Antimicrobial susceptibilities and comparative whole genome analysis of two isolates of the probiotic bacterium Lactiplantibacillus plantarum, strain ATCC 202195

A synbiotic containing Lactiplantibacillus plantarum [American Type Culture Collection (ATCC) strain identifier 202195] and fructooligosaccharide was reported to reduce the risk of sepsis in young infants in rural India. Here, the whole genome of two isolates of L. plantarum ATCC 202195, which were deposited to the ATCC approximately 20 years apart, were sequenced and analyzed to verify their taxonomic and strain-level identities, identify potential antimicrobial resistant genes and virulence factors, and identify genetic characteristics that may explain the observed clinical effects of L. plantarum ATCC 202195. Minimum inhibitory concentrations for selected antimicrobial agents were determined using broth dilution and gradient strip diffusion techniques. The two L. plantarum ATCC 202195 isolates were genetically identical with only three high-quality single nucleotides polymorphisms identified, and with an average nucleotide identity of 99.99%. In contrast to previously published reports, this study determined that each isolate contained two putative plasmids. No concerning acquired or transferable antimicrobial resistance genes or virulence factors were identified. Both isolates were sensitive to several clinically important antibiotics including penicillin, ampicillin and gentamicin, but resistant to vancomycin. Genes involved in stress response, cellular adhesion, carbohydrate metabolism and vitamin biosynthesis are consistent with features of probiotic organisms.


S a bc
Clinical breakpoint interpretation consistent with previous disc diffusion findings for the L. plantarum ATCC 202195 strain 5 Observed MIC value within one doubling dilution of the upper end of the MIC range that was previously reported for 46 other isolates of L. plantarum (0.5-2 mg/l) 22 Ampicillin 2 2 S a,b,c Strain-specific susceptibility data were not previously available for ampicillin Observed MIC value within the range of values previously reported for 46 other strains of L. plantarum (0.125-2 mg/l) 22 and 10 other L. plantarum strains (0.5-32 µg/ml) 23 Oxacillin + 2% NaCl 4 4 -a,b,c Strain-specific susceptibility data were not previously available for oxacillin. Oxacillin activity against other strains of L. plantarum has not previously been reported Strain-specific susceptibility data were not previously available for piperacillin/tazobactam Piperacillin/tazobactam activity against other strains of L. plantarum have not been described previously Lincosamide Clindamycin ≤ 0.12 ≤ 0.12 S a,b,c Strain-specific susceptibility data were not previously available MIC value was within the range of values independently reported by two different groups for 46 (0.032-1 mg/L) 22 and 10 (0.03-32 µg/ml) 23  Strain-specific susceptibility data were not previously available MIC value was within the range of values independently reported by two different groups for 10 (0.25-16 µg/ml) 23 and 46 (0.016-0.5 mg/l) 22 Table 1). The MIC for tetracycline as determined by broth dilution ( ≥ 32 µg/ml) was four times higher than the value determined by E-testing (8 µg/ml) for L. plantarum ATCC 202195-A. In microbroth dilution assays, the concentration of tetracycline only went up to 16 µg/ml (MIC reported as equal to or greater than twotimes the antibiotic concentration in the well where growth occurred). The range of tetracycline concentrations evaluated by E-testing was 0.016-256.0 µg/ml and growth inhibition intersected the side of the E-strip at 8 µg/ ml (Supplementary Table 1). Thus, the true MIC value for L. plantarum ATCC 202195 and tetracycline may lie in between the values described by microbroth dilution assays and E-testing. MICs determined by E-tests were identical (penicillin, erythromycin and vancomycin) or within one doubling dilution (ceftriaxone (MIC = 1 µg/ ml), chloramphenicol (4 µg/ml) and gentamicin (1 µg/ml)) to those determined by broth dilution assays for six of the eight antimicrobials assayed ( Table 1). The MIC for ciprofloxacin ( ≥ 32 µg/ml) as determined by E-testing was 8 times higher than the value determined using the broth dilution assay, which likely reflects the antibiotic concentration range tested in each assay. L. plantarum ATCC 202195-B was negative for beta-lactamase activity.
Comparisons of the L. plantarum ATCC 202195-A and L. plantarum ATCC 202195-B genomes. To corroborate our phenotypic data, the genomes of both isolates were sequenced and compared.   and, (c) Unnamed plasmid 2 (1815 base pairs, G + C content = 37.4%). Images were generated using CGViewer 24 and gene annotation was prepared using RASTtk 25 . Forward strand genes are denoted as red arrows, reverse strand genes are denoted as blue arrows, RNA genes are denoted as orange arrows, and repeat regions are denoted as aqua coloured arrows. The original image was generated using GC Viewer (https:// pauls totha rd. github. io/ cgview/) and the text was modified in Inkscape v1.0.1 (https:// inksc ape. org/ relea se/ inksc ape-1. 0.1/). www.nature.com/scientificreports/ A, with coverage of greater than 1000×, indicating that this discrepancy in genome length is likely a result of variation in sequence methodology and genome assemblers. Comparison between the two bacterial genomes for single nucleotide polymorphisms (SNPs) identified just three high-quality variants ( Table 2). Two of the three variants were localized within one gene encoding a putative surface-layer protein (LPXTG motif), one of which was a missense mutation and the other was a stop-gain mutation. The remaining SNP was located within a transposase, IS4 family gene and was classified as a deletion variant ( Table 2).
The average nucleotide identity (ANI) between the two genomes was 99.99% and global alignment identified 10 locally colinear blocks with a minimum weight of 109 (Fig. 2a). Comparing the genome of L. plantarum ATCC 202195-A with a previously released complete genome assembly for L. plantarum strain ATCC 202195 (accession number: GCA_010586945.1) resulted in an ANI of 99.99% and global alignment identified four locally colinear blocks with a minimum weight of 19,941 (Fig. 2b). In a comparison of our complete genome of L. plantarum ATCC 202195-A with the partial draft L. plantarum ATCC 202195 genome released by Wright et al. 13 (accession number: GCA_004354995.1), we found an ANI of 99.98%. Although Wright et al. did not report any associated plasmid sequences, we found 100% sequence homology between both plasmids described here with contigs 10 and contigs 16 of the GCA_004354995.1 genome assembly. L. plantarum ATCC 202195 Table 2. Summary of single nucleotide variants between L. plantarum ATCC 202195-A and L. plantarum ATCC 202195 -B. a Based on an estimated probability that the alternate allele is present at the loci, as calculated using Freebayes 26 . b Strand balance probability for the alternate allele, as calculated using Freebayes 26 . c Strand balance probability for the reference genome, as calculated using Freebayes 26 .

Nucleotide position
Quality score a  Antimicrobial resistance and virulence genes. As genomic comparison revealed near identical genome sequences, screening for AMR and virulence genes was limited to the most complete genome (L. plantarum ATCC 202195-A). Initial high stringency screening of AMR databases did not identify potential antimicrobial resistance genes. By reducing the screening threshold stringency, three partial matches to the AMR genes LmrD, LmrC and rpoB from CARD 28 and 12 partial matches to known virulence factors in VFDB 29 were identified (Table 3). Based on observed phenotypic resistance patterns, additional targeted screening was performed against known resistance genes for vancomycin (ddl), tetracycline [tet(M), tet(S), tet(W), tet(O), and tet(Q)], and ciprofloxacin (gyrA) ( Table 3). L. plantarum ATCC 202195-A genome was found to contain an intrinsic active site mutation (F261Y) in the ddl gene that confers resistance to vancomycin. A single partial match to a tetracycline resistance gene, tetM, was found, and while gyrA gene was found to be present in L. plantarum ATCC 202195-A genome, no mutations responsible for conferring fluoroquinolone resistance were identified.
Genome features of L. plantarum ATCC 202195-A. Annotation of the genome, including two putative plasmids, using the RASTtk pipeline revealed a total of 3286 coding sequences, including 2287 and 999 sequences assigned as functional and hypothetical proteins, respectively. A total of 72 transfer RNA genes and 16 ribosomal RNA genes were identified. Unnamed plasmid 1 likely represents a conjugative plasmid with a total of 56,486 base pairs (Fig. 1b) that encode 64 genes, 50 of which have assigned putative functions. Unnamed plasmid 2, a non-conjugative plasmid (Fig. 1c), contains just 2 genes: one gene encodes a putative replication protein and the other codes for a copy number control protein. Evaluating the assembled plasmid sequences for sequence homology revealed an almost exact match of unnamed plasmid 2 with that of Pediococcus claussenii ATCC BAA-344 plasmid pPECL-1 32 (sequence identity of 99%, and 100% coverage by BLASTn). Unnamed plasmid 1 has high homology to the unnamed plasmid (60,765 base pair in size) identified in the previously released L. plantarum ATCC 202195 genome assembly GCA_010586945.1 (92% query coverage and 100% identity).  Functional annotation of L. plantarum ATCC 202195-A genome. Using the eggNOG database 33 , 2860 genes were assigned to at least one of the cluster orthologous group families comprising 20 functional groups. Of those 2860 genes, 10.9% were assigned to the transcription functional group, followed by 9.3% of genes assigned to carbohydrate transport and metabolism, and 7.8% of genes assigned to amino acid transport and metabolism (Fig. 4). Further functional characterization by KEGG Pathways analysis, revealed a similar functional distribution of genes, with the majority of identified pathways attributed to nucleotide, carbohydrate and amino acid metabolism. A PATRIC subsystem analysis revealed that 32% of identified genes from L. plantarum ATCC 202195-A mapped to 10 of the 11 possible superclasses, including metabolism (38.4%), protein processing (14.9%), and stress response, defence and virulence (9.3%).
Stress response and adhesion. Based on subsystem analyses, a total of 33 unique genes were sub-categorized as stress response ( Table 4). The L. plantarum ATCC 202195-A genome encodes genes involved in acid tolerance including four bile salt acid hydrolases (bsh1, bsh2, bsh3 and bsh4), and eight sodium-proton antiport genes. A total of 16 genes were subclassified as stress response genes, including nine genes that encode putative universal stress response proteins and seven genes that play a role in the oxidative stress response including glutathione peroxidase, NADH peroxidase and catalase. Several genes responsible for responses to heat and cold stress were also identified, including a number of putative protein-folding chaperones within the well characterized Clp protein family (clpC, clpP, clpL, clpX, clpB and clpE), and four homologous members of the cold shock protein family (cspP, cspL, cspR and cspC). Two of the three small heat shock proteins (sHSP) genes encoded by the L. plantarum reference strain WCFS1 34 were also identified in L. plantarum ATCC 202195-A. The L. plantarum ATCC 202195-A genome also contains several genes that facilitate cellular adhesion, including two genes that encode fibronectin binding proteins, one gene that encodes a mucin binding protein, two mucus adhesion promoting protein genes (mapA), two enolases, two surface layer LPXTG anchored proteins and four putative LPXTG internallins.
Carbohydrate processing and utilization of fructo-oligosaccharides (FOS). Of the 266 genes encoded by L. plantarum ATCC 202195-A that were identified as being involved in carbohydrate metabolism according to COG annotation, only 94 of these genes were annotated as carbohydrate-active enzymes by the CAZy database 36 . Specifically, L. plantarum ATCC 202195-A was found to encode 52 glycoside hydrolases, 35 glycosyltransferases, three carbohydrate binding modules, two carbohydrate esterases and two auxiliary activity enzymes, indicating a strong metabolic capability to degrade and process complex carbohydrates. Embedded within the pool of carbohydrate processing genes was a conserved pts1BCA operon 37 , which is responsible for Table 3. Antimicrobial resistance genes and putative virulence genes present in the genomes of L. plantarum ATCC 2021295-A and L. plantarum ATCC 202195-B. a Virulence genes were identified by comparing DNA sequences from both isolates to the Virulence Factor Database (VFDB) 29 . b Antimicrobial resistance gene were identified using ABRicate 30 and The Comprehensive Antibiotic Resistance Database (CARD) 28  www.nature.com/scientificreports/   35 , and Rapid Annotations Using Subsystems Technology (RASTtk) 25 for annotations.

Stress response
Glutathione: Redox cycle 2 2 Universal stress protein family 9 1 Cluster containing Glutathione synthetase 2 2 Hydroxy-fatty acid production as stress response 1 1 Glutathione: Biosynthesis and gamma-glutamyl cycle 1 1 www.nature.com/scientificreports/ the import of short-chain FOS (scFOS) into the cytosol. The L. plantarum ATCC 202195-A genome also contains three gene clusters critical to the production of the short-chain fatty acid butyrate from acetyl-CoA.

Biosynthesis of B complex vitamins.
Evaluating the L. plantarum ATCC 202195-A genome for genes involved in the biosynthesis of B vitamins, revealed a complete gene cluster (folA, folB, folC1, folC2, folD, folE, folK, folP and folQ) that is involved in folate (vitamin B9) biosynthesis and utilization (Table 5). In contrast to L. plantarum strain WCFS1 38 , the L. plantarum ATCC 202195-A genome contains a complete riboflavin operon, including the ribA, ribB, ribH, ribE and ribG genes, required for riboflavin biosynthesis (Table 5). L. plantarum ATCC 202195-A also encodes genes involved in thiamine and biotin utilization and salvage; however, based on the genome sequences, the microbe appears incapable of de novo synthesis of either of these vitamins.
Bacteriocins. The L. plantarum ATCC 202195-A genome encodes three plantaricin specific operons, including the regulator operon plnABCD, the bacteriocin operon plnEFI and the transport operon plnGHTUVW, which are required to produce the class IIb bacteriocins, plantaricin plnEF and plnA. Bacteriocins are a heterogeneous group of bioactive bacterial peptides that act as antimicrobial agents against closely related susceptible bacterial species 39 . Similar to the reference strain L. plantarum WCFS1 38 , L. plantarum ATCC 202195-A also encodes the plantaricin immunity protein plnL, directly upstream from the regulatory operon.

Discussion
Prior to the widespread use of a probiotic agent in humans, it is imperative to delineate microbial susceptibility to antimicrobial agents, strain level identification and establish if antimicrobial resistance is intrinsic or has the potential to be transferred to other microorganisms 40 . Here, we present the first comparative genome analysis of two isolates of L. plantarum ATCC 202195, which were procured from separate ATCC deposits that occurred approximately 20 years apart. Our analysis revealed previously unreported features of the L. plantarum ATCC 202195 genome including two putative plasmid sequences. In addition, we have provided the most complete antimicrobial susceptibility characterization of this clinically important strain to date. To be considered safe for human consumption, it has been suggested that probiotics should be susceptible to at least two major classes of currently available antibiotics 41 . While transferable resistance is uncommon among lactic acid-producing bacteria, acquired antibiotic resistance has been identified in isolates considered for probiotic or nutritional uses 22 . Herein, we evaluated the genomic sequence of L. plantarum ATCC 202195 for the presence of AMR genes and tested the susceptibility of each isolate against a panel of 20 antimicrobial agents across 16 classes of antibiotics and corroborated these findings with the genome sequence data. The two isolates had identical in vitro antimicrobial susceptibility profiles, were both found to be sensitive to eight of the newly tested antimicrobials (Table 1), and with two exceptions (rifampin and penicillin), the observed MICs fell within the range of values previously reported for other strains of L. plantarum 22,23 . MICs observed for rifampin and penicillin were both within one doubling dilution of the upper range of MICs previously reported for other L. plantarum isolates 23 .
Screening the genome of both L. plantarum ATCC 202195 isolates for sequence homology to AMR genes revealed only partial hits, with low sequence identity and coverage, to the multidrug efflux heterodimers LmrCD and rpoB (β-subunit of RNA polymerase) for which a mutation is known to confer resistance to rifampin 22,23 . In vitro susceptibility testing for rifampin using a broth dilution assay demonstrated growth at the highest rifampin concentration tested, which may be supportive of rifampin resistance; however, we were unable to make a conclusive determination in this regard due to the lack of established clinical breakpoints. Additionally, a partial match to a tetracycline resistance gene tetM, was identified which may be responsible for the phenotypic resistance observed in this strain. Both isolates were resistant to vancomycin, consistent with previously published findings for this strain generated by disc diffusion assay 42 . The signature active site mutation in the ddl gene, which confers resistance to vancomycin, was conserved within the L. plantarum ATCC 202195 genome, corroborating our phenotypic findings. While a clinical breakpoint interpretation was not available for ciprofloxacin activity against lactobacilli, previous disc diffusion assays with L. plantarum ATCC 202195 suggested that the strain is resistant to ciprofloxacin 5 . Herein, we report ciprofloxacin MICs suggestive of resistance for L. plantarum ATCC 202195 using the broth dilution and E-testing assays, respectively. Notably, while the target gyrA gene was present in the L. plantarum ATCC 202195 genome, no fluoroquinolone resistant mutations were identified in this gene and thus, ciprofloxacin resistance may likely be attributed to efflux mechanisms. Both isolates tested in the present study were sensitive to penicillin (MIC = 4 µg/ml) and gentamicin (MIC ≤ 2 µg/ ml), consistent with previous findings for L. plantarum ATCC 202195 obtained using disc diffusion 5 . Since the World Health Organization currently recommends the use of ampicillin plus gentamicin for the initial empiric treatment of neonates with suspected sepsis 43 , evidence that L. plantarum ATCC 202195 is susceptible to these agents further supports the safety of this probiotic strain in the unlikely event that an infant administered L. plantarum ATCC 202195 were to develop probiotic-related bacteremia or infection.
Verifying the genetic lineage of these two isolates is of critical importance to any future clinical work using this probiotic strain. Herein, our comparative genome analyses revealed that L. plantarum ATCC 202195-A and L. plantarum ATCC 202195-B are identical with only three high-quality SNPs identified between the two genomes, and an ANI of 99.99%. Relative to other strains of L. plantarum, L. plantarum ATCC 202195 forms a distinct branch on the phylogenetic tree and its closest phylogenetic neighbours were L. plantarum ATCC 202195 (GCA_010586945.1) and L. plantarum JBE245, a food-related bacterium 44 . In contrast to previous reports, which indicated that L. plantarum ATCC 202195 did not contain a plasmid 5,13 , we identified two putative plasmids in both isolates of L. plantarum ATCC 202195; however, these plasmids were only assembled into complete sequences in the L. plantarum ATCC 202195-A genome.
Plasmid sequence assembly and identification is challenging due to the presence of repeat sequences and the use of short reads for genome sequencing 45 . Herein, we resolved plasmid sequences by combining whole genome sequencing data with sequences that were generated using DNA obtained from a plasmid extraction. Often, plasmid sequences identified in L. plantarum strains do not encode any genes that functionally impact the host 46 . However, we annotated several plasmid-related genes that encode potentially impactful biological properties, including metal transport, amino acid synthesis and carbohydrate processing. Specifically, BglA-2 (6-phospho-beta-glucosidase) encoded on unnamed plasmid 1 is responsible for catalyzing the conversion of cellbiose-6P to glucose and glucose-6P, which then can enter the glycolic pathway to generate further energy for the cell 47 . Transfer of this gene through a conjugative plasmid could potentially provide an adaptive advantage to the plasmid recipient microbe 47 .
Evaluating the functional attributes of a probiotic strain by annotation of the complete genome sequence sheds light onto mechanistic underpinnings of documented clinical outcomes. Previous work has found that oral administration of L. plantarum ATCC 202195 leads to sustained colonization 5 . Here we report strain specific functional gene annotation related to clinical phenotypic findings. Notably, the L. plantarum ATCC 202195 genome was found to contain an array of genes involved in environmental stress responses, including genes related to cellular adhesion, bile acid tolerance, and responses to heat and cold stress, as well as oxidative stress. Since bile acids have known antimicrobial properties 48 , it is advantageous to the bacterium that L. plantarum ATCC 202195 encodes several bile acid hydrolase genes, the products of which can metabolize conjugated bile www.nature.com/scientificreports/ salts 49 . The L. plantarum ATCC 202195 genome also contains the heat shock protein genes: clpE, clpX and clpP. The production of the ATPases ClpE and ClpX and the ClpP protease increase in response to heat-shock and other microenvironmental stresses 50 , and play an important role in maintaining protein quality by regulating proteolysis 51 . Microbial production of secondary metabolites is a potential mechanism by which probiotics can improve host health. Microbial production of the short-chain fatty acid butyrate has been linked to enhanced intestinal epithelial barrier function 52 , modulation of inflammatory status 52,53 , regulation of colonic T-cell differentiation 54 and overall homeostasis in the intestinal tract. Multiple studies have linked the production of short-chain fatty acids to the utilization of prebiotics, including fructo-oligosaccharides and galacto-oligosaccharides 55 . The L. plantarum ATCC 202195 genome encodes three clusters of genes involved in butyrate production, as well as an operon for fructo-oligosaccharide metabolism. In other strains of L. plantarum, butyrate production is upregulated in the presence of fructo-oligosaccharides 56,57 . A potential relationship between the metabolism of fructooligosaccharides and the production of butyric acid by L. plantarum ATCC 202195 is intriguing, since FOS could well have played a role in the positive clinical outcomes in newborns reported by Panigrahi et al. 5 Moreover, the identification of a diverse repertoire of carbohydrate active enzymes suggests that L. plantarum ATCC 202195 is able to degrade, utilize and synthesize both simple and complex saccharides. The functional annotation of the complete genome of L. plantarum ATCC 202195 provides the foundation for future mechanistic investigations; in vitro and in vivo studies are needed to verify the impact of each highlighted functional characteristic on the microbe and host.
Advances in next generation sequencing have increased the availability of high-quality genomic data and improved our ability to identify phylogenetic relationships; however, technical differences in genome sequencing protocols and genome assembly tools can result in artifactual variance 58 . Here, we sequenced, assembled and compared genomes of two isolates from the same strain, L. plantarum ATCC 202195. Our methodology used a combination of short-and long-read sequences for ATCC 202195-A genome assembly and only shortread sequences for ATCC 202195-B, as well as two different assembly tools. To limit technical bias in our comparative analyses, we compared unassembled sequencing reads from L. plantarum ATCC 202195-B against the complete assembled L. plantarum ATCC 202195-A genome, and utilizing genome alignment tools that allow for rearrangement 27 ; however, technical variation may have played a role in the identification of 3 SNPs and the discrepancy in genome length between the two isolates.
This study confirms that L. plantarum ATCC 202195-B, which was only recently made commercially available, is genetically identical to the isolate of L. plantarum ATCC 202195 first deposited into the ATCC over 20 years ago and presumably identical to the isolate of L. plantarum ATCC 202195 used in the hospital and community-based trials conducted in India 5,7 . L. plantarum ATCC 202195 does not contain any unexpected AMR patterns and it is susceptible to multiple clinically important groups of antimicrobial agents. We show that L. plantarum ATCC 202195 contains two plasmids, but since there are no concerning plasmid-encoded antimicrobial or virulence genes, L. plantarum ATCC 202195 does not pose a material threat for the transfer of AMR or virulence factors to other microorganisms. While the probiotic potential of L. plantarum ATCC 202195 has been described in an initial clinical trial 5 , the genomic characteristics that were identified in the current work, including the identification of genes involved in stress responses, cellular adhesion, carbohydrate metabolism, and vitamin biosynthesis, provide further evidence in support of the probiotic properties of L. plantarum ATCC 202195 and shed light on potential mechanisms by which the strain exerts its biological effects on the human host. Taken together, the findings arising from this study will inform the design of future clinical trials and programs to employ L. plantarum ATCC 202195 for use as either a probiotic or, together with FOS, as a synbiotic 59 . Susceptibility testing was performed according to the Clinical and Laboratory Standards Institute (CLSI), M07, 11th edition and M45, 3rd edition, guideline for Lactobacillus spp. 19,60 . Direct colony suspensions, prepared to an equivalent of 0.5 McFarland standard in cation-adjusted Mueller-Hinton broth supplemented with lysed horse blood (CAMHB-LHB), were inoculated into three different commercially available Sensititre plates: Gram Positive MIC, Streptococcus STP6F AST and Gram Negative GN4F AST plates (ThermoFisher). Sensititre plates were incubated for 48 h at 35 °C, 5% CO 2 and MICs were manually recorded and interpreted using clinical breakpoints established by CLSI 19 , the European Committee on Antimicrobial Susceptibility Testing (EUCAST) 20  www.nature.com/scientificreports/ ("S"), intermediate ("I"), resistant ("R"), no interpretive criteria available "("-") and insufficient evidence ("IE").

Materials and methods
To assess the potential for variability in observed MICs, all assays performed using the Sensititre Gram Positive MIC plate were repeated in triplicate for both L. plantarum ATCC 202195-A and L. plantarum ATCC 202195-B.
To minimize bias during interpretation, laboratory technicians and analysts were blinded to the ' A' or 'B' identity of each isolate. Resistance of L. plantarum ATCC 202195-B to eight of the 20 antimicrobial agents assayed using broth dilution assays (ciprofloxacin, erythromycin, ceftriaxone, gentamicin, penicillin, chloramphenicol, tetracycline and vancomycin) was also tested using gradient strip diffusion (Epsilometer test) at the Child Health Research Foundation (CHRF) in Dhaka, Bangladesh. E-strips were procured from AB Biodisk, Sweden (penicillin (0.002-32.0 µg/ml), ceftriaxone (0.002-32.0 µg/ml), gentamicin (0.016-256.0 µg/ml), tetracycline (0.016-256.0 µg/ml) and chloramphenicol (0.016-256.0 µg/ml), Hi Media, India [ciprofloxacin (0.002-32.0 µg/ml), erythromycin (0.016-256.0 µg/ ml)] and Liofilchem, Italy (vancomycin (0.016-256.0 µg/ml). E-testing was performed as recommended by the CLSI 19 . In brief, a single colony of L. plantarum ATCC 202195-B was emulsified in normal saline to achieve turbidity to an equivalent of 0.5 McFarland standard. The inoculum was swabbed over the entire surface of a Mueller-Hinton agar plate supplemented with 5% sheep blood. After the plate had dried (approximately 10 min), E-strips were placed onto the surface of the agar. MICs, which were determined based on the edge where growth inhibition intersected the side of the E-strip, were recorded after incubation for ~ 24 h at 37 °C and then interpreted using breakpoints from CLSI 19 , EUCAST 20 and EFSA 21 . MIC interpretations were categorized using the same criteria applied to the broth dilution assays. L. plantarum ATCC 202195-B was also tested for beta-lactamase activity using Nitrocefin (BD BBL, New Jersey, USA).

Bacterial propagation and nucleic acid extraction. Lactiplantibacillus plantarum ATCC 202195-A
was cultured anaerobically using the BD GasPak EZ container system (BD 260001) in 4 ml of Man-Rogosa-Sharpe (MRS) broth at 37 °C for 16-18 h. Genomic DNA was extracted using the DNeasy Blood and Tissue Kit (Qiagen, Model: 69506) following the pre-treatment protocol optimized for Gram-positive bacteria. In brief, 2 ml aliquots of L. plantarum ATCC 202195-A were centrifuged at 10,000×g for 5 min to harvest bacterial cells. Cells were resuspended in 180 µl of 1× Tris/EDTA buffer at pH 8.0 (BP2473-1, ThermoFisher) containing lysozyme (20 mg/ml) and then incubated for 30 min at 37 °C. Genomic DNA was eluted into 125 µl of elution buffer. RNA was removed by adding RNase (5 µg/µl) followed by incubation for 30 min at 37 °C. DNA yield and quality were assessed by electrophoresis in a 2% agarose gel and using a Qubit 2.0 fluorometer, respectively.
L. plantarum ATCC 202195-B was cultured in 5 ml MRS broth, which was incubated without shaking under aerobic conditions for 16-18 h at 37 °C and aerated with 5% CO 2 . Genomic DNA was purified using the One-4-All Genomic DNA Miniprep Kit (Bio Basic, Markham, Ontario, Canada) according to the manufacturer's protocol, with the addition of an enzymatic lysis step where bacterial cells were resuspended in lysis buffer (20 mM Tris-HCl, pH 8.0, 2 mM Na ETDA, 1.2% Triton X-100, 20 mg/ml lysozyme) and then incubated at 37 °C for 60 min. DNA yield and quality were assessed using a ratio of absorbances measured at 260 nm and 280 nm and 230 nm and 260 nm, respectively, using a NanoDrop 2000c spectrophotometer (ThermoFisher Scientific, Wilmington, MA).
Plasmid DNA extraction. Lactiplantibacillus plantarum ATCC 202195-A was cultured in 5 ml MRS broth, which was incubated without shaking under aerobic conditions for 16-18 h at 37 °C and aerated with 5% CO 2 . Plasmid DNA was isolated using the QIAprep Spin Miniprep Kit (Qiagen, Germany), following the manufacturer's guidelines. Extracted plasmid DNA yield and quality was assessed by Nanodrop, as described above, and agarose gel electrophoresis.
Whole genome sequencing and de novo genome assembly. Whole-genome shotgun sequencing of L. plantarum ATCC 202195-A was performed at the Roy J. Carver Biotechnology Center, University of Illinois Urbana-Champaign, using methods previously described 61 . In brief, Illumina reads were prepared using the Hyper library preparation kit (Kapa Biosystems, Roche, Basel, Switzerland) and sequenced on Illumina MiSeq with paired-end reads 250 nucleotides in length. For Oxford Nanopore (ONT) sequencing, 1 µg of DNA was sheared in a g-Tube and barcoded with 1D Native barcoding genomic DNA kit (EXP-NBD103 and SQK-LSK108). Ten libraries were pooled and sequenced on a GridION X5 sequencer. ONT reads were base-called using Albacore v. 2.1.10 (Oxford Nanopore Technologies, Oxford, United Kingdom). Raw paired-end reads were trimmed using Sickle (v1.3) 62 . Hybrid assembly of the Illumina and ONT reads was performed using Unicycler (v0.4.7) 63 , and the initial genome assembly was refined using Pilon 64 . To inspect the fidelity of the assembly, Illumina reads were mapped to assembled reference contigs using the Burrows-Wheeler Aligner (BWA) (v0.7.17) 65 and Samtools (v1.9) 66 , and visualized using the Integrative Genomics Viewer (IGV) (v2.5.3) 67 . No manual corrections were made.
Whole-genome shotgun sequencing of L. plantarum ATCC 202195-B was performed at the Clinical Genomics Centre at Mount Sinai Hospital (Toronto, Ontario, Canada) using the NexteraXT platform (Illumina) with sequencing libraries prepared with the Nextera XT DNA library preparation kit. Paired-read sequence libraries were generated at 150 nucleotides in length, according to the manufacturer's instructions. Raw sequenced reads were trimmed using Trimmomatic 68 ; nucleotides at the beginning and end of each sequencing read were discarded if the quality score was below 20 Phred. Reads shorter than 30 bases in length were also discarded. FastQ Screen 69 was used to identify human genomic DNA contamination using PhiX as a calibration control. Reads associated with the human genome were removed from all subsequent analyses. BBNorm (v37.02) 70 was used to normalize coverage in high-depth regions of the genome. SPAdes genome assembler (v3.13.3) 71 was used to reconstruct the genome and the quality of the genome assembly was assessed using QUAST (v5.0.2) 72 . www.nature.com/scientificreports/ Next generation sequencing of the extracted plasmid DNA from L. plantarum ATCC 202195-A was performed at the Clinical Genomics Centre at Mount Sinai Hospital (Toronto, Canada) using the same methodology employed for L. plantarum ATCC 202195-B whole genome sequencing. A stricter filtering and trimming protocol was utilized to process plasmid sequencing reads as per Gallegos 73 . In brief, raw reads were trimmed using Trimmomatic 68 , and discarded with a quality score below 30 and a minimum length of 50 base pairs. Trimmed reads were assembled using Unicycler (v0.4.7) 63 , with bold contig bridging. Assembled contigs were flagged as putative plasmid sequences if the assembled multiplicity was ≥ 10× or if the contigs were denoted as complete and circular. The resulting identified putative plasmid sequences were screened against the NCBI nucleotide database using BLASTn 31 and PLSDB 74 . Comparative genome analysis. Overall similarity between the assembled genomes of L. plantarum ATCC 202195-A and L. plantarum ATCC 202195-B and other publicly available assembled genomes for L. plantarum ATCC 202195 (accession number: GCA_010586945.1 and GCA_004354995.1) were assessed by comparing the average nucleotide identities (ANI) using the Orthologous Average Nucleotide Identity Tool (OAT) 75 . A global alignment was performed using progressiveMauve 27 and the contig mover feature was employed to account for variation introduced due to differing genome assembly tools used for L. plantarum ATCC 202195-A and L. plantarum ATCC 202195-B. The two bacterial genomes were then compared for SNPs. Paired-end normalized read libraries from the genome of L. plantarum ATCC 202195-B were aligned with the assembled genome of L. plantarum ATCC 202195-A using the BWA-MEM algorithm 65 with default parameters. SNPs were identified using the variant caller FreeBayes (v1.2.0) 26 . For aligned reads to be evaluated as a potential variant, a minimum mapping quality of 20 was used and ploidy was set to 1. Filtering thresholds were employed to detect variants with a quality score > 10 and a read depth > 5. Variants that remained after filtering were assessed for their potential functional impact using SnpEFF 76 . Functional genome annotation of L. plantarum ATCC 202195-A and identification of antimicrobial resistance and virulence genes in L. plantarum ATCC 202195-A and L. plantarum ATCC 202195-B. The de novo assembled L. plantarum ATCC 202195-A genome was annotated using RASTtk (v1.3.0) 25 and the evolutionary genealogy of genes: Non-supervised Orthologous Groups () Database 33 . Annotated genes were compared against the Clusters of Orthologous Groups (COG) 77 and Kyoto Encyclopedia of Genes and Genomes (KEGG) databases 78 to make inferences about higher order biological functions. Carbohydrate Active Enzymes (CAZymes) were identified using the meta server for automated carbohydrate-active enzyme annotation (dbCAN) 79 . CAZyme annotation of a putative L. plantarum ATCC 202195 gene was based on agreement between at least two of the three annotation tools employed by dbCAN, including HMMER which searched against the CAZy hidden Markov models database, DIAMOND which screened against the preannotated CAZY database, and Hotpep, which screened against a conserved CAZyme short peptide sequence database.
Lactiplantibacillus plantarum ATCC 202195-A was screened against five antimicrobial and virulence factor databases: Comprehensive Antibiotic Resistance Database (CARD) 28 , ResFinder 80 , Antibiotic Resistance Gene-ANNOTation (ARG-annot) 81 , Virulence Factor Database (VFDB) 29 and the NCBI Bacterial antimicrobial resistance reference gene database using ABRicate (v0.5) 30,82 . Two different stringency thresholds for AMR gene detection were used: a high stringency threshold, which used a sequence identity and query coverage cut-offs of > 80% and a low stringency threshold employing a sequence identity cut-off of > 50% and a query coverage cut-off of > 10%. Targeted screening for resistance genes related to phenotypic resistance patterns was performed using BLASTp.