Comprehensive Evaluation of Safety and Probiotic Potential of Lactiplantibacillus plantarum HY7715 Isolated from Korean kimchi Authors: Ju–Eun Lee, Tae–Hyun Lim, Ji–Hyun Kim, Yong–Tae Kim, Jae Jung Shim, Jae–Hwan Lee Published: 2026 ABSTRACT Lactiplantibacillus plantarum HY7715, isolated from Korean kimchi, has been previously reported to overproduce riboflavin in vitro and in vivo, indicating its functional potential as a probiotic candidate. Given this characteristic, a comprehensive safety and genomic evaluation is required prior to application. In this study, we assessed the safety, genomic stability, and probiotic–related properties of HY7715 through whole–genome sequencing, comparative genomics, and phenotypic analyses. The genome (3,348,486 bp; 44.29% GC content) was assembled into seven contigs and encoded 3,224 protein–coding sequences, of which 2,933 functionally annotated. No virulence factors or acquired antimicrobial resistance genes were detected, and mobile genetic elements did not harbor undesirable determinants. Although phenotypic resistance to kanamycin was observed, the absence of corresponding resistance genes indicates intrinsic resistance. HY7715 also exhibited γ-hemolysis and lacked genes associated with cytotoxic secondary metabolites or biogenic amine production. The rib operon was identified in the genome, and in silico analysis of the FMN riboswitch region revealed a unique nucleotide substitution potentially affecting regulatory conformation, which may contribute to the riboflavin–overproducing phenotype. Comparative pangenome analysis of 60 L. plantarum genomes revealed that HY7715 shares a conserved genomic backbone dominated by core and soft–core genes. Genes associated with stress tolerance, adhesion, carbohydrate utilization, and bacteriocin production were identified, consistent with probiotic–related traits. Collectively, these findings indicate that HY7715 is a genomically stable and safe strain with regulatory features supporting its functional probiotic potential.Supplementary InformationThe online version contains supplementary material available at 10.1007/s12602-026-10972-5. Introduction Probiotics are defined by the Food and Agriculture Organization (FAO) as “live microorganisms that, when administered in adequate amounts, confer a health benefit on the host” [1]. They exert their beneficial effects primarily by modulating the gut microbiota, enhancing intestinal barrier integrity, and regulating host immune responses [2–4]. In recent years, extensive clinical and mechanistic evidence has demonstrated their positive influence on multiple health domains, including improvement of gut symptoms in irritable bowel syndrome [5], metabolic modulation in type 2 diabetes and non–alcoholic fatty liver disease [6, 7], as well as adjunctive effects on dermatologic, oral, and mental health conditions [8–10]. However, probiotic effects are strain–specific and context–dependent. Moreover, probiotic use may be critical in vulnerable populations, such as preterm infants, immunocompromised patients, and critically ill individuals, as rare cases have been reported of probiotic–associated bacteremia or sepsis [11, 12]. Accordingly, current regulatory frameworks such as the European Food Safety Authority (EFSA) Qualified Presumption of Safety (QPS) system emphasize that probiotic safety must be demonstrated at the strain level [13]. Lactiplantibacillus plantarum is a gram-positive, non-spore-forming, non–motile, microaerophilic, and mesophilic bacterium [14]. The species is frequently isolated from diverse ecological niches, including gastrointestinal and vaginal tracts, meat, fish, fermented vegetables, and dairy products, reflecting its broad ecological adaptability [15–17]. L. plantarum is also typical lactic acid bacteria (LAB) found in Korean kimchi, which is known a main player in the late stage of kimchi fermentation [18]. It is among the most extensively characterized and commercially utilized probiotic species, supported by growing evidence of its beneficial effects on host health and technological robustness in food fermentation [19–21]. Owing to its long history of safe use, L. plantarum is included in the EFSA’s QPS list, which establishes a presumption of safety at the species level [13]. Moreover, multiple L. plantarum strains have received Generally Recognized as Safe (GRAS) status from the U.S. Food and Drug Administration (FDA), confirming their suitability for use in foods and dietary supplements. Probiotic candidates are commonly evaluated based on their capacity to survive gastrointestinal conditions, adhere to intestinal epithelial cells, and exhibit acceptable antibiotic susceptibility and non–hemolytic behavior. Additionally, in order to select competitive probiotics, functional characteristics need to be identified. Whole–genome–based bioinformatics analysis is widely used to quickly and efficiently screen for safe and functionally beneficial probiotic strains [22]. Specific probiotic traits can be investigated through functional gene prediction [23–27], including cell membrane composition, epithelial adhesion, gastric and bile salt tolerance, pathogenicity, hemolytic activity, bacteriocin production. In a previous study, it was shown that L. plantarum HY7715, a strain isolated from Korean kimchi, has a riboflavin–overproducing ability through in vitro and in vivo analysis [28]. Building upon the beneficial functional trait, the present study aimed to obtain a comprehensive evaluation of the strain’s safety and probiotic potential via in silico and in vitro analyses. We confirmed the strain’s genomic and phenotypic safety based on the EFSA guidelines [13, 29]. Based on complete whole–genome sequence, in silico screening of adverse genetic elements, including genes encoding antimicrobial resistance, virulence factors, biogenic amine production, secondary metabolites, and mobile genetic elements potentially involved in the horizontal transfer of undesirable genetic determinants, was conducted. Through functional annotation and bioinformatic analyses, functional advantages of HY7715 were also investigated. Additionally, to characterize the probiotic features of the strain, including carbohydrate utilization, acid and bile resistance, antimicrobial susceptibility, biogenic amine production, and hemolysis, in vitro analyses were conducted. Materials and Methods Bacterial Strains and Culture Conditions L. plantarum HY7715, isolated from Korean kimchi, was obtained from hy Co., Ltd. The strain was cultured in de Man Rogosa Sharp broth (MRS; Difco Corp.) at 37 °C for 15–20 h. The reference strains, L. plantarum KCTC 3108T, Staphylococcus aureus ATCC 33519 and Escherichia coli ATCC 25922 were obtained from the Korean Collection for Type Cultures (KCTC). L. plantarum KCTC 3108T was cultured in MRS broth (Difco Corp.) at 37 °C for 15–20 h. S. aureus ATCC 33519 and E. coli ATCC 25922 were cultured in tryptic soy broth (TSB; Difco Corp.) at 37 °C for 24 h. All bacterial strains were preserved in 10% (w/v) skim milk at − 80 °C. For whole–genome sequencing, L. plantarum HY7715 was first streaked on MRS agar (Difco Corp.) and incubated anaerobically at 37 °C for 48 h. A single fresh colony was sub–cultured twice on MRS agar to obtain pure isolate. A colony from the final plate was inoculated in MRS broth and incubated under the same conditions. The resulting culture was centrifuged at 3000 × g for 15 min, and the supernatants were discarded. The resulting cell pellets were stored at 4 °C until further use. Genome Sequencing, Assembly and Annotation The whole genome of strain HY7715 was sequenced by the CJ Bioscience Whole–Genome Sequencing Service (CJ Bioscience Inc.). Genomic DNA was extracted for long– and short–read sequencing using the Qiagen MagAttract HMW DNA kit (Qiagen) according to the manufacturer’s instructions. DNA quantity and quality were assessed using TapeStation 4200 system (Agilent Technologies) and the Qubit 2.0 fluorometer (Invitrogen). A minimum of 6 µg of genomic DNA with an OD260/OD280 ratio between 1.8 and 2.0 was used for sequencing. Long–read and short–read sequencing were performed using the PacBio Revio System (Pacific Biosciences) and the Illumina NovaSeq 6000 platform (Illumina), respectively. De novo assembly and annotation were performed by Senigen Whole-Genome Sequencing Analysis Service (Senigen). A short–read–first hybrid assembly was generated using Unicycler v0.5.0 [30]. Genome annotation was conducted using Prokka v1.14.6 with default settings [31]. The predicted protein coding sequences (CDSs) were functionally classified according to clusters of orthologous groups (COG) using EggNOG-mapper v2.1.12 (EggNOG v5.0.2; http://eggnogdb.embl.de) [32]. Clustered regularly interspaced short palindromic repeats (CRISPRs) and associated cas genes were detected using CRISPRCasFinder online tool (https://crisprcas.i2bc.parissaclay.fr/CrisprCasFinder/Index) with default settings [33]. Genome Contamination Analysis A taxonomic classification of raw reads obtained from Illumina and PacBio sequencing was performed using Kraken 2 (database version of 2024–01–12) with confidence threshold of 0.1 [34]. To evaluate potential contamination, ContEst16S (http://www.ezbiocloud.net/tools/content16S) was applied by extracting 16 S rRNA gene sequences from the assembled genome and conducting taxonomic classification [35]. Genome completeness and possible contamination were further assessed using UBCG (Up–to–date Bacterial Core Gene), which is based on 92 universally conserved single–copy bacterial core genes [36]. The resulting score ranges from 0 to 100, with a value of 100 indicating that presence of all 92 core genes. BUSCO (Benchmarking Universal Single–Copy Orthologs) v5.2.7 was also used to evaluate genome completeness [37]. The analysis was conducted in genome mode with the Bacteria_odb10 lineage dataset. Taxonomic Identification Taxonomic assignment of L. plantarum HY7715 was performed using 16 S rRNA gene sequence alignments and pairwise average nucleotide identity (ANI) calculations based on whole–genome sequence data from the test strain and selected reference species [38]. The 16 S rRNA gene sequence of HY7715 was compared with the those in the EzBioCloud database to identify the most closely related species [39]. Multiple sequence alignments were generated using the ClustalW algorithm implemented in MEGA12, and a maximum-likelihood phylogenetic tree was constructed with 1,000 bootstrap replications [40]. Evolutionary distances were calculated by the Jukes and Cantor model. ANI analysis between HY7715 and the genomes of ten species selected according to 16 S rRNA gene similarity was performed using the Orthologous ANI Tool (OAT) software, and a phylogenetic tree was constructed based on the resulting ANI values using the UPGMA clustering method [41]. Species-level identification was determined according to the EFSA guideline, using ANI threshold of > 94% [29]. Genome Stability Test The genomic stability of L. plantarum HY7715 was evaluated by comparing the whole–genome sequences of the mother culture and the 200th generation of the strain using ANI analysis. The whole–genome sequence of the test strain obtained in a previous analysis was used to represent the mother culture. The strain was inoculated into MRS broth at 1% (v/v) and incubated at 37 °C for 24 h to reach a cell concentration of approximately 109 cfu/mL, which was defined as one culture cycle. The cell concentration increased approximately 100–fold in each cycle, corresponding to about 6.64 generations of growth per cycle (log2 100 ≈ 6.64) [42]. To obtain the 200th generation, sequential subculturing was performed by transferring 1% (v/v) of the previous culture into fresh MRS broth and incubating under the same conditions. Whole–genome sequencing of the 200th generation was performed by Senigen Whole–Genome Sequencing Analysis Service (Senigen). Long–read and short–read sequencing were carried out using the Oxford Nanopore GridION platform with R10.4.1 flow cells (Oxford Nanopore Technologies) and the Illumina Nextseq 2000 platform (Illumina), respectively. Genome assembly and annotation were conducted as described above. In addition to ANI analysis, whole–genome alignment between the mother culture and the 200th generation assembly was performed using MUMmer v4.0 [43]. Pairwise alignment was conducted with the nucmer program, one–to–one best alignments were retained using delta-filuer (−1 option), and single nucleotide polymorphisms (SNPs) were identified using show-snps. Screening of Virulence Factor (VF) and Antimicrobial Resistance (AMR) Genes ABRicate v1.0.1 with the Virulence Factor Database (VFDB; version 2025–09–12) was used to identify virulence factor genes [44]. Antimicrobial resistance genes were detected using AMRFinderPlus v4.0.22 (database version of 2025–07–16.1) [45], ResFinder 4.1 [46], and ABRicate v1.0.1 with MEGARes (version 2025–09–12) [47]. AMR and VF screening was initially conducted using the minimum coverage threshold (≥ 50%) permitted by the software to allow comprehensive detection, as recommended for exploratory screening. However, the final interpretation of positive hits followed the EFSA guideline criteria (≥ 80% sequence similarity and ≥ 70% coverage). Mobile Genetic Elements (MGE) Analysis Plasmids were detected within the assembled genome using MOB–suite v3.1.9 [48]. To evaluate the potential horizontal transfer of AMR genes, MGEs were identified using MobileElementFinder v1.1.2 [49]. In addition, prophage regions were predicted with PHASTEST (version 2025–01–27) [50]. Protein–coding sequences located within predicted prophage regions were queried against the NCBI non–redundant (nr) database using protein BLAST (blastp) to evaluate potential horizontally acquired genes. Sequences exhibiting ≥ 80% amino acid identity and ≥ 80% query coverage to homologs derived from bacterial taxa other than L. plantarum were regarded as putative horizontal gene transfer (HGT) candidates, in accordance with previously reported approaches [51, 52]. Biosynthetic Gene Screening The biosynthetic potential of the genome was assessed by identifying secondary metabolite gene clusters (BGCs) using antiSMASH v8.0 (MIBiG 4.0 database) with default settings [53]. In antiSMASH v8.0, similarity is not expressed as raw percentages of homologous genes but in instead classified into three levels: “high” (≥ 0.75), “medium” (0.5–0.75), and “low” (0.15–0.5). Clusters with < 15% similarity are considered insufficiently conserved and are not displayed in the overview. Addition to antiSMASH analysis, BAGEL4 webserver (http://bagel4.molgenrug.nl/) was used to identify the type of bacteriocins [54]. The theoretical isoelectric point (pI) and molecular mass (MW) of identified peptides were computed using the Expasy Compute pI/MW online tool (https://web.expasy.org/compute_pi/). BlastKOALA search tool with the KEGG database (Release 115.1; https://www.kegg.jp/) was used to identify genes related to biogenic amine (BA) and riboflavin biosynthesis [55]. All identified genes and open reading frames (ORF) were further confirmed using the blastp against the nr database (http://blast.ncbi.nlm.nih.gov/Blast.cgi) [56]. Comparative genomic Analysis A total of 60 L. plantarum genomes, including HY7715, KCTC 3108T, and 58 publicly available genomes retrieved from NCBI, were included in the comparative analysis. Pairwise ANI values among the 60 strains were calculated using FastANI v1.34 with default parameters in an all–vs–all comparison [57]. The ANI matrix was converted to a distance matrix (100 - ANI) and subjected to hierarchical clustering using the complete linkage algorithm. To ensure uniform annotation, all genomes were re–annotated using Prokka v1.15.6. The resulting GFF files were analyzed with Roary v3.13.0 to construct the species pangenome, applying a 95% BLASTp identity threshold for ortholog clustering. Genes were categorized into ‘core’ (99%≤strains ≤ 100%), ‘soft core’ (95%≤strains < 99%), ‘shell’ (15%≤strains < 95%) and ‘cloud’ (0%≤strains < 15%) genes according to Roary’s classification scheme. The concatenated core gene alignment generated by Roary was used for phylogenetic reconstruction using IQ–TREE 2 [58]. The optimal substitution model was determined by ModelFinder [59], and branch support was evaluated with 1,000 ultrafast bootstrap replicates and 1,000 SH–aLRT replicates. Carbohydrate–Active Enzyme (CAZyme) Gene Screening CAZyme genes and predicted substrates were characterized through dbCAN3 v4.0, the tool incorporates DIAMOND, HMMER, and dbCAN_sub for annotating CAZyme families and substrates prediction [60]. We considered only CAZyme entries supported by ≥ 2 annotation tools. Carbohydrate Utilization Assay Carbohydrate utilization assay was performed with API® 50CH strips (bioMérieux) according to the manufacturer’s instructions. HY7715 and KCTC 3108T were streaked on MRS agar incubated anaerobically at 37 °C for 48 h. Colonies of strains were suspended in an API® 50CHL medium (bioMérieux) with a turbidity equivalent to 2 McFarland. The tubes of API 50CH strips were filled with the prepared suspension, covered with mineral oil, and incubated at 37 °C for 48 h. The medium contains bromocresol purple as a pH indicator, changing the color from purple to yellow in response to acid production, thereby indicating carbon source utilization. For the esculin test (tube no. 25), a change in color from purple to black is considered as positive. Acid and Bile Tolerance Acid and bile tolerance of L. plantarum HY7715 and KCTC 3108T were evaluated as described by Chung et al. [61]. For acid tolerance test, 100 µL of cultured strains were inoculated into 10 mL of phosphate buffered saline (PBS; Welgene) supplemented with 1% pepsin. pH was adjusted to 2.5 using 1 M HCl (Merck) and incubated at 37 °C for 2 h. Bile tolerance was tested by inoculation of 100 µL of cultured strains into 10 mL of PBS supplemented with 0.3% bile. pH was adjusted to 8.0 using 1 M NaOH (Merck) and incubated at 37 °C for 2 h. As a positive control, the strains were incubated in PBS (pH 6.2) at 37 °C for 2 h. After incubation in each condition, counting of bacterial cell population was performed by plating them on MRS agar and incubating anaerobically at 37 °C for 48 h. The survival rate was calculated as follows.\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm{S}\mathrm{u}\mathrm{r}\mathrm{v}\mathrm{i}\mathrm{v}\mathrm{a}\mathrm{l}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{e}\left(\mathrm{\%}\right)=\frac{{\mathrm{C}\mathrm{F}\mathrm{U}}_{\mathrm{t}\mathrm{e}\mathrm{s}\mathrm{t}}}{{\mathrm{C}\mathrm{F}\mathrm{U}}_{\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}}}\times100$$\end{document} CFUtest: plate count number (cfu/mL) after incubation at each test condition (acid or bile). CFUcontrol: plate count number (cfu/mL) of positive control. Antimicrobial Susceptibility Test Antimicrobial susceptibility of HY7715 was evaluated using The Liofilchem® MIC Test Strips (Liofilchem). The minimum inhibitory concentrations (MIC) of nine antibiotics, including ampicillin, chloramphenicol, clindamycin, erythromycin, gentamycin, kanamycin, streptomycin, tetracycline, and vancomycin, were determined and compared with the EFSA cut–off criteria for probiotics [62, 63]. HY7715 suspension in a 0.85% (w/v) NaCl solution with a turbidity equivalent to 0.5 McFarland was swabbed evenly on LSM agar plates (90% Iso–Sensitest broth, 10% MRS broth, 1.5% agar) with a sterile cotton swab [64]. After drying the surfaces of the inoculated plates, the test strips were applied, and incubated anaerobically at 37 °C for 48 h. MICs for each antibiotic were read from the test strip according to the manufacturer’s instructions. Biogenic Amine Production Test Biogenic amine production was assessed following the method of Bover–Cid and Holzapfel [65]. Prior to the screening assay, L. plantarum HY7715 and E. coli ATCC 25922 were sub–cultured twice in MRS broth and TSB, respectively, at 37 °C for 18 h and 24 h to activate the cultures. The reference strain E. coli ATCC 25922 was used as a positive control for biogenic amine production. The optimal grown cultures were streaked in duplicate onto the screening medium (Table 1), each supplemented with a precursor amino acid (Sigma–Aldrich, St. Louis, MO, USA), including L-tyrosine, L-histidine monohydrochloride, L-ornithine monohydrochloride, and L-lysine, and incubated at 37 °C for 4 days. The results were considered positive when the medium changed to a purple color. This color change was caused by the action of the pH indicator bromocresol purple, reflecting alkalinity generated by the conversion of the supplemented amino acids into BAs. Table 1Composition of biogenic amine producing ability screening mediumComponentContents (% w/v)Tryptone0.5Yeast extract0.5Meat extract0.5Sodium chloride0.25Glucose0.05Tween 80®0.1MgSO4·7 H₂O0.02MnSO4·H₂O0.005FeSO4·7 H₂O0.004Ammonium citrate0.2Thiamine0.001K2HPO40.2CaCO30.01Pyridoxal–5–phosphate0.005Precursor amino acid1.0Bromocresol purple0.006Agar2.0pH5.3 Table 1Composition of biogenic amine producing ability screening mediumComponentContents (% w/v)Tryptone0.5Yeast extract0.5Meat extract0.5Sodium chloride0.25Glucose0.05Tween 80®0.1MgSO4·7 H₂O0.02MnSO4·H₂O0.005FeSO4·7 H₂O0.004Ammonium citrate0.2Thiamine0.001K2HPO40.2CaCO30.01Pyridoxal–5–phosphate0.005Precursor amino acid1.0Bromocresol purple0.006Agar2.0pH5.3 Composition of biogenic amine producing ability screening medium Hemolysis Test For hemolysis test, HY7715 and S. aureus ATCC 33519 were streaked on blood agar plate (BAP; nutrient agar supplemented with 5% sheep blood; Synergy Innovation) and incubated at 37 °C for 48 h. Hemolytic activity was determined based on the color and clarity of medium around colonies. Complete clear and transparent zone indicates complete hemolysis (β-hemolysis). A partial, greenish or brownish discoloration indicates incomplete hemolysis (α-hemolysis). No apparent changes in the color of medium indicates that the strain do not have hemolytic activity (γ-hemolysis). Statistical Analysis All data were presented as the mean ± standard deviation (SD) from three independent experiments. Statistical analyses were performed using GraphPad Prism v10.6.1 (GraphPad Software). Prior to applying parametric tests, data normality was assessed using the Shaprio–Wilk test. For the acid and bile tolerance assays, HY7715 was compared with KCTC 3108T using multiple unpaired t–tests with Holm–Šídák multiple comparisons correction. Statistical significance was defined as p < 0.05. Results Genome Features of HY7715 The genome of L. plantarum HY7715 was sequenced using the PacBio and Illumina platforms. The average sequencing coverage was 1,277× for PacBio and 280× for Illumina. Short–read first hybrid assembly resulted in seven contigs with a total length of 3,348,486 bp and a DNA GC content of 44.29%. All contigs were predicted to be circular, with contig 1 predicted to represent the chromosome and the remaining six contigs identified as plasmids (Fig. 1).Fig. 1Circular representation of the HY7715 genome comprising seven contigs. The largest contig corresponds to the chromosome, while contigs 2–7 represent plasmid sequences shown as individual circular maps. From the outside to the innermost circles: annotated genes (CDSs), rRNAs, and tRNAs on the reverse and forward strand; GC content (black); and GC skews (green, GC skew+; purple, GC skew-). The total genome size is 3,348,486 bp and includes 3,224 CDSs, 16 rRNAs, 70 tRNAs, and 1 tmRNA Circular representation of the HY7715 genome comprising seven contigs. The largest contig corresponds to the chromosome, while contigs 2–7 represent plasmid sequences shown as individual circular maps. From the outside to the innermost circles: annotated genes (CDSs), rRNAs, and tRNAs on the reverse and forward strand; GC content (black); and GC skews (green, GC skew+; purple, GC skew-). The total genome size is 3,348,486 bp and includes 3,224 CDSs, 16 rRNAs, 70 tRNAs, and 1 tmRNA Annotation of the assembled contigs identified 3,224 CDSs, 16 rRNA genes, 70 tRNA genes, and 1 tmRNA. The distribution of annotated genes across the chromosome and individual plasmid contigs is summarized in Table 2. Functional annotation of the identified CDSs was performed using EggNOG–mapper. Among 3,224 CDSs, 2,933 genes were functionally annotated and classified into 19 COG categories (Fig. 2). The three most abundant categories were transcription (K), carbohydrate transport and metabolism (G), and amino acid transport and metabolism (E).Fig. 2Functional classification of CDSs of strain HY7715 based on COG analysis. Among 3,224 CDSs, 2,933 (≈ 90.9%) were assigned to 19 COG functional categories. Categories are grouped into major functional classes (Information storage and processing; Cellular processes and signaling; Metabolism; Poorly characterized; and Other), and bars represent absolute gene counts for each COG category. The most abundant categories were transcription (K), carbohydrate transport and metabolism (G), and amino acid transport and metabolism (E), indicating prominent regulator and metabolic capacities of the strainTable 2Annotation summary of the chromosome and plasmid contigs of Lactiplantibacillus plantarum HY7715Contig no.ClassificationSize (bp)GeneCDSrRNAtRNAtmRNA1Chromosome3,232,8423,1823,095167012Plasmid58,98764640003Plasmid21,93226260004Plasmid11,37114140005Plasmid10,251990006Plasmid9,61012120007Plasmid3,49344000 Functional classification of CDSs of strain HY7715 based on COG analysis. Among 3,224 CDSs, 2,933 (≈ 90.9%) were assigned to 19 COG functional categories. Categories are grouped into major functional classes (Information storage and processing; Cellular processes and signaling; Metabolism; Poorly characterized; and Other), and bars represent absolute gene counts for each COG category. The most abundant categories were transcription (K), carbohydrate transport and metabolism (G), and amino acid transport and metabolism (E), indicating prominent regulator and metabolic capacities of the strain Table 2Annotation summary of the chromosome and plasmid contigs of Lactiplantibacillus plantarum HY7715Contig no.ClassificationSize (bp)GeneCDSrRNAtRNAtmRNA1Chromosome3,232,8423,1823,095167012Plasmid58,98764640003Plasmid21,93226260004Plasmid11,37114140005Plasmid10,251990006Plasmid9,61012120007Plasmid3,49344000 Annotation summary of the chromosome and plasmid contigs of Lactiplantibacillus plantarum HY7715 Genome Contamination Analysis Read–level contamination analysis was performed using Kraken2. The classification results showed that 90.55% of Illumina reads and 97.01% of PacBio reads were assigned to genus Lactiplantibacillus. At the species level, 55.88% of Illumina reads and 88.97% of PacBio reads were assigned to Lactiplantibacillus plantarum. The full taxonomic classification profiles generated by Kraken2 for each sequencing platform are summarized in Table S1 (Illumina) and Table S2 (PacBio). Contamination in the 16 S rRNA genes extracted from the genome of strain HY7715 was evaluated using ContEst16S. Five 16 S rRNA gene copies were identified, all showing the highest sequence similarity to the type strain Lactiplantibacillus pentosus DSM 20314T. Each sequence exhibited 99.93% identity with only one nucleotide over 1,489 bp. Genome completeness and potential contamination were evaluated using UBCG, which detects the presence of 92 universally conserved single–copy bacterial core genes. All 92 genes were identified, indicating that the genome was complete and free of contamination. Further assessment with BUSCO (v5.2.7) with Bacteria_odb10 lineage dataset revealed that 100% of the 124 marker genes were complete, including 123 single–copy and one duplicated gene, with no fragmented or missing BUSCOs. Taxonomic Identification Taxonomic analysis based on 16 S rRNA gene sequence was conducted. The 16 S rRNA gene sequence extracted from the genome of L. plantarum HY7715 was compared with reference sequences in the EzBioCloud 16 S rRNA gene database using BLASTn. The results confirmed that the strain HY7715 belonged to the genus Lactiplantibacillus and showed the highest similarity to L. pentosus DSM 20314T (99.93%), followed by L. argentoratensis DSM 16365T (99.87%), L. paraplantarum DSM 10667T (99.87%), L. plantarum DSM 20174T (99.87%), L. fabifermentans DSM 21115T (99.06%), L. herbarum TCF032–E4T (98.99%), L. xiangfangensis LMG 26013T (98.93%), L. plajomi NBRC 107333T (98.72%), and L. modestisalitolerans NBRC 107235T (98.46%). The maximum–likelihood phylogenetic tree based on 16 S rRNA gene sequences demonstrated that strain HY7715 was positioned within the L. plantarum–L. pentosus complex but formed a separate branch adjacent to L. herbarum (Fig. 3).Fig. 3Maximum–likelihood tree based on 16 S rRNA gene sequences of strain HY7715 and the eleven closely related Lactiplantibacillus species derived from EzBioCloud database (GenBank accession numbers are shown in parentheses). Bootstrap values (expressed as the percentage of 1,000 replications) are shown at the branch points; only values greater than 50% are shown Maximum–likelihood tree based on 16 S rRNA gene sequences of strain HY7715 and the eleven closely related Lactiplantibacillus species derived from EzBioCloud database (GenBank accession numbers are shown in parentheses). Bootstrap values (expressed as the percentage of 1,000 replications) are shown at the branch points; only values greater than 50% are shown Whole–genome–based ANI analysis revealed the highest similarity to L. plantarum DSM 20174T, with 99.12% identity (Table 3). A phylogenetic tree constructed using the UPGMA method based on the ANI values showed that HY7715 formed a stable cluster with L. plantarum DSM 20174T, establishing a distinct phylogenetic lineage within the genus Lactiplantibacillus (Fig. 4).Fig. 4UPGMA phylogenetic tree based on ANI values derived from whole–genome sequences of strain HY7715 and the ten closely related Lactiplantibacillus species (GenBank accession numbers are shown in parentheses). The tree was constructed using OAT with the UPGMA method based on pairwise ANI values. Branch lengths represent genomic distance inferred from ANI. Strain HY7715 clustered most closely with L. plantarum DSM 20174T, forming a distinct phylogenetic lineage within the genusTable 3Whole–genome ANI similarity between Lactiplantibacillus plantarum HY7715 and ten closely related speciesRankReference speciesANI similarity (%)GenBank accession no.1Lactiplantibacillus plantarum DSM 20174T99.12GCA 014131735.12Lactiplantibacillus argentoratensis DSM 16365T95.51GCA 003641165.13Lactiplantibacillus paraplantarum DSM 10667T86.07GCA 003641145.14Lactiplantibacillus pentosus DSM 20314T80.31GCA 003641185.15Lactiplantibacillus herbarum TCF032-E4T77.71GCA 001038045.16Lactiplantibacillus xiangfangensis LMG 26013T76.55GCA 001438845.17Lactiplantibacillus fabifermentans DSM 21115T75.71GCA 000498955.28Lactiplantibacillus plajomi NBRC 107333T75.37GCA 005405405.19Lactiplantibacillus daoliensis 116-1AT75.28GCA 005405005.110Lactiplantibacillus modestisalitolerans NBRC 107235T74.90GCA 005405425.1 UPGMA phylogenetic tree based on ANI values derived from whole–genome sequences of strain HY7715 and the ten closely related Lactiplantibacillus species (GenBank accession numbers are shown in parentheses). The tree was constructed using OAT with the UPGMA method based on pairwise ANI values. Branch lengths represent genomic distance inferred from ANI. Strain HY7715 clustered most closely with L. plantarum DSM 20174T, forming a distinct phylogenetic lineage within the genus Table 3Whole–genome ANI similarity between Lactiplantibacillus plantarum HY7715 and ten closely related speciesRankReference speciesANI similarity (%)GenBank accession no.1Lactiplantibacillus plantarum DSM 20174T99.12GCA 014131735.12Lactiplantibacillus argentoratensis DSM 16365T95.51GCA 003641165.13Lactiplantibacillus paraplantarum DSM 10667T86.07GCA 003641145.14Lactiplantibacillus pentosus DSM 20314T80.31GCA 003641185.15Lactiplantibacillus herbarum TCF032-E4T77.71GCA 001038045.16Lactiplantibacillus xiangfangensis LMG 26013T76.55GCA 001438845.17Lactiplantibacillus fabifermentans DSM 21115T75.71GCA 000498955.28Lactiplantibacillus plajomi NBRC 107333T75.37GCA 005405405.19Lactiplantibacillus daoliensis 116-1AT75.28GCA 005405005.110Lactiplantibacillus modestisalitolerans NBRC 107235T74.90GCA 005405425.1 Whole–genome ANI similarity between Lactiplantibacillus plantarum HY7715 and ten closely related species Genome Stability The genomic stability of L. plantarum HY7715 was assessed through comparative genomic analysis between the mother culture and the 200th generation of the strain (Table 4). The ANI value between the two genomes was 99.99%. In addition, whole–genome alignment using MUMmer v4.0 identified a total of 15 SNPs across the approximately 3.3 Mb genome, with no large–scale structural variations detected. These results indicate a high level of genomic stability during serial subculturing. Table 4Genome characteristics of the whole genome sequence of the mother culture and 200th generations of strain HY7715StrainMother culture200th generationAssembly statusCompleteCompleteGenome length (bp)3,348,4863,347,601No. of contigs77GC content (%)44.2944.29CDS3,2243,226rRNA1615tRNA7068tmRNA11Functionally annotated genes2,9332,933ANI similarity (%)99.99Total SNPs (pairwise comparison)15 Table 4Genome characteristics of the whole genome sequence of the mother culture and 200th generations of strain HY7715StrainMother culture200th generationAssembly statusCompleteCompleteGenome length (bp)3,348,4863,347,601No. of contigs77GC content (%)44.2944.29CDS3,2243,226rRNA1615tRNA7068tmRNA11Functionally annotated genes2,9332,933ANI similarity (%)99.99Total SNPs (pairwise comparison)15 Genome characteristics of the whole genome sequence of the mother culture and 200th generations of strain HY7715 In Silico Identification of Virulence and Antimicrobial Resistance The complete genome sequence of L. plantarum HY7715 was analyzed to identify potential VF and AMR genes. Screening of VFs using ABRicate against VFDB detected no VF genes exceeding the thresholds of 80% identity and 70% coverage, as recommended by the EFSA guidelines [29]. For AMR gene detection, three different databases—AMRFinderPlus, ResFinder, and MEGARes—were used, and no AMR genes were detected under the same threshold criteria. Further analyses of MGEs were conducted to evaluate the potential for horizontal transfer of antimicrobial resistance determinants to other microorganisms. The analysis identified four plasmids and 25 transposase-related elements, including insertion sequences and composite transposons, within the genome of HY7715 (Tables 5, 6 and 7). Additionally, five genomic regions were predicted as potential phage–derived sequences (Table 8). Sequence homology screening using blastp indicated that most prophage–associated genes in HY7715 showed highest similarity to L. plantarum. Only 114 genes (3.53% of the total predicted genes in the genome) met the defined criteria for HGT (≥ 80% amino acid identity and ≥ 80% query coverage to non-L. plantarum taxa), suggesting limited acquisition of prophage genes from other bacterial species (Table S3).Table 5Plasmid typing results derived from the complete genome of strain HY7715Sample_idContigSizeGC content (%)Observed host rangeNCBI rankNCBI nameHY7715:AA490contig 258,98739.07orderLactobacillalesHY7715:AD132contig 321,93237.99genusLactiplantibacillusHY7715:AA698contig 510,25136.07familyLactobacillaceaeHY7715:AF706contig 73,49338.16familyLactobacillaceaeTable 6Insertion sequence identified in the complete genome of strain HY7715No.NamePredictionTypeIdentifyCoverageSubstitutionContigStartEnd1ISP2predictedinsertion sequence0.98712312,478,6322,480,4272ISLsa1predictedinsertion sequence0.9970.99931810,397811,4303ISLpl1predictedinsertion sequence0.9931711,728,6311,729,6734ISLpl1predictedinsertion sequence0.994161364,137365,1795ISLpl3predictedinsertion sequence0.9930.999513,081,6193,082,4716ISLpl3predictedinsertion sequence0.9930.99951204,608205,4607ISLpl3predictedinsertion sequence0.9930.99951426,858427,7108ISLpl3predictedinsertion sequence0.9930.999512,176,2572,177,1099ISLpl3predictedinsertion sequence0.9940.99941257,681258,53310ISLpl3predictedinsertion sequence0.9940.999412,766,3542,767,20611ISLpl3predictedinsertion sequence0.9940.99941724,196725,04812ISLpl3predictedinsertion sequence0.9940.999412,314,5922,315,44413ISLpl3predictedinsertion sequence0.9940.999411,152,5721,153,42414ISLpl3predictedinsertion sequence0.9940.999412,271,8172,272,66915ISLpl3predictedinsertion sequence0.9940.99941806,220807,07216ISLpl3predictedinsertion sequence0.9940.99941287,381288,23317ISLpl3predictedinsertion sequence0.9940.999413,111,0373,111,88918ISLpl1predictedinsertion sequence0.8440.987149247,47048,49819ISLpl1predictedinsertion sequence0.988112256,53057,57220ISLpl4predictedinsertion sequence0.9921824,4375,421Table 7Composite transposons identified in the complete genome of strain HY7715No.NamePredictionTypeIdentifyCoverageSubstitutionContigStartEnd18_16_9cn_30553_ISLpl3putativecomposite transposon0.9940.99941257,680288,23319_12_14cn_43628_ISLpl3putativecomposite transposon0.9940.999412,271,8162,315,44420_17_5cn_30271_ISLpl3putativecomposite transposon0.9930.999513,081,6183,111,88920_18cn_13377_ISLpl1putativecomposite transposon0.8440.987149235,12148,49821_18_19cn_10103_ISLpl1putativecomposite transposon0.8440.987149247,46957,572Table 8Potential prophage regions identified in the complete genome of strain HY7715Contig no.Region lengthCompletenessScoreTotal proteinsRegion positionMost common phage(hit genes count)117.0 kbquestionable802435,969–53,049PHAGE_Strept_315.2_NC_004585(3)144.2 kbintact150521,206,685–1,250,909PHAGE_Lactob_Sha1_NC_019489(33)143.3 kbintact150691,845,760–1,889,142PHAGE_Lactob_Sha1_NC_019489(13)144.3 kbintact120542,171,917–2,216,285PHAGE_Oenoco_phiS13_NC_023560(15)122.4 kbquestionable80222,221,952–2,244,374PHAGE_Lactob_Lc_NC_007501(2) Table 5Plasmid typing results derived from the complete genome of strain HY7715Sample_idContigSizeGC content (%)Observed host rangeNCBI rankNCBI nameHY7715:AA490contig 258,98739.07orderLactobacillalesHY7715:AD132contig 321,93237.99genusLactiplantibacillusHY7715:AA698contig 510,25136.07familyLactobacillaceaeHY7715:AF706contig 73,49338.16familyLactobacillaceae Plasmid typing results derived from the complete genome of strain HY7715 Table 6Insertion sequence identified in the complete genome of strain HY7715No.NamePredictionTypeIdentifyCoverageSubstitutionContigStartEnd1ISP2predictedinsertion sequence0.98712312,478,6322,480,4272ISLsa1predictedinsertion sequence0.9970.99931810,397811,4303ISLpl1predictedinsertion sequence0.9931711,728,6311,729,6734ISLpl1predictedinsertion sequence0.994161364,137365,1795ISLpl3predictedinsertion sequence0.9930.999513,081,6193,082,4716ISLpl3predictedinsertion sequence0.9930.99951204,608205,4607ISLpl3predictedinsertion sequence0.9930.99951426,858427,7108ISLpl3predictedinsertion sequence0.9930.999512,176,2572,177,1099ISLpl3predictedinsertion sequence0.9940.99941257,681258,53310ISLpl3predictedinsertion sequence0.9940.999412,766,3542,767,20611ISLpl3predictedinsertion sequence0.9940.99941724,196725,04812ISLpl3predictedinsertion sequence0.9940.999412,314,5922,315,44413ISLpl3predictedinsertion sequence0.9940.999411,152,5721,153,42414ISLpl3predictedinsertion sequence0.9940.999412,271,8172,272,66915ISLpl3predictedinsertion sequence0.9940.99941806,220807,07216ISLpl3predictedinsertion sequence0.9940.99941287,381288,23317ISLpl3predictedinsertion sequence0.9940.999413,111,0373,111,88918ISLpl1predictedinsertion sequence0.8440.987149247,47048,49819ISLpl1predictedinsertion sequence0.988112256,53057,57220ISLpl4predictedinsertion sequence0.9921824,4375,421 Insertion sequence identified in the complete genome of strain HY7715 Table 7Composite transposons identified in the complete genome of strain HY7715No.NamePredictionTypeIdentifyCoverageSubstitutionContigStartEnd18_16_9cn_30553_ISLpl3putativecomposite transposon0.9940.99941257,680288,23319_12_14cn_43628_ISLpl3putativecomposite transposon0.9940.999412,271,8162,315,44420_17_5cn_30271_ISLpl3putativecomposite transposon0.9930.999513,081,6183,111,88920_18cn_13377_ISLpl1putativecomposite transposon0.8440.987149235,12148,49821_18_19cn_10103_ISLpl1putativecomposite transposon0.8440.987149247,46957,572 Composite transposons identified in the complete genome of strain HY7715 Table 8Potential prophage regions identified in the complete genome of strain HY7715Contig no.Region lengthCompletenessScoreTotal proteinsRegion positionMost common phage(hit genes count)117.0 kbquestionable802435,969–53,049PHAGE_Strept_315.2_NC_004585(3)144.2 kbintact150521,206,685–1,250,909PHAGE_Lactob_Sha1_NC_019489(33)143.3 kbintact150691,845,760–1,889,142PHAGE_Lactob_Sha1_NC_019489(13)144.3 kbintact120542,171,917–2,216,285PHAGE_Oenoco_phiS13_NC_023560(15)122.4 kbquestionable80222,221,952–2,244,374PHAGE_Lactob_Lc_NC_007501(2) Potential prophage regions identified in the complete genome of strain HY7715 Screening of the CRISPR array revealed that the genome harbors one CRISPR locus with associated spacer and Cas genes. The identified CRISPR–Cas system was classified as type IIA, containing cas1, cas2, cas9, and csn2 genes (Table 9).Table 9Potential CRISPR arrays and Cas genes identified in the complete genome of strain HY7715. “Evidence level” indicates the confidence category defined by CRISPRCasFinder for predicted CRISPR loci, based on structural features and associated cas genes. Level 4 corresponds to the highest confidence assignmentElementCRISPR ID/Cas typeFromToSpacer/GeneRepeat consensus/cas genesDirectionEvidence levelCas clusterCAS–Type IIA126,748132,8794cas9 (Type II), cas1 (Type II), cas2 (Type I–II–III), csn2 (type IIA)--CRISPRcontig1_1132,904133,4017GTCTTGAATAGTAGTCATATCAAACAGGTTTAGAAC+4 Table 9Potential CRISPR arrays and Cas genes identified in the complete genome of strain HY7715. “Evidence level” indicates the confidence category defined by CRISPRCasFinder for predicted CRISPR loci, based on structural features and associated cas genes. Level 4 corresponds to the highest confidence assignmentElementCRISPR ID/Cas typeFromToSpacer/GeneRepeat consensus/cas genesDirectionEvidence levelCas clusterCAS–Type IIA126,748132,8794cas9 (Type II), cas1 (Type II), cas2 (Type I–II–III), csn2 (type IIA)--CRISPRcontig1_1132,904133,4017GTCTTGAATAGTAGTCATATCAAACAGGTTTAGAAC+4 Potential CRISPR arrays and Cas genes identified in the complete genome of strain HY7715. “Evidence level” indicates the confidence category defined by CRISPRCasFinder for predicted CRISPR loci, based on structural features and associated cas genes. Level 4 corresponds to the highest confidence assignment Taken together, although multiple MGEs were identified in the genome of HY7715, none of these elements harbored AMR or VF genes, indicating a low risk of horizontal gene transfer associated with undesirable traits. Secondary Metabolite Gene Clusters Analysis of the complete genome sequence of L. plantarum HY7715 using antiSMASH analysis indicated five BGCs (Fig. 5). None of these clusters exhibited significant similarity to known secondary metabolite clusters in the MIBiG database, suggesting that they represent novel or strain–specific gene clusters.Fig. 5Predicted biosynthetic gene clusters (BGCs) identified in the whole genome of HY7715 using antiSMASH v8.0. The predicted BGC classes and similarity levels to the closest MIBiG reference clusters are summarized in the inset table. No clusters showed significant similarity to known toxin–associated BGCs Predicted biosynthetic gene clusters (BGCs) identified in the whole genome of HY7715 using antiSMASH v8.0. The predicted BGC classes and similarity levels to the closest MIBiG reference clusters are summarized in the inset table. No clusters showed significant similarity to known toxin–associated BGCs Region 1.1 was annotated as ribosomally synthesized and post–translationally modified peptide (RiPP)–like cluster with low similarity to bacteriocin–producing clusters, including colicin X, carnobacteriocin XY, microcin M/L, coagulin, and salivricin CRL1328. Region 1.2 was predicted to encode terpene–precursor gene clusters putatively involved in the biosynthesis of astallatene, boleracene, carotenoids, and gibberellins, although no significant similarity was observed. Region 1.3 corresponded to a type III polyketide synthase (PKS) cluster showing weak similarity to known clusters associated with enterocin F4–9, indolmycin, SapB peptide, pseudomycoicidin, phenazines, and bryostatin. Region 1.4 exhibited moderate similarity (0.75) to a carotenoid–type terpene cluster from Rhodobacter sphaeroides and was associated with the β–carotene and zeaxanthin biosynthetic pathways. Region 1.5 was matched to clusters potentially involved in quorum sensing and intercellular signaling, showing low similarity to known clusters such as carnobacteriocin XY, estericin A, paenibacillin, fumigaclavine C, and lipopolysaccharides biosynthesis. The result of the BAGEL4 webserver with the HY7715 genome identified one bacteriocin–related gene cluster as Area of interest (AOI) at chromosome (contig 1) (Fig. 6). The AOI encodes the Plantaricin EF, which is a two–peptide bacteriocin belongs to the class IIb bacteriocin. The mature peptide sequences of these bacteriocin and their computed pI and MW are provided in Table S4.Fig. 6Predicted bacteriocin gene cluster in whole–genome sequence of strain HY7715 through the BAGEL4 webserver. The area of interest at contig 1 indicates Plantaricin EF Predicted bacteriocin gene cluster in whole–genome sequence of strain HY7715 through the BAGEL4 webserver. The area of interest at contig 1 indicates Plantaricin EF The amino acid sequences of all members in the gene cluster were confirmed by blastp against the nr database (Table S5). In addition to two core peptides, the bacteriocin gene cluster consists of three immunity protein, a transporter LanT, HlyD family secretion protein, and ABC–transporter. The other ORFs are described in Table S5. Biogenic Amine Production BlastKOALA analysis was performed to identify genes involved in BA formation from the complete genome of L. plantarum HY7715. The list of genes retrieved from the KEGG database is provided in Table 10. No genes associated with the biosynthesis of major BAs—such as cadaverine, putrescine, spermidine, spermine, ornithine, histamine, tyramine, or tryptamine—were detected. Additionally, a BA production assay was conducted to verify whether the strain can actually produce them. The results confirmed that strain HY7715 did not produce any tested BAs, including tyramine, putrescine, histamine, and cadaverine, compared to a positive control, E. coli ATCC 25922 (Fig. 7).Fig. 7BA production assay. L. plantarum HY7715 did not produce BAs on the screening media containing (a) L-tyrosine, (b) L-ornithine monohydrochloride, (c) L-histidine monohydrochloride, and (d) L-lysine. In contrast, the positive control, E. coli ATCC 25922, exhibited a clear color change to purple on media containing (e) L-tyrosine, (f) L-ornithine monohydrochloride, (g) L-histidine monohydrochloride, and (h) L-lysine, thereby validating the assay performanceTable 10List of the KEGG pathway related to biogenic amine formation examined in the genome analysis of strain HY7715Map IDNameGene IDSymbolPresentin HY7715Enzyme nameFunction00310Lysine degradationK01582ldcC, cadAAbsentLysine decarboxylase [EC:4.1.1.18]production of cadaverineK23385AbsentD-ornithine/D-lysine decarboxylase [EC:4.1.1.116]production of cadaverine00330Arginine and prolinemetabolismK01583AbsentArginine decarboxylase [EC:4.1.1.19]arginine -> agmatineK01584adiAAbsentK01585speAAbsentK02626pdaDAbsentK01480speBAbsentAgmatinase [EC:3.5.3.11]agmatine -> putrescineK00797speE, SRM, SPE3AbsentSpermidine synthase [EC:2.5.1.16]putrescine -> spermidine, spermineK01476rocF, argAbsentArginse [EC:3.5.3.1]arginine -> ornithineK01581ODC1, speC, speFAbsentOrnithine decarboxylase [EC:4.1.1.17]ornithine -> putrescine00340Histidine metabolismK01590HDCAbsentHistidine decarboxylase [EC:4.1.1.22]histidine -> histamine00350Tyrosine metabolismK22329TDC-1AbsentTyrosine decarboxylase [EC:4.1.1.25]tyrosine -> tyramineK22330tyrDCAbsentK01592TYDCAbsentK18933mfnA, adcAbsent00380Tryptophan metabolismK01593DDC, TDCAbsentTryptophan decarboxylase [EC:4.1.1.28]tryptophan -> tryptamine BA production assay. L. plantarum HY7715 did not produce BAs on the screening media containing (a) L-tyrosine, (b) L-ornithine monohydrochloride, (c) L-histidine monohydrochloride, and (d) L-lysine. In contrast, the positive control, E. coli ATCC 25922, exhibited a clear color change to purple on media containing (e) L-tyrosine, (f) L-ornithine monohydrochloride, (g) L-histidine monohydrochloride, and (h) L-lysine, thereby validating the assay performance Table 10List of the KEGG pathway related to biogenic amine formation examined in the genome analysis of strain HY7715Map IDNameGene IDSymbolPresentin HY7715Enzyme nameFunction00310Lysine degradationK01582ldcC, cadAAbsentLysine decarboxylase [EC:4.1.1.18]production of cadaverineK23385AbsentD-ornithine/D-lysine decarboxylase [EC:4.1.1.116]production of cadaverine00330Arginine and prolinemetabolismK01583AbsentArginine decarboxylase [EC:4.1.1.19]arginine -> agmatineK01584adiAAbsentK01585speAAbsentK02626pdaDAbsentK01480speBAbsentAgmatinase [EC:3.5.3.11]agmatine -> putrescineK00797speE, SRM, SPE3AbsentSpermidine synthase [EC:2.5.1.16]putrescine -> spermidine, spermineK01476rocF, argAbsentArginse [EC:3.5.3.1]arginine -> ornithineK01581ODC1, speC, speFAbsentOrnithine decarboxylase [EC:4.1.1.17]ornithine -> putrescine00340Histidine metabolismK01590HDCAbsentHistidine decarboxylase [EC:4.1.1.22]histidine -> histamine00350Tyrosine metabolismK22329TDC-1AbsentTyrosine decarboxylase [EC:4.1.1.25]tyrosine -> tyramineK22330tyrDCAbsentK01592TYDCAbsentK18933mfnA, adcAbsent00380Tryptophan metabolismK01593DDC, TDCAbsentTryptophan decarboxylase [EC:4.1.1.28]tryptophan -> tryptamine List of the KEGG pathway related to biogenic amine formation examined in the genome analysis of strain HY7715 Arginine and proline metabolism Genetic Insights of Riboflavin–Overproducing Properties Genome analysis performed using BlastKOALA with the KEGG database revealed that the genome of L. plantarum HY7715 harbors a complete functional rib operon, comprising the ribG, ribB, ribA, and ribH genes (Table 11).Table 11Riboflavin biosynthesis pathway genes identified from the KEGG database using the genome of strain HY7715NameSymbolKEGG Ontology (KO)Gene Ontology (GO)ContigFromToLength (bp)hydroxyethylthiazole kinase [EC:2.7.1.50]thiMK00878GO: 00044171112,916113,707792FMN reductase (NADPH) [EC:1.5.1.38]nfrA1K19285GO: 00528731723,288724,040753diaminohydroxyphosphoribosylaminopyrimidine deaminase [E:3.5.4.26]/5-amino-6-(5-phosphoribosylamino)uracil reductase [EC:1.1.1.193]ribD, ribGK11752GO: 0008835/GO: 000870311,314,1801,315,2471,068Riboflavin synthase [EC:2.5.1.9]ribE, ribB, RIB5K00793GO: 000474611,315,2481,315,8506033,4-dihydroxy 2-butanone 4-phosphate synthase [EC:4.1.99.12]/GTP cyclohydrolase II [EC:3.5.4.25]ribBA, ribAK14652GO: 0008686/GO: 000393511,315,8521,317,0661,2156,7-dimethyl-8-ribityllumazine synthase [EC:2.5.1.78]ribHK00794GO: 000090611,317,0631,317,5424805-amino-6-(5-phospho-D-ribitylamino)uracil phosphatase [EC:3.1.3.104]ycsE, yitU, ywtEK21064GO: 004372611,532,1601,532,975816riboflavin kinase [EC:2.7.1.26]/FMN adenylyltransferase [EC:2.7.7.2]ribFK11753GO: 0008531/GO: 000391911,800,3101,801,3111,002ADP-ribose diphosphatase [EC:3.6.1.13] [E:3.6.1.-]nudFK01515GO: 004763112,474,1942,474,760566 Table 11Riboflavin biosynthesis pathway genes identified from the KEGG database using the genome of strain HY7715NameSymbolKEGG Ontology (KO)Gene Ontology (GO)ContigFromToLength (bp)hydroxyethylthiazole kinase [EC:2.7.1.50]thiMK00878GO: 00044171112,916113,707792FMN reductase (NADPH) [EC:1.5.1.38]nfrA1K19285GO: 00528731723,288724,040753diaminohydroxyphosphoribosylaminopyrimidine deaminase [E:3.5.4.26]/5-amino-6-(5-phosphoribosylamino)uracil reductase [EC:1.1.1.193]ribD, ribGK11752GO: 0008835/GO: 000870311,314,1801,315,2471,068Riboflavin synthase [EC:2.5.1.9]ribE, ribB, RIB5K00793GO: 000474611,315,2481,315,8506033,4-dihydroxy 2-butanone 4-phosphate synthase [EC:4.1.99.12]/GTP cyclohydrolase II [EC:3.5.4.25]ribBA, ribAK14652GO: 0008686/GO: 000393511,315,8521,317,0661,2156,7-dimethyl-8-ribityllumazine synthase [EC:2.5.1.78]ribHK00794GO: 000090611,317,0631,317,5424805-amino-6-(5-phospho-D-ribitylamino)uracil phosphatase [EC:3.1.3.104]ycsE, yitU, ywtEK21064GO: 004372611,532,1601,532,975816riboflavin kinase [EC:2.7.1.26]/FMN adenylyltransferase [EC:2.7.7.2]ribFK11753GO: 0008531/GO: 000391911,800,3101,801,3111,002ADP-ribose diphosphatase [EC:3.6.1.13] [E:3.6.1.-]nudFK01515GO: 004763112,474,1942,474,760566 Riboflavin biosynthesis pathway genes identified from the KEGG database using the genome of strain HY7715 diaminohydroxyphosphoribosylaminopyrimidine deaminase [E:3.5.4.26]/ 5-amino-6-(5-phosphoribosylamino)uracil reductase [EC:1.1.1.193] 3,4-dihydroxy 2-butanone 4-phosphate synthase [EC:4.1.99.12]/ GTP cyclohydrolase II [EC:3.5.4.25] The upstream region of the operon was further examined to identify the RFN element, a conserved regulatory RNA motif in prokaryotes that controls the expression of rib operon genes involved in riboflavin biosynthesis. Sequence inspection revealed the presence of a putative promoter region (Fig. 8a). The secondary structure of the FMN riboswitch aptamer predicted using the RNAfold web server closely resembled the conserved FMN aptamer structure (Fig. 8b, c) [66]. Based on the predicted secondary structure, a single nucleotide substitution (G123) unique to HY7715 compared with L. plantarum NCDO1752 and previously reported by Kim et al. [28], was located within the L6 loop of the FMN aptamer. Comparison with the well–characterized FMN aptamer of Fusobacterium nucleatum suggested that the corresponding nucleotide residue (G123) in HY7715 pairs with U55 (corresponding to A92 and U24 in F. nucleatum), participating in the loop–loop (L2–L6) interaction (Fig. 8d, e).Fig. 8Predicted FMN riboswitch aptamer structure of strain HY7715. (a) Nucleotide sequence of the leader region upstream of the rib operon. The − 35 and extended − 10 promoter elements (italicized), two possible transcription start sites (+ 1, asterisks), and the ribosomal binding site (RBS) of the first gene are indicated. (b) Predicted secondary structure of the FMN riboswitch aptamer obtained using the RNAfold web server. (c) Schematic representation of the conserved secondary structure of the FMN aptamer, showing the major stems (P1-P6), hairpin loops (L2–L6) and long–range RNA tertiary interactions forming the FMN binding pocket (scheme adapted from [66]). (d) Three–dimensional cartoon model of the FMN riboswitch in complex with FMN, derived from Fusobacterium nucleatum (PDB ID: 3F2Q). The T–loop (TL) motifs are shown in red, the two nucleotides immediately 3’ of the TL in magenta, and the T-loop receptors (TLRs) in blue [67]. The magnified inset shows the U24–A92 base pairing within the L2–L6 interaction region. (e) Sequence alignment of the FMN aptamer from strain HY7715, L. plantarum CB300 [68] and F. nucleatum (FN). Colors correspond to the secondary–structure elements shown in (c); TL/TLR regions follow the color scheme in (d), limited to the L2–L6 interaction. Red arrows in (a), (b) and the underline in (e) indicate the single nucleotide difference relative to L. plantarum NCDO1752 [28, 68] Predicted FMN riboswitch aptamer structure of strain HY7715. (a) Nucleotide sequence of the leader region upstream of the rib operon. The − 35 and extended − 10 promoter elements (italicized), two possible transcription start sites (+ 1, asterisks), and the ribosomal binding site (RBS) of the first gene are indicated. (b) Predicted secondary structure of the FMN riboswitch aptamer obtained using the RNAfold web server. (c) Schematic representation of the conserved secondary structure of the FMN aptamer, showing the major stems (P1-P6), hairpin loops (L2–L6) and long–range RNA tertiary interactions forming the FMN binding pocket (scheme adapted from [66]). (d) Three–dimensional cartoon model of the FMN riboswitch in complex with FMN, derived from Fusobacterium nucleatum (PDB ID: 3F2Q). The T–loop (TL) motifs are shown in red, the two nucleotides immediately 3’ of the TL in magenta, and the T-loop receptors (TLRs) in blue [67]. The magnified inset shows the U24–A92 base pairing within the L2–L6 interaction region. (e) Sequence alignment of the FMN aptamer from strain HY7715, L. plantarum CB300 [68] and F. nucleatum (FN). Colors correspond to the secondary–structure elements shown in (c); TL/TLR regions follow the color scheme in (d), limited to the L2–L6 interaction. Red arrows in (a), (b) and the underline in (e) indicate the single nucleotide difference relative to L. plantarum NCDO1752 [28, 68] Pangenome analysis To place HY7715 in a species–wide genomic context, a comparative pangenome analysis was conducted using 60 L. plantarum genomes, including HY7715, the type strain KCTC 3108T, and 58 publicly available genomes (Table S6). Roary analysis identified a total of 12,031 orthologous gene clusters, comprising 1,663 core genes, 441 soft core genes, 1,641 shell genes, and 8,286 cloud genes (Fig. 9a). Within this dataset, core genes accounted for approximately 13.8% of the total pangenome, whereas accessory genes (soft core, shell, and cloud genes) represented the majority, indicating substantial genomic variability among the L. plantarum strains considered.Fig. 9Pangenome analysis of 60 L. plantarum genomes, including HY7715. (a) Distribution of gene clusters across core, soft core, shell, and cloud categories as defined by Roary; (b) frequency distribution of gene clusters according to the number of genomes in which each gene is present; (c) pan–genome (blue) and core–genome (red) accumulation curves generated by iterative genome addition Pangenome analysis of 60 L. plantarum genomes, including HY7715. (a) Distribution of gene clusters across core, soft core, shell, and cloud categories as defined by Roary; (b) frequency distribution of gene clusters according to the number of genomes in which each gene is present; (c) pan–genome (blue) and core–genome (red) accumulation curves generated by iterative genome addition The gene frequency distribution exhibited a bimodal pattern (Fig. 9b), with a large proportion of strain–specific or low–frequency genes and a distinct peak corresponding to highly conserved core genes. This distribution pattern reflects the substantial accessory gene content observed within the analyzed strain set. Pan– and core–genome accumulation curves further illustrated the dynamic nature of the pangenome (Fig. 9c). The pan–genome size continued to increase with the sequential addition of genomes and did not reach a plateau, whereas the core–genome size gradually decreased and approached stabilization. These observations suggest that, within the analyzed dataset, L. plantarum displays features consistent with an open pangenome structure. A maximum–likelihood phylogenetic tree was constructed based on the concatenated alignment of core genes shared among the 60 analyzed genomes (Fig. 10). Among the analyzed strains, ATG-K2 formed a relatively distinct branch within the overall topology. The remaining strains were grouped into two major phylogenetic clusters. Within this topology, HY7715 was positioned in one of the two major lineages and formed a closely related subgroup with strains X7022, KF511, and K1. This placement indicates that HY7715 shares a conserved genomic background with closely related strains rather than representing a phylogenetically distinct lineage.Fig. 10Core genome–based phylogenetic analysis of 60 L. plantarum strains, including HY7715. The maximum–likelihood tree was reconstructed from the concatenated core gene alignment generated by Roary and inferred using IQ–TREE 2. Branch support values were estimated using 1,000 ultrafast bootstrap replicates. Colored circles indicate the isolation source of each strain, as defined in the legend. The scale bar represents the number of nucleotide substitutions per site Core genome–based phylogenetic analysis of 60 L. plantarum strains, including HY7715. The maximum–likelihood tree was reconstructed from the concatenated core gene alignment generated by Roary and inferred using IQ–TREE 2. Branch support values were estimated using 1,000 ultrafast bootstrap replicates. Colored circles indicate the isolation source of each strain, as defined in the legend. The scale bar represents the number of nucleotide substitutions per site Although strains were isolated from diverse ecological niches, no strict clustering pattern based solely on isolation source was observed. These results demonstrate substantial phylogenetic diversity within the analyzed strain set while situating HY7715 within a genomically conserved lineage. All–vs–all ANI analysis across the 60 L. plantarum genomes revealed several genome–wide similarity clusters (Fig. 11). Consistent with the core–genome phylogeny, strain ATG–K2 formed a clearly separated branch in the ANI–based hierarchical clustering, supporting its relatively divergent genomic position within the dataset. In contrast, HY7715 grouped most closely with strain K35, indicating high whole–genome relatedness between these two strains.Fig. 11FastANI–based genome–wide similarity of 60 L. plantarum strains. Pairwise ANI values were computed using FastANI in an all–vs–all comparison and visualized as a clustered heatmap. Dendrograms represent hierarchical clustering based on the ANI–derived distance matrix. HY7715 clustered most closely with strain K35, indicating high whole–genome relatedness within the dataset FastANI–based genome–wide similarity of 60 L. plantarum strains. Pairwise ANI values were computed using FastANI in an all–vs–all comparison and visualized as a clustered heatmap. Dendrograms represent hierarchical clustering based on the ANI–derived distance matrix. HY7715 clustered most closely with strain K35, indicating high whole–genome relatedness within the dataset Pangenome analysis identified 127 genes unique to HY7715 compared with the other 59 L. plantarum strains included in the dataset. The gene cluster composition of HY7715 and the remaining strains is summarized in Fig. 12. Across all strains, gene clusters were predominantly classified as core and soft–core categories.Fig. 12Pangenome composition of 60 L. plantarum strains. Stacked bar plot showing the proportional distribution of gene clusters across pangenome categories (core, soft–core, shell, and cloud) based on comparative analysis of 60 L. plantarum genomes. Strains are arranged according to the core-genome phylogenetic tree order. Each bar represents the relative proportion of gene clusters within a given strain, summing to 100%. The HY7715 strain is highlighted with a black outline, and the percentage values of each gene class are indicated within its bar. Overall, HY7715 exhibits a pangenome composition comparable to that of other strains, with the majority of gene clusters belonging to the conserved core and soft-core genome fractions Among 127 HY7715–specific genes, 81 genes were functionally annotated using EggNOG–mapper, whereas 46 genes remained unannotated (Fig. 13). COG classification of the annotated subset revealed that genes were distributed across multiple functional categories. The largest number of assignments corresponded to category L (replication, recombination, and repair; 28 assignments), followed by category S (function unknown; 17 assignments). Additional assignments were observed in transcription (K; 8), cell wall/membrane/envelope biogenesis (M; 7), carbohydrate transport and metabolism (G; 7), defense mechanisms (V; 5), amino acid transport and metabolism (E; 4), and inorganic ion transport and metabolism (P; 4). Smaller numbers of assignments were found in nucleotide transport and metabolism (F; 3), coenzyme transport and metabolism (H; 2), signal transduction mechanisms (T; 2), cell cycle control (D; 1), intracellular trafficking and secretion (U; 1), and secondary metabolite biosynthesis, transport and catabolism (Q; 1). Because some genes were assigned to more than one COG category, the total number of COG assignments exceeded the number of annotated genes. Overall, the HY7715-specific genes were broadly distributed across functional categories, with no single category accounting for a dominant proportion of the annotated gene set. Fig. 13Functional classification of HY7715 strain–specific genes based on COG categories. Bars represent the number of genes assigned to each COG functional class. Genes without COG assignment are categorized as “unclassified”. The majority of genes were annotated as hypothetical proteins or lacked functional annotation Pangenome composition of 60 L. plantarum strains. Stacked bar plot showing the proportional distribution of gene clusters across pangenome categories (core, soft–core, shell, and cloud) based on comparative analysis of 60 L. plantarum genomes. Strains are arranged according to the core-genome phylogenetic tree order. Each bar represents the relative proportion of gene clusters within a given strain, summing to 100%. The HY7715 strain is highlighted with a black outline, and the percentage values of each gene class are indicated within its bar. Overall, HY7715 exhibits a pangenome composition comparable to that of other strains, with the majority of gene clusters belonging to the conserved core and soft-core genome fractions Functional classification of HY7715 strain–specific genes based on COG categories. Bars represent the number of genes assigned to each COG functional class. Genes without COG assignment are categorized as “unclassified”. The majority of genes were annotated as hypothetical proteins or lacked functional annotation In Silico Identification of Probiotic Properties To evaluate the probiotic potential of HY7715 at the genomic level, we screened its genome for probiotic marker genes (PMGs), including those related to acid and bile tolerance, stress resistance, adhesion, and immunomodulation. The 191 PMGs were compiled based on previously published genome–based evaluations of probiotic strains [69–72]. The complete list of PMGs, along with their literature sources and the corresponding target species reported in previous studies, is provided in Table S7. Among the 191 PMGs surveyed, 138 genes were identified in the HY7715 genome. These comprised 68 genes related to acid and bile tolerance, 26 associated with oxidative stress, 14 encoding resistance to temperature stress, 12 involved in adhesion, 6 related to osmotic stress, and 5 associated with biofilm formation, as well as additional genes linked to immunomodulation, gut persistence, and ionic and heavy metal stress (Table 12).Table 12List of probiotic marker genes identified in the complete genome of strain HY7715. PMG count indicates the number of PMGs identified in HY7715 relative to the total number of PMGs defined in each functional categoryCategoryGenePMG count(HY7715/total)Acid and bile toleranceackA_1, ackA_3, arcB, copA, dnaJ, dnaK, eno, fabH_1, fabH_2, groES, grpE, htrA, pepF, pgk, ppk15/21Acid stressatpA, atpB, atpC, atpD, atpE, atpF, atpG, atpH, clpB, clpE, dltA, dltD, gadB, gap, gpmB, groEL, guaA_1, guaA_2, ldh_1, ldh_2, ldh_3, nhaC_1, nhaC_2, pgi, plsC, pyk, recA, relA, tpiA, uvrA_1, uvrA_2, uvrA_3, yjbM, ywaC34/40Bile resistanceargS, cfa_1, cfa_2, dps_2, glf_1, glnA, nagB, pdhD_1, pdhD_2, pepO, ponA, ppaC, pyrG, rplD, rplE, rplF, rpsC, rpsE, rpsS19/26Oxidative stressfnr_1, fnr_2, mntB_1, mntB_2, mntH_1, mntH_2, mntH_3, msrA_1, msrA_2, msrA_3, msrB, msrC, nox_1, nox_2, nox_3, nox_4, npr_1, npr_2, nrdH, tpx, trxA_1, trxA_2, trxA_3, trxA_4, trxB, trxB_226/43Temperature stressclpC, clpL, clpP_1, clpP_2, clpP_3, clpX, csp, cspL, cspLA, ctsR, hrcA, hslO, hslV, htpX14/17Adhesioneno, epsH, exoA, glnH_1, glnH_2, gpr_1, gpr_2, lspA, pgaC, pyc, strA, tuf12/19Osmotic stressgla, glpF_1, glpF_2, opuCA, opuCB, osmV6/9Biofilm formationcomC, comEA, comEC_1, comEC_2, luxS5/7ImmunomodulationdltB, dltC1_1, dltC1_2, dltD4/4Gut persistencecelB, treA2/5Ionic and heavy metal stressczcD1/1 Table 12List of probiotic marker genes identified in the complete genome of strain HY7715. PMG count indicates the number of PMGs identified in HY7715 relative to the total number of PMGs defined in each functional categoryCategoryGenePMG count(HY7715/total)Acid and bile toleranceackA_1, ackA_3, arcB, copA, dnaJ, dnaK, eno, fabH_1, fabH_2, groES, grpE, htrA, pepF, pgk, ppk15/21Acid stressatpA, atpB, atpC, atpD, atpE, atpF, atpG, atpH, clpB, clpE, dltA, dltD, gadB, gap, gpmB, groEL, guaA_1, guaA_2, ldh_1, ldh_2, ldh_3, nhaC_1, nhaC_2, pgi, plsC, pyk, recA, relA, tpiA, uvrA_1, uvrA_2, uvrA_3, yjbM, ywaC34/40Bile resistanceargS, cfa_1, cfa_2, dps_2, glf_1, glnA, nagB, pdhD_1, pdhD_2, pepO, ponA, ppaC, pyrG, rplD, rplE, rplF, rpsC, rpsE, rpsS19/26Oxidative stressfnr_1, fnr_2, mntB_1, mntB_2, mntH_1, mntH_2, mntH_3, msrA_1, msrA_2, msrA_3, msrB, msrC, nox_1, nox_2, nox_3, nox_4, npr_1, npr_2, nrdH, tpx, trxA_1, trxA_2, trxA_3, trxA_4, trxB, trxB_226/43Temperature stressclpC, clpL, clpP_1, clpP_2, clpP_3, clpX, csp, cspL, cspLA, ctsR, hrcA, hslO, hslV, htpX14/17Adhesioneno, epsH, exoA, glnH_1, glnH_2, gpr_1, gpr_2, lspA, pgaC, pyc, strA, tuf12/19Osmotic stressgla, glpF_1, glpF_2, opuCA, opuCB, osmV6/9Biofilm formationcomC, comEA, comEC_1, comEC_2, luxS5/7ImmunomodulationdltB, dltC1_1, dltC1_2, dltD4/4Gut persistencecelB, treA2/5Ionic and heavy metal stressczcD1/1 List of probiotic marker genes identified in the complete genome of strain HY7715. PMG count indicates the number of PMGs identified in HY7715 relative to the total number of PMGs defined in each functional category The prevalence of acid– and bile–tolerance–related genes is consistent with the high survival rates observed in the corresponding phenotypic assays (Sect. 3.13). Pangenome mapping revealed that more than 90% of the PMGs identified in HY7715 were classified as core or soft–core genes within the L. plantarum pangenome. Only a small fraction belonged to shell (7%) or cloud (1%) categories (Fig. 14). The shell and cloud PMGs were primarily associated with oxidative stress and bile tolerance functions. These findings suggest that HY7715 harbors a largely conserved probiotic–associated genomic repertoire characteristic of the species, with a small accessory component that may contribute to strain–specific ecological adaptation.Fig. 14Distribution of PMGs (n = 138) of HY7715 across pangenome categories. The pie chart summarizes the classification of PMGs into core, soft–core, shell, and cloud gene clusters based on pangenome analysis of 60 L. plantarum genomes. The majority of PMGs identified in HY7715 were assigned to core (71%) or soft–core (21%) categories. The table lists PMGs belonging to shell and cloud categories, including their functional classification and prevalence among the analyzed strains. This distribution indicates that most probiotic–associated traits of HY7715 correspond to conserved species–level features Distribution of PMGs (n = 138) of HY7715 across pangenome categories. The pie chart summarizes the classification of PMGs into core, soft–core, shell, and cloud gene clusters based on pangenome analysis of 60 L. plantarum genomes. The majority of PMGs identified in HY7715 were assigned to core (71%) or soft–core (21%) categories. The table lists PMGs belonging to shell and cloud categories, including their functional classification and prevalence among the analyzed strains. This distribution indicates that most probiotic–associated traits of HY7715 correspond to conserved species–level features Carbohydrate Utilization Profiling Carbohydrate utilization capacity of HY7715 were evaluated using the API 50CH system and compared with the type strain KCTC 3108T. In addition, previously reported API 50CH results for several L. plantarum strains (ATG-K2, ATG-K6, ATG-K8, DY46, TRA56, and Y44) were included for comparative analysis based on published studies [73–75] (Fig. 15). HY7715 showed positive fermentation of L–arabinose, ribose, galactose, D–glucose, D–fructose, D–mannose, mannitol, sorbitol, α–methyl–D–mannoside, α–methyl–D–glucosamine, N–acetyl–glucosamine, amygdalin, arbutin, esculin, salicin, cellobiose, maltose, lactose, melibiose, saccharose, trehalose, melezitose, α–gentiobiose, and D–turanose, while rhamnose was weakly positive. Overall, HY7715 exhibited a carbohydrate utilization pattern broadly consistent with previously reported L. plantarum strains. However, minor differences were observed in the utilization of rhamnose and α–methyl–D–glucosamine, and gluconate, indicating strain–level metabolic variability within the species.Fig. 15Carbohydrate utilization profiling of L. plantarum HY7715 and reference strains based on API 50CH assays. Fermentation results are presented as a heatmap, where blue indicates negative, green indicates weak, and yellow indicates positive reactions (scale bar, fermentation degree). API 50CH data for strains ATG–K2, ATG–K6, ATG–K8, DY46, TRA56, and Y44 were obtained from previously published studies [73–75], while results for HY7715 and KCTC 3108T were generated in the present study Carbohydrate utilization profiling of L. plantarum HY7715 and reference strains based on API 50CH assays. Fermentation results are presented as a heatmap, where blue indicates negative, green indicates weak, and yellow indicates positive reactions (scale bar, fermentation degree). API 50CH data for strains ATG–K2, ATG–K6, ATG–K8, DY46, TRA56, and Y44 were obtained from previously published studies [73–75], while results for HY7715 and KCTC 3108T were generated in the present study CAZyme genes were annotated from the genomes of HY7715, KCTC 3108T, and 58 reference strains using dbCAN3 (Table S8). The HY7715 genome contained 102 CAZy families, comprising 52 glycoside hydrolase (GH) families, 35 glycosyltransferase (GT) families, 4 carbohydrate esterase (CE) families, 9 carbohydrate–binding module (CBM) families, and 2 auxiliary activities (AA) families. To investigate CAZyme distribution patterns, CAZy family profiles were compared across all 60 strains. Strain clustering in the CAZyme heatmap was performed based on hierarchical clustering of CAZy family profiles (Fig. 16). Across the dataset, GH and GT families were the most abundant CAZy classes. In particular, GH1, GH13, GH25, GT2, and GT4 were widely distributed and exhibited relatively consistent copy numbers across strains.Fig. 16CAZy family profiling of L. plantarum HY7715 and reference strains. In silico prediction of CAZy family members, including glycoside hydrolases (GH), glycosyltransferases (GT), carbohydrate esterases (CE), carbohydrate–binding modules (CBM), and auxiliary activities (AA), shown as the number of hits per genome (scale bar, number of hits). For CAZy profiling, genome sequences of KCTC 3108T and 58 publicly available L. plantarum strains isolated from diverse ecological niches were retrieved from GenBank and analyzed together with HY7715. Strains were clustered using hierarchical clustering based on CAZy profiles. CAZy families were not clustered to facilitate direct family-wise comparison CAZy family profiling of L. plantarum HY7715 and reference strains. In silico prediction of CAZy family members, including glycoside hydrolases (GH), glycosyltransferases (GT), carbohydrate esterases (CE), carbohydrate–binding modules (CBM), and auxiliary activities (AA), shown as the number of hits per genome (scale bar, number of hits). For CAZy profiling, genome sequences of KCTC 3108T and 58 publicly available L. plantarum strains isolated from diverse ecological niches were retrieved from GenBank and analyzed together with HY7715. Strains were clustered using hierarchical clustering based on CAZy profiles. CAZy families were not clustered to facilitate direct family-wise comparison Overall, HY7715 displayed a CAZy composition comparable to that of other L. plantarum strains, without evident enrichment or depletion of specific CAZy families relative to the clustered groups. Antimicrobial Susceptibility Antimicrobial susceptibility of HY7715 was evaluated based on MIC test according to the EFSA guideline [63]. The MIC values for ampicillin, gentamicin, erythromycin, clindamycin, tetracycline, and chloramphenicol were within the established EFSA cut–off limits established for L. plantarum. However, the MIC for kanamycin exceeded the corresponding cut–off value, indicating that the strain is resistant to kanamycin (Table 13, Figure S1).Table 13Minimum inhibitory concentrations (MICs) of antibiotics for L. plantarum HY7715AntibioticsMIC(µg/mL)Cut–off values* (µg/mL)InterpretationAmpicillin0.0942SVancomycin> 256n.r.n.r.Gentamicin816SKanamycin> 25664RStreptomycin128n.r.n.r.Erythromycin11SClindamycin1.54STetracycline1232SChloramphenicol68S* Cut–off values were applied according to the EFSA FEEDAP Panel guidance (2018) [63]; n.r., not required; S, sensitive; R, resistance Table 13Minimum inhibitory concentrations (MICs) of antibiotics for L. plantarum HY7715AntibioticsMIC(µg/mL)Cut–off values* (µg/mL)InterpretationAmpicillin0.0942SVancomycin> 256n.r.n.r.Gentamicin816SKanamycin> 25664RStreptomycin128n.r.n.r.Erythromycin11SClindamycin1.54STetracycline1232SChloramphenicol68S* Cut–off values were applied according to the EFSA FEEDAP Panel guidance (2018) [63]; n.r., not required; S, sensitive; R, resistance Minimum inhibitory concentrations (MICs) of antibiotics for L. plantarum HY7715 * Cut–off values were applied according to the EFSA FEEDAP Panel guidance (2018) [63]; n.r., not required; S, sensitive; R, resistance Acid and Bile Tolerance Acid and bile tolerance of HY7715 and KCTC 3108T were evaluated by determining the survival rate under simulated gastrointestinal conditions (Fig. 17). Both strains demonstrated high survival rates in acidic and bile conditions. Under acidic condition (1% pepsin, pH 2.5), more than 60% of both HY7715 and KCTC 3108T survived, with HY7715 exhibiting a significantly higher survival rate than KCTC 3108T (p < 0.05). In bile condition (0.3% bile, pH 8.0), although no significant difference was observed between strains (p > 0.05), both strains showed survival rates exceeding 70%.Fig. 17Acid and bile tolerance of L. plantarum HY7715 and KCTC 3108T. The survival rate was determined by comparing viable cell counts (cfu/mL) after 2 h of exposure to acidic (1% pepsin, pH 2.5) or bile (0.3% bile, pH 8.0) conditions relative to the control (pH 6.2). Data are presented as mean ± SD (n = 3). Comparative analyses between HY7715 and the type strain were performed using multiple unpaired t–tests with Holm–Šídák multiple comparisons correction. Statistical significance was defined as p < 0.05 and is indicated by asterisks in the figure Acid and bile tolerance of L. plantarum HY7715 and KCTC 3108T. The survival rate was determined by comparing viable cell counts (cfu/mL) after 2 h of exposure to acidic (1% pepsin, pH 2.5) or bile (0.3% bile, pH 8.0) conditions relative to the control (pH 6.2). Data are presented as mean ± SD (n = 3). Comparative analyses between HY7715 and the type strain were performed using multiple unpaired t–tests with Holm–Šídák multiple comparisons correction. Statistical significance was defined as p < 0.05 and is indicated by asterisks in the figure Hemolytic Activity As shown in Fig. 18, L. plantarum HY7715 did not show significant hemolytic activity on BAP, whereas S. aureus ATCC 33519 exhibited rapid growth and a completely clear zone around the cells.Fig. 18Hemolytic activity of S. aureus ATCC 33519 and L. plantarum HY7715. (a) S. aureus ATCC 33519 exhibited complete hemolysis with a clear zone surronding, indicating β–hemolysis. (b) HY7715 showed no visible hemolysis (γ–hemolysis) Hemolytic activity of S. aureus ATCC 33519 and L. plantarum HY7715. (a) S. aureus ATCC 33519 exhibited complete hemolysis with a clear zone surronding, indicating β–hemolysis. (b) HY7715 showed no visible hemolysis (γ–hemolysis) Discussion Comprehensive safety assessment at the strain level is a critical prerequisite for the commercialization of probiotic candidates. Although L. plantarum is included in the EFSA’s QPS list, EFSA clearly emphasizes that safety must be confirmed at the strain level [13]. In this context, the present study provides an integrated genome–based and phenotypic evaluation of HY7715, a riboflavin–overproducing strain with demonstrated functional potential [28]. Collectively, our findings indicate that HY7715 fulfills current safety criteria while retaining functional traits relevant to probiotic performance. Genome–wide screening revealed no detectable VFs or transferable antimicrobial resistance determinants, and MGEs did not harbor genes associated with undesirable traits. These genomic observations were supported by phenotypic analyses demonstrating antibiotic susceptibility (with intrinsic resistance to kanamycin), absence of hemolytic activity, high tolerance to acidic and bile conditions, and lack of biogenic amine production. Taken together, these data support the conclusion that HY7715 represents a functionally promising yet genomically stable strain suitable for probiotic development. The complete genome of strain HY7715 was identified through hybrid assembly combining short–read (Illumina) and long–read (PacBio) sequencing data (Fig. 1). Read–level contamination analysis performed using Kraken2 based on both datasets did not fully comply with the EFSA guideline, which recommends that reads assigned to unexpected taxa should remain below 5% at species level [29]. Kraken2 constructs its database from NCBI RefSeq genomes by fragmenting sequences into k–mers and mapping them onto nodes of a phylogenetic tree [34]. During classification, sequencing reads are split into k–mers and matched against the database, with taxonomic assignment based on the lowest common ancestor (LCA) algorithm. Although this method enables ultra–fast and memory–efficient classification, it can have false–positive or ambiguous assignments, particularly among closely related taxa sharing high genomic similarity, because the method relies on exact k–mer matching [34, 76, 77]. Moreover, continuous database growth and compositional updates can alter Kraken2’s k–mer database and thus influence the reproducibility of taxonomic assignment. Several previous studies have also reported that Illumina short–read data tend to yield higher rates of false–positive and false–negative classifications compared with long–read sequencing platforms such as Oxford Nanopore or PacBio [78, 79]. This bias arises from the fact that the shorter read length limits the number of unique k–mers, which consequently hinders accurate mapping, especially in genomes with high sequence homology among related species. Therefore, the exceedance above the EFSA limit is considered a methodological artifact rather than biological contamination. Supporting this interpretation, genome contamination and completeness analyses performed using ContEst16S, UBCG, and BUSCO all confirmed that the genome sequence of strain HY7715 was completely assembled and free of contamination. All five 16 S rRNA gene copies extracted from the assembled genome were identical, indicating the absence of sequence heterogeneity or exogenous DNA. In addition, both UBCG and BUSCO analyses showed 100% completeness, further supporting the high assembly quality and genomic integrity of strain HY7715. Functional annotation of the identified CDSs using EggNOG–mapper provided a comprehensive overview of the genomic repertoire and inferred physiological capacities of HY7715 (Fig. 2). As shown in the COG–based classification, genes were distributed across major functional classes, with the highest proportions observed in transcription (K), carbohydrate transport and metabolism (G), and amino acid transport and metabolism (E). The predominance of transcription–related genes (K) suggests a well–developed regulatory network capable of modulating gene expression in response to fluctuating environmental conditions, thereby supporting adaptive regulation of metabolic and stress–response pathways during growth and ecological transitions [80, 81]. This was followed by carbohydrate transport and metabolism (G), and amino acid transport and metabolism (E). These groups include enzymes and transporters involved in the uptake, degradation, and interconversion of diverse carbohydrate and amino acids, thereby linking carbon and nitrogen metabolism. Such a genomic configuration is commonly interpreted as evidence of metabolic versatility, enabling microorganisms to thrive under variable nutrient conditions and to exploit complex resources within their ecological niches [82, 83]. Taxonomic identification of strain HY7715 was performed using both 16 S rRNA gene sequence comparison and whole–genome ANI. BLASTn search of the 16 S rRNA gene sequence against the EzBioCloud database showed that the closest match was L. pentosus DSM 20314T (99.93% similarity), followed by L. plantarum DSM 20174T (99.87%). However, these two species share nearly identical 16 S rRNA gene sequences, differing by only a nucleotide, which makes it difficult to achieve species–level discrimination within the L. plantarum–L. pentosus complex. Such minimal sequence divergence often results in ambiguous classification and is a well–recognized limitation of ribosomal gene–based taxonomy for closely related LAB [84]. Consistent with this observation, the maximum–likelihood phylogenetic tree based on 16 S rRNA gene sequences positioned strain HY7715 within the L. plantarum–L. pentosus complex, forming a distinct branch closely related to L. herbarum (Fig. 3). This topology indicates that HY7715 is phylogenetically close to the L. plantarum group, while harboring minor sequence variations that distinguish it from other members of the complex. Such topological separation likely reflects microevolutionary divergence within the L. plantarum clade rather than true taxonomic distinctness. Whole–genome ANI analysis resolved this ambiguity, clearly identified HY7715 as L. plantarum (99.12% identity to L. plantarum DSM 20174T) and showing only 80.31% to L. pentosus DSM 20314T, far below the species–level threshold (> 94%) (Table 3). Consistently, a phylogenetic tree based on ANI values demonstrated that HY7715 clustered tightly with L. plantarum DSM 20174T, confirming its taxonomic position within the Lactiplantibacillus plantarum lineage (Fig. 4). Ensuring genomic stability is a key prerequisite for sustaining the functionality and safety of probiotic strains. To evaluate this aspect, the genomic stability of strain HY7715 was assessed by comparing the whole–genome sequences of the mother culture and the strain after 200 generations. The ANI value of 99.99%, combined with the detection of only 15 SNPs across the approximately 3.3 Mb genome, suggests that HY7715 remained stable during prolonged cultivation, demonstrating its high genomic stability (Table 4). According to the EFSA guidelines, screening for VF and AMR genes should be performed using the lowest available coverage threshold, and all hits showing at least 80% sequence identity and 70% coverage must be reported [29]. Moreover, if multiple fragments collectively cover less than 70% but individually exhibit ≥ 80% identity to the same reference sequence, these should also be reported to verify the full gene’s presence. The use of at least two well–maintained AMR databases is further recommended to ensure reliability and cross–validation of results. Following these guidelines, the genome of HY7715 was screened against the VFDB for VF genes, and against three independent databases —AMRFinderPlus, ResFinder, and MEGARes—for AMR genes, using thresholds of 80% identity and 50% minimum coverage. No virulence factors or antimicrobial resistance genes were detected in the HY7715 genome. The potential for HGT was further assessed by analyzing mobile genetic elements (MGEs), which are known vehicles of genetic exchange. HGT can occur through vector–mediated mechanisms involving plasmids, transposons, or prophages that facilitate gene movement among microorganisms [85, 86]. MGEs are ubiquitous components of bacterial genomes and play key roles in evolution and environmental adaptation. The presence of MGEs is a common feature among lactic acid bacteria, including probiotic candidates, and the potential risk associated with HGT largely depends on whether these elements harbor AMR or VF genes, as emphasized in recent assessments of probiotic safety and mobilome structure [87, 88]. MGE analysis of the HY7715 genome identified four plasmids, 25 transposase-related elements, and five potential prophage regions (Tables 5, 6, 7 and 8). Importantly, none of these elements carried AMR or VF genes. As the risk associated with MGEs primarily depends on their genetic cargo rather than their mere presence, the likelihood of HGT–mediated dissemination of undesirable traits from HY7715 is considered negligible. These findings are consistent with previous whole–genome–based safety assessments of L. plantarum strains, in which MGEs such as plasmids, transposons, and prophage regions were identified but did not harbor AMR or VF genes, and were therefore interpreted as posing a low risk of HGT–mediated dissemination of undesirable traits [89, 90]. Through MIC testing, HY7715 was susceptible to multiple antibiotics but resistant to kanamycin (Table 13). Genome–wide screening did not detect any acquired antimicrobial resistance genes. Moreover, none of the identified MGEs, including plasmids, transposase–related elements, or prophage regions, harbored AMR determinants. Reduced susceptibility or intrinsic resistance to aminoglycosides, including kanamycin and streptomycin, has been widely reported in L. plantarum and other lactobacilli in the absence of acquired resistance genes [91–95]. Such intrinsic aminoglycoside resistance is generally attributed to physiological characteristics of LAB, including low membrane potential and limited respiratory activity, which reduce aminoglycoside uptake [96–98]. Taken together, the absence of detectable AMR gene and the lack of association with MGEs provide direct genomic evidence that the observed kanamycin resistance in HY7715 represents an intrinsic, non–transferable phenotype rather than an acquired resistance trait with horizontal transfer potential. CRISPR array screening identified a complete Type IIA CRISPR–Cas system in the genome of HY7715, comprising one CRISPR array and four associated cas genes (cas1, cas2, cas9, and csn2) (Table 9). The close arrangement of the cas operon and CRISPR array suggests that the system is structurally intact and potentially functional [99–101]. CRISPR–Cas systems are known to provide sequence–specific defense against foreign genetic elements such as bacteriophages and plasmids [102–104]. In probiotic candidates, such systems are often considered beneficial because they may limit the acquisition of exogenous DNA. Accordingly, the presence of an intact CRISPR–Cas system may further contribute to genomic stability by reducing the likelihood of horizontal transfer of virulence or antimicrobial resistance genes. Bacterial secondary metabolites are not essential for their growth but play important roles in microbial communication, stress response, and ecological competition. These compounds are produced during the stationary phase in response to nutrient limitation and environmental stress [105]. They exhibit diverse biological activities such as antimicrobial, anticancer, anti–inflammatory, antioxidant, and immunomodulatory effects, thus are considered as candidates of clinical application [106–108]. However, since certain secondary metabolites are cytotoxic or related to virulence, the evaluation of the presence of biosynthetic gene clusters (BGCs) encode such metabolites is important for food–grade microorganisms [109]. Five BGCs, predicted as RiPP–like, terpene, and T3PKS clusters, were identified from the genome of HY7715 using antiSMASH v8.0 (Fig. 5). These were potentially related to antimicrobial activity, antioxidative defense, or ecological interactions. Importantly, none of the predicted BGCs showed significant similarity to known toxin–associated clusters in the MIBiG database based on antiSMASH analysis, suggesting that HY7715 does not harbor BGCs associated with harmful secondary metabolites. Additionally, BAGEL4 analysis confirmed that region 1.1 within these BGCs is part of the bacteriocin gene cluster (Fig. 6). The predicted gene cluster in HY7715 encodes plantaricin EF, a representative Class IIb bacteriocin of L. plantarum. Plantaricin EF consists of two complementary peptides (PlnE and PlnF) that form transmembrane pores in target bacteria, leading to disruption of intracellular pH homeostasis and membrane potential [110–112]. Members of this bacteriocin class are generally considered to exhibit low cytotoxicity toward eukaryotic cells while maintaining antimicrobial activity against foodborne pathogens and antibiotic–resistant bacteria [113]. In addition, plantaricin EF has been reported to retain activity under a wide range of temperature and pH conditions, supporting its functional stability in gastrointestinal environments [114]. Importantly, comparative genomic analyses have demonstrated that the plnE/F locus is widely distributed and conserved among L. plantarum strains isolated from diverse ecological niches [115]. These findings indicate that plantaricin EF is not a rare or strain–specific acquisition, but rather part of the conserved bacteriocin repertoire of the species. In this context, the presence of the plantaricin EF cluster in HY7715 appears to reflect of species–level functional trait. BAs are bioactive compounds with low molecular weight and produced through the decarboxylation of precursor amino acids. Among them, histamine and tyramine represent the primary food–safety concerns, while putrescine and cadaverine can potentiate their toxic effects [116, 117]. To evaluate the genetic potential for BA formation, the genome of HY7715 was analyzed using BlastKOALA with the KEGG database, which encompasses the major pathways responsible for BA biosynthesis (Table 10). There were no genes related to the production of histidine, tyramine, putrescine, cadaverine, spermidine, spermine, ornithine, and tryptamine, indicating that HY7715 lacks the genetic potential to produce BAs via amino acid decarboxylation. To experimentally confirm these genomic findings, BA production was assessed using screening media supplemented with the specific precursor amino acids, including L–histidine monohydrochloride, L–tyrosine, L–ornithine monohydrochloride, and L–lysine, corresponding to the formation of histamine, tyramine, putrescine, and cadaverine, respectively. HY7715 exhibited no color change indicative of BA formation, confirming that it can be regarded as a non–BA–producing strain from a food–safety perspective (Fig. 7). The FMN riboswitch consists of two distinct domains; an aptamer domain, which specifically recognizes FMN (ligand) and forms a tertiary structure that provides a ligand–binding pocket, and an expression platform that functions as a regulatory switch, modulating transcription or translation of downstream rib operon [118]. The aptamer’s tertiary structure, FMN–binding pocket, accommodates the isoalloxazine ring and phosphate moieties of FMN through an intricate network of hydrogen bonds, stacking interactions, and Mg²⁺-mediated contacts [119, 120]. The overall structure is stabilized by two long-range loop–loop interactions, L2–L6 and L3–L5 (Fig. 8c), which known as T–loop/T–loop receptor (TL/TLR) interaction [67]. These long–range interactions are important for proper alignment of the helices surrounding the FMN binding site, and therefore induces proper conformational changes on expression platform, regulating transcription or translation of rib operon [67, 121]. The riboflavin–overproducing phenotype of HY7715 were previously reported by Kim et al. [28], where rib operon genes were identified through expression analysis using PCR. In the present study, genome–wide KEGG pathway analysis reconfirmed that HY7715 contains all genes of the rib operon (Table 11). Notably, a unique nucleotide substitution (G123) compared to L. plantarum NCDO1752, which harbors a complete rib operon, was located within the L6 loop of the FMN aptamer, corresponding to a residue participating in the L2–L6 tertiary contact observed in the F. nucleatum structure. Given the critical role of this loop–loop interaction in maintaining aptamer stability and ligand affinity, this substitution may subtly alter stacking geometry or weaken interloop hydrogen–bonding, thereby modulating the efficiency of FMN–dependent conformational switching. Consequently, the G123 substitution could influence riboswitch performance and regulatory sensitivity, potentially contributing to the enhanced riboflavin–overproducing phenotype of strain HY7715. This finding explains the previous observations that HY7715 produced similar amounts of riboflavin regardless of the presence or absence of riboflavin in the culture medium [28]. Burgess et al. [68] reported mutation sites of riboflavin–overproducing mutants of L. plantarum NCDO1752, strain CB300 exhibited a point mutation within the L2–L6 interaction region, which is similar to our findings (Fig. 8e). Moreover, several studies reported that a single nucleotide mutation in the RFN region alters their rib operon regulation ability, inducing higher riboflavin production [68, 122, 123]. Comparative pangenome analysis of 60 L. plantarum genomes revealed a typical open pangenome structure, characterized by a continuous increase in pan–genome size with the addition of new genomes and a gradual stabilization of the core genome (Fig. 9). This pattern is consistent with previous large–scale comparative genomic studies of L. plantarum, which have described the species as genetically diverse and possessing an open pangenome architecture [69, 124]. Such genomic plasticity has been proposed to underline the ecological flexibility of L. plantarum, enabling adaptation to a wide range of environments. The relatively small proportion of core genes and the predominance of accessory genes observed in this study further support the notion that L. plantarum exhibits substantial intra–species variability. This genomic configuration is in agreement with earlier reports describing L. plantarum as a “nomadic” species that does not display strict niche specialization at the genomic level [124]. Consistently, phylogenetic clustering based on core genome alignment did not reveal clear grouping patterns strictly associated with isolation source, suggesting that ecological origin alone does not fully determine genomic relatedness within the species (Fig. 10). Within this framework, HY7715 was positioned in one of two major phylogenetic lineages and clustered closely with strains X7022, KF511, and K1, indicating that it shares a conserved genomic backbone with closely related strains rather than representing a distinct evolutionary lineage. This finding aligns with the broader understanding that many probiotic candidate strains of L. plantarum derive their functional properties from a conserved species–wide genomic repertoire rather than from unique phylogenetic divergence [124, 125]. An independent whole–genome similarity assessment based on all–vs–all FastANI analysis further supported this phylogenetic framework (Fig. 11). Consistent with the core–genome phylogeny, strain ATG–K2 appeared as a relatively distinct branch in the ANI–based hierarchical clustering, reinforcing its divergent genomic position within the dataset. In contrast, HY7715 grouped most closely with strain K35, indicating high overall genomic similarity. Minor differences between the ANI–based clustering and the core–gene phylogeny are expected, as FastANI estimates genome–wide average nucleotide identity across shared genomic regions, whereas phylogenetic reconstruction is inferred from concatenated alignments of conserved core genes. Collectively, these complementary analyses position HY7715 within a genomically conserved lineage of L. plantarum. Although L. plantarum is widely recognized as possessing an open pangenome, the gene cluster composition within individual genomes was dominated by core and soft–core genes, with cloud genes representing only a minor fraction (Fig. 12). HY7715 followed this species–wide pattern. This observation highlights that, despite extensive pan–genomic diversity at the species level, individual genomes remain largely structured around a conserved genetic backbone. Pangenome analysis identified 127 genes unique to HY7715, of which 46 remained unannotated, highlighting the limited functional characterization of a substantial fraction of strain-specific genes (Fig. 13). Among the annotated subset, COG classification revealed that the largest number of assignments corresponded to replication, recombination, and repair (L), followed by genes of unknown function (S). Additional assignments were distributed across diverse functional categories, including transcription, membrane biogenesis, carbohydrate metabolism, and defense mechanisms. Importantly, no single functional category dominated the annotated gene set. Furthermore, the predominance of replication- and recombination-related assignments, together with a considerable proportion of unannotated genes, is consistent with the typical composition of accessory gene pools observed in bacterial pangenomes [126]. These findings suggest that the HY7715-specific gene set does not represent the acquisition of large, functionally specialized clusters, but rather reflects dispersed accessory elements within the broader species-level genomic framework of L. plantarum. The PMGs analyzed in the present study represent functional determinants that have been repeatedly associated with probiotic–related phenotypes in genome–guided evaluations of LAB, including stress tolerance, adhesion, metabolic adaptability, and host interaction, rather than genes exclusively restricted to probiotic strains [51, 52, 74]. Recent in silico and integrative genomic studies have similarly emphasized that many probiotic–associated traits are broadly distributed within species and reflect conserved adaptive capacities rather than strain–specific signatures [127]. Consistent with this concept, most PMGs identified in HY7715 were classified as core or soft–core genes within the L. plantarum dataset, whereas only approximately 8% belonged to the accessory (shell or cloud) genome (Fig. 14). This distribution indicates that probiotic–associated capacities observed in HY7715 are primarily embedded within the conserved genomic backbone of the species. Accordingly, the stress adaptation and gastrointestinal survival capacities identified in HY7715 appear to reflect a representative genomic profile of L. plantarum, rather than highly strain–specific innovations. In line with this interpretation, several comparative studies have reported that L. plantarum strains frequently exhibit strong tolerance to acidic and bile conditions relative to other LAB species commonly used as probiotics, reflecting their adaptability to gastrointestinal environments [128, 129]. The HY7715 genome harbored numerous genes associated with acid and bile tolerance (Table 12). The observed tolerance to acidic and bile conditions is likely supported by coordinated mechanisms, including intracellular pH homeostasis mediated by active proton extrusion systems such as the F0F1–ATP synthase complex (atpA–atpH) and the Na+/H+ antiporter NhaC (nhaC); amino acid–dependent pH buffering and nitrogen metabolism (arcB, argS, glnA); sustained energy production (ackA, eno, pgk, pdhD, ppaC); protein quality control system (dnaK–dnaJ–grpE, htrA, pepO); membrane and cell envelop remodeling (fabH, cfa, glf, ponA); and DNA protection and oxidative stress defense (dps, copA, pyrG) [130–135]. Collectively, these systems provide a mechanistic basis for the high survival rate of HY7715 under simulated gastrointestinal conditions (Fig. 17), a key attribute for probiotic functionality. Oxidative stress tolerance further contributes to gastrointestinal persistence, functional stability, and industrial robustness. The HY7715 genome contains multiple genes involved in oxidative stress responses, including fnr, a global transcriptional regulator responsive to oxygen and redox changes [136]; manganese transporter genes (mntB and mntH), which enhance resistance to superoxide stress through intracellular Mn accumulation [137]; methionine sulfoxide reductase genes (msrA, msrB, and msrC), which repair oxidatively damaged proteins [138]; nox, encoding NADH oxidase that reduces intracellular oxygen levels [139]; and thioredoxin–dependent redox network genes (trxA, trxB, tpx, and nrdH), which maintain thiol–redox homeostasis [140–142]. These coordinated redox defense systems likely reinforce cellular resilience under fluctuating oxygen and oxidative conditions encountered during gastrointestinal transit. In addition, several adhesion–related genes were identified, including exopolysaccharide (EPS) biosynthesis genes (epsH, exoA) [143]; moonlighting surface protein genes (eno, tuf), which facilitate initial attachment to epithelial or mucosal surfaces [144–146]; sortase (strA) [129]; and the competence–stimulating peptide gene comC, which links quorum sensing to biofilm formation [147]. Together, these determinants may facilitate mucosal adherence, biofilm formation, and persistence within the gastrointestinal tract. Carbohydrate utilization profiling demonstrated that HY7715 is capable of fermenting a broad range of carbohydrates in the API 50CH assay. Such versatility is consistent with the well–recognized metabolic flexibility of L. plantarum, a species known for its relatively large genome and adaptation to carbohydrate–rich ecological niches, including fermented foods and plant–associated environments [148]. In our comparative analysis, HY7715 fermented a slightly broader set of substrates than some previously reported strains; however, the overall utilization pattern remained largely consistent with that described for other L. plantarum strains (Fig. 15). A notable observation was the inability of HY7715 to ferment gluconate in the API 50CH assay, despite the presence of genes predicted to participate in gluconate metabolism via alternative pathways, including the pentose phosphate pathway (PPP) or an Entner–Doudoroff–like pathway (Table S9) [149]. This discrepancy may reflect regulatory constraints, pathway flux partitioning, or assay–specific limitations. The potential redirection of gluconate into the PPP could theoretically enhance NADPH generation, thereby supporting redox homeostasis and oxidative stress defense [150]. Nevertheless, this hypothesis is based on genomic inference and would require targeted experimental validation. Comparative CAZyme profiling further contextualized these phenotypic findings. Across the 60 L. plantarum genomes analyzed, GH and GT families constituted the dominant CAZy classes (Fig. 16). Major families such as GH1 (including β–glucosidases and β–galactosidases), GH13 (α–amylase family), GH25 (including chitosanase and chitinase), GT2 (primarily involved into cell wall biosynthesis), and GT4 (mainly responsible for sucrose biosynthesis) were broadly distributed and showed relatively consistent copy numbers across strains. Consistent with the clustering pattern observed in the CAZy heatmap, HY7715 did not exhibit pronounced enrichment or depletion of major GH or GT families relative to other strains. Although minor variation was detected in certain families, including CBMs, these differences fell within the expected range of intra–species variability. Taken together, the concordance between API 50CH phenotyping and CAZyme composition indicates that HY7715 harbors a carbohydrate metabolic capacity typical of L. plantarum. Rather than reflecting extensive CAZy expansion, the data suggest that HY7715 operates within the conserved enzymatic framework characteristic of the species, with limited accessory variation that may fine–tune ecological adaptation. Although many L. plantarum strains have been extensively studied, the present work underscores a distinctive aspect of HY7715 within the broader genomic landscape of the species. Comparative pangenome analysis demonstrated that HY7715 largely conforms to the conserved genomic backbone of L. plantarum, rather than representing a highly divergent lineage. Notably, despite this genomic typicality, HY7715 combines a previously reported riboflavin–overproducing phenotype with a comprehensive genome–based safety profile and robust tolerance to gastrointestinal stress conditions. Thus, rather than introducing a singular novel genetic mechanism, this study highlights how conserved species–level capacities can be functionally integrated within a single, genomically stable strain to achieve both enhanced metabolic functionality and regulatory safety compliance. This integrative framework — linking pangenomic context, functional genomics, and phenotypic validation — strengthens the positioning of HY7715 as a well-characterized and industrially relevant probiotic candidate. Conclusion A comprehensive genome–based evaluation demonstrated that Lactiplantibacillus plantarum HY7715 is a genomically stable and safe probiotic candidate. Whole–genome analyses confirmed its taxonomic identity, high stability during prolonged cultivation, and the absence of virulence factors or acquired antimicrobial resistance genes. Mobile genetic elements did not harbor undesirable determinants, and the observed kanamycin resistance was consistent with an intrinsic, non-transferable phenotype. Comparative pangenome analysis further positioned HY7715 within the conserved genomic framework of L. plantarum, indicating that its probiotic–associated traits largely derive from species–level adaptive capacities rather than extensive strain-specific innovations. The concordance between genomic predictions and phenotypic validation—including gastrointestinal stress tolerance and the absence of biogenic amine production—supports its suitability for probiotic development. Collectively, this study provides an integrated safety and functional characterization of HY7715 within a species–wide genomic context. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 (ZIP 3.24 MB) Supplementary Material 1 (ZIP 3.24 MB)