]_c homeostasis [23,31]. To further characterize the loss of the antifungal activity of PAF^D19S, we analysed the influence of this PAF variant on [Ca²⁺]_c homeostasis. To measure [Ca²⁺]_c changes, we used the transgenic N. crassa-AEQ strain that expresses the gene aeqS, coding for the Ca²⁺-sensitive photoprotein aequorin [27,32]. This strain has been previously shown to exhibit the same susceptibility to PAF as the untransformed N. crassa wild-type strain [23]. As expected, PAF elicited a fast and sustained elevation of [Ca²⁺]_c, which corresponded well with our previous studies [23,31]. After 60 min of measurement the [Ca²⁺]_c maintained a steady level of 0.35-0.45 μM for 3.2 μM and 32 μM PAF (Fig 6A). In contrast, the [Ca²⁺]_c in the PAF^D19S-treated samples remained virtually unchanged at the same level as that of the media-treated control (0.15-0.2 μM Ca²⁺) (Fig 6C). The addition of the extracellular Ca²⁺-chelator BAPTA (5 mM) to the test system prevented the PAF-specific elevation of [Ca²⁺]_c, indicating that an influx of extracellular Ca²⁺ ions was probably responsible for the detected [Ca²⁺]_c increase (Fig 6B). Predictably, BAPTA had no effect on the Ca²⁺-response to PAF^D19S (Fig 6D).

Download:

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

Fig 6. The effect of PAF and PAF^D19S on the [Ca²⁺]_c resting level of 6 h-old N. crassa-AEQ germlings.

Proteins were added (arrow at time point 0) at a final concentration of 3.2 and 32 μM to the growth medium (A, C) and in medium pre-treated with 5 mM BAPTA (B, D). Top panels for PAF and bottom panels for PAF^D19S. Measurements were taken every 40 sec over a period of 60 min. Samples treated with medium without the antifungal protein served as controls. Values represent the mean of three samples ± standard errors.

https://doi.org/10.1371/journal.pone.0169920.g006

PAF and PAF^D19S act in a fungicidal, concentration and time dependent manner

To further dissect the functional difference between PAF and PAF^D19S, both proteins were tested for fungistatic and fungicidal activities. To this end, N. crassa conidia were incubated with the antifungal proteins for 1, 4 and 24 h and appropriate dilutions were plated onto solid Vogel's agar to allow vital conidia to germinate and establish colonies. The survival rates were significantly reduced when exposing the conidia to high doses of PAF (32 μM) for 1 h (52 ± 2.6%, p = 0.00005) and 4 h (44 ± 12%, p = 0.03). This was further diminished after 24 h of incubation (3 ± 0.5%, p = 0.03) (Fig 7A). Instead, no significant reduction in the spore survival rate was detected with 0.13 μM PAF compared to the control (data not shown). This indicated that PAF acted in a fungistatic way at the concentration that corresponded to its MEC, but fungicidally at a 250-fold higher concentration in a time dependent way.

Download:

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

Fig 7. Viability of N. crassa conidia exposed to 32 μM PAF and PAF^D19S.

(A) Colony forming units were determined by counting the colonies emerging on agar plates after conidia had been treated for 1, 4 and 24 h, respectively, plated in appropriate dilutions and incubated for 24 h (untreated controls) or 48 h. Values are given in %-survival (untreated controls were set to be 100%) and represent the mean of three experiments ± standard errors. (B) Untreated conidia and conidia exposed to ethanol served as negative and positive staining controls, respectively. Conidia exposed to BODIPY-PAF (C) and BODIPY-PAF^D19S (D) for 1, 4 and 24 h were co-stained with PI. Scale bar = 30 μm.

https://doi.org/10.1371/journal.pone.0169920.g007

Instead, PAF^D19S acted fungicidal on conidia (survival rate: 56 ± 0.9%, p = 0.006) after 1 h of incubation at its MEC (32 μM). This effect was extenuated with prolonged incubation more rapidly as with PAF: the survival rate was reduced to 18 ± 5.3% (p = 0.004) after 4 h and to 5 ± 0.1% (p = 0.003) after 24 h of exposure (Fig 7A). No colony formation was detected with ethanol-treated controls.

Importantly, ungerminated N. crassa conidia exposed to the fungicidal concentration of 32 μM PAF or PAF^D19S showed no signs of cell death at any incubation time tested, as evaluated with PI staining (Fig 7B–7D). We therefore assume that PAF-induced cell death occurred exclusively in hyphae. To prove our hypothesis, N. crassa conidia were germinated in the presence of PAF and PAF^D19S for 6-30 h and then stained with PI. No increase in PI fluorescence was detected after 6 h of incubation with PAF (0.13 and 32 μM) compared to the untreated control. Instead, PI signals increased in hyphae after 24 h and reached a maximum after 30 h of exposure to 32 μM PAF (S10 Fig). Similar results were obtained with 32 μM PAF^D19S (data not shown).

The structure, dynamics and thermal unfolding of PAF^D19S and their impact on antifungal function

We designed PAF^D19S as a part of our extensive mutation studies by which we wish to disclose unknown structure-function relations. Out of these attempts [7], PAF^D19S proved to be an extraordinary example where the mutation did not have a dramatic influence on the overall molecular structure, but caused a dramatic loss of antifungal activity. In our detailed NMR analyses we could show that the exchange of the solvent exposed and negatively charged aspartic acid 19 against a neutral serine perturbed internal molecular motions and the surface electrostatic distribution not only at the neighbourhood of the mutation, but at distant regions as well: for example in loop 3 (from Lys27 to Lys42) that involves the positively charged and mutation sensitive residues Lys34, Lys35 and Lys38 [7,31]. We propose that this is the consequence of the redistribution of charged side chains in absence of the negative charge of Asp19 in a “domino-like” manner: lysines near the mutated site are orienting towards other negative charges and this effect propagates via repulsion between lysines further on the surface. We suggest that this kind of “electrostatic allostery” is responsible for the observed changes in the surface and binding properties of the protein, as the lysines affected, together with Lys9 in loop 1 are highly conserved and located in near proximity in the folded protein, thus contributing to a cationic domain that is essential for full protein function [7,31]. A significant change in the charge of the protein surface may have severe impact on the interaction of PAF with its target molecule(s) and may explain the loss of antifungal function.

Another result raised our interest: the apparent difference between the thermal unfolding of PAF and PAF^D19S. While ¹⁵N-relaxation data (which is sensitive to ps-ns molecular motions) revealed only slight local changes in the internal dynamics around the modified residue, the changes of the thermal unfolding parameters of the two protein forms extended into more regions. In our previous studies we demonstrated that PAF is a three-state folder and not a simple two-state folder: only a few residues could be fit with a two-state unfolding model, and this is attributed to the different timescale motions by which PAF interconverts between a few low-populated transitional states [8]. According to our present study PAF^D19S is rather a two-state folder and possesses less or less extended transition states than PAF. These NMR-invisible transition states may be essential for full antifungal action by enabling optimal target recognition, corroborating a putative role of intermediate states in the biological function of PAF [33]. Finally, our study provides for the first time a deeper understanding for the importance of an unconserved region, namely the Asn-Asp sequence in loop 2, for antifungal activity of PAF. To affirm our assumptions, we are currently generating PAF mutants with other amino acid substitutions in this loop to further dissect in detail its role in structure-function relation.

PAF and PAF^D19S function is particularly Ca²⁺ ion sensitive

Current models for the mechanistic function of numerous antibacterial and antifungal acting cationic AMPs suggest that positively charged motifs on the protein surface are electrostatically attracted by the negatively charged phospholipid heads in the plasma membrane of microorganisms before the proteins insert into the membrane to evoke membrane perturbation and pore formation to ultimately trigger cell death [34,35]. This electrostatic interaction can be cation sensitive [36,37].

Indeed, the interaction of PAF and PAF^D19S with N. crassa was disturbed by elevated amounts of divalent cations, whereby the activity of both protein variants was most sensitive to Ca²⁺ ions. The specificity of the interference with divalent cations is further corroborated by the observation that monovalent cations like Na⁺ had no/little effect on the antifungal activity of PAF and PAF^D19S. Whereas the role of Mg²⁺ ions in the activity of antifungal proteins awaits further investigations, the relevance of Ca²⁺ ions in the mechanistic function of PAF and other related antifungal proteins and peptides was studied in detail [23,28,31,38] and let assume that their role goes beyond the disruption of electrostatic interactions. We assume that Ca²⁺ ions may influence the PAF-response of the sensitive fungus in one or more of the following ways: (i) competition with PAF for binding to a Ca²⁺-sensitive interaction molecule, e.g. a receptor, pump, transporter or a binding protein, (ii) modulation of the distribution and/or activity of such binding molecules in the fungal cell membrane, or (iii) activation of resistance mechanisms, all of which ultimately impede PAF interaction.

The killing mechanism of PAF in N. crassa

Cationic surface motifs of antifungal proteins and peptides that have no pore forming activity and are taken up into fungal cells to reach their site of action may determine not only the electrostatic attraction and interaction with the fungal cell but also regulate the intracellular transport triggering cell death [29]. Indeed, the uptake and the induction of a specific Ca²⁺ signature in susceptible fungal cells are prerequisites for the killing mechanism of PAF [23,30]. PAF and PAF^D19S were first attracted by the conidial cell envelope before they were internalized. This interaction per se was not detrimental to N. crassa as conidia showed no PI staining and still germinated in the presence of toxic concentrations of PAF and PAF^D19S. Only following conidial germination and colony establishment the antifungal proteins fully acted on the actively growing hyphae in a concentration and time dependent manner. However, the primary killing mechanism of PAF and PAF^D19S was not the formation of pores in the plasma membrane because treated germlings were PI positive only after long incubation times. Such membrane damage could be a secondary effect after prolonged exposure to the antifungal proteins, for example as a result of increased oxidative stress induction in the cell [39]. A similar mode of action was proposed for antifungal plant defensins [40]. Instead, the increased vacuolation in PAF-treated hyphae suggested the onset of apoptotic events. Extensive hyphal vacuolation and other specific markers for apoptosis, e.g. elevated levels of reactive oxygen species, were previously identified in PAF-treated Aspergillus nidulans hyphae, which further proceeded to massive membrane damage [39]. We therefore favour the view that PAF induces the activation of specific signal transduction pathways that regulate growth inhibition and apoptosis, as proposed earlier [23,31,39,41].

Indeed, we could show that PAF^D19S failed to trigger a specific Ca²⁺ response even when applied at its MEC (32 μM) in contrast to PAF. In one of our previous studies we demonstrated that the PAF-specific rapid and sustained increase in [Ca²⁺]_c was inhibited when the test medium was supplemented with high Ca²⁺ concentrations [23,31]. This observation can be now directly linked to a significantly reduced interaction of PAF with fungal germlings in the presence of elevated amounts of extracellular Ca²⁺. Similarly, it has to be considered that PAF^D19S might not have reached a concentration on/in the fungal cells fast and/or high enough to induce a Ca²⁺ response even when tested in 0.2 x Vogel's medium with a low Ca²⁺ concentration (0.03 mM CaCl₂). This assumption is supported by the observation of considerably reduced PAF^D19S binding and internalization in conidia and fungal cells compared to PAF. However, PAF^D19S acted fungicidal in a time dependent manner when applied at its MEC (32 μM) suggesting that fungal killing is mediated by other motifs that regulate mechanisms independently from the elevation of [Ca²⁺]_c. Our observation parallels with recent studies on plant defensins: by using defensin fragments amino acid sequences were identified that exhibit distinct antifungal features of their parental proteins [28].

In summary, our study shows that the exchange of one critical negatively charged amino acid that resides in a non-conserved and dynamically flexible loop has severe impact on the protein surface charge and unfolding and affects the dynamics of distant protein motifs that are sensitive to surface modifications and essential for full PAF activity. This has to be considered in approaches that apply protein engineering for the development of natural bio-molecules with improved antifungal properties.