Functional aspects of the solution structure and dynamics of PAF--a highly-stable antifungal protein from Penicillium chrysogenum

Abstract

Penicillium antifungal protein (PAF) is a promising antimycotic without toxic effects on mammalian cells and therefore may represent a drug candidate against the often lethal Aspergillus infections that occur in humans. The pathogenesis of PAF on sensitive fungi involves G-protein coupled signalling followed by apoptosis. In the present study, the solution structure of this small, cationic, antifungal protein from Penicillium chrysogenum is determined by NMR. We demonstrate that PAF belongs to the structural classification of proteins fold class of its closest homologue antifungal protein from Aspergillus giganteus. PAF comprises five beta-strands forming two orthogonally packed beta-sheets that share a common interface. The ambiguity in the assignment of two disulfide bonds out of three was investigated by NMR dynamics, together with restrained molecular dynamics calculations. The clue could not be resolved: the two ensembles with different disulfide patterns and the one with no S-S bond exhibit essentially the same fold. (15)N relaxation dispersion and interference experiments did not reveal disulfide bond rearrangements via slow exchange. The measured order parameters and the 3.0 ns correlation time are appropriate for a compact monomeric protein of this size. Using site-directed mutagenesis, we demonstrate that the highly-conserved and positively-charged lysine-rich surface region enhances the toxicity of PAF. However, the binding capability of the oligosaccharide/oligonucleotide binding fold is reduced in PAF compared to antifungal protein as a result of less solvent-exposed aromatic regions, thus explaining the absence of chitobiose binding. The present study lends further support to the understanding of the documented substantial differences between the mode of action of two highly homologous antifungal proteins.

Fig. 1

Microtiterplate activity assay for the determination of the growth of A. niger in the presence of PAF that had been exposed to various test conditions. 10⁴ conidia/ml were incubated with 1-5 μg/ml PAF for 24 h at 30°C. (A) PAF was reduced by DTT as described in Material and Methods. (B) PAF was exposed to 60°C, 80°C, 90°C and 100°C for 10, 30, 60 min, respectively. (C) PAF was digested with proteinase K for 3, 9 and 24 h or with pronase for 3, 9, and 12 h. Please note, that the asterisk in (C) indicates different exposure times of PAF with pronase (12 h instead of 24 h). Values represent the precent growth (%) of A. niger in the presence of PAF that had been exposed to various test conditions compared to A. niger left untreated (=100%).

Fig. 2

Sequence alignment of PAF and AFP with three highly conserved regions marked in yellow that are putatively assigned to chitin binding (3-9), DNA binding (12-17) domains and cation channel forming (34-39) capabilities. Arrows and S labels stand for β strands.

Fig. 3

Super secondary structure of PAF

Fig. 4

Deuteration rates measured for amide NH groups in PAF (note the logarithmic scale). Missing bars represent fast deuteration rates, i.e. those NH signals disappeared within 10 minutes. Slow deuteration of amides protons correlates with solvent protection in β-sheet regions.

Fig. 5

S² order parameters reflecting internal mobility of the NH residues obtained from the Lipari-Szabo analysis of ¹⁵N T₁, T₂ and NOE relaxation parameters. Slightly enhanced mobility is clearly detected at the N-terminus and in the loop regions as shown by the dips of the bar plot. For comparison, the S² values calculated from the assigned chemical shifts is shown at the bottom using the RCI index and the program of http://wishart.biology.ualberta.ca/rci/cgi-bin/rci_cgi_1_e.py. The average of S²_exp/S²_RCI = 0.96 ± 0.07. Residue 29 is proline and consequently not shown in experimental data.

Fig. 6

Different CSA/DD relaxation interference rates displayed as a pile-up bar graph. Instead of extracting site specific ¹⁵N chemical-shift anisotropies from all relaxation data, this kind of straight visualisation of ¹H- ¹⁵N CSA/ DD transversal cross-correlated relaxation rates (Hxy) is sensitive to secondary structure elements.

Figure 7

Final MUMO ensembles (80 conformers) of PAF calculated with different disulfide pairings labelled as abbacc, abcabc and no SS bond. The average NMR structure (red line) with no disulfide bonds is overlayed with the ensembles.

Fig. 8

Analysis of the antifungal activity of mutated PAF protein versions on A. niger. (A) Microscopic analysis of A. niger exposed to 100 μg/ml PAF for 24 h. The recombinant mPAF protein (panel C) exhibited comparable growth inhibition potency to the native PAF (panel B). In panel A hyphae of the untreated control are shown. Panels A-C are microscopic overviews (×20), panels a-c are details of A-C (×63). (B) The increase in proliferation of A. niger when exposed to mutated PAF protein versions was correlated to the proliferation in the presence of recombinant mPAF at the corresponding protein concentrations of 5 μg/ml (light grey) and 100 μg/ml (dark grey). The proliferation of the PAF-unexposed A. niger control cells was 2.4(±0.2)-fold and 7.4(±1.1)-fold compared to the growth of the samples treated with recombinant mPAF at the respective concentrations of 5 μg/ml and 100 μg/ml.

Fig. 9

Ribbon diagram of the mean PAF structure without disulfide bond constraints. The congested hydrophobic core with the cysteines labelled is shown in the excerpt.

Fig. 10

The electrostatic surface potenctial of AFP (left) and PAF (right) structures representing the orientation of the side chain Lysine resiues, Lys9, Lys35, Lys 38 of PAF and Lys9, Lys32 and Arg35 of AFP are the corresponding conservative mutated lysines.