New Antimicrobial Potential and Structural Properties of PAFB: A Cationic, Cysteine-Rich Protein from Penicillium chrysogenum Q176

Small, cysteine-rich and cationic proteins with antimicrobial activity are produced by diverse organisms of all kingdoms and represent promising molecules for drug development. The ancestor of all industrial penicillin producing strains, the ascomycete Penicillium chryosgenum Q176, secretes the extensively studied antifungal protein PAF. However, the genome of this strain harbours at least two more genes that code for other small, cysteine-rich and cationic proteins with potential antifungal activity. In this study, we characterized the pafB gene product that shows high similarity to PgAFP from P. chrysogenum R42C. Although abundant and timely regulated pafB gene transcripts were detected, we could not identify PAFB in the culture broth of P. chrysogenum Q176. Therefore, we applied a P. chrysogenum-based expression system to produce sufficient amounts of recombinant PAFB to address unanswered questions concerning the structure and antimicrobial function. Nuclear magnetic resonance (NMR)-based analyses revealed a compact β-folded structure, comprising five β-strands connected by four solvent exposed and flexible loops and an “abcabc” disulphide bond pattern. We identified PAFB as an inhibitor of growth of human pathogenic moulds and yeasts. Furthermore, we document for the first time an anti-viral activity for two members of the small, cysteine-rich and cationic protein group from ascomycetes.


Synthesis of cDNA and verification of the pafB coding sequence. The GoScript Reverse
Transcription System kit for cDNA synthesis from total RNA was used according to the manufacturer's instructions (Promega). The cDNA served to amplify the coding sequence (CDS) of pafB by PCR with the primer pair opafB_fw/opafB_rev (Supplementary Table S4). The PCR fragment was cloned into the plasmid pGEM-T (Promega) and the CDS was determined by Sanger sequencing at Eurofins genomics (Germany).
Construction of expression vector. The Q5 High Fidelity DNA Polymerase (New England Biolabs) was used for all PCR reactions. For PAFB overexpression the P. chrysogenum-based expression system was used 1 . To put the pafB gene under the control of the strong paf promoter, the paf gene in the expression vector pSK275paf was replaced by pafB genomic DNA that codes for the pre-pro PAFB. To this end, the pafB-gene (404 bp) was amplified from genomic DNA of P. chryosgenum Q176 using the primers pafB_BglII_fw/pafB_SmaI_rev (Supplementary Table S4). 1,292 bp of the paf 5´UTR and 386 bp of the paf 3´UTR were amplified using the primer pairs 5UTR_PstI_fw/5UTR_SmaI_BglII_rev and 3UTR_SmaI_fw/ 3UTR_SpeI_rev, respectively (Supplementary Table S4). After digestion of the PCR products with the respective restriction enzymes (SmaI, BglII, PstI, SpeI, New England Biolabs) the resulted fragments were ligated using the T4-DNA-Ligase (Promega). The ligation product (2,082 bp) was subsequently cloned into the SpeI and PstI digested pSK275paf vector exchanging the paf-coding sequence and resulting in the expression vector pSK275pafB.
Southern blotting. For Southern blotting, genomic DNA extraction was carried out according to Zadra et al. (2000) 2 . SacII digested DNA (2 µg per lane) was fractionated on a 0.8% (w/v) agarose gel. The DNA was transferred on Hybond-N membranes and hybridized with DIG-labelled probes, specific for the pyrithiamine (ptrA) resistance gene present on the plasmid pSK275pafB. The probes were generated from pSK275pafB by PCR using the oligonucleotides ptrA_southern_fw/ptrA_southern_rev (Supplementary Table S4). The Southern blot experiments proved multiple random integration of the transforming DNA into the fungal genome ( Supplementary Fig. S2). A 1.3 kb DIG-labelled PCR probe spanning a part of the ptrA resistance cassette was used for plasmid detection. The presence of the transforming plasmid carrying the recombinant PAFB encoding gene in the genome of P. chrysogenum pafB strain was proven by the detection of a 3.7 kb hybridizing fragment, which corresponded to a SacII digested part of the transformed plasmid. The additional signals varied in sizes and intensities, which further proved multiple-copy random plasmid integration ( Supplementary Fig. S2).
In silico modeling and analysis of full-length PAFB. The full-length PAFB structure was modeled by manually building and adding a leucine-serine segment to the N-terminus of the first model in the 2NC2 structure using lsqman 3 for initial superposition and standard text editing for generating the Protein Data Bank submission of the full-length model. All subsequent calculations were performed with an in-house modified version of GROMACS 4 4 . Modifications including the handling of NOE distance restraints in a pairwise manner over multiple replicas 5 were described earlier 6 and the source files are freely available at http://users.itk.ppke.hu/~gaszo. The AMBER99SB-ILDN force field 7 was used with the GBSA implicit water model using the Onufriev-Bashford-Case method 8 . The initial model was energy minimized to the 1,000 kJ/mol/nm forrce limit using steepest descents. Short exploratory molecular dynamics calculations were run with and without NOE restraints. Restrained calculations were run using the distance restraints deposited in the Protein Data Bank for sfPAFB (Protein Data Bank ID: 2NC2) on two replicas for 100 ps with a step of 1fs and NOE force constant of 1,000 kJ/mol/nm 2 . In this simulation, only the final two conformations were analyzed. Unrestrained simulations were run for 3 ns and structures were sampled every 50 ps resulting in 61 models. Secondary structure content was calculated by running DSSPcont 9 on each model and then averaging the probabilities of the structural states for each residue. The state with the highest probability was chosen, in the case of equal probabilities (50-50%), preference was given to 'E' over 'L' and 'L' over 'S'.
Comparison of PAF and sfPAFB. The structures of PAF and sfPAFB were aligned with MAMMOTH-Mult 10 . Using the corresponding residue positions form the structure alignment, the NMR ensembles of the two molecules were superimposed with MOLMOL 11 that was also used for local and global RMSD calculations as well as visualization. Electrostatic potential was generated using the APBS method 12 and visualized with Schrödinger Maestro 13 .
Cytotoxicity testing. The haemolytic activity of PAFB and PAF was tested on Columbia Blood Agar plates (VWR). Ten µL of protein solution (2 µg/µL) were pipetted on sterile filter discs (6 mm) placed on the agar plates. As controls 10 µL ddH 2 O (negative) or 10 µL five-fold diluted Triton X-100 (positive) were used. The plates were incubated at 37 °C for 24 h. The cytotoxicity of PAFB and PAF was evaluated on the human epithelial cell line L132 (ATCC CCL-5) according to Mosmann (1983) 14 . The assays were performed in 96-well flat bottom tissue culture plates (Sarstedt). One hundred microliters of medium (DMEM supplemented with 2% FBS) containing proteins in increasing concentrations were added to the monolayers of L132 cells (10 4 /well) in triplicates. The plates were incubated at 37 °C in a 5% CO 2 atmosphere. After 72 h, the supernatant was removed and the cells were washed with phosphate buffered saline (PBS, Invitrogen, France). Ten microliters of MTT-solution (2 mg/mL), prepared in PBS, were added to each well and the plates were incubated for 4 h at 37 °C. Then, 100 µl of SDS (100 µg/mL) was added to the wells to solubilize the MTT crystals. The plates were incubated at 37 °C for 4 h, agitated until complete crystal dissolution and evaluated by measuring the OD 570 using a Multiskan EX 96-well plate ELISA reader (Thermo Electron Corporation). The 50% cytotoxic concentration (CC50) was then determined using trend function analysis in Microsoft Excel 2010 (Microsoft Corp.).

Activity of PAFB against P. chrysogenum.
To determine any toxic effects of PAFB on P. chrysogenum mycelia, conidia (10 4 /mL) were seeded in 100 µl aliquots of MM or 0.1 x PDB into 96well plates in triplicates and incubated at 25 °C for 24 h to reach an OD 620 of 0.1-0.2. Protein was added at increasing concentrations (0-8 µM) and the plates were further incubated for 24 h. The OD 620 was measured spectrophotometrically to determine further proliferation. Experiments were repeated at least twice. Tables   Table S1. Amino acid sequence and in silico predicted physical and chemical properties of the mature, full-length PAFB (which is identical to PgAFP 15 ) and its N-terminal short-forms PAFB-L and PAFB-LS.

Protein
Number    Polyclonal antibodies specific for PAFB and PAF, respectively, were used for protein detection. c, purified PAFB or PAF (1 µg, respectively) were loaded as controls.