Two small, cysteine-rich and cationic antifungal proteins from Penicillium chrysogenum: A comparative study of PAF and PAFB

Abstract

The filamentous fungus Penicillium chrysogenum Q176 secretes the antimicrobial proteins (AMPs) PAF and PAFB, which share a compact disulfide-bond mediated, β-fold structure rendering them highly stable. These two AMPs effectively inhibit the growth of human pathogenic fungi in micromolar concentrations and exhibit antiviral potential without causing cytotoxic effects on mammalian cells in vitro and in vivo. The antifungal mechanism of action of both AMPs is closely linked to - but not solely dependent on - the lipid composition of the fungal cell membrane and requires a strictly regulated protein uptake into the cell, indicating that PAF and PAFB are not canonical membrane active proteins. Variations in their antifungal spectrum and their killing dynamics point towards a divergent mode of action related to their physicochemical properties and surface charge distribution. In this review, we relate characteristic features of PAF and PAFB to the current knowledge about other AMPs of different sources. In addition, we present original data that have never been published before to substantiate our assumptions and provide evidences that help to explain and understand better the mechanistic function of PAF and PAFB. Finally, we underline the promising potential of PAF and PAFB as future antifungal therapeutics.

Keywords: Antimicrobial proteins and peptides; Apoptosis; Endocytosis; Fungal membrane lipids; Penicillium chrysogenum; β-Fold structure; γ-Core.

Graphical abstract

Fig. 1

(A) Clustal W multiple alignment of the putative PAF-like antifungal AMPs from Penicillium spp. from the UniProt database [21]. Alignment was generated by BioEdit [22] and visualized by Jalview 2.11.0. [23]. The cleavage of the preprosequence and the first aa of the mature protein is marked by brown dotted line and asterisks. After the species name the accession number of the respective AMP (see Table 1) is indicated. The Clustal X default color scheme was applied (http://www.jalview.org/help/html/colourSchemes/clustal.html). (B) Maximum-likelihood (ML) tree of putative PAF-like AMPs from Penicillium spp. from UniProt database [21]. The alignment of the full-length proteins was generated with PRANK v. 140,110 [24] with default settings for phylogenetic analysis. ML analysis was performed with RAxML v. 8.2.10 [25] under the GAMMA distributed rate heterogeneity empirical frequencies model with DCmut substitution matrix in 1000 through bootstrap replicates. ML bootstrap values >60% are shown next to branches. After the species name the accession number of the respective AMP (see Table 1) is indicated. “#” marks PAF and PAFB from P. rubens Wisconsin 54-1255, which is the descendent strain of P. chrysogenum Q176 [19].

Fig. 2

Growth inhibition of N. crassa by increasing concentrations of PAF and PAFB. The antifungal activity was tested in a microdilution broth assay by inoculating conidia (10⁴/mL) with increasing concentrations of PAF or PAFB in undiluted Vogel's medium and incubation for 30 h at 25 °C without shaking. Growth was determined spectrophotometrically (FLUOstar Omega, BMG Labtech) measuring the optical density at λ = 620 nm. Values represent the mean ± SD (n = 3, technical triplicates) growth (%) in the presence of AMPs in comparison to the untreated control which was set to be 100%. The result of one representative experiment of two biological repeats is shown.

Fig. 3

Killing of C. albicans exposed to PAF and PAFB. Planktonic cells (10⁴/mL) were mixed with 0.25 ×, 1 × and 4 × the IC₉₀ of PAF or PAFB in 0.1 × PDB and incubated for 6 h at 30 °C. Samples with appropriate dilutions were plated on 0.1 × PDB agar and the colony number was determined after 24 h of incubation at 30 °C. Values represent the mean ± SD (n = 3, technical triplicates) colony number (%) in the presence of AMPs in comparison to the number of colonies from the untreated control at time point 0 h, which was set to be 100%. The result of one representative experiment of two biological repeats is shown.

Fig. 4

Structure of PAF (PDB ID: 2mhv, red) and sfPAFB (PDB ID: 2nc2, blue) and structural alignment of both proteins. sfPAFB is a truncated variant of PAFB, shorter with two aa residues at the N-terminus. Antiparallel β-strands represented by arrows form two overlapping β-sheets. Consecutive strands are connected by short turns or longer loop regions. The positions of cysteines and cationic residues are indicated in one letter code. Disulfide bonds are marked in yellow. PyMol 1.4.1 (Schrödinger, Inc.) software was used for structure alignment.

Fig. 5

Chemical synthesis of PAFB^opt by regio-selective disulfide bond formation. The synthesized protein was detached from the solid support and protecting groups of the aa except Acm and Mob were removed by a trifluoroacetic acid (TFA)/water/dithiothreitol (95:5:3) (v/v/w) cleavage cocktail. The first disulfide bond was formed with O₂ of air in a pH 7.5 buffer (1). Iodine treatment cleaved Acm and oxidized free thiols of the second pair of cysteines in one step (2). After cleavage of Mob by a trifluoromethanesulfonic acid (TFMSA)/TFA/anisole mixture (3), iodine was used to form the third disulfide bond (4).

Fig. 6

Stability testing of PAFB wt and PAFB^opt in cell-free supernatants of A. fumigatus and C. albicans. Spores or yeast cells (10⁴/mL) were grown for 48 h in 0.05 × PDB medium as stationary cultures at 37 °C and 30 °C, respectively. The hyphae and yeast cells were removed by centrifugation and 100 μL of the cell free supernatant were transferred to an Eppendorf tube together with 100 μg/mL of PAFB wt and PAFB^opt. The samples were incubated for 1 h at 30 °C (C. albicans supernatant) and 37 °C (A. fumigatus supernatant). The proteins incubated in water served as controls (control). Samples were loaded in 10 μL aliquots (corresponding to 1 μg protein per sample) on a 18% (w/v) SDS polyacrylamide gel and size fractionated. Silver staining was performed to visualize the proteins.

Fig. 7

Cellular localization of PAF and PAFB and cell death induction in C. albicans in different media. Yeast cells were shaken at 200 rpm for 3 h at 30 °C in 0.1 × PDB or 0.1 × YPD in the presence of 8 μM BODIPY-labelled PAF and PAFB. Co-staining with 5 μg/mL propidium iodide was performed for 10 min before imaging. Images were taken with the same exposition time (1.500 ms). BF, Brightfield; BP, BODIPY-labelled proteins; PI, Propidium iodide.

Fig. 8

Susceptibility of N. crassa towards PAF and PAFB in the presence of heparin. (A) Uptake of PAF and PAFB: N. crassa conidia (5 × 10⁵/mL) were grown for 4 h in 0.2 × Vogel's medium (25 °C, shaking at 200 rpm). 8 μM BODIPY-labelled PAF (BP-PAF) and PAFB (BP-PAFB) were added alone or in combination with 50 μg/mL heparin to the germlings and samples were further incubated for 2.5 h at 25 °C until microscopic evaluation. (B) Control experiments: N. crassa conidia (5 × 10⁵/mL) were grown for 4 h in 0.2 × Vogel's medium in the presence of 50 μg/mL heparin or without heparin. Then, all samples were washed three-times with fresh medium and 8 μM BP-PAFB was added to the germlings alone (heparin removed and - heparin) or in combination with 50 μg/mL heparin (+ heparin). The samples were further incubated for 2.5 h at 25 °C until microscopic evaluation. Images were taken with the same exposition time (1.500 ms). BF, Brightfield; BP, BODIPY-labelled proteins.

Fig. 9

Impact of PAF and PAFB on the thermotrophic behavior. DSC scan of DPPC in absence and presence of PAF and PAFB as observed by heating scans. The concentration of PAF and PAFB is indicated in relation to lipid to peptide (L:P) molar ratio of 25:1.

Fig. 10

Impact of PAF and PAFB treatment on the phospholipid composition of N. crassa. Conidia (10⁷) were inoculated in 20 mL Vogel's medium for 18 h (25 °C, shaking at 200 rpm). Hyphae were then treated with 4 × IC₉₀ of PAF (6 μM) and PAFB (3 μM) for additional 3 h until harvesting for total lipid extraction and phospholipidomics analysis. Sample preparation and LC-MS/MS measurement were performed as described in [56] in negative ESI mode. Different lipid features were quantified, normalized to total protein content and summed per class (Cer: ceramide; CL: cardiolipin; PA: phosphatidic acid; PC: phosphatidylcholine; PE: phosphatidylethanolamine; PI: phosphatidylinositol; PS: phosphatidylserine). The obtained values were normalized to the respective mean control values. CLs, PCs and PEs show a trend to be reduced by PAF and PAFB treatment, while the total ceramide content was significantly increased when treated with PAFB compared to PAF and the controls. Values are shown as mean ± SD (n = 3). Normality of fold changes was tested with Shapiro-Wilk test class wise (p > 0.05, Holm multiple testing adjustment). Homogeneity of variances of the changes by class and treatment was assessed by Levene's test (p = 0.638). Fold changes were found significantly associated with treatment and lipid class, F(12, 42) = 5.517, p < 0.001. Post-hoc Tukey HSD analysis was applied, and p-values were adjusted for multiple testing (**: p = 0.01; ***: p = 0.001).

Fig. 11

ROS induction by PAF and PAFB in N. crassa. Conidia (5 × 10⁵/mL) were grown for 4.5 h in 0.2 × Vogel's medium at 25 °C as stationary cultures. Germlings were treated with 5 × IC₉₀ PAF or 5 × IC₉₀ PAFB for 1.5 h, respectively. ROS production was detected by incubating the samples with 10 μM of the fluorescent dye 2′,7′-dichlorofluorescin diacetate for 30 min. Images were taken with the same exposition time (1.500 ms). BF, Brightfield; DCFH-DA, 2′,7′-dichlorofluorescin diacetate.