Safety Evaluation of Weissella cibaria JW15 by Phenotypic and Genotypic Property Analysis

Weissella cibaria is one of the bacteria in charge of the initial fermentation of kimchi and has beneficial effects such as immune-modulating, antagonistic, and antioxidant activities. In our study, we aimed to estimate the safety of W. cibaria JW15 for the use of probiotics according to international standards based on phenotypic (antibiotic resistance, hemolysis, and toxic metabolite production) and genotypic analysis (virulence genes including antibiotic resistance genes). The results of the safety assessment on W. cibaria JW15 were as follows; (1) antibiotic resistance genes (ARGs) (kanamycin and vancomycin etc.) were intrinsic characteristics; (2) There were no acquired virulence genes including Cytolysin (cylA), aggregation substance (asa1), Hyaluronidase (hyl), and Gelatinase (gelE); (3) this strain also lacked β-hemolysis and the production of toxic metabolites (D-lactate and bile salt deconjugation). Consequently, W. cibaria JW15 is expected to be applied as a functional food ingredient in the food market.

Commercial starter culture products have been constantly consumed in the fermented food market as microbial food cultures (MFC) including LAB. Probiotic strains of LAB are also used in diverse medical and health-related areas, including the treatment of infections during pregnancy; management of allergic diseases; alleviation of intestinal inflammation; halt of antibiotic-related diarrhea, and prevention of urinary tract infections [5]. Of these, several probiotics such as L. rhamnosus GG (LGG), B. animalis subsp. lactis BB-12, and L. casei Shirota, etc. have been developed by global companies in the field of probiotics and they have also recently been marketed in the form of tablets or powders [6,7].
Probiotics are well-known and generally classified as safe (GRAS) because of their long-term safety in dairy products or fermented foods. The Lactic Acid Bacteria Industrial Platform (LABIP) has reported that the risk of infection caused by LAB occurs very rarely except for enterococci and streptococci [7]. However, in recent years, many controversies

Genomic Analysis
We performed the genomic analysis of the JW15 strain using Macrogen service (Macrogen Inc., Seoul, Korea). The manufacturer's instructions were as follows; DNA samples were sequenced using the PacBio RS II platform and Illumina HiseqXten platform, and then the subheads generated from PacBio RS II were assembled using the hierarchical genome assembly process (HGAP) [21] with default options. For error correction, the Illumina raw reads were filtered by quality at a level of 90% of bases had a phred score of 30 or higher. The assembly was corrected using high-quality HiseqXten reads by Pilon v1.21 [22]. Prokka v1.13 [23] was used for gene prediction and basic annotation. For additional annotation, the predicted protein sets were subjected to InterProScan v5.30-69.0 [24] and psiblast v2.4.0 [25] with EggNOG DB v4.5 [26]. Circular maps depicting each contig were generated using Circos v0.69.3 [27].

Bioinformatic Analysis of Virulence Factor-Related Genes
The VF-and toxin genes in the genome of the strain JW15, including Contig 1 to 4, were searched through BlastX analysis using Diamond software (ver. 0.9.26.127) [28] based on the virulence factor database (VFDB, http://www.mgc.ac.cn/VFs/, accessed on 12 August 2021) [29], which is an integrated and comprehensive online resource for curating information about virulence factors of bacterial pathogens. Thresholds for percent identity (% ID) and minimum length were set at 50% and 70%, respectively. In detail, VF-related genes, including those associated with enterotoxin, leukotoxin, cytolysin, cytotoxin K, hemolysis, biogenic amine production, hyaluronidase, aggregation, enterococcal surface protein, endocarditis antigen, collagen adhesion, cereulide, sex pheromone, and serine protease. These genes were additionally confirmed through BlastX analysis using experimentally-verified VF and toxin genes in the UniRef90 database.
2.6. Hemolytic Activity W. cibaria strains, LGG, and Bacillus cereus KACC 10004 were used as positive controls for hemolytic activity. The strains were aerobically cultured in blood agar supplemented with 5% sheep blood at 37 • C for 2 days. The plates were then analyzed for microbial hemolytic properties by illuminating and observing the plate. Colonies that revealed greenhue zones (α-hemolysis) or did not reveal any hemolysis (γ-hemolysis) were considered non-hemolytic strains. Colonies that displayed blood lyses zones (clear zones) were classified as hemolytic (β-hemolysis) strains.

D-Lactic Acid Measurement
The production of D-lactic acid by W. cibaria strains and LGG was measured using the d-lactate colorimetric assay kit from BioVision Research (Mountain View, CA, USA). The LAB strains were cultured in MRS broth for 24 h at 37 • C, and the supernatant was used for this experiment.

Bile Salt Deconjugation
Bile salt deconjugation was carried out according to the plate assays of Dashkevicz and Feifhner [30]. W. cibaria strains and LGG were cultured for 24 h at 37 • C on an MRS agar plate containing 0.5% taurodeoxycholic acid (TDCA; Sigma, St. Louis, MO, USA). The results were interpreted as positive in the case of the formation of a halo of sediment or opaque granular white colonies around the colonies.

Enzymatic Profiles by API ZYM
Use of the API ZYM kit (bioMérieux, Marcy l'Etoile, France) was based on a substrate availability of a total of 19 enzymes. The bacterial suspension was adjusted with McFarland no. 5 being dropped in each tube. After incubation at 37 • C for 4 h, the results were determined to be positive if the color intensity was more than three following the manufacturer's instructions.

Bacterial Reverse Mutation Assay
A bacterial reverse mutation assay was performed to evaluate the mutagenicity of W. cibaria JW15 with or without the S9 mix, following the principles of OECD Guideline 471 (2020) [31]. The assay was carried out using Salmonella typhimurium histidine-auxotrophic strains TA98, TA100, TA1535, TA1537, and Escherichia coli tryptophan-auxotrophic strain WP2uvrA (Molecular Toxicology, Boone, INC, USA). The S9 mix was used as a metabolic activation system (ORIENTAL YEAST Co., Ltd., Tokyo, Japan) and was prepared at the time of use in the required amount. Freeze-stored S9 (Lot No.: 20121110) and Cofactor A (Lot No.: A20120810) were thawed and prepared by mixing at a ratio of 1:9. Different dilutions of W. cibaria JW15 samples (5000, 2500, 1250, 625, and 313 µg/plate) were used for all tests under the same conditions. After being cultured at 37 • C for 48 h, the number of colonies in each tested group was counted per plate. This result was determined to be positive when the revertant colonies in the subject group were more than doubled. The data of historical control is presented in Table S1 as Supplementary Information.

Results and Discussion
The aim of this study was to verify the safety of W. cibaria JW15 based on phenotypic (antibiotic resistance, hemolysis, and toxic metabolite production) and genotypic analyses (virulence genes including antibiotic resistance genes). Currently, W. cibaria has no use as a probiotic ingredient, and the species is reported on antibiotic resistance such as kanamycin and vancomycin. Nevertheless, they have been frequently isolated from fermented foods and human feces and are well-known for their beneficial effects such as probiotic properties, antimicrobial-, antagonistic-, and antioxidant activities etc. Many researchers or consumers expect higher functional or healthy foods made from lactic acid bacteria with novel activity.

Determination of Minimum Inhibitory Concentrations
Weissella spp. has not been cleared on the cut-off values of MIC against antibiotics by EFSA in 2012. Accordingly, we determined an antibiotic susceptibility test of the W. cibaria JW15 strain corresponding to a Leuconostoc spp., based on the EFSA cut-off value, which reflects the phylogenetic and phenotypic characterization of the JW15 strain.
To ensure safety, the phenotypic antibiotic susceptibility of W. cibaria JW15 was investigated against 9 antibiotics, including ampicillin (AM), chloramphenicol (CL), clindamycin (CM), erythromycin (EM), gentamicin (GM), kanamycin (KM), streptomycin (SM), tetracycline (TC), and vancomycin (VA) using the E-test method [32]. As shown in Table 2, W. cibaria JW15 was susceptible to 7 kinds of antibiotics, including ampicillin (AM), chloramphenicol (CL), clindamycin (CM), erythromycin (EM), gentamicin (GM), streptomycin (SM), and tetracycline (TC), which were found below the cut-off value (µg/mL) within the safe range. However, the JW 15 strain was shown to be resistant to kanamycin (KM). Recent studies have shown that the antibiotic susceptibility profile of W. cibaria differs between each strain [33,34]. W. cibaria CMU was found to be sensitive to AM, CL, CM, EM, GM, SM, and TC, except for KM corresponding to an obligate hetero-fermentative lactobacilli [33]. In our result, W. cibaria strains showed MICs ≥ 256 mg/L for kanamycin and vancomycin, suggesting that the resistance against kanamycin and vancomycin could be considered an intrinsic property. Antibiotic resistance was found not only in the genus Weissella, but also in many lactic acid bacteria used as food ingredients. Lactobacillus sp. shows high resistance to antibiotics reported as endogenous with strong resistance to antibiotics such as kanamycin and vancomycin [17,34]. It has been reported that lactic acid bacteria derived from fermented food are resistant to antibiotics [35,36]. Therefore, it has been speculated that the characteristic that the JW15 strain isolated from kimchi is resistant to some antibiotics may be common.

Detection of Antibiotic Resistance Genes
The transferability of antibiotic resistance (AR) genes and plasmids present in bacteria is associated with human health. Here, we confirmed the existence of AR genes and plasmids in the W. cibaria JW15 strain on four antibiotics (clindamycin, kanamycin, streptomycin, and vancomycin) showing high MIC cut-off value presented in Table 2. PCR analysis for four antibiotic resistance genes such as streptomycin (aadA, aadE, and strB), tetracycline (tet (K)), kanamycin (aph (3")-III and ant (2")-I), and clindamycin (Inu (A) and Inu (B)) were conducted. Although there was detected amplicons in several samples, they were not antibiotic resistant genes based on sequencing analysis. The PCR results are  Table 3. There was no expected amplicon in the chromosome and plasmid DNA of W. cibaria JW15, W. cibaria LMG 21843, W. cibaria LMG 17699, and L. rhamnosus ATCC 53103 used in this study.  For antibiotic resistance, the MIC cut-off value of kanamycin was exceeded, which is a phenotypic evaluation, but the antibiotic resistance target gene was not detected in the chromosome and plasmid of the JW15 strain, which is a genotype evaluation. Sharma et al. (2014) reported that antibiotics intrinsic strains were phenotypically resistant may be genotypically susceptible [37]. We found several studies showing this characteristic, and strains that also had specific antibiotic resistance, but no gene was detected [33,38]. Therefore, the results of antibiotic resistance to KM and detection of their target genes are similar to those seen in antibiotic intrinsic strains according to previous reports. In addition, the phenotypic property of the JW15 strain that exhibits resistance to kanamycin may be due to four endogenous-related mechanisms such as enzyme inactivation or modification, alteration of bacterial target sites, antibiotic efflux pump and outer membrane permeability change, and intracellular metabolic rearrangement [37].
Moreover, for the transferability of antibiotic resistance, the plasmid plays a major role in the ARG gene transfer method (HGT) [37]. In our results, as shown in Table 3, kanamycin resistance gene (aph (3 )-III and ant (2 )-I) were not detected in the plasmid of JW15, thus the transferability is considered low.

Genomic Features of JW15 Strain
The key genomic features of W. cibaria JW15, including GC skew, protein-coding sequences (CDSs), COG categories, and G+C contents, are graphically depicted in Figure 1. The genome of the JW15 strain was a single circular chromosome of 2,472,214 bp with 3 plasmids (30,944 bp, 17,267 bp, and 14,411 bp). The genome of strain JW15 contains a total of 2315 CDS, 42 tRNA genes, and 28 rRNA genes. The result of COG-assigned proteins in the genomes of strain JW15 and their distributions into COG categories was not abbreviated. As a result, the COGs were classified into 26 functional categories except for Nohit against the COG database and of the 2556 protein-coding genes, 2259 genes (88.39%) were assigned to COGs categories. The W. cibaria UTNGt21O strain (1635 genes) reported by Tenea and Hurtado [39] was less than the W. cibaria JW15 strain. We found that the essential genes from the functional subcategories with the COG codes G (Carbohydrate transport and metabolism, 7.75%), J (Translation, ribosomal structure, and biogenesis, 7.67%), K (Transcription, 5.95%), L (Replication, recombination and repair, 4.38%), H (Coenzyme transport and metabolism, 3.44%), and I (Lipid transport and metabolism, Microorganisms 2021, 9, 2450 7 of 13 3.4%). However, the distribution of functional annotation of W. cibaria UTNGt21O strain was differently expressed in the order of R (General function prediction only, 8.99%), J (Translation, ribosomal structure, and biogenesis, 8.07%), K (Transcription, 6.60%), and L (Replication, recombination and repair, 6.54%) in comparison with W. cibaria JW15 strain. Owing to different genes, depending on strain-specificity, the information of functional genes mentioned here will help additional studies of this strain and demonstrate its potential property for the use of probiotics.
against the COG database and of the 2556 protein-coding genes, 2259 genes (88.39%) were assigned to COGs categories. The W. cibaria UTNGt21O strain (1635 genes) reported by Tenea and Hurtado [39] was less than the W. cibaria JW15 strain. We found that the essential genes from the functional subcategories with the COG codes G (Carbohydrate transport and metabolism, 7.75%), J (Translation, ribosomal structure, and biogenesis, 7.67%), K (Transcription, 5.95%), L (Replication, recombination and repair, 4.38%), H (Coenzyme transport and metabolism, 3.44%), and I (Lipid transport and metabolism, 3.4%). However, the distribution of functional annotation of W. cibaria UTNGt21O strain was differently expressed in the order of R (General function prediction only, 8.99%), J (Translation, ribosomal structure, and biogenesis, 8.07%), K (Transcription, 6.60%), and L (Replication, recombination and repair, 6.54%) in comparison with W. cibaria JW15 strain. Owing to different genes, depending on strain-specificity, the information of functional genes mentioned here will help additional studies of this strain and demonstrate its potential property for the use of probiotics.

Bioinformatic Analysis of VF-Related Genes
In our study, we did not discover virulence-related genes in the chromosomal and plasmid genomes of the W. cibaria JW15 strain as shown in Table 4. However, two genes in JW15_contig1 (Table 5) were identified with low homology (loose identity, <95%) with the virulence factor database (VFDB) containing information on the virulence genes of bacterial pathogens online. Gene JW15-00598 showed a homology of 57.4% to gene efaA involved in endocarditis antigen, and gene JW15-00853 showed a homology of 53.1% to gene CD1208 (or CVF417) involved in RNA methyltransferase (or hemolysin A). In addition, the two genes (gene JW15-00598 and JW15-00853) researched in NCBI and Uniprot were found to have high homology (>95%) with transporter substrate-binding protein, RNA methyltransferase or cell division, respectively. In particular gene JW15-00853, which was identified as TlyA, is known to be not, on its own, a potent hemolysin [40]. Therefore, the two genes are presumed to be general transporters and transferase genes that are not related to toxic genes such as endocarditis antigen and hemolysin. In addition, the W. cibaria JW15 strain was negative in the hemolysis test, which was consistent with the bioinformatic analysis. The gene sequences are shown in Table S1 in Supplementary Materials. The VF-related gene information that was used to confirm the safety of the strain should help further probiotic studies. Table 4. Bioinformatic analysis for the presence of putative virulence factor-related genes in the genomes of strain JW15.

Hemolytic Activity
Hemolysin is a toxic enzyme of pathogenic bacteria such as Bacillus cereus and has hemolytic activity to destroy red blood cells in the host, as well as the possibility for edema and anemia [38,41]. Generally, β-hemolysis is associated with microbial pathogenicity. In our study, B. cereus KACC 10004 as a positive control showed clear zones (expressed as β-hemolysis) around the colonies, whereas W. cibaria strains and LGG did not show β-hemolysis activity (Figure 2).

Hemolytic Activity
Hemolysin is a toxic enzyme of pathogenic bacteria such as Bacillus cereus and has hemolytic activity to destroy red blood cells in the host, as well as the possibility for edema and anemia [38,41]. Generally, β-hemolysis is associated with microbial pathogenicity. In our study, B. cereus KACC 10004 as a positive control showed clear zones (expressed as βhemolysis) around the colonies, whereas W. cibaria strains and LGG did not show β-hemolysis activity (Figure 2).

D-Lactic Acid Production
Various bacterial species are known to produce D-lactate or both D-and L-lactates are produced in fermentation. Of them, the genus Lactobacillus produces D-and L-lactates, the genus Pediococcus produces L-, and the genera Leuconostoc, Oenococcus, and Weissella produce D-lactic acid [42]. As shown in Table 6, the productivity of D-lactic acid by W. cibaria was measured by enzymatic assays concerning D-lactate dehydrogenase. W. cibaria strains did not produce D-lactic acid like the commercial probiotic strain LGG. Our result was similar to the report showing that the W. cibaria CMU strain was unable to produce D-lactic acid [33]. Table 6. Enzymatic profiles and assay of toxic metabolic production.

D-Lactic Acid Production
Various bacterial species are known to produce D-lactate or both D-and L-lactates are produced in fermentation. Of them, the genus Lactobacillus produces D-and L-lactates, the genus Pediococcus produces L-, and the genera Leuconostoc, Oenococcus, and Weissella produce D-lactic acid [42]. As shown in Table 6, the productivity of D-lactic acid by W. cibaria was measured by enzymatic assays concerning D-lactate dehydrogenase. W. cibaria strains did not produce D-lactic acid like the commercial probiotic strain LGG. Our result was similar to the report showing that the W. cibaria CMU strain was unable to produce D-lactic acid [33]. Table 6. Enzymatic profiles and assay of toxic metabolic production.

Enzymatic Profiles JW15
LGG LMG 21843 LMG 17699 Bile salts are less capable of solubilizing and absorbing lipids in the gut. All strains used in this study were able to grow in the presence or absence of sodium taurodeoxycholate (0.5%) and did not show the precipitate halos or the opaque white colonies after growth in MRS with TDCA (Table 6). These results show the lack of ability to deconjugate sodium taurodeoxycholate and agree that W. cibaria could not convert to secondary bile acid as previously published report [36].

Enzymatic Profile by API ZYM
The enzyme profile of the JW15 strain was similar to that of the LMG 28143 strain isolated from fermented kimchi, while the LMG 17699 strain was different from the βgalactosidase β-glucosidase enzymes ( Table 6). In Muñoz-Atienza et al. (2013), it was observed that leucine arylamidase, valine arylamidase, β-galactosidase, and β-glucosidase showed different patterns among the 15 kinds of Weissella spp. [36]. In the case of the β-glucuronidase, no generation of potential carcinogenic metabolites [43] was detected in any of the W. cibaria strains and LGG, indicating that they are safe.

Bacterial Reverse Mutation Assay
The bacterial reverse mutation assay, developed by Bruce Ames in 1973 [44], was performed for mutagenicity testing of probiotics such as L. rhamnosus, B. adolescentis, L. paracasei, L. mali, and P. acidilactici [45][46][47]. The genotoxicity was conducted by this assay with different doses of W. cibaria JW15 against four mutant S. typhimurium strains (TA98, TA100, TA1535, and TA1537) and a mutant E. coli strain (WP2uvrA), respectively. An expected increase of revertant colonies was observed in all positive groups after induction of the mutants. After exposure of bacterial strain to different concentrations of W. cibaria JW15, the number of revertant colonies, regardless of the presence or absence of S9 mix, did not exceed twice that of the negative control group (Table 7). Therefore, the W. cibaria JW15 treatment groups were considered not to have mutagenic activity in the histidine auxotrophy of the S. typhimurium strains or the tryptophan auxotrophy of E. coli.
Consequently, in this study, we verified the safety of the W. cibaria JW15 strain by phenotypic and genotypic property analysis according to the international guidelines by FAO/WHO. The safety was evaluated by a minimum inhibitory concentration assay for 9 antibiotics, chromosomal and plasmid DNA analysis for 12 antibiotic resistance genes (ARGs) on 4 antibiotics, virulence gene analysis, beta-hemolysis, toxic metabolite production, and bacterial reverse mutation assay. The strain W. cibaria JW15 was susceptible to all antibiotics except for kanamycin and vancomycin. We confirmed that there was no harboring of antibiotic resistance target genes and virulence-related genes in the genome of strain JW15. We therefore considered that antibiotic resistance (e.g., kanamycin, vancomycin) was an intrinsic property of W. cibaria JW15. Additionally, the strain JW15 lacked β-hemolysis, β-glucuronidase, toxic metabolites such as D-lactate and bile salt deconjugation, and bacterial reverse mutagenic activity. Accordingly, we believe that W. cibaria JW15 could be commercially applied as a probiotic strain in the future.  Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.
Data Availability Statement: This genome sequence was deposited in GenBank (BioProject number PRJNA639573; and GenBank accession numbers CP058237-CP058240). The version described in this paper is the first version.