A frog‐derived antimicrobial peptide as a potential anti‐biofilm agent in combating Staphylococcus aureus skin infection

Abstract Staphylococcus aureus (S. aureus), one of the most prevalent bacteria found in atopic dermatitis lesions, can induce ongoing infections and inflammation by downregulating the expression of host defence peptides in the skin. In addition, the emergence of the ‘superbug’ Methicillin‐resistant S. aureus (MRSA) has made the treatment of these infections more challenging. Antimicrobial peptides (AMPs), due to their potent antimicrobial activity, limited evidence of resistance development, and potential immunomodulatory effects, have gained increasing attention as potential therapeutic agents for atopic dermatitis. In this study, we report a novel AMP, brevinin‐1E‐OG9, isolated from the skin secretions of Odorrana grahami, which shows potent antibacterial activity, especially against S. aureus. Based on the characteristics of the ‘Rana Box’, we designed a set of brevinin‐1E‐OG9 analogues to explore its structure–activity relationship. Brevinin‐1E‐OG9c‐De‐NH2 exhibited the most potent antimicrobial efficacy in both in vitro and ex vivo studies and attenuated inflammatory responses induced by lipoteichoic acid and heat‐killed microbes. As a result, brevinin‐1E‐OG9c‐De‐NH2 might represent a promising candidate for the treatment of S. aureus skin infections.

dysregulated by the microbes and inflammatory molecules present in AD lesions. [7][8][9] Recently, research on human HDPs indicate they have some potential advantages for AD treatment, such as their broad spectrum antibacterial action and their immunomodulatory effects, making them a promising candidates in AD patients' management. 6 Amphibians, found in both terrestrial and aquatic environments rich in microbes, are considered excellent natural sources of HDPs/ AMPs. [10][11][12] From their skin secretions, some HDPs with great pharmacological values have been identified and developed to treat skin infections. For example, Magainin, from Xenopus laevis, has been modified and studied for their potential to treat infected diabetic foot ulcers. 13,14 Brevinins are an influential group of AMPs from frog skin secretions, which have been found in various species of frogs, and which display potent broad spectrum antimicrobial activity, including activity against antibiotic-resistant strains. Brevinins have also been found to display antibiofilm, anticancer, immunomodulatory and wound healing activities. These bioactivities make brevinins stand out in the crowd of the multitudinous AMPs. [15][16][17][18][19] However, studies using brevinins against skin infections are limited. Therefore, in this study, we characterised, modified and explored the antibacterial potential of a novel brevinin-1 AMP, brevinin-1E-OG9, using a porcine skin infection model. This novel peptide was identified from the skin secretions of Odorrana grahami and exhibited potent antibacterial properties with test pathogens, especially gram-positive bacteria. As a characteristic part of brevinin-1 peptides, the role of the 'Rana Box' has been widely discussed in previous studies, although its functions still remain elusive. In our work, we designed a set of analogues of brevinin-1E-OG9 by modifying the cationicity or truncating the loop structure to explore further the function of the 'Rana Box' in this peptide.

| Acquisition of skin secretions from Odorrana grahami
Adult Odorrana grahami frogs were captured in Yunnan province, PRC. Skin secretion was obtained from the dorsal skin using mild electrical stimulation. The skin secretion was collected and lyoph-

| 'Shotgun' cloning of the novel peptide from Odorrana grahami skin secretion
Brevinin-1E-OG9 Precursor-Encoding cDNA was acquired from the cDNA library of Odorrana grahami using a NUP primer (nested universal primer; 5′-AAGCA GTG GTA TCA ACG CAGAGT-3′) and a sense primer (5′-GAWYY ARA GCC YAA ADATG-3′) (W = A or T, Y = C or T, R = A or G, D = A, G or T) that was designed to match highly conserved domains of the 5′-untranslated regions.

| Peptide synthesis, purification and characterisation
Brevinin-1E-OG9 and its analogues were synthesised through solidphase peptide synthesis by Tribute Peptide Synthesizer (Protein Technologies). The crude peptide was purified by employing Reverse-Phase High-Performance Liquid Chromatograph (RP-HPLC; Jupiter C-18, 250 × 10 mm, Phenomenex). Matrix-assisted laser desorption/ionisation-time of flight (MALDI-TOF) was employed to measure the molecular weight of the crude and purified peptides.

| Bioinformatics analysis
The multiple sequence alignment was performed using Clustal  20 The secondary structure of peptides was predicted using I-TASSER (https://zhang group.org/I-TASSE R/). 21 UCSF Chimera (version 1.15) was utilised to visualise predicted 3D models of peptides.

| Circular dichroism spectrum
The secondary structure of peptides was determined using a JASCO J-815 circular dichroism (CD) spectrometer (JASCO, UK), as mentioned previously. 11 The wavelength was set from 190 to 260 nm.
The scanning speed was 200 nm/min. The bandwidth and data pitches were 1 nm and 0.5 nm, respectively. Peptides (final concentration = 50 μM) were prepared in 10 mM NH 4 AC and 50% tetrafluoroethylene (TFE; v/v, in 10 mM NH 4 AC) to mimic water and membrane-like environments, respectively.

| Antimicrobial activity screening
Antimicrobial assays were conducted using standard broth microdilutions, as mentioned previously. 11

| Haemolytic activity
The haemolytic activity of peptides was conducted using horse erythrocytes (2%, v/v). Peptide solutions were mixed with washed horse erythrocyte suspension at 37°C for 2 h. Triton X-100 (0.1%) was used as a positive control. The mixture was then centrifuged at 930 g for 5 min. The absorbance of the supernatants was detected at 570 nm. The haemolysis rate was calculated using the formula: This experiment was performed in three independent runs.

| Kinetics time-killing
Bacteria in the log phase were diluted using corresponding broths and mixed peptide solutions (final concentration 1, 2, 4 × MIC). At different time intervals, 10 μL of the mixture was diluted in phosphatebuffered saline (PBS) and spotted on the agar plates. The bacterial colony number was counted after overnight incubation. The experiments were conducted in three independent runs.

| Lipoteichoic acid/lipopolysaccharide binding assay
The binding affinity of peptides to lipoteichoic acid (LTA) and lipopolysaccharide (LPS) was examined using the fluorescent dye BODIPY TR cadaverine (BC) (ThermoFisher) displacement assay. BC

| Outer membrane permeability assay
Escherichia coli ATCC 8739 at the logarithmic phase was washed with HEPES (containing 5 mM Glucose, pH 7.4) and diluted to OD 600 = 0.5.
Peptide solutions were mixed with bacterial suspension and incubated at 37°C for 2 h. Then, the fluorescent dye N-phenyl-1-naphthylamine (10 μM) was added to each well. The fluorescence was recorded at excitation 350 nm/emission 420. Melittin (10 μM) was set as a positive control. The experiments were conducted in three independent runs.

| Cytoplasmic membrane permeability assay
The bacterial cytoplasmic membrane permeability of peptides was detected using the fluorescent dye SYTOX Green (ThermoFisher). 23 Bacteria were incubated with peptide solutions for 2 h and the mix-

| Cytoplasmic membrane potential
The impact of peptides on the bacterial membrane potential were de-

| Live/Dead staining for the evaluation of the effect of peptides on bacteria
Bacteria were mixed with peptide solutions for 1 h at 37°C. After centrifuged at 3000 g for 15 min, bacterial cell pellets were resuspended and stained with propidium iodide (PI; Sigma) for 15 min.
Unbound PI subsequently was washed away with PBS, and bacterial cells were further stained with SYTO 9 (ThermoFisher) for another 15 min. The stained bacteria were observed under Leica DMi8 fluorescence microscopy (Leica) using the 100 × oil-immersion objective.

| Biofilm inhibition assay
Peptide solutions were first mixed with bacterial cultures (5 × 10 5 CFU/mL) and incubated at 37°C for 24 h. The planktonic bacterial cells were gently removed by washing with PBS the next day, and methanol was utilised to fix the biofilm. Crystal violet (0.1% v/v), PBS and acetic acid (30%, v/v) were subsequently used to stain biofilm, remove remaining stains and re-dissolve stains, respectively.
The absorbance was determined at 595 nm. The experiments were conducted in three independent runs.

| Biofilm eradication assay
The bacterial culture (5 × 10 5 CFU/mL) was cultured in a sterile 96well plate for 24 h to generate mature biofilms. The planktonic bacteria were removed by washing three times with sterile PBS buffer.
Then, fresh broth with different peptide concentrations was added to each well and further incubated for another 24 h. Then, the plate was stained with crystal violet. The stain was solubilised in 30% acetic acid, and the absorbance was measured at 595 nm. The experiments were conducted in three independent runs.

| Anti-persister cells assay
The antibacterial activity of peptides against persister cells was per-

| Detection of biofilm components using a fluorescent microscope
The variations of components in the MRSA NCTC 12493 biofilm stained at 1/4 and 1/2 MIC of OG9 and OG9c-De-NH 2 were detected, as mentioned previously. 25 Bacteria (5 × 10 5 CFU/mL) were first inoculated in a 24-well plate whose bottom of each well was covered with a sterile glass coverslip (13 mm, VMR) and treated with sub-MIC concentrations of peptide solutions. On the next day, liquid in each well was removed, and PBS was utilised to wash to clean planktonic cells. Three fluorescent dyes, DAPI, SYPRO Ruby (ThermoFisher) and WGA-488 (ThermoFisher), were used to stain nucleic acids, proteins and polysaccharides in EPS, respectively. The stained biofilms were observed under Leica DMi8 fluorescence microscopy (Leica) using the 20 × objective.

| MTT assay
The cytotoxicity of peptides against human keratinocyte cells (HaCaT) was studied using the MTT assay, as reported previously. 26 HaCaT cells (1 × 10 4 cells/mL) were seeded into a 96-well plate with After that, cells were treated with a corresponding serum-free culture medium for 4 h. Cells were treated with peptides in concentrations ranging from 25 to 100 nM. Then, 10 μL of MTT reagent (5 mg/ mL) was added to each sample after overnight incubation and cells were further incubated at 37°C for another 4 h. All liquids were removed, and 100 μL of dimethyl sulfoxide was added. The absorbance of each sample was at 570 nm. The experiments were conducted in three independent runs.

| Ex vivo skin infection model
The

| Reactive oxygen species detection
The anti-inflammatory capacity of peptides was detected by measuring the production of reactive oxygen species.

| Statistical analysis
Prism (Version 9.0; GraphPad Software Inc.) was used to analyse data and generate diagrams. The standard error of the mean (SEM) for each set of data in three replicates from three experiments was shown by the error bars surrounding the mean data points. Oneway anova analysis, with multiple comparisons of the mean of each column with the mean of every other column, was used to evaluate the statistical significance (*p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001).

| The discovery and identification of brevinin-1-OG9 by 'shotgun' cloning
The

| Database-based peptide design and characterisation of OG9 and its analogues
Brevinin-1 peptides display broad spectrum antimicrobial activity and has been speculated that this is due to the interaction of their α-helical structures with the anionic lipid bilayers of bacterial membranes. 18 The significant role of the 'Rana Box' in the brevinin-1 family and other peptide families has been examined previously. 29 However, the role of the cationic residues, especially in the 'Rana Box', has not been examined, although the positively charged residues are thought to play a significant role in attracting the anionic charged bacteria cell membranes. 30 Therefore, here we examined the cationic residues of the 'Rana Box' in the following function-structure relationship studies.
The exploration of cationic residues in the 'Rana Box' was based on the frequency of residues in all brevinin-1 sequences in the Uniport database. The amino acid frequency results aided in the selection of second-best residues or in avoiding changes to conserved sites.  Figure S1).

| Conformational analysis of OG9 and its analogues
The prediction results showed that most analogues inherited the alpha-helical structure from the parent peptide, OG9 (Figure 2A-C and Figure S2). CD results further confirmed that all analogues adopted an alpha-helical structure in 50% TFE solution, except for OG9-De and OG9-De-NH 2 , which had fewer positive charges compared with other analogues ( Figure 2D). The α-helical degrees of OG9 and its analogues were predicted with K2D3 online server (Table 2).

| Haemolysis of OG9 and its analogues
OG9 shows obvious haemolytic activity in a concentrationdependent manner (Figure 3)  with other similar peptides. The red letters mean nonpolar amino acids with hydrophobic residues; the green letters mean polar amino acids with hydrophilic residues; the blue letters mean acidic amino acids with negatively charged; the pink letters mean essential amino acids with positively charged. The number at the end of each line represents the number of amino acids in the sequence. The '.' and ':', respectively, indicate the low and high similarity of amino acid residues in each row. The asterisks indicate the identical amino acid residues in each sequence; (C) Frequency of existing amino acids in all the brevinin-1 peptides based on the UniProt database. The 'Rana-Box' was shown in red.
However, the amidation of OG9c-De improved the haemolytic activity of OG9c-De-NH 2 , especially when the concentration was above 64 μM.

| The screening of antimicrobial activity of OG9 and its analogues
OG9 showed a broad spectrum antibacterial activity, but no an-

| Saline tolerance of OG9 and OG9c-De-NH 2
To assess the tolerance of peptides under the background of salt   ( Figure 5A,B). Based on this result, we further evaluated the possible disruptive effects of these two peptides on the outer or/and intracellular membrane of bacteria. OG9 showed an outer membranedisruptive impact in a concentration-dependent manner, while the permeability rate of OG9c-De-NH 2 was kept at around 65% under different test concentrations ( Figure 5C). Subsequently, the impact of peptides on the intracellular membrane of bacteria was further evaluated with a SYTOX Green staining assay. Both test peptides showed concentration-dependent effects on the intracellular membrane of S. aureus ATCC 6538 and E. coli ATCC 8739 ( Figure 5D,E).

F I G U R E 2
Characterisation of OG9 and OG9c-De-NH 2 . (A) Three-dimensional models of OG9 and OG9c-De-NH 2 ; (B) helical-wheel projections of peptides analysed using HeliQuest server. Amino acids with a positive charge are shown in blue, while yellow represents hydrophobic amino acids. Alanine and glycine are in grey, proline is in proline and glutamine is pink; (C) Electrostatic potential surface of peptide OG9 (left) and OG9c-De-NH 2 (right); (D) CD spectra of OG9 and its analogues in 10 mM NH 4 AC solution and 50% TFE/10 mM NH 4 AC solution.
TA B L E 2 Estimated α-helical degrees of OG9 and its analogues in 10 mM NH 4 AC solution and 50% TFE/10 mM NH 4 AC solution.  Figure 5F). The electric potential (ΔΨ), as one of the essential parts of bacterial transmembrane potentials, plays a critical role in fundamental cellular functions. 33 Results shown in Figure 5G indi-  (5). The data shown are the means ± SEM of three independent experiments using three replicates. The significance is indicated by ****p < 0.0001.

| 1577
FEI et al. The data are derived from three independent experiments and shown as the means ± SEM. The significance is indicated by ****p < 0.0001, ***p < 0.001, **p < 0.01 and *p < 0.05. higher than their MBIC values (Table S1, Figure 6A,B showed potent anti-persister effects compared to OG9, and most persisters were eradicated at 1 μM (1/2 MIC) ( Figure 6C). The effects of OG9 and OG9c-De-NH 2 on extracellular DNA (eDNA), protein and polysaccharide formation in MRSA NCTC 12493 biofilms were also investigated. As shown in Figure 6D,E, the presence of sub-MIC concentrations of OG9 and OG9c-De-NH 2 markedly reduced the levels of these compounds compared to untreated growth controls. In particular, 1/2 MIC OG9c-De-NH 2 (1 μM) resulted in a near complete loss of these three components, indicating its potent antibiofilm capacity.

| Ex vivo antibacterial activity of OG9 and OG9c-De-NH 2
As shown in Figure 6F,

| Inhibition on inflammation response in human keratinocytes HaCaT cells
High numbers of S. aureus in skin of AD patients often induce skin inflammation. As key signalling indicators, ROS play a critical role in the pathogenesis of inflammation. Herein, we employed human keratinocytes HaCaT cells and measured the effects of peptides on the production of ROS in the response of LTA and heat-killed MRSA.
OG9c-De-NH 2 exhibited more potent inhibitory effects on the production of ROS induced by LTA or heat-killed microbe compared to the parent peptide ( Figure 7).

| DISCUSS ION
Antimicrobial peptides, due to their potent antimicrobial capacity and low possibility of developing resistance, have gained increasing attention and been given great expectations as novel antibacterial agents to cover the shortage of effective antibiotics. 34,35 Herein, we identified a novel brevinin-1 type AMP, OG9, and designed a set of analogues to study its structure-activity relationship.
OG9c-De-NH 2 , which had the most potent antibacterial activity, was selected to further study the potential of these peptides for use in S. aureus-induced skin infections.
Modification of the OG9 'Rana Box' indicates that the loop structure does play an important role in the antibacterial activity of this peptide, and this is more likely to be contributed by the cationic residues within the loop, since the truncation of the 'Rana Box' after removing two cationic residues out did not deprive the antibacterial activity of OG9c-De and OG9c-De-NH 2 . The haemolysis results indicate that it might be the loop structure, rather than the cationic residues inside the 'Box' that affect the haemolytic activity of brevinin-1 s.
OG9c-De-NH 2 showed potent binding with LTA and LPS and can significantly induce membrane permeabilisation of bacteria. However, given the low killing rate, OG9c-De-NH 2 might not kill bacteria via rapid lysis, although its ability to cause membrane permeability may contribute to antimicrobial efficacy. In addition, OG9c-De-NH 2 exhibited potent antibiofilm activity and inhibitory effect on ROS production stimulated by LTA or heat-killed microbe.
Overall, our studies identified a novel AMP and developed an analogue with potential as an antimicrobial agent.

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.