Isolation and Identification of an Extracellular Subtilisin-Like Serine Protease Secreted by the Bat Pathogen Pseudogymnoascus destructans

White nose syndrome (WNS) is a cutaneous fungal disease of bats. WNS is responsible for unprecedented mortalities in North American cave bat populations. There have been few descriptions of enzyme activities that may function in WNS host/pathogen interactions, while no study has isolated and described secreted proteases. To address the hypothesis that Pseudogymnoascus destructans secretes extracellular proteases that function in wing necrosis during WNS infection, the object of this study was to culture P. destructans on various media, then isolate and structurally identify those proteases accumulated stably in the culture medium. We found a single dominant protease activity on minimal nutrient broth enriched with protein substrates, which was strongly inhibited by phenylmethylsulfonyl fluoride. This P. destructans serine protease (PdSP1) was isolated by preparative isoelectric focusing and concanavalin A lectin affinity chromatography. PdSP1 showed a molecular weight 27,900 (estimated by SDS-PAGE), broad pH optimum 6-8, and temperature optimum 60°C. Structural characterization of PdSP1 by MALDI-TOF MS, Orbitrap MS/MS, and Edman amino-terminal peptide sequencing matched it directly to a hypothetical protein accession from the sequenced P. destructans genome that is further identified as a MEROPS family S8A subtilisin-like serine peptidase. Two additional isoforms, PdSP2 and PdSP3, were identified in the P. destructans genome with 90% and 53% homology, respectively. P. destructans S8A serine proteases showed closer sequence conservation to P. pannorum and plant pathogenic fungi than to human pathogenic dermatophytes. Peptide-specific polyclonal antibodies developed from the PdSP1 sequence detected the protein in western blots. These subtilisin-like serine proteases are candidates for further functional studies in WNS host-pathogen interaction.


Materials and Methods Fungal Cultivation and Secreted Protein Production
To prepare inocula for protein-production cultures, P. destructans was grown on malt broth for two months. Mycelial plugs (4 of ca. 2 cm dia.) were inoculated in liquid culture media (100 ml in 2 L Fernbach flask for high surface area to volume ratio) and grown statically in the dark at 8°C. Total protease production was determined in complex culture media (Difco): tryptone peptone, brain/heart infusion, proteose peptone, tryptic soy broth (TSB), and compared with minimal nutrient broth (MNB) (NH 4  supplemented with 0.5% protein: gelatin (G), keratin (K), casein (C), or elastin (E). Cultures were grown for 6-8 weeks then harvested when mycelia covered the entire liquid surface. Culture supernatants were recovered by centrifugation at 10,000 x g for 30 min, filtered through a 0.45 μm membrane, and then concentrated by lyophilization or ultrafiltration. Concentrated media were exchanged into deionized water (containing 0.02% w/v NaN 3 ) with Econo-Pac 10 DG desalting columns (Biorad # 732-2010).

Preparative Isoelectric Focusing
Preparative isoelectric focusing (IEF) was performed for six hours at 12 W with a BioRad Rotofor cell using 1% BioRad pH 3-10 ampholytes added to salt-free protein samples following manufacturer's instructions. The pH of each fraction was recorded and adjusted to pH 7.0 with HCl/NaOH. Fractions were brought to an equal volume with ultrapure H 2 O (18 mO), and protein and protease activity (measured at pH 7.0) was determined for each.

Lectin Affinity Chromatography
Glycosylated proteins were separated by lectin affinity chromatography with concanavalin A Sepharose (ConA; Sigma # C7911) following manufacturer's recommendation. Briefly, ConA (ca. 10 ml) packed in small columns (1.5 cm ID x 12 cm) was washed with five bed volumes of binding buffer (20 mM Tris, 0.5 M NaCl, pH 8), and then sample was applied twice. Following washing the ConA column with five bed volumes of binding buffer, the glycoproteins were eluted with five bed volumes of elution buffer (20 mM α-D-methylglucoside, 0.5 M NaCl, 20 mM Tris, pH 4). The eluted fraction was neutralized with Tris buffer (pH 9). ConA samples were exchanged into ultrapure H 2 O with Bio-Rad Econopak 10DG column prior to further analysis by protein and enzyme assay or gel electrophoresis. All samples were stored with 0.02% (w/v) NaN 3 at 4°C.

Protein Determination and Electrophoretic Analyses
The total protein concentration was determined using Bradford's Coomassie Brilliant Blue reagent and BSA as a standard [27]. Samples were prepared for gel electrophoresis using trichloroacetic acid (TCA) precipitation. TCA pellets were washed with ice-cold acetone, dried in vacuo and reconstituted in 50 μl SDS Laemmli sample buffer. Proteins in culture medium samples were separated using the BioRad Mini-PROTEAN TCX 12% pre-cast gels with the Laemmli Tris/glycine buffer system and staining with Coomassie Brilliant Blue reagent according to manufacturer's instructions [28]. Protein size was estimated with Precision Plus Protein All Blue standards (BioRad # 161-0373) using U-SCAN-IT Graph Digitizing Software (v 5.1, Silk Scientific Corp, USA). Percentage protein composition stained bands were also estimated with U-SCAN-IT software.

Enzyme Activity Assays
Protease activity in culture media was initially determined with fluorescein isothiocyanate (FITC) casein according to the manufacturer's instructions and compared to a trypsin standard (Thermo Fisher Scientific, kit # 23266). Values are expressed as units/ml, where one unit is defined as 1.0 mg / 1.0 ml trypsin standard. For determining optimal pH, protease activities were determined with azocoll (Sigma # A4341) added in 5.0 mg portions to 1.0 ml 100 mM potassium phosphate buffer, pH 7.0 at 37°C. Affinity-purified protein samples were added in 200 μl aliquots, the reaction was continued at 37°C for 30 min with gentle shaking, and absorbance was measured at 520 nm [29].
Serine protease activity of affinity-purified proteins was quantified using a peptide-pNA (Suc-Ala-Ala-Pro-Phe-NHPhNO 2 ) colorimetric substrate (Sigma # S7388; 0.1 mM). Proteinase K (E.C. 3.4.21.64; Sigma # P6556) was used as a positive control to determine relative serine protease activity. Substrate was prepared in 0.1 M Tris-HCl, 0.01 M CaCl 2 buffer at pH 7.5. The protein samples (100 μl) were mixed with peptide-pNA substrate (600 μl) and incubated for 10 min at room temperature, then product absorbance was determined at 410 nm and compared to a 4-nitroaniline standard curve. One unit serine protease activity was defined as 1 μmol 4-nitroaniline released per minute at 25°C.

Amino-Terminal Sequence Determination
Amino-terminal protein sequences for individual proteins blotted to PVDF membrane were determined by Edman sequencing chemistry with an AB Procise 494 instrument at the Protein Core Facility of the Iowa State University (www.protein.iastate.edu/nsequence494). Individual proteins resolved by SDS-PAGE were electroblotted to 0.45 μm Immobilon-P PVDF membrane with 10 mM CAPS, pH 11, with 10% methanol (v/v) transfer buffer for 1 hr at 150 V constant as previously described [30].

Peptide Mass Fingerprint by MALDI-TOF MS
Structure-based protein identification was performed by peptide mass fingerprinting using trypsin digestions and separation by MALDI-TOF MS [31]. Briefly, SDS-PAGE-resolved protein bands were excised from gels, processed with 0.1% w/v Rapigest (Waters, Milford, MA, USA), and digested overnight with Trypsin-Gold (Promega, Madison, WI, USA) according to manufacturer's instructions. Tryptic peptides were recovered and desalted with C 18 ZipTips. Peptides were mixed with α-cyano-4-hydroxy-cinnamic acid before being spotted onto a MALDI plate. Spectra were obtained with a Waters MALDI Micro MX MS in positive reflectron mode (pulse voltage, 2,000 V; reflectron, 5,200 V; source, 15,000 V) with a 20 Hz N 2 laser at 337 nm. Spectral data were processed with MassLynx v 4.1 and ProteinLynx Global Server 2.5 (Waters).

De novo Peptide Sequencing by Orbitrap MS/MS
Sequence data from trypsin digests from affinity-purified proteins was obtained from the University of Arkansas for Medical Sciences Proteomics Core Facility (tri.uams.edu/researchresources-services-directory/core-facilities-technical-services/proteomics-core/) using a Thermo Scientific LTQ-Orbitrap Velos mass spectrometer [32]. The five most abundant peptides were selected for high-resolution tandem MS, with a 99.0% confidence protein threshold and 95.0% confidence peptide threshold.

Bioinformatics
The peptide ion list generated by MALDI-TOF MS was used for putative fungal enzyme identification by the Mascot search engine (www.matrixscience.com). Processed spectra were submitted to MASCOT (Fixed Modification: Carbamidomethyl, Variable Modification: Oxidation, Peptide tol. ± 200 ppm) for comparison to in silica digested proteins. The BLASTp program at NCBI (www.ncbi.nlm.nih.gov/blast/) was used to determine gene sequence identity and correlate identity to known fungal proteases [33]. Putative enzyme function was determined with the Universal Protein Resource (UniProt; www.uniprot.org) [34]. The presence of a signal peptide was determined with the SignalP 4.1 server (www.cbs.dtu.dk/services/SignalP/). Phylogenetic analysis of P. destructans enzyme to dermatophytic and plant pathogenic fungi was performed by www.phylogeny.fr/ [35,36]. A 3-D structure of the protein was generated by FirstGlance in Jmol (bioinformatics.org/firstglance/fgij/). Multiple sequence alignments for the three homologous PdSP proteins were performed with Clustal Omega v 1.2.0 (www.ebi.ac.uk/Tools/msa/ clustalo/).

Fungal cultivation and enzyme production
Representative P. destructans cultures comparing growth on MNB (minimal nutrient broth with supplemented protein, e.g., MNB+G) media, on complex fungal culture media (e.g., TSB), and extracellular protein profiles are shown in Fig. 1. More diffuse and slower hyphal growth was observed on defined MNB, requiring~6 weeks to form a mycelial mat across the surface. Hyphal-growth occurred more quickly on complex media (~4 weeks for considerable hyphal mass), but the mycelial mass formed thick localized clumps rather than the thin mycelial mat observed on MNB surfaces. Growth on MNB resulted in consistent secreted protein profiles (observed by SDS-PAGE) regardless of protein supplement used (G, E, C, or K), while growth on complex media produced variable protein profiles. Protease zymogram screening of culture media showed two different activity profiles represented by MNB+G and TSB, which indicated differences in the dominant protease activities present. The location of protease activity at the top of the gel MNB media indicated the protein migrated into the gel poorly, whereas the dominant activity in the TSB migrated midway down the gel, indicating distinct differences in size and charge for the associated protein.
Qualitative differences were observed in total protease activity (determined with FITC casein substrate assay) secreted into culture media (Fig. 2). P. destructans grown on MNB with supplemented proteins accumulated~3X more total protease activity compared to the complex media treatments. MNB samples treated with representative protease inhibitors (EDTA, PMSF, and E-64) showed a dramatic reduction in activity occurred with PMSF, indicating dominant serine protease activity. E-64 treatment resulted in no protease activity inhibition, indicating measurable cysteine proteases were not secreted in media tested. EDTA treatment resulted in pronounced inhibition in MNB media only. While EDTA generally inhibits metalloproteases, it can also de-stabilize other broad class activities, including certain serine proteases. Though metallo-proteases may be secreted in MNB media, they likely represent a minor class compared to serine proteases due to the singular activity band observed by zymogram analysis (Fig. 1A). MNB+G provided the highest total protease activity yield, with nearly complete inhibition of serine protease activity with PMSF. Further experimentation to isolate serine proteases focused on culturing P. destructans with MNB+G.

Isolation of the serine protease activity
While the extracellular protein profile in MNB+G culture media is simple, the considerable solution viscosity (possibly due to secreted polysaccharides) of concentrated samples was problematic for isolating the serine protease present. To address this issue, two separation methods were evaluated-preparative IEF and affinity chromatography. For the first, the concentrated enzyme solution was mixed with broad range ampholytes (pH 3-10) and fractionated in a Rotofor cell. Protease activity was resolved as a peak over fractions 10-13 (pH 7-9; Fig. 3). PMSF completely inactivated protease activity in the pooled fraction, confirming presence of the serine protease (data not shown). The SDS-PAGE profile for proteins in pooled peak fractions showed enrichment of two major bands at 27.9 kDa and 55.0 kDa in the pooled fraction (their co-isolation indicates a similar pI for both proteins).
Further treatment of concentrated culture medium evaluated affinity binding to benzamidine-Sepharose and ConA chromatography media. No protein or activity was adsorbed to benzamidine-Sepharose, suggesting the serine protease is not a family S1-type serine peptidase [38]. In contrast, serine protease activity in extracellular preparations bound to the ConA lectin affinity column, indicating it contains an N-linked glycan. SDS-PAGE of proteins eluted from ConA (Fig. 4) showed the 27.9 kDa band was greatly enriched by this separation, and the 55.0 kDa band being eliminated. The specific activity of the ConA-bound fraction was increased 33fold, implicating the 27.9 kDa protein as the P. destructans serine protease (PdSP).
Electrophoretic and biochemical properties determined for the ConA-treated PdSP are summarized in Table 1. The enzyme showed a broad pH optimum covering pH 6 to 8. Its temperature optimum was~60°C, suggesting modest thermal stability. Addition or DTT or SDS to substrate (casein) solution showed increased enzyme activity, while divalent cation addition did not further increase activity.
MALDI-TOF MS screening to identify P. destructans proteins There were 7 major bands observed consistently by SDS-PAGE for proteins isolated from P. destructans MNB+G culture medium (Fig. 1). All 7 proteins were excised from SDS-PAGE gels, trypsin-digested, and then analyzed by MALDI-TOF MS to establish identity based on sequence by matching tryptic peptide sets using the MASCOT search engine with the Broad Institute's P. destructans genomic sequence database [Geomyces destructans Sequencing Project, Broad Institute of Harvard and MIT (www.broadinstitute.org)] and NCBI's Reference Sequence database [39]. Tentative identifications were initially made for 5 proteins ( Table 2). The largest band, 115 kD, matched to a putative catalase/peroxidase HPI. The dominant 106 kD protein, which co-separated by ConA binding, is without annotated function, but structural elements and sequence homology implicate it to be a FAD-binding oxidoreductase. A minor 84 kD protein is also putatively identified as a C-N ligase. The 55 kDa protein, enriched in the IEF peak fractions but not ConA, is putatively identified as a glycoside family 18 chitinase. The smallest protein identified, 12.5 kDa, is putatively identified as a phosphatidylglycerol / phosphatidylinositol transfer protein. The 70 kDa protein remains unidentified, but is  eliminated as the serine protease since it was separated from the isolated enzyme activity by both IEF and affinity chromatography treatments. The tryptic peptide mass spectrum (by MALDI-TOF MS) for the P. destructans 27.9 kDa protein is shown in Fig. 5A. While the overall sample quality appeared good, the tryptic peptide ion data was not able to provide statistically significant matching to any existing protein. Several exogenous peptides are identified in the mass spectrum-the autolytic trypsin peptide, m/z 842.50 and three peptides derived from ConA, m/z 1318.63, m/z 2104.06, and m/z 2832.34. Further analysis of tryptic peptides from the 27.9 kDa protein was performed with an Orbitrap (ThermoScientific LTQ Velos) mass spectrometer to obtain sequence data from peptide ions to establish identity as the PdSP.  Table 3. A representative MS/MS spectrum is shown for peptide ion m/z 908.49 (SVISM Ox SLR) (Fig. 5B). The combined MS and MS/MS data matched unequivocally to a specific open reading frame (UniProt GMDG_08491) from the P. destructans 20631-21 genome, confirming structure with functional serine protease activity for PdSP1. The calculated tryptic peptide ions from the translated protein sequence (GenBank ELR07576.1) are summarized in Table 3.

Sequence analysis for the translated PdSP1 gene
The complete translated amino acid sequence for the P. destructans gene product ELR07576.1, represented by PdSP1, is shown in Fig. 6. BLASTp database searching (NCBI nrDB) indicated two conserved structural domains for PdSP1. The first includes a 99 amino acid propeptide sequence representing a MEROPS family I9 inhibitor domain, followed by the 256 amino acid sequence representing a MEROPS proteinase K-like peptidase S8 family (SB clan) domain [20]. Two serine protease homologues are present in the P. destructans genome. The first, PdSP2, shows 90% sequence identity with PdSP1, and a second homologue, PdSP3, showed a much lower 53% sequence identity. There were BLASTp hits for 43 putative S8 serine proteases in the P. pannorum genome, with sequence similarity falling between the two P. destructans homologues. The primary sequences for the three PdSP homologues from P. destructans and the highest matching homologue from P. pannorum are included in Fig. 6.
Features of PdSP primary sequences in Fig. 6 include a 20 amino acid secretory signal peptide and the location for the N-terminal amino acid in the mature protein (determined directly  (Table 1). There is a single N-glycosylation sequon (N 187 -Y 188 -S 189 ), which is occupied based on ConA binding, but the glycan appears to contribute little to the apparent mass by SDS-PAGE (Fig. 4). A single cysteine (C 196 ) present in the mature protein sequence indicates no internal disulfide bonds in PdSP1 structure. A phylogenetic tree was constructed with the three P. destructans PdSP sequences and selected dermatophytic and plant pathogenic subtilisin-like serine proteases (S8A) with high sequence relationship (identified from BLASTp search), Endgyodontium album proteinase K, Bacillus licheniformis subtilisin (Carlsberg), and with Aspergillus clavatus serine protease chosen arbitrarily as an outgroup (Fig. 7). P. pannorum S8A serine protease sequences were not included due to the very large number of genes (43) and their close sequence conservation (as indicated by BLASTp) compared to the P. destructans PdSPs. PdSP2 and PdSP3 were 90% and 53% homologous (by sequence identity) to PdSP1, respectively. P. destructans S8 serine proteases grouped closer to plant pathogenic fungal proteases [e.g., Botryotinia fuckeliana (gray mold disease), Sclerotinia sclerotiorum (white mold), Marssonina brunnea (Marssonina leaf spot)] than to human pathogenic dermatophytes (i.e., Arthroderma, Trichophyton, and Microsporum).

Western Blotting
The peptide GSVDSTDTRASSSN (corresponding to amino acid positions 400-413 in Fig. 6) was synthesized and used to generate a PdSP1-specific antiserum for subsequent use in western blotting detection of the protein. The resulting PdSP1 antisera showed suitable titer and sensitivity for detecting PdSP1 (27.9 kDa band) in blots from total protein isolated from P. destructans culture medium. Bovine trypsin, proteinase K, or other P. destructans proteins did not cross-react (data not shown).

Discussion
WNS is an emergent fungal disease with devastating consequences for North American cavedwelling bat populations, and little is known about P. destructans' ecology or relative pathogenicity [1,2,26]. Bats play integral roles in our ecosystem as keystone species, pollinators, and pest control [40,41]. The unprecedented mass mortality of cave bats may lead to increases in insect pests, pesticide usage, and insect vector diseases [42]. The loss of bats and their function in natural pest control is estimated to cost at least $3.7 billion/year in damages to agriculture [43]. A clear understanding of the biochemical ecology between P. destructans and bats is paramount in determining proper control strategies to alleviate disease effects. Extracellular proteases play roles in fungal pathogenesis and are hypothesized to function in WNS [23]. Epidermal wing necrosis observed in WNS may be partially attributed to protease activities secreted by P. destructans. Secreted serine proteases are common in saprophytic fungi and are documented in species pathogenic to plants, insects, mammals, and other fungi [44][45][46]. In this study we isolated a 27.9 kDa protein consistently secreted by P. destructans into liquid culture media and identified it as a family S8A subtilisin-like serine proteinase (PdSP1). We confirmed structural identification by matching peptide ion masses and de novo peptide sequences obtained by MS (Table 3 and Fig. 5) to the hypothetical protein (GenBank ELR07576.1), translated from the gene GMDG_08491 present in the draft genome sequence of P. destructans isolate 20631-21 (www.broadinstitute.org). The properties determined experimentally for PdSP1 (Table 1 and Fig. 5) matched values predicted for the 281 AA mature protease having a theoretical molecular weight of 27,882.7 and pI of 7.20. The nearly complete inhibition of total protease activity by PMSF indicates serine protease is the predominant protease type produced by P. destructans cultured with protein as sole nitrogen source. Subtilisinlike serine proteases are dominant secreted enzymes in fungi associated with symbiotic interactions, particularly with animals [46]. While PdSP1 is isolated as an extracellular protein produced in vitro, it may function as a key protease in P. destructions' virulence.
P. destructans has three homologous genes for S8 subtilisin-like serine proteases. The translated sequence for PdSP2 (NCBI accession# ELR03877.1) exhibits high sequence identity (90%) with PdSP1, while PdSP3 (NCBI accession# ELR10046.1) retains 53% sequence identity. Considering the high sequence conservation between PdSP1 and PdSP2, there are only 4 tryptic peptides (> m/z 750) in common, with two of these observed experimentally (m/z 892.49 and 1435.70; Table 3). Four tryptic peptide ions unique to PdSP1 are observed in sample spectra, readily confirming identity matching. A recent analysis by Muszewska et al. (2011) suggested independent expansions of subtilisin-like serine proteases in fungi may be related to use of mammalian substrates, rather than pathogenicity [46]. Three isoforms in P. destructans may reflect a more obligate relationship with its host(s), whereas the entomopathogen M. anisopliae can have as many as 25 isoforms and P. pannorum has 43 S8A serine protease genes present [47], with the latter reflecting a general saprophytic lifestyle [26,44]. The low number of additional isoforms may support previous results that P. destructans exhibits reduced saprophytic growth [26]. Phylogenetic tree generated from S8 serine proteases primary sequence from P. destructans and other select microbial organisms. Included are three serine proteases from P. destructans, selected serine protease sequences (S8A proteinase-K subfamily) from plant pathogenic and human dermatophytic fungi, proteinase K, Carlsberg subtilisin, and Aspergillus cavatus serine protease. Phylogenetic tree generated with program Phylogeny.fr (S8A accessions are listed in S1 Fig.) doi:10.1371/journal.pone.0120508.g007 To facilitate preparation of PdSP1-specific antisera for future in vivo studies, we selected the peptide sequence GSVDSTDTRASSSN for synthesizing a peptide antigen. This peptide shows a suitable hydrophilicity profile, avoids the single N-glycosylation site, and appears to represent an exterior loop (determined from a 3-D model generated by FirstGlance in Jmol). We obtained a monospecific polycolonal antiserum (rabbits) from the synthetic peptide that detected the 27.9 kDa PdSP1 in western blots with good sensitivity (Fig. 4). While PdSP2 may cross-react with this antiserum, the more divergent sequence for the corresponding peptide in PdSP3 suggests it will not cross-react. Pending further characterization of the antiserum, it will prove useful service to detect the presence of PdSP1 in tissue washings or in histological sections from bats displaying WNS. Antisera may also be generated for the other two PdSP isoforms in a similar approach.
Proteases functioning as virulence factors in fungal epithelial infections may provide targets for vaccines and disease prophylactics [23]. While extracellularly-secreted proteases can be molecular biomarkers in disease processes, non-pathogenic fungi secrete many similar enzymes. Currently, the differential expression of PdSPs in P. destructans upon host tissue infection and establishment is unknown. Here, we provide two new tools for PdSP1 detection, PdSP-specific gene sequences that can be used for selecting PCR primer probes and an antiserum to directly detect the expressed product from P. destructans genes. Current evidence suggests pathogenesis is a complex process involving host, pathogen, and environmental interactions [48], hence many factors other than simply enzyme secretion contribute to disease manifestation. Furthermore, if certain enzymes involved in disease processes are inhibited, fungi may compensate by expressing alternative enzymes. In vitro fermentation systems provide a starting point for identifying fungal-specific extracellular enzymes. The temporal patterns of such enzymes, the presence of their active vs. precursor forms at different time points, and their secretion in situ within bat tissue remains an area available for future research. The availability of the P. destructans genomic sequence and application of highly accurate mass spectrometric techniques will facilitate such studies. In conclusion, we isolated and identified a dominant subtilisin-like serine protease produced in vitro by P. destructans, providing new tools to determine its putative function in bat wing necrosis [6], as well as for assessing expression of other PdSP isoforms during WNS.
Supporting Information S1 Fig. Accession numbers for S8A sequences used with P. destructans proteases for preparing phylogenetic tree (Fig. 7). (DOCX) S1 Table. Summary data for total protease activity measured in culture media and response to protease inhibitor treatments (Fig. 2). (XLSX)