Subspecies Classification and Comparative Genomic Analysis of Lactobacillus kefiranofaciens HL1 and M1 for Potential Niche-Specific Genes and Pathways

(1) Background: Strains HL1 and M1, isolated from kefir grains, have been tentatively identified, based on their partial 16S rRNA gene sequences, as Lactobacillus kefiranofaciens. The two strains demonstrated different health benefits. Therefore, not only the genetic factors exerting diverse functionalities in different L. kefiranofaciens strains, but also the potential niche-specific genes and pathways among the L. kefiranofaciens strains, should be identified. (2) Methods: Phenotypic and genotypic approaches were employed to identify strains HL1 and M1 at the subspecies level. For the further characterization of the probiotic properties of both strains, comparative genomic analyses were used. (3) Results: Both strains were identified as L. kefiranofaciens subsp. kefirgranum. According to the COG function category, dTDP-rhamnose and rhamnose-containing glycans were specifically detected in the L. kefiranofaciens subsp. Kefirgranum genomes. Three unique genes (epsI, epsJ, and epsK) encoding glycosyltransferase in the EPS gene cluster, and the ImpB/MucB/SamB family protein encoding gene were found in HL1 and M1. The specific ability to degrade arginine via the ADI pathway was found in HL1. The presence of the complete glycogen metabolism (glg) operon in the L. kefiranofaciens strains suggested the importance of glycogen synthesis to enable colonization in kefir grains and extend survival under environmental stresses. (4) Conclusions: The obtained novel information on the potential genes and pathways for polysaccharide synthesis and other functionalities in our HL1 and M1 strains could be applied for further functionality predictions for potential probiotic screening.


Introduction
Lactobacillus kefiranofaciens was first described in 1988 by Fujisawa et al. [1] for homofermentative lactobacilli strains isolated from kefir grains. This species has been reported as a kefiran (exopolysaccharide, EPS) producer in kefir grains. Kefiran can be used as a food grade additive to obtain fermented products due to its rheological properties, which enhance the apparent viscosity, storage and loss modulus of chemically acidified skim milk gels [2]. This phenomenon was strengthened by the heat treatment usually applied in the manufacturing of yogurts [3].
In 1994, Lactobacillus kefirgranum has published validly as a new species among the homofermentative lactobacilli strains from kefir grains [4]. However, Vacanneyt et al. (2004) reclassified L. kefirgranum as L. kefiranofaciens subsp. kefirgranum, since L. kefirgranum and L. kefiranofaciens show 100% 16S rRNA gene sequence similarity, DNA-DNA hybridization under anaerobic conditions for 72 h. Carbohydrate fermentation was determined using API 50 CHL system (bioMérieux, Marcy-l'Etoile, France), according to the manufacturer's instructions. The whole cell protein profile was analyzed as described previously [15], in three steps: cell protein extraction, protein quantification using a protein assay kit (Bio-Rad Protein Assay Kit, Bio-Rad, Hercules, CA, USA) and protein profiling using SDS-PAGE. The banding patterns were clustered together using the unweighted pair group method with arithmetic mean (UPGMA) algorithm. The evolutionary distances were computed using the p-distance method and are in the unit of the number of base differences per site.

Genotypic Characterization
Genotypic characterization was performed by 16S rRNA and housekeeping gene sequence analysis, enterobacterial repetitive intergenic consensus polymerase chain reaction (ERIC-PCR) and randomly amplified polymorphic DNA (RAPD) fingerprinting [16], as well as by whole genome sequence-based methods, e.g., based on the average nucleotide identity (ANI) values, digital DNA-DNA hybridization (dDDH) and phylogenomic analysis.

ERIC-PCR
The genomic DNA of four strains were amplified using the ERIC 1 and ERIC 2 pair of primers, as described previously [23] (Table S1). The PCR products were electrophoresed on 1.5% (wt/vol) agarose (Fisher Biotech, Fair Lawn, NJ, USA) gel electrophoresis (BioDoc-It R 220 Imaging System, UVP LLC., Upland, CA, USA) with ethidium bromide staining. The process was repeated twice to verify the accuracy of the results.

RAPD
The extracted genomic DNA was used as a template in subsequent PCR amplifications. Five primers [24], i.e., RAPD-A, RAPD-B, RAPD-E, RAPD-G, and RAPD-I, with arbitrary nucleotide sequences were used (Table S1). The RAPD products were electrophoresed on 1.5% (wt/vol) agarose gel. The process was performed twice. The banding patterns were clustered using the UPGMA algorithm with Dice coefficients using the Dolphin-1D software (Wealtec Corp., Sparks, NV, USA).

Genome Sequencing and Assembly
The whole genomes of L. kefiranofaciens M1 and HL1 were sequenced with Nanopore (MinION, Oxford Nonopore Technologies, Oxford, UK) and Illumina MiSeq (Illumina, San Diego, CA, USA) (301 base, paired end reads). The Illumina raw data were trimmed to remove adapters, low quality sequences (Q20) and ambiguous bases. The nanopore reads were used to perform de novo assembly using NECAT program (https://github.com/xiaochunle/necat, accessed on 8 October 2021), and the contig with trimmed NovaSeq reads was corrected using CLC Genomics Workbench. Gap closing was performed using PCR and Sanger sequencing. The genome information of HL1 and M1 was deposited in the GenBank database under the accession nos. GCA_023674385.1 and GCA_023674405.1, respectively.

Subspecies Identification of HL1 and M1
For the classification of L. kefiranofaciens strains HL1 and M1 at the subspecies level, phenotypic and genotypic characterizations were conducted with two reference strains (L. kefiranofaciens subsp. kefiranofaciens BCRC 16059 T and L. kefiranofaciens subsp. kefirgranum BCRC 80410 T ).

Phenotypic Characterization
First, we observed the cell morphology of L. kefiranofaciens HL1 and M1, as well as those of two reference strains, by microscopy. Cells of all four strains were Gram-positive rods ranging from 2 to 30 µm in length with no significant difference in morphology ( Figure 1A). When cultured on MRL agar (replaced 1% glucose by 1% lactose), strains HL1, M1 and BCRC 80410 T demonstrated opaque and yellowish colonies with protrusions, whereas BCRC 16059 T showed a semi-transparent, white sticky surface (data not shown). In MRS broth, HL1, M1 and BCRC 80410 T showed powdery bacterial chunks with flocculation, while BCRC 16059 T showed a sticky appearance, indicative of high EPS production ( Figure 1B). Our findings corresponded well with those of previous studies [28], i.e., that L. kefiranofaciens subsp. kefirgranum form dry, compact, dull bulging colonies, whereas L. kefiranofaciens subsp. kefiranofaciens have transparent, glossy, convex and extremely slimy colonies.   The carbohydrate fermentation characteristics of four strains, as determined using an API 50 CHL system, demonstrated diversity among strains in terms of the presence/contents of eleven carbohydrates (amygdalin, arbutin, D-cellobiose, gentibiose, Dmaltose, D-melibiose, D-raffinose, salicin, D-sucrose, D-trehalose, and aesculin) ( Table 1). All strains produced acid from D-fructose, D-galactose, D-glucose, D-lactose, D-mannose and N-acetylglucosamine, whereas none produced acid from the remaining 32 substrates according to the API 50 CHL system. Strains HL1, M1 and L. kefiranofaciens subsp. kefirgranum BCRC 80410 T hydrolyzed aesculin, whereas L. kefiranofaciens subsp. kefiranofaciens BCRC 16059 T did not. The result regarding aesculin hydrolysis was consistent with previous studies [1,4,28]. The fermentation patterns of carbohydrates suggested that L. kefiranofaciens strains HL1 and M1 may belong to the kefirgranum subspecies.
The SDS-PAGE whole cell protein profiles revealed that HL1 and M1 were closely related to each other in terms of the composition of their cell wall proteins. Additionally, strains HL1, M1 and L. kefiranofaciens subsp. kefirgranum BCRC 80410 T were bundled in a cluster and distinct from L. kefiranofaciens subsp. kefiranofaciens BCRC 16059 T on the basis of the three unique banding patterns in regions of 15-20, 30-35 and 170 kDa ( Figure 1C and Supplementary Figure S1). This finding corresponded well to a previous study [5] which noted that SDS-PAGE profiles of whole-cell proteins could be used to differentiate the strains of L. kefiranofaciens at the subspecies level into two subspecies, i.e., L. kefiranofaciens subsp. kefiranofaciens and L. kefiranofaciens subsp. kefirgranum.

Genotypic Characterization
The ERIC-PCR and RAPD fingerprinting methods are considered convenient discriminatory tools for measuring biodiversity in the genomes of bacterial strains at the strain level. To investigate the taxonomic position of HL1 and M1, we carried out genotypic characterizations, including sequence analyses of 16S rRNA and two housekeeping genes (pheS and rpoA), ERIC-PCR and RAPD fingerprinting and phylogenomic and core genome multilocus sequence typing (cgMLST) analyses. The average nucleotide identity (ANI) values and the digital DNA-DNA hybridization (dDDH) values were also calculated. HL1 shared 100% 16S rRNA, pheS and rpoA gene sequence similarities with M1 and the type strains of L. kefiranofaciens subsp. kefiranofaciens and L. kefiranofaciens subsp. kefirgranum. Through phylogenetic analyses based on these three gene sequences together with the two types strains, HL1 and M1 were found to be located in an independent cluster among the species in the genus Lactobacillus (Supplementary Figures S2 and S3). The phylogenomic tree based on whole genome sequences showed that the six L. keifanofaciens strains were included in the same cluster (Supplementary Figure S4). We also identified the HL1 and M1 strains based on the overall genome related index (ORGI), e.g., the ANI and dDDH values. All strains of L. kefiranofaciens (HL1, M1, ATCC 43761 T , DSM 10550 T , ZW3 and KR) shared >99.2% ANI values and >93.7% dDDH values, indicating that these six strains represent the same species (see Supplementary Table S2). However, based on a core gene multilocus sequence typing (cgMLST) analysis of the 1674 core genes, the six L. kefiranofaciens strains could be clearly divided into two clusters: Cluster A (comprising two L. kefiranofaciens subsp. kefiranofaciens strains, ATCC 43761 T and ZW3), and Cluster B (comprising HL1 and M1, and two L. kefiranofaciens subsp. kefirgranum strains, DSM 10550 T and KR) (Figure 2A). For further subspecies identification of HL1 and M1, the ERIC-PCR and RAPD fingerprinting approaches were applied. Using dendrogram analysis based on the concatenated ERIC-PCR and five RAPD profiles, it was found that HL1 shares 100% similarity with M1, with these two strains forming a distinct cluster with BCRC 80410 T , demonstrating that HL1 and M1 belong to L. kefiranofaciens subsp. kefirgranum. This result was consistent with the result obtained by SDS-PAGE protein profiling ( Figure 2B). The results from a previous study using various phylogenetic and genotypic approaches, including 16S rRNA gene sequence analysis and DNA-DNA hybridizations, did not find discriminating power for subspecies identification of L. kefiranofaciens. However, we successfully differentiated the strains of L. kefiranofaciens subsp. kefirgranum from L. kefiranofaciens subsp. kefiranofaciens using SDS-PAGE whole-cell protein profiling and the RAPD typing method, as well as cgMLST analysis.

Genome Features
The assembled complete genome sizes of strains HL1 and M1 were 2,216,505 bp and 2,179,135 bp, respectively, with 37.5% of the same G+C contents. They comprised a circular chromosome of 2,156,113 bp and 2,180,483 bp, respectively, and a circular plasmid of 36,022 bp and 23,022 bp, respectively. For the HL1 genome, a total of 2225 predicted protein coding sequences (CDSs) were found, with 15 ribosomal RNAs (rRNAs) and 64 transfer RNAs (tRNAs). Meanwhile, the M1 genome had 2208 CDSs, 15 rRNAs, and 64 tRNAs ( Figure 3A and Table 2). The general genomic features were almost the same in all six strains. The differences in genomic information may be a result of the genetic backgrounds of the different subspecies or strains. The interplay of sequencing quality, read length, sequencing depth and the assembler could also have affected the sequencing results [29]. The findings from this study demonstrated that phenotypic-and genotypic-based strain identification methods were extremely effective for the classification of L. kefiranofaciens into two subspecies. Consequently, we confirmed that our strains, HL1 and M1, were indeed L. kefiranofaciens subsp. kefirgranum.

Genome Features
The assembled complete genome sizes of strains HL1 and M1 were 2,216,505 bp and 2,179,135 bp, respectively, with 37.5% of the same G+C contents. They comprised a circular chromosome of 2,156,113 bp and 2,180,483 bp, respectively, and a circular plasmid of 36,022 bp and 23,022 bp, respectively. For the HL1 genome, a total of 2225 predicted protein coding sequences (CDSs) were found, with 15 ribosomal RNAs (rRNAs) and 64 transfer RNAs (tRNAs). Meanwhile, the M1 genome had 2208 CDSs, 15 rRNAs, and 64 tRNAs ( Figure 3A and Table 2). The general genomic features were almost the same in all six strains. The differences in genomic information may be a result of the genetic backgrounds of the different subspecies or strains. The interplay of sequencing quality, read length, sequencing depth and the assembler could also have affected the sequencing results [29]. It is worth noting that HL1 and M1 possessed seven clustered, regularly interspaced, short palindromic repeats (CRISPR), whereas DSM 10550 T , KR, ATCC 43761 T and ZW3 had six, two, one and one, respectively. The CRISPR-Cas system cleaves phage and plasmid DNA, showing promise for self-defense [30]. Higher repeated CRISPR in HL1 and M1 than other strains suggested that CRISPR may play an important role in providing immunity against phages and plasmids. showing promise for self-defense [30]. Higher repeated CRISPR in HL1 and M1 than other strains suggested that CRISPR may play an important role in providing immunity against phages and plasmids. The COG function of the gene showed that the top ten functions (Classes) of HL1 and M1 were as follows: replication, recombination and repair (Class-L); carbohydrate transport and metabolism (Class-G); transcription (Class-K); translation, ribosomal structure and biogenesis (Class-J); amino acid transport and metabolism (Class-E); inorganic ion transport and metabolism (Class-P); nucleotide transport and metabolism (Class-F); cell wall/membrane/envelope biogenesis (Class-M); and energy production and conversion (Class-C) ( Table 3). These were similar to other L. kefiranofaciens strains [30], showing no difference between the two subspecies or the strains.    The COG function of the gene showed that the top ten functions (Classes) of HL1 and M1 were as follows: replication, recombination and repair (Class-L); carbohydrate transport and metabolism (Class-G); transcription (Class-K); translation, ribosomal structure and biogenesis (Class-J); amino acid transport and metabolism (Class-E); inorganic ion transport and metabolism (Class-P); nucleotide transport and metabolism (Class-F); cell wall/membrane/envelope biogenesis (Class-M); and energy production and conversion (Class-C) ( Table 3). These were similar to other L. kefiranofaciens strains [30], showing no difference between the two subspecies or the strains.  Figure 3B shows Venn diagrams and an Upset plot of the coding sequences of the six L. kefiranofaciens strains and four L. kefiranofaciens subsp. kefirgranum strains, respectively. The numbers of unique genes in HL1, M1, DSM 10550 T , KR, ATCC 43761 T and ZW3 were 39 (1.7%), 17 (0.8%), 99 (4.7%), 132 (6.1%), 39 (1.6%) and 95 (3.9%), respectively. The HL1 genome shared 98.3% of the gene with M1. Comparing the HL1 and M1 genomes with the four L. kefiranofaciens subsp. kefirgranum strains, approximately 85% of the genes were orthologous. The unique genes could provide information related to the various properties and functionalities of the two subspecies and strains of L. kefiranofaciens.

Polysaccharide Synthesis
L. kefiranofaciens is a polysaccharide kefiran-producing species which is responsible for the formation of the kefir grains matrix and the viscous property of kefir milk [31]. Thus, polysaccharide synthesis-related genes were analyzed.

The Cluster of Orthologous Groups Function of Genes in EPS Related Subsystems
The cluster of orthologous groups (COG) function of genes in the SEED subsystem was first analyzed; it showed that except for the sortase enzyme in the "Gram-positive cell wall components" subcategory, the gene numbers of HL1 and M1 in the "capsular and extracellular polysaccharides" subcategory, "no subcategory" and "Gram-positive cell wall components" were identical (Table 4 and Supplementary Table S3). Compare with other L. kefiranofaciens strains, differences in gene numbers were observed in the "capsular and extracellular polysaccharides" subcategory. Three L. kefiranofaciens subsp. kefirgranum strains, i.e., HL1, M1, and DSM 10550 T , demonstrated similar gene numbers in the "capsular and extracellular polysaccharides" subcategory with the genes involved in the "dTDP-rhamnose synthesis" and "rhamnose-containing glycans" subsystems (Table 4). dTDP-rhamnose is an important precursor of cell wall polysaccharides and rhamnose-containing EPS [32]. Various lactic acid bacteria [33][34][35][36] possess rhamnose in their cell walls; this may serve as the primary binding site for certain bacteriophages [37]. HL1 and M1, with dTDP-rhamnose synthesis genes and rhamnose-containing glycans, verified our previous study, in which we determined that the M1 cell wall contained rhamnose (unpublished data). The finding regarding genetic COG functions not only suggested that rhamnose in the cell wall and kefiran were strain-dependent, but also provided a possible explanation for the previous CRISPR discovery. The higher repeated CRISPR in HL1 and M1 might be needed for self-defense against bacteriophage due to the presence of rhamnose in the cell wall.  Lipoteichoic acid biosynthesis  3  3  3  3  3  3  Sortase  1  -1  1  1  1  Teichoic and lipoteichoic acids biosynthesis  13  13  13  13  13  13   Total  39  38  39  28  28  28 Microorganisms 2022, 10, 1637 11 of 18

Identification of the HL1 and M1EPS Biosynthetic Gene Cluster
A genomic comparison between the organization of EPS gene clusters in L. kefiranofaciens subsp. kefirgranum HL1 and M1, based on the putative or established functions of these products, is provided in Figure 4A. Four other L. kefiranofaciens strains (DSM 10550 T , KR, ATCC 43761 T , and ZW3) were used as references for the DNA sequences of the putative EPS gene clusters. The results indicated that HL1 and M1 possessed 13 genes (see Supplementary Table S4) which were located in the same orientation ( Figure 4A). However, the EPS gene cluster in Wzy (polysaccharide polymerase) was different in HL1 and M1; this cluster encodes the functional protein related to the biosynthesis of repeating units. Wzy polysaccharide polymerase exhibits low sequence conservation in species with no Wzy homologues and with X-ray crystal structures [38]. Additionally, the protein encoded by epsE in strains HL1 and M1 demonstrated 94% identity with Lactobacillus helveticus. It was annotated as a priming glycosyltransferase (EC 2.7.8.6) which transfers the first sugar of each subunit of an EPS molecule. This enzyme plays an important role in EPS biosynthesis in Gram-positive lactic acid bacteria [39,40].
We also found three genes (epsI, epsJ and epsK) which were capable of encoding putative glycosyltransferases in the central portion of the putative EPS locus of HL1 and M1; there were considered to be distinct in the genomes, compared to those of other L. kefiranofaciens strains (see Figure 4A, Supplementary Table S4). An earlier study [40] revealed that the function of genes encoding glycosyltransferases in Lactobacillus was to transfer the monosaccharides of the EPS subunit in a sugar-and glycoside linkage-dependent manner. The three unique genes encoding glycosyltransferases in L. kefiranofaciens HL1 and M1 are probably responsible for the key enzymes producing unique EPS.
Based on our bioinformatic analysis, a biosynthetic model of EPS in L. kefiranofaciens HL1 and M1 is proposed ( Figure 4B). The full biosynthetic process can be divided into two separate steps. The first involved the generation of activated sugar precursors from the metabolism of carbon in the cytoplasm. These enzymes, with the corresponding genes, indicated that L. kefiranofaciens HL1 and M1 possess multi-metabolic routes, including phosphoenolpyruvate, the sugar phosphotransferase system (PTS) and the Leloir pathway, which is involved in the generation of activated sugar precursors for EPS synthesis during the catabolism of glucose/lactose.
Among the aforementioned enzymes, fifteen were involved in UDP-glucose, UDPgalactose, UDP-mannose and TPD-glucosamine in the HL1 and M1 genomes. The number and type of monosaccharide nucleotides influence the composition and production of EPS [41,42]. This finding may also partially explain the differences in EPS yield and compositions between the strains of L. kefiranofaciens subsp. kefirgranum and those of L. kefiranofaciens subsp. kefiranofaciens. However, the gene encoded β-phospho-glucomutase (β-PGM) was not found in the Leloir pathway of the HL1 and M1 genomes. A previous study [43] which deleted β-phosphoglucomutase of Lactococcus lactis showed that the mutation did not influence growth, cell composition or product formation when glucose/lactose was used as the carbon source, but significantly reduced the maximum specific growth rates with maltose or trehalose as the carbon source. Thus, the lack of β-phospho-glucomutase in HL1 and M1 may affect the utilization of maltose/trehalose; this was consistent with the API 50 CHL result.
The second step ( Figure 4B) was the Wzy pathway, connected to committed cell membrane-associated assembly and the polymerization of polysaccharides. L. kefiranofaciens M1 and HL1 possessed the following enzymes, characterized into three functional groups: (1) polysaccharide assembly function, including priming glycosyltransferase (epsE), flippase (wzx), polysaccharide polymerase (wzy) and phosphotransferase (epsA); (2) glycosyltransferase (epsF, epsG, epsH, epsI, epsJ, epsK); and (3) the phosphoregulatory system, including tyrosine kinase (epsB, epsC) and phosphotyrosine phosphatase (epsD), that regulate the polysaccharide assembly process. Both strains demonstrated a similar Wzy pathway to those of other L. kefiranofaciens, suggesting that this pathway was the conserved region in the eps genetic cluster. However, other than the conserved region, different re-gions of the eps genetic cluster in Lactobacillus could form EPSs with varied structures and molecular weights [44]. ever, the EPS gene cluster in Wzy (polysaccharide polymerase) was different in HL1 and M1; this cluster encodes the functional protein related to the biosynthesis of repeating units. Wzy polysaccharide polymerase exhibits low sequence conservation in species with no Wzy homologues and with X-ray crystal structures [38]. Additionally, the protein encoded by epsE in strains HL1 and M1 demonstrated 94% identity with Lactobacillus helveticus. It was annotated as a priming glycosyltransferase (EC 2.7.8.6) which transfers the first sugar of each subunit of an EPS molecule. This enzyme plays an important role in EPS biosynthesis in Gram-positive lactic acid bacteria [39,40].
Additionally, one gene, HL1_0495 and M1_0491 in HL1 and M1 (NCBI Ref. KRL28325.1), respectively, coded the ImpB/MucB/SamB family protein, i.e., a a family of error-prone DNA polymerases involved in DNA repair [53]. The ImpB/MucB/SamB family of proteins has been reported to protect DNA from oxidative damage by directly binding to DNA [54]. The presence of the complete glg operon and the ImpB/MucB/SamB proteins in L. kefiranofaciens HL1 and M1 provide the potential for both strains to survive in harsh environments. This finding corresponds to our previous stress adaptation data [55]. The adaptation of L. kefiranofaciens M1 to heat, cold, acid and bile salts induced homologous tolerance and cross-protection against heterologous challenge through the increased synthesis of stress proteins.

Cell Surface Adhesins
By the BLASTx analysis of the complete genomes of L. kefiranofaciens HL1 and M1, three mucus-binding proteins (NCBI Ref. AEG40448.1, KRL28865.1 and WP_013854242.1) and two LPXTG cell wall anchor domain-containing proteins (NCBI Ref. WP_126096172.1 and WP_054640578.1) were identified in each strain with high homology with mucusbinding domains, suggesting some functional similarities. LPXTG cell wall anchor domaincontaining proteins have been reported to contain a C-terminal cell wall sorting signal with a sequence of amino acids; these proteins are connected to the cell wall by sortase A (SrtA) in lactic acid bacteria [56]. Other adhesion related genes, such as glycosylated streptococcal protein B (GspB), with affinity for sialic acid residues in mucins, and the mucus adhesionpromoting protein (mapA), were not found in HL1 and M1. In our previous study in germ free mice, L. kefiranofaciens M1 did not demonstrate a strong adhesion ability [57], probably due to the lack of certain adhesion-related genes.

The Unique Genes in HL1 and M1
A comparison of the full chromosome alignments of HL1 and M1 revealed a significant amount of genetic information about the two strains. The number of unique genes in L. kefiranofaciens HL1 and M1 were 72 and 32, respectively. The unique genes in HL1 comprised 52 hypothetical protein genes and 20 encoded genes (see Table 5, Supplementary Figure S5). The unique genes in M1 comprised 31 hypothetical protein genes and one encoded gene with the function of producing cold shock protein 2.
Adenine phosphoribosyltransferase, arginine deiminase, ornithine carbamoyltransferase and carbamate kinase 1, all of which were identified in HL1, are important enzymes in the arginine deiminase (ADI) pathway for arginine degradation. This pathway has been reported to contribute to ATP and ammonia production, resulting in enhanced viability under anaerobiosis with arginine induction in Lactobacillus sakei [58]. Additionally, arginine deiminase (arcA gene) has been considered as a potential anticancer agent [59] and an inhibitor of cell proliferation in various cancer cell lines [58,60]. The ability of L. kefiranofaciens to degrade arginine by the ADI pathway has never been described in the literature, and its physiological role remains unclear. The presence of the ADI pathway in HL1 suggests that this strain may increase stress tolerance under harsh environments, as well as providing certain health benefits.