Rice Defensin OsAFP1 is a New Drug Candidate against Human Pathogenic Fungi

Fungal infections, such as candidiasis and aspergillosis, are some of the most frequent infections in humans. Although antifungal drugs are available for the treatment of these infections, antifungal agents with new mechanisms of action should be developed because of the increasing incidence of drug-resistant pathogens in recent years. In this study, a basic functional analysis of rice defensin OsAFP1, a novel antifungal drug candidate, was conducted. OsAFP1 exerted fungicidal activity against Candida albicans, the most common pathogenic fungus in humans, at 4 μM concentration, but it did not inhibit the growth of human pathogenic bacteria. In addition, OsAFP1 retained structural stability after heat treatment at 100 °C for 10 min and after serum treatment at 37 °C for 24 h. A propidium iodide (PI) uptake assay and mutational analysis revealed that amino acid residues within the C-terminal γ-core motif of OsAFP1, particularly Leu-39 and Lys-41, play an important role in its antifungal activity. Further, PI uptake and apoptosis assays suggested that OsAFP1 exerts its antifungal activity by inducing apoptosis of target cells. Immunohistochemistry showed that the OsAFP1 target molecule was located in the cell wall. These findings indicate that OsAFP1 may be developed into a potent antifungal drug.

In this study, antifungal activity assays against Candida albicans, the most common pathogenic fungal species that causes candidiasis in humans 20 , revealed that 4 μM DEF7 completely inhibited its proliferation. In a previous study, DEF7 (designated OsAFP1 in subsequent papers) was reported to exhibit antimicrobial activity against rice pathogenic microbes, such as Pyricularia oryzae 21 and Helminthosporium oryzae 22 . Based on this latest knowledge, basic functional analyses of O. sativa OsAFP1 were conducted to assess its potency as a novel antifungal drug. Specifically, the antimicrobial spectrum of OsAFP1 against various human pathogenic microorganisms, stability upon heat and serum treatments, mechanism of action, and structural factors responsible for its antifungal activity were analysed.

Results
Antimicrobial activity of OsAFP1. We cloned the OsAFP1-encoding gene, then produced the recombinant peptide in Escherichia coli and purified it. Antifungal activity assays revealed that OsAFP1 significantly suppresses the growth of C. albicans CAI4 (IFM 61937) cells in a concentration-dependent manner; the MIC and half maximal inhibitory concentration (IC 50 ) were determined to be 4 μM and 2 μM, respectively (Fig. 1a). Next, the antimicrobial activities of OsAFP1 against several human pathogenic gram-negative or gram-positive bacteria and other fungi were examined. OsAFP1 exhibited antifungal activity against two yeast strains, Saccharomyces cerevisiae BY4742 (MIC of 4 μM; Fig. 1b) and S. cerevisiae S288C (MIC of 16 μM). The defensin, at concentrations up to 32 μM, did not show any growth inhibition against five bacterial strains, namely, E. coli K-12, Porphyromonas gingivalis ATCC 33277, Streptococcus mutans JCM 5705, Staphylococcus aureus NBRC 12732, and Propionibacterium acnes JCM 6473 (Table 1). This suggested that OsAFP1 is a fungi-specific antibiotic.
A subsequent fungicidal assay was performed to evaluate whether OsAFP1 is a fungicidal or fungistatic agent (Fig. 1c). After treatment with OsAFP1, about 50% of C. albicans cells died in 30 min, and almost all cells died in 4 h, indicating that OsAFP1 exhibits antifungal activity against target cells in a fungicidal manner.
Thermal and serum stability of OsAFP1. Because of multiple disulphide bonds, the structures of some cysteine-rich peptides, including defensins, have been reported to be highly rigid, with excellent thermal stability and protease resistance 23,24 . Molecular masses of reduced and oxidized forms of recombinant OsAFP1 ([M + H] + ) were determined to be 5963.879 and 5956.508 by matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF/MS) (Fig. S1). Considering that the calculated molecular mass of OsAFP1 is 5963.87, recombinant OsAFP1 is suggested to have four disulphide bonds in its tertiary structure. Therefore, since OsAFP1 is predicted to have high structural stability, the thermal and serum stability of this peptide were evaluated.
First, antifungal activity against C. albicans CAI4 was examined using heat-treated OsAFP1. The heat-treated peptide inhibited the growth of C. albicans cells in a concentration-dependent manner with a MIC of 8 μM, i.e., a negligible loss of activity even after heating at 100 °C for 10 min (Fig. 1d). Then, the degree of biodegradation of OsAFP1 incubated for up to 24 h in human serum at 37 °C was analysed using Tricine-sodium dodecyl sulphate-polyacrylamide gel electrophoresis (Tricine-SDS-PAGE). No significant degradation of OsAFP1 was observed even upon conditioning in serum for 24 h (Fig. 1e). These results demonstrated that, as anticipated, OsAFP1 possesses extremely high stability upon heat and serum treatment.
The effect of OsAFP1 on the C. albicans cell membrane. Many antimicrobial peptides and proteins cause cell death by acting on the cytoplasmic membrane of target cells, causing disruption and increased permeation of the membrane 25 . Therefore, the effect of OsAFP1 on C. albicans CAI4 cytoplasmic membrane was assessed by a propidium iodide (PI) uptake assay that enabled evaluation of cell membrane permeability by flow cytometry 26 . After treatment with 4 × IC 50 concentration of melittin, which exerts an antibacterial effect by disrupting the cytoplasmic membrane 27 , 58.3% of cells were stained with PI, indicating that their membranes were disrupted. On the other hand, after OsAFP1 treatment (4 × IC 50 ), only 3.6% of cells were stained with PI, suggesting that the peptide does not act on the cytoplasmic membrane to a great extent (Fig. 2a-c).
Sequence alignment of plant defensins. The results of the PI uptake assay with OsAFP1 were not entirely unanticipated. It is well known that many vertebrate and invertebrate defensins (e.g. human HNP1-4 and HD5) exert antibacterial activity by binding to the cytoplasmic membrane and subsequently increasing the membrane permeability of target cells [28][29][30] . Some plant defensins, such as NaD1 from Nicotiana alata and TPP3 from Solanum lycopersicum, mainly act by increasing the membrane permeability of their target fungal cells [31][32][33] . On the other hand, others, such as RsAFP2 from Raphanus sativus, MsDef1 from Medicago sativa, and MtDef4 from Medicago truncatula, not only increase fungal membrane permeability, but also bind to specific components of cell walls or cytoplasmic membranes and exert antifungal activity by inducing apoptosis or disrupting Ca 2+ signalling and homeostasis 31,32,34,35 . Psd1 from Pisum sativum also binds fungal cell walls, induces cell cycle arrest, and blocks K + channels 31,36 . An alignment of the amino acid sequences of these defensins and OsAFP1 is shown in Fig. 3a. As shown, all cysteine residues are well conserved in these defensins; they are presumed to form disulphide bonds as inferred from the three-dimensional structures of NaD1 and MtDef4 37,38 . Furthermore, the γ-core motif (GXCX 3-9 C, boxed in Fig. 3a), reported to be important for antifungal activity of MtDef4 38 , is also conserved. However, other amino acid residues, even those within the γ-core motif, are not conserved in any of the sequences.  Specific regions of OsAFP1 essential for its antifungal activity. To identify the regions of OsAFP1 associated with antifungal activity, eight overlapping peptides were designed (peptides 1-8). Their amino acid sequences corresponded to different regions of OsAFP1 and collectively covered the entire OsAFP1 sequence; further, each two adjacent peptides overlapped (Fig. 3b). Since MICs could not be determined at a concentration of 25 μM in all the peptides, their antifungal activity was assessed using IC 50 . Antifungal assays revealed that peptide 1 exhibited pronounced antifungal activity; peptides 2, 7, and 8 showed antifungal activity somewhat weaker than that of peptide 1, suggesting that approximately 10 N-terminal and C-terminal amino acid residues are important for the activity ( Table 2). C. albicans CAI4 PI uptake assays with peptides 1 and 7 (4 × IC 50 ) were conducted (Fig. 2d,e). Interestingly, 5.5% of cells emitted fluorescence after treatment with peptide 7, similarly to OsAFP1 treatment, whereas 17.6% of cells emitted fluorescence after peptide 1 treatment. These observations indicated that the mechanism underpinning the antifungal activity of peptide 1 was at least partially different from that of OsAFP1 and did not reflect its properties. Based on the described peptide analysis, the C-terminal region of OsAFP1 was considered to play an important role in its antifungal activity. This was next verified using site-specific mutagenesis. Six amino acid residues in the peptide 7 region of OsAFP1 that had bulky side chains (Lys-35, His-37, Leu-39, Glu-40, Arg-41, and Lys-42) were separately substituted with an alanine residue (yielding OsAFP1 mutants K35A, H37A, L39A, E40A, R41A, and K42A); they were then overproduced in E. coli and purified. Because of the low expression levels of the E40A mutant in recombinant E. coli cells, the role of Glu-40 in the antifungal activity of the defensin could not be verified. Antifungal activity evaluations of the remaining mutants revealed a large drop in activity (MIC > 32 μM) of all mutant defensins. To compare the reduction of activity of the mutants, IC 50 values were calculated. The activities of L39A and K41A mutants were so low that the IC 50 values could not be calculated, in contrast with the remaining mutants (Table 3).
Taken together, these results clearly indicated that amino acid residues within the γ-core motif, in particular Leu-39 and Lys-41, play critical roles in the antifungal activity of OsAFP1.
Apoptosis-inducing effect of OsAFP1. As described above, the results of the PI uptake experiments indicated that OsAFP1 did not disrupt the plasma membrane of the fungal cells or increase membrane permeability. Therefore, we proposed that the antifungal activity of OsAFP1 might involve the induction of apoptosis (as seen for RsAFP2) 34 and performed an annexin V assay to detect apoptosis (Fig. 4). Annexin V-FITC, an . Amino acid sequences were aligned using the Clustal Omega program. Identical, strongly conserved, and weakly conserved amino acid residues are denoted by asterisks, colons, and dots, respectively. Putative secondary structural elements of OsAFP1 are displayed above the sequences. Cysteine residues that are presumed to form disulphide bonds are highlighted in black. The C-terminal γ-core motif is enclosed within a solid box. (b) Designed peptides corresponding to partial amino acid sequences of OsAFP1. The peptides correspond to eight overlapping OsAFP1 sequence fragments.

Peptide
Sequence a IC 50 (μM)  Table 2. Antifungal activity of peptides corresponding to partial sequences of OsAFP1 against C. albicans. a OsAFP1 regions corresponding to each peptide are shown in Fig. 3b. apoptosis marker, was used to verify this hypothesis. It specifically binds to phosphatidylserine exposed on the apoptotic cell surface in the presence of calcium ions and emits fluorescence 39 , which may be detected using an epifluorescence microscope. Most C. albicans CAI4 cells did not emit fluorescence after up to 4 h of treatment with phosphate-buffered saline (PBS), i.e., apoptosis was not induced in those cells. In contrast, almost all fungal cells emitted fluorescence after 0.5 h treatment with OsAFP1, and in particular, FITC fluorescence was detected in the vicinity of the plasma membrane, indicating that early apoptotic morphology was induced. Necrotic cells that only emit PI fluorescence were not detected at any treatment time. In addition, the proportion of dead cells as estimated by PI fluorescence coincided with the results of the fungicidal assay described above. These results suggested that OsAFP1 exhibits antifungal activity by inducing apoptosis of target cells.
Binding of OsAFP1 on the cell wall. Immunocytochemical (ICC) staining using an OsAFP1 antibody was performed to identify its target molecule in C. albicans cells. The analysis showed that OsAFP1 localised around the cell wall after 0.5 h treatment (Fig. 5). The fungicidal and apoptosis assay described above showed that early apoptosis was induced in all cells 0.5 h after OsAFP1 treatment, while half of the cells were already dead. Taken together, these analyses suggested that OsAFP1 induces apoptosis in C. albicans cells by binding to target molecules present around the cell wall, resulting in cell death.

Discussion
In this study, we systematically examined the antimicrobial properties and mechanism of action of the rice defensin OsAFP1. The peptide is a novel antifungal drug candidate. Antimicrobial activity assays with various microorganisms, including human pathogenic fungi and bacteria, indicated that OsAFP1 possesses pronounced antifungal activity against C. albicans and S. cerevisiae, with no activity against the tested gram-negative and gram-positive bacteria. A subsequent fungicidal assay indicated that OsAFP1 exhibits antifungal activity in a fungicidal manner. In addition, OsAFP1 maintained structural stability upon heat treatment (100 °C for 10 min) and serum treatment (at 37 °C for 24 h). These findings indicated that OsAFP1 could be developed into a fungi-specific drug against human pathogens.
Antifungal activity assays, PI uptake assays with eight peptides corresponding to different regions of OsAFP1, and subsequent mutational analysis revealed that the C-terminal region plays an important role in the antifungal activity of this defensin. Peptide 1, corresponding to the N-terminal fragment of OsAFP1, slightly disrupted the membranes of C. albicans CAI4 cells; no further mutational analysis of peptide 1 region was performed in this   Our data demonstrated that the γ-core motif, which is preserved in many plant defensins and has been reported to play a role in their antibacterial activity 38,40,41 , is important for the antifungal activity of OsAFP1. Binding to the target molecule has been reported to constitute one of the functions of the γ-core motif. Because most amino acid residues of this motif are not highly conserved, it is reasonable to anticipate that their roles in the γ-core motifs of different defensins differ. Indeed, it has been shown that the ' 35 RGFRRR 40 ' loop, a region within the γ-core motif of MtDef4, binds to its target molecule, and Phe-37 and Arg-38 are especially important for antifungal activity 41 . In comparison with the MtDef4 γ-core motif, in OsAFP1, two amino acid residues (Ser-36 and His-37) have been inserted, and two (Leu-39 and Glu-40) have been substituted (for Phe-37 and Arg-38 in MtDef4, respectively) (Fig. 3a). Our analysis of OsAFP1 mutants showed that two of these four residues are important for antifungal activity (Table 3). It has been suggested that MtDef4 binds to cytoplasmic phosphatidic acid and kills target cells by destroying Ca 2+ signalling and homeostasis 38,41 . The above sequence difference in the γ-core motif may be associated with the difference in the mechanisms of action of the two defensins. Furthermore, since RsAFP2, Psd1, and MsDef1 bind to glucosylceramide (GlcCer) and do not exert antifungal activity against GlcCer-lacking strains, it has been suggested that these defensins exhibit antifungal activity by binding to GlcCer in the fungal cell wall 35,[42][43][44] . Since OsAFP1 exhibited antifungal activity against S. cerevisiae, which lacks GlcCer 45 (Fig. 1b and Table 1), it is possible that it induces apoptosis via a target molecule in the fungal cell wall other than GlcCer. ICC staining analysis strongly supported this hypothesis: OsAFP1 was These analyses also clarify the localization of defensin in the target fungus cell and give strong clues to elucidate its mechanism of action. In the future, we will explore the target of OsAFP1 using a combination of these fluorescent analyses and fungal mutants lacking specific membrane components. The calculated homology model of OsAFP1 is shown in Fig. 6a. The γ-core motif corresponding to peptide 7 is located in a site that likely interacts with the target molecule. In addition, sequence diversity within the motif is considered to contribute to the diversity of the recognized target molecules, as if the motif were a variable region of an antibody (Fig. 6b).

Conclusion
In the present study, functional analyses of the antifungal activity of the rice-derived defensin OsAFP1 were performed. OsAFP1 completely inhibited the cell growth of C. albicans CAI4 at a concentration of 4 μM in a fungicidal manner. Since MIC values of the existing antifungal drugs, such as fluconazole, amphotericin B, and caspofungin, are in the 0.01-1 μM range 48 , further engineering to enhance its function may be required before OsAFP1 can be used as a drug. On the other hand, OsAFP1 was extremely stable upon heat and serum treatments and exerted its antifungal activity by inducing apoptosis, i.e., a mechanism different from that of existing antifungal drugs. These features render OsAFP1 a useful new antifungal drug candidate. Thus, further study of the antifungal mechanism of action of OsAFP1 and verification of its safety towards human cells are warranted.

Materials and Methods
Microorganisms and culture conditions. Seven microbial strains were used for OsAFP1 antimicrobial activity testing. P. gingivalis ATCC 33277 cells were cultured in modified Gifu anaerobic medium (GAM) (Nissui, Tokyo, Japan) in an anaerobic jar (Mitsubishi Gas Chemical, Tokyo, Japan) in the presence of a deoxygenating regent (AnaeroPack, Mitsubishi Gas Chemical) for 48 h at 37 °C. E. coli K-12 NBRC3301, S. mutans JCM 5705, S. aureus NBRC 12732, and P. acnes JCM 6473 cells were cultured in Luria-Bertani (LB) broth 49  Antibacterial assays. Recombinant OsAFP1 and its mutants used in this study were prepared as described in "Supplementary Information". Antibacterial assays to determine the MIC of OsAFP1 were performed using a modified BacTiter-Glo Luminescent cell viability assay kit (Promega, Madison, WI, USA). This assay quantifies ATP in bacterial cultures to determine the number of live cells. All procedures were performed following the manufacturer's instructions. Briefly, bacterial cells (1 × 10 4 cells/ml) were cultured in 100 μl of the appropriate medium (as specified in the "Microorganisms and culture conditions" section) in the presence of 0-32 μM OsAFP1 in 96-well Costar cell culture plates. After appropriate incubation for 4-40 h, 10 μl of Lucifell ATP-eliminating reagent (Kikkoman, Chiba, Japan) was added to each well to reduce medium-derived ATP and mixed for 10 min using an orbital shaker. The cell culture (50 μl) was mixed gently with BacTiter-Glo reagent (50 μl) in a well of white Optiplate-96 plates. After incubation for 15 s, luminescence at 560 nm was measured using a plate reader. MIC was defined as the minimum concentration of OsAFP1 that reduced the sample ATP luminescence by >99% compared with that of the control culture. Experiments were independently repeated at least three times.
Antifungal assays. Antifungal assays to determine the MIC of OsAFP1 were performed as follows. Fungal cells (1 × 10 5 cells) were suspended in 100 μl of the appropriate medium containing 0-32 μM OsAFP1 in 96-well Costar cell culture plates. After incubation, OD 650 was measured using a plate reader. MIC was defined as the minimum concentration of OsAFP1 that resulted in >99% reduction of OD 650 compared with that in the control culture. In some measurements, because of aggregation of medium components, it was difficult to determine MIC values. In those cases, the cultures were plated on agar plates (0.5 × PDB), and the peptide concentration at which viable cell numbers were 1% or less than that in the control culture was considered the MIC. GraphPad Prism version 6.00 for Macintosh (GraphPad Software, La Jolla, CA, USA) was used to calculate IC 50 values. Experiments were independently repeated at least three times. Detailed growth conditions for each fungus are provided in the "Microorganisms and culture conditions" section above.
Fungicidal assays. C. albicans CAI4 cells (2 × 10 5 cells/ml) were treated with OsAFP1 and incubated at 30 °C for 0, 0.17, 0.5, 1, 2, and 4 h. After being washed three times with KPB, the cells were plated on agar plates (0.5 × PDB). Calculated CFUs in the plate were used as indicators of viable cells. OsAFP1 was found to exhibit its fungicidal activity with a delayed effect in the course of examining various conditions. Therefore, in the present assay, 4 × MIC concentration was used, because long-term OsAFP1 treatment may produce various undesirable effects other than fungicidal activity and make the assay difficult to interpret. 4 × MIC concentration was also applied in the subsequent annexin V assay and ICC staining analysis. Stability assays. Thermal stability of OsAFP1 was evaluated by comparing the MIC of heat-treated (10 min at 100 °C) peptide to the MIC of untreated peptide.
A serum stability assay was used to determine the resistance of OsAFP1 to biodegradation. An OsAFP1 solution (2 μl, 0.9 mM) and 8 μl of human serum (BioWest, Nuaillé, France) were mixed in sterile microtubes; the mixtures were incubated at 37 °C for 0, 6, 12, and 24 h. At each time point, serum proteins were denatured by the addition of equal volumes of 3 M TCA and 6 M urea. In a control experiment, PBS was used instead of human serum, and the mixture was treated as above. After denaturing, the mixtures were centrifuged at 10,000 × g for 10 min, and the upper lipid phase was removed. OsAFP1 degradation was analysed by Tricine-SDS-PAGE followed by Coomassie Brilliant Blue staining, as described previously 50 . The peptide band intensity was determined with Image J 1.48 51 , and the time-dependent rate of its degradation was calculated. Experiments were independently repeated at least three times. PI uptake assay. C. albicans CAI4 cells (2 × 10 5 cells/ml) were treated with OsAFP1, peptide 1, peptide 7, or melittin at 4 × IC 50 concentrations (8 μM for OsAFP1, 24 μM for peptide 1, 40 μM for peptide 7, and 6 μM for melittin 52,53 ) and incubated at 30 °C for 3 h. After the addition of PI (5 μg/ml) and incubation at room temperature in the dark for 30 min, the cells were analysed using a flow cytometer equipped with a 488 nm air-cooled argon laser. At least 10,000 events per sample were recorded. As a negative control, PBS was used instead of antifungal agents, and the mixtures were treated in the same way.
Peptide synthesis. Peptides with partial amino acid sequences of OsAFP1 were synthesized using the fluorenylmethyloxycarbonyl solid-phase method and a PSSM-8 automated peptide synthesizer and purified to >98% purity using reversed-phase high performance liquid chromatography with a Cadenza CD-C18 column. Molecular masses of purified peptides were confirmed by MALDI-TOF/MS analysis with an Autoflex III TOF/ TOF. Stock solutions of peptides (10 mg/ml) were prepared in sterile ultra-pure water.
Annexin V assay. C. albicans CAI4 cells (2 × 10 5 cells/ml) were treated with OsAFP1 at 16 μM (4 × MIC) and incubated at 30 °C for 0, 0.5, 1, 2, and 4 h. After being washed three times with KPB, the cells were resuspended in 500 μl of 1 × binding buffer and mixed with 5 μl each of annexin V-FITC and PI reagent (BioVision, Mountain View, CA, USA). The mixtures were incubated at room temperature in the dark for 5 min, and the cells were then observed under a TE2000-S epifluorescence microscope (Nikon, Tokyo, Japan) equipped with an appropriate objective lens [CFI Plan Fluor 100 × oil (Nikon)] and bandpass filters with 542 nm excitation and 620 nm emission for PI and 494 nm excitation and 518 nm emission for FITC. ICC staining. The custom rabbit anti-OsAFP1 antibody used in this assay was purchased from Eurofins Genomics (Tokyo, Japan). For ICC staining, C. albicans CAI4 cells (2 × 10 5 cells/ml) were treated with or without OsAFP1 at 16 μM (4 × MIC) and incubated at 30 °C for 0.5 h. The cells were collected, fixed for 2 h in 4% paraformaldehyde phosphate buffer solution, and stored in 1% formalin neutral buffer solution. Sliced sections from each 25 × 35 mm paraffin block containing cells were placed on slides and used for subsequent ICC staining. After washing the slides once with distilled water and three times with PBS, antigen retrieval was carried out in a high-pH antigen retrieval solution (1 mM EDTA, pH 8.0) for 40 min at 95 °C. The slides were kept at room temperature and then immersed in 3% H 2 O 2 for 5 min to block endogenous peroxidase activity. Sections were treated with anti-OsAFP1 primary antibody (100 × diluted solution) overnight at 4 °C, and then with secondary antibody (Super Sensitive One-Step Polymer-HRP, BioGenex Laboratories, Fremont, CA, USA) for 15 min at room temperature. In between each of the above steps, the sections were washed thrice for 5 min in PBS. The antigen-antibody complex was visualised by diaminobenzidine (Super Sensitive DAB, BioGenex Laboratories) as a chromogen for 1 min at room temperature. The sections were counterstained with Mayer's haematoxylin solution (Muto Pure Chemicals, Tokyo, Japan) for 1 min, dehydrated, cleared, and mounted.
Homology modelling of OsAFP1. The homology model of OsAFP1 was generated using the MODELLER program 54 . Three-dimensional structures of γ-thionin from Hordeum vulgare (PDB ID: 1GPT), defensin DEF4 from M. truncatula (PDB ID: 2LR3), and defensin NaD1 from N. alata (PDB ID: 4CQK) were used as template structures. It is presumed that eight cysteine residues in these peptides (and in OsAFP1) are completely conserved and that they have similar tertiary structures (data not shown). The primary structures of OsAFP1 and the template peptides were first aligned using the Clustal Omega program 55 . A homology model of OsAFP1 was then generated by executing a script file in MODELLER based on the alignment; the validity of the structure was verified using the PROCHECK program 56 .