A total of 1720 proteins were retrieved from NCBI Identical Protein Groups (IPG) by applying the following filters: Search Streptomyces Filters: RefSeq; Prokaryotes; >100. Only the sequences showing hits in ≥ 100 Streptomyces genomes were retained for further analysis. Ribosomal proteins and sequences of ≥100 aa were discarded. Protein sequences for the resulting 125 IPGs were clustered using mcl (v14-137) (Li et al., 2003) to reduce the dimension of search space. The analysis yielded 44 clusters, three of which were further separated into two subclusters based on manual inspection, yielding 47 total clusters ( Supplementary Table 1 ). The prediction of potential AMPs was performed using Antimicrobial Peptide Scanner (v2), a prediction tool implemented in the database of antimicrobial activity and peptide structure (DBAASP v3.0), and AMPDiscover (Veltri et al., 2018; Pinacho-Castellanos et al., 2021; Pirtskhalava et al., 2021). Nucleotide sequences belonging to each subcluster were aligned by MUSCLE (v5.1) (Edgar, 2021) using default parameters. Hidden Markov model (HMM) sequence profiles were built and used to search 2061 Streptomyces genomes and 761 selected Actinobacterial genomes ( Supplementary Tables 2, 3 ) (Eddy, 2011; Wheeler and Eddy, 2013). In visualizing the results of homology searches, a phylogenomic tree based on whole proteome sequences was created by using CVTree (Qi et al., 2004). The K-tuple length was set to six, as determined in the previous study (Zuo et al., 2010). The distribution of protein homology was visualized by a phylogenomic tree using Graphical Phylogenetic Analysis (GraPhlAn v1.1.4) (Asnicar et al., 2015).

The PacBio reads were assembled by RS HGAP Assembly (v3.0; Pacific Biosciences; https://www.pacb.com/products-and-services/analytical-software/smrt-analysis/). Pilon (v1.21) (Walker et al., 2014) was used to polish the draft assembly with the filtered Illumina reads. The quality of gene prediction was evaluated using Benchmarking Universal Single-Copy Orthologs (BUSCO v5.3.2; actinobacteria_phylum_odb10 dataset) (Manni et al., 2021). The prediction of protein-coding and RNA genes was performed using Prokka (v1.13) (Seemann, 2014) and RNAmmer (v1.2) (Lagesen et al., 2007), respectively. Visual representation of genomic features, including gene prediction was performed using Circos (v0.69-9) (Krzywinski et al., 2009). The assignment of Clusters of Orthologous Groups (COGs) for the predicted genes was conducted using eggNOG-mapper (Cantalapiedra et al., 2021).

The 16S rRNA gene sequence (1513 bp) retrieved from the genome was analyzed using the EZBioCloud 16S database (Yoon et al., 2017). Genomic relatedness indices, including digital DNA–DNA hybridization (dDDH) and average nucleotide identity, were calculated using the recommended settings of the Genome-to-Genome Distance Calculator (GGDC v3.0) (Meier-Kolthoff et al., 2013; Meier-Kolthoff et al., 2022) and Orthologous Average Nucleotide Identity Tool (v1.40) (Lee et al., 2016), respectively. Genomic relatedness was visualized in a scatter plot using ggplot2, ggExtra, and ggthemes packages (Wickham, 2016; Arnold, 2021; Baker, 2022) in R (v4.1.2) (Team, 2021).

Based on the identification pipeline for potential AMPs, peptide KPP03845_1_03451 named SHC-AMP was selected and then amplified by PCR using DNA of strain KPP03845 as a template and the following set of primers that contain EcoRI and XhoI restriction sites for subsequent insertion into the pGEX-4T-1 plasmid (Addgene, Cambridge, MA): 3451F (5′-AAGAATTCGTGGGCTCTGTTATCAAG-3′) and 3451R (5′-AACTCGAGTTACTTCTTGTTACGGCG-3′). Expand High Fidelity enzyme mix (Roche Diagnostics, Mannheim, Germany) was added to the reaction. Amplification was performed with an initial denaturation step at 95°C for 5 min, followed by 25 cycles of denaturation at 94°C for 1 min, annealing at 60°C for 1 min, extension at 72°C for 1 min and 30 s, and a final extension at 72°C for 3 min. The amplified peptide SHC-AMP was cloned into the pGEX-4T-1 plasmid vector, and then the recombinant plasmid was transformed into E. coli strain BL21 (DE3) (DYNE BIO, Seoul, Korea). For protein expression, the cell was cultured in Luria–Bertani (LB) broth (Difco Laboratories, Detroit, Mich.) containing ampicillin at 37°C overnight in a shaking incubator at 200 rpm. Twenty milliliters of overnight culture was transferred into 2 L of LB broth containing 100 µg mL⁻¹ of ampicillin. Isopropyl-β-D-1-thiogalactopyranoside (Sigma Chemical Co., St Louis, MO, USA) was added to the culture to a final concentration of 1 mM during the late log phase (OD_{600 =} 0.6–0.8). Then, the culture was shaken further for 16 h at 18°C. After harvesting by centrifugation at 6900 ×g for 20 min, the cells were resuspended in buffer A containing 20 mM Tris-HCl (pH 8.0) and 100 mM NaCl and then disrupted by sonication. Cell debris was removed by centrifugation at 9500 ×g and 4°C for 1 h. The supernatant was loaded onto a glutathione-Sepharose column (GE Healthcare) equilibrated with buffer A for purification. After the column was washed with buffer A, the protein binding to the column was eluted using elution buffer containing 20 mM Tris-HCl (pH 8.0) and 20 mM reduced glutathione (Sigma Chemical Co., St Louis, MO, USA). Eluted protein was cleaved with thrombin at 4°C for 16 h. The eluted fractions (flow through, wash with buffer A, GST-KPP03845_1_03451, GST-cleaved KPP03845_1_03451 and GST) were separated using the SDS-PAGE on 15% gel (data not shown). The protein expression was confirmed by the size of the SDS-PAGE gel. The expressed KPP03845_1_03451 protein (69.6 µg) was examined for antifungal activity against F. oxysporum f. sp. lycopersici. The KPP03845_1_03451 amino acid sequence was synthesized as ¹MGSVIKKRRKRMAKKKHRKLLKRTRVQRRNKK³² (Peptron, Daejeon, Republic of Korea). The synthesized protein sequence was determined using a high-performance liquid chromatography (HPLC) system (SHIMADZU Prominence HPLC System) equipped with Shiseido capcell pak C18 (4.6 × 50 mm, 5 μm, 120 Å) and mass spectrometer (SHIMADZU LCMS-2020 system). Solvent A consisted of water containing 0.1% trifluoroacetic acid, and solvent B consisted of acetonitrile containing 0.1% trifluoroacetic acid. The HPLC analysis was conducted at a flow rate of 1 mL/min using a linear gradient elution of 3–10% B (2 min), 10–40% B (10 min) and 40–60% B (1 min). Synthesized protein was monitored at 220 nm.

Three-dimensional structure of the peptide, SHC-AMP, was modelled by using the SWISS-MODEL homology modeling server (https://swissmodel.expasy.org) (Waterhouse et al., 2018). The amino acid sequence of SHC-AMP was graphically represented by using iCn3D (Wang et al., 2020).

The MIC values of the synthesized peptide were evaluated against various plant pathogens using the modified CLSI protocol in a 96-well plate (Rex, 2008). Then, 25 µL of conidial suspensions (final concentration of 4 × 10⁵ spores mL^-1) of Aspergillus oryzae, Alternaria brassicicola, Botrytis cinerea, Colletotrichum orbiculare, F. oxysporum f. sp. lycopersici, and Rhizopus stolonifer var. stolonifer or bacterial suspensions (final concentration of 4 × 10⁵ CFU mL^-1) of Erwinia carotovora subsp. atroseptica BAA672, E. carotovora subsp. carotovora ATCC39048, Pseudomonas syringae pv. tomato DC3000, and Xanthomonas campestris pv. vesicatoria Ds1 were added to 96-well plates (SPL life Sciences, Pocheon, Korea), followed by the addition of 49 µL of distilled water. The growth medium contained 25 µL of 4 × PDB or 4 × LB for fungi and bacteria, respectively. Subsequently, the synthesized peptide was added to each well at final concentrations of 0.5–512 µg mL^-1. MICs of synthesized peptide were determined as the lowest concentration that caused complete growth inhibition by visual examination after 48 h.

Bacterial growth of Pst DC3000 in response to SHC-AMP treatment was confirmed in Col-0 as described in the previously reported method with minor modifications (Tornero and Dangl, 2001). Pst DC3000 suspension (1 × 10⁵ CFU mL^-1) was inoculated into 5-week-old A. thaliana leaves using needleless syringe, until the suspension is spread to whole leaf. A series of concentrations of SHC-AMP (32 μg, 64 μg, and 128 μg mL^-1) were infiltrated 1 day post-Pst DC3000 infection in the same way with Pst DC3000. Two and four days after bacterial infection, four leaf discs (6 mm in diameter) were collected for each sampling with three replicates to measure bacterial growth. In case of 0 dpi, leaves were collected after water soaking area disappeared. Leaf discs were ground in homogenizer with 300 μl of 10 mM of MgCl₂. 700 μl of distilled water was added to meet 1 mL, and ten-fold serial dilution was performed with distilled water. Diluted samples (10 μl) were spotted on King’s B agar plate supplemented with rifampicin and cycloheximide at a final concentration of 25 μg mL^-1. After culturing bacteria in 28°C incubator overnight, the number of CFU was counted for each sample. Disease damage was calculated as damaged leaf area/total leaf area × 100%. The effect of SHC-AMP treatment in Pst DC3000 infected plant leaves was analyzed by one-way analysis of variance (ANOVA) using SAS 9.4 (SAS Inst., Cary, NC, USA). When ANOVA indicated significance, means were separated using Least Significant Difference (LSD) test at α = 0.05.

An identification pipeline was constructed to search for potential antimicrobial proteins that are highly conserved in the genus Streptomyces. The NCBI IPG database was used as the initial database to identify candidate sequences. A total of 1720 protein sequences were found in more than 100 prokaryotic genomes, including those belonging to Streptomyces in the RefSeq database. Of these protein sequences, only 375 were found in more than 100 Streptomyces genomes, yielding 272 IPGs after filtering out ribosomal proteins. Ribosomal proteins were identified on the basis of protein annotation from the NCBI Gene database and discarded because of their conserved structure and function during evolution (Timsit et al., 2021). Only sequences shorter than 100 amino acids (aa) in length were retained for further analysis because the majority (97.54% or 3089/3167) of AMPs in the Antimicrobial Peptide Database (APD3; https://aps.unmc.edu/downloads; last accessed on July 18, 2022) (Wang et al., 2016) were shorter than 100 aa in length. AMP prediction was performed using three recently developed tools, resulting in 125 sequences (Veltri et al., 2018; Pinacho-Castellanos et al., 2021; Pirtskhalava et al., 2021). Of the 125 sequences, 47 were predicted to be AMPs by at least one predictor, 11 sequences by two predictors, and one sequence by all three ( Supplementary Table 4 ). The 47 proteins belonged to 17 clusters (19 subclusters) based on protein clustering analysis of the 125 sequences. Subsequently, HMM sequence profiles built for the 19 subclusters were used to scan 2062 Streptomyces genomes and 762 selected Actinobacterial genomes. The homology distribution showed intricate patterns, showing wide degrees of sequence conservation ( Figure 1 , Supplementary Figure 1 ). Only few genes encoding highly homologous proteins were functionally characterized, playing regulatory roles in sporulation, morphology, and antibiotic production ( Supplementary Table 5 ). Cluster43, representing a SHC-AMP, was selected for experimental validation because significant sequence similarities were found in 2054 out of 2062 Streptomyces genomes, and all of the three tools were predicted it to be an AMP.

Several reports have been found novel AMPs identified in Streptomyces strains, all of which exhibited strong antimicrobial activity (Vasilchenko et al., 2020; Zhang et al., 2020). Meanwhile, Streptomyces strain KPP03845 was isolated from soil, and its high degree of antifungal activity against plant pathogenic fungi was demonstrated ( Supplementary Figure 2 ). The culture filtrate (2 mL) of strain KPP03845 showed inhibition zone diameters of 35 mm, 25 mm, 28 mm, 24 mm, 30 mm and 32 mm for A. brassicicola, A. oryzae, C. gloeosporioides, C. orbiculare, F. oxysporum f. sp. lycopersici and F. oxysporum f. sp. lycopersici, respectively. The culture filtrate did not show a clear inhibition zone against R. stolonifer var. stolonifera. Thus, the strain KPP03845 was sequenced for use in comparative genome analysis and as a genetic template for heterologous expression and peptide synthesis.

The complete genome of the strain KPP03845 was assembled in a single chromosome of 8204848 bp with 72.31% GC content. In addition, two plasmids were assembled into molecules of 289669 and 101118 bp ( Table 1 , Figure 2 ). A total of 6262 protein-coding genes were assigned with COGs, yielding 7018 hits ( Supplementary Table 6 ). In addition to gene matches with unknown functions, the top five largest groups included genes involved in transcription (859), amino acid transport and metabolism (680), signal transduction mechanism (518), carbohydrate transport and metabolism (487), and inorganic ion transport and metabolism (390) ( Supplementary Table 6 ).

The 16S rRNA gene of the strain KPP03845 (1513 bp) showed 100% similarity spanning 1447 bp with that of Streptomyces xanthophaeus NRRL B-5414^T, S. nojiriensis LMG 20094^T, and S. spororaveus LMG 20313^T. In addition, 11 other strains had similarity greater than 99%, hindering accurate species identification ( Supplementary Table 7 ). Indices of genomic relatedness, dDDH and OrthoANI, were calculated for 2061 Streptomyces genomes ( Supplementary Table 2 ) to identify the KPP03845 strain. The widely accepted criteria for species demarcation (Auch et al., 2010; Lee et al., 2016) were applied to the calculations. Two S. xanthophaeus genomes with OrthoANI values greater than 98% and dDDH values of 83.5% were identified ( Supplementary Figure 3 , Supplementary Table 2 ). Collectively, the strain KPP03845 was identified as S. xanthophaeus.

Based on the prediction results from the identification pipeline, the KPP03845_1_03451 gene was selected to test for antimicrobial activity. The full-length open reading frame of KPP03845_1_03451 (M1 to K32) from the gDNA of S. xanthophaeus strain KPP03845 was cloned into the expression vector, pGEX-4T-1, between the EcoRI and XhoI restriction enzyme sites. Then, the recombinant protein was expressed in E. coli BL21 (DE3). After the purification of the protein by affinity chromatography using glutathione-Sepharose, 69.6 µg of the purified protein was shown to have antifungal activity against F. oxysporum f. sp. lycopersici ( Figure 3 ). In addition, 300 µg of the synthesized peptide SHC-AMP exhibited antifungal activity against F. oxysporum f. sp. lycopersici ( Figure 3 ).

The antimicrobial activity of the synthesized SHC-AMP was evaluated against various plant pathogens by using modified CLSI microdilution method. The synthesized peptide exhibited antimicrobial activity against several plant pathogenic bacteria and fungi used in this assay ( Table 2 ). The growth of X. campestris pv. vesicatoria Ds1 and P. syringae pv. tomato DC3000 was inhibited at 256 and 128 µg mL^-1, respectively, whereas the antibacterial activity against E. carotovora subsp. atroseptica BAA672 and E. carotovora subsp. carotovora was not observed. The synthesized peptide inhibited the mycelial growth of A. brassicicola, B. cinerea, and F. oxysporum f. sp. lycopersici at concentrations of 32 to 128 µg mL^-1, whereas the mycelial growth of A. oryzae, C. orbiculare, or R. stolonifer var. stolonifer was not inhibited even at a concentration of 512 µg mL^-1.

The synthesized SHC-AMP was evaluated with regard to its ability to suppress Pst DC3000 infection in A. thaliana leaves. The symptoms developed at 2 days after bacterial infection in the SHC-AMP-untreated control. Four days after infection, A. thaliana leaves treated with the SHC-AMP at a concentration of 64 µg mL^-1 exhibited attenuated Pst DC3000-infected symptoms compared to the untreated control ( Supplementary Figure 4 ). However, A. thaliana leaves treated with the 32 µg mL^-1 and 128 µg mL^-1 were observed to have similar symptoms of chlorosis and necrosis to the control leaves at 4 days after infection. A. thaliana leaves treated with the 0 µg mL^-1, 32 µg mL^-1, 64 µg mL^-1, and 128 µg mL^-1 exhibited infection damage of 36.07%, 29.38%, 1.21%, and 26.59%, respectively. For further confirmation of the reduced Pst DC3000 growth in the SHC-AMP-treated A. thaliana leaves, the number of bacterial colonies grown in planta was monitored ( Figure 4 ). The number of colonies in the SHC-AMP-untreated leaves were observed to be 39 × 10⁵ colony-forming units (CFU)/cm² and 93 × 10⁵ CFU/cm² on average at 2 days and 4 days after inoculation of Pst DC3000. Compared to the SHC-AMP-untreated control at 2 and 4 days after inoculation, bacterial growth was reduced in leaves treated with 64 µg mL^-1 of the peptide. Treatment with 64 µg mL^-1 of the SHC-AMP reduced the bacterial growth of Pst DC3000 by 8.54 and 2.82 times compared to control, respectively. Initially, treatment with 32 µg mL^-1 of the SHC-AMP suppressed the growth of Pst DC3000, but suppression of bacterial growth was not observed at 4 days after infection. The treatment with 128 µg mL^-1 of the SHC-AMP resulted in a decrease in bacterial growth until 4 days after inoculation, but statistically significant results were not obtained. Thus, the most effective concentration of SHC-AMP was determined to be 64 µg mL^-1 with less necrotic symptoms.