Probiogenomic analysis of functional potential and safety of L. plantarum 8p-a3 and DMC-S1 strains: in silico vs in vitro and in vivo data

ABSTRACT

The molecular basis of the beneficial effects and the causes of the negative effects of probiotics are not entirely clear. Clarifying these issues is important for understanding the biology and assessing the safety of the microbes. Omics technologies have opened up new resources for obtaining relevant knowledge. Here, for the first time, we present the results of a comparative analysis of the functional potential and safety of two L. plantarum strains: the approved probiotic 8p-a3 and the Drosophila intestinal resident, which exhibit opposite effects on D. melanogaster as the model host organism. Through genomic analysis, extracellular vesicle studies, and in vitro and in vivo assays, we have identified the common and specific characteristics of the strains. The strains proved to be similar in a set of genes that determine benefits to the host organism, as well as in the presence of some risk factors. Significant differences between the strains are related to genes responsible for adhesion, sialic acid metabolism, mucin degradation, antimicrobial peptides, tannin resistance, and immunomodulation. In silico data correlated with in vitro and in vivo data, with the exception of antimicrobial sensitivity. Pronounced differences between the strains were found in terms of the composition and biological effects of their vesicles. In vivo data on the effects of the strains correlate with the corresponding data of their vesicles in the fruit fly model. The results obtained open up new facets in L. plantarum strains relevant for evaluating the functionality and safety of probiotics.

IMPORTANCE

Using a probiogenomic approach, common and specific features regarding functionality and safety were identified in the strains (the approved probiotic strain L. plantarum 8p-a3 and the Drosophila intestinal bacterium L. plantarum DMC-S1), which exhibit opposite effects on the model host organism (D. melanogaster). The genomic analysis was supplemented by the analysis of extracellular vesicles of the strains. Comparative analysis of in silico data in combination with in vitro and in vivo studies was performed, and unexpected capabilities of the strains were discovered. Novel factors, essential for evaluating the safety of probiotics, were identified. New facets in the interplay of probiotic bacterium with host organism have been revealed.

INTRODUCTION

Probiotics are attracting increasing attention due to their ability to positively influence intestinal homeostasis and the immune system of the host, regulate the structure of microbial communities, and inhibit the growth of pathogenic bacteria (1). The antimicrobial activity of these microorganisms is especially relevant nowadays, given the rapid spread of antibiotic resistance and the antimicrobial drug crisis background (2). The creation of new-generation antimicrobials, immunomodulators, and precision postbiotics is associated with bacteria with GRAS (Generally Recognized As Safe) status, in particular probiotic strains. Specific compounds produced by bacteria (antimicrobial peptides), as well as bacterial extracellular vesicles transporting bioactive molecules and mediating intercellular communication and interactions of bacteria with pro- and eukaryotic cells, are considered candidates for the role of novel postbiotics (3–5). The realization of the relevant prospects involves a comprehensive analysis of the strains’ efficiency, both in terms of functionality and safety. Meanwhile, the molecular mechanisms of interplay between probiotic bacteria and cells of micro- and macroorganisms, including those providing beneficial properties for the host organism, remain largely unknown (6, 7). Moreover, while differential outcomes of the interaction of various probiotic strains with the host organism, as well as the manifestation of negative effects of probiotics, have been reported, the causes of these phenomena remain unclear (6, 8). The lack of standardization of the analysis of probiotics (and probiotic candidates) is a significant obstacle to addressing these issues (1).

The development of high-resolution methods, advancing genomic and postgenomic technologies, and progressing bioinformatic tools has determined the emergence of a new scientific field named probiogenomics (9, 10). This area aims to ensure that the potential of omics technologies, along with the capabilities of classical microbiology, is used for the comprehensive characterization of probiotic strains. The new knowledge gained deepens our understanding of the biology of bacteria and advances probiogenomics. The establishment of the fact that extracellular vesicles are key intermediaries of bacterial interaction with cells of macro- and microorganisms determines the relevance of analyzing the properties of these nanostructures for understanding the biology of intestinal commensals/symbionts, as well as for evaluating the functionality and safety of probiotic strains (3, 11).

Genome-wide analysis of strains and targeted in vitro/ex vivo/in vivo studies are used today to assess crucial elements—probiotic features and risk factors in “probiotic candidates” (10). Such analysis is also relevant for long-used commercial probiotics, especially if negative effects have been reported for them.

We have shown previously that the approved probiotic strain Lactiplantibacillus plantarum 8p-a3 (isolated from the commercial probiotic drug Lactobacterin) exhibits negative effects against the model host organism Drosophila melanogaster (Canton-S) (12). On the contrary, the strain isolated from D. melanogaster intestine, L. plantarum DMC-S1, exhibits an exceptionally positive effect on the host. The existence of two phylogenetically close strains that differentially affect the host, in combination with probiogenomics tools, opens a unique possibility to reveal genetic features determining the observed effects and to conduct a comparative study of these bacteria from the point of view of their functionality, safety, and identification of candidate factors associated with the negative effects of the probiotic strain.

Here, we present a comparative analysis of in silico, in vitro, and in vivo data reflecting the functional potential and safety of two L. plantarum strains: the approved probiotic 8p-a3 and the Drosophila gut resident DMC-S1, which exhibit opposite effects on the model host organism D. melanogaster (Canton-S). The genome-wide analysis of the strains, supplemented with data characterizing features and biological activity of the strains’ vesicles, allowed the identification of novel factors essential for evaluating the functionality and safety of probiotics.

RESULTS

The general properties of genomes of L. plantarum strains

The whole genome of L. plantarum DMC-S1 has been sequenced using Illumina MiSeq and Oxford Nanopore MinION platforms with an overall coverage of 180× (75× on Illumina MiSeq and 105× on Oxford Nanopore MinION) and sequence reads quality of 99.9%. The L. plantarum DMC-S1 genome includes a circular chromosome consisting of 3,376,068 bp and a GC content of 44.5% (Table 1; Fig. 1), and two plasmids with a total length of 63,256 and 35,273 base pairs, respectively (2.92% of the genome), and GC contents of 37.5% and 39.5%. In total, the genome contains 3,349 predicted open reading frames, as well as genes encoding six copies of 5S rRNA, five copies of 16S rRNA, five copies of 23S rRNA, 82 tRNA genes, and four ncRNA molecules.

TABLE 1.

Comparison of L. plantarum 8p-a3 and DMC-S1 genome features^a

	L. plantarum 8p-a3*	L. plantarum DMC-S1
Genome length (bp)	3,301,244	3,376,068
GC content (%)	44.5	44.5
Plasmids	ND (**)	2 (63,256 bp and 35,273 bp)
Number of ORF	3,154	3,349
5S rRNA	5	6
16S rRNA	2	5
23S rRNA	1	5
tRNA	73	82
ncRNA	4	4

^a (*) ANI L. plantarum 8p-a3 and L. plantarum 8P-A3 (GCF_009762745.1), 99.997%; ND, no data; (**) plasmid of L. plantarum 8P-A3, 9,480 bp.

Fig 1.

Circular maps of L. plantarum 8p-a3 and DMC-S1 genomes visualized using the Proksee tool. (A and B) Chromosomes of L. plantarum 8p-a3 and DMC-S1, respectively. (C and D) Plasmids of L. plantarum DMC-S1. See COG category abbreviations in Table S1.

The genome of L. plantarum 8p-a3 has been sequenced earlier with an overall coverage of 175× (12). The chromosomal DNA consists of 43 scaffolds with the total length of 3,301,244 base pairs and the GC content of 44.5% (Table 1; Fig. 1). The genome contains 3,154 predicted open reading frames, as well as genes of five copies of 5S rRNA, two copies of 16S rRNA, one copy of 23S rRNA, 73 tRNA genes, and four ncRNA molecules. The main characteristics of the 8p-a3 strain genome correspond to those of strain 8P-A3 (GenBank NCBI CP046726; CP046727), isolated from the commercial probiotic drug Lactobacterin and analyzed by independent researchers. L. plantarum 8P-A3 also carries a plasmid DNA of 9,480 bp in length and a GC content of 44.5%. The average nucleotide identity (ANI) between L. plantarum 8p-a3 and 8P-A3 is 99.997%. In total, 50 mutations differentiating these strains from each other have been confirmed by both raw whole genome and read alignment, including 14 (28.0%) mutations in the intergenic region, 22 (44.0%) in genes encoding hypothetical proteins, and 14 (28.0%) in genes with a known function (msrA, rbsR, rpoN, rpoD, mltG, atpF, genes encoding 2-oxoacid dehydrogenase subunit E2, ABC transporter permease, C40 family peptidase, Cof-type HAD-IIB family hydrolase, Dps family protein, LysR family transcription regulator, and uracil-xanthine permease family protein) (see ST_snp.xlsx). Thus, the strains 8p-a3 and 8P-A3 are very close, which allows data to be extrapolated.

Functional annotation of genomes

To evaluate the functional potential of L. plantarum strains, their genomes and gene functions have been annotated using SEED, COG, KEGG, Pfam, and TCDB databases. Results are summarized in Table 2; Tables S1 to S3 and Fig. 2. Most genes are involved in the metabolism of carbohydrates and amino acids, genetic machinery, and environmental information processing. A substantial fraction of genes belongs to the category of “unknown function.” Both strains have pathways for the synthesis of vitamins (riboflavin, thiamine, pyridoxine, and folic acid), which are important for the metabolism of the host, the lactic acid bacterium itself, and other bacteria in the consortium. While the genes for sialic acid metabolism were found in both genomes, three genes associated with the catabolism of sialic acids (N-acetylneuraminate lyase, putative sialic acid transporter, and sialic acid utilization regulator) were found in the genome of L. plantarum DMC-S1 and were absent in the 8p-a3 strain (Fig. 3). Thus, the functional annotation of genomes indicates high similarity among strains in terms of functional potential, including the pool of genes relevant for survival under critical conditions and the implementation of functions beneficial to the host. At the same time, both strains contain genes for sialic acid catabolism, although with some differences, suggesting an advantage of L. plantarum DMC-S1.

TABLE 2.

Functional profiling of L. plantarum 8p-a3 and DMC-S1 genomes^a

	Database
	COG	KEGG	SEED	Pfam	TCDB
L. plantarum 8p-a3	2,554	1,259	930	2,528	368
L. plantarum DMC-S1	2,665	1,284	970	2,618	389

^a In case of multiple annotation results, the best ratio is chosen for gene annotation.

Fig 2.

Comparison of distribution of COG (Cluster of Orthologous Groups) functional categories of the proteins in L. plantarum 8p-a3 and DMC-S1 (according to COG annotation).

Fig 3.

The presence of genes for sialic acid metabolism in L. plantarum DMC-S1 and 8p-a3 (according to SEED annotation).

Safety assessment of strains in silico and in vitro

Genome instability

The genomes of L. plantarum 8p-a3 and DMC-S1 contained features associated with genomic instability, including IS, transposons (Table S4), prophages (Table S5), plasmids (Tables S6 to S8), CRISPR-Cas systems (Table S9), and genomic islands (ST_islands.xlsx). The genes within genomic islands are associated with bacteriocins, prophages, insertion elements, transcriptional regulation, transporters, and virulence, in particular, hemolysins of the XhlA family. While CRISPR/Cas systems of bacteria contribute to their protection from viruses, CRISPR/Cas systems in both strains have been identified with an index of 1.0, indicating their apparent inactivity (Table S9), increasing the possibility of foreign DNA acquisitions associated with horizontal transfer, virulence factors, and antibiotic resistance determinants. These data suggest similarity between the strains in terms of the potential for genomic instability.

Hemolysins and toxic metabolites

Using the RefSeq annotation results, genes for hemolysins of the XhlA family have been identified in both genomes. Therefore, the expression of these genes was evaluated in vitro by qRT-PCR (Fig. S1). A twofold higher expression of the хhlA gene at 37°C than at 25°C has been observed in both strains, suggesting their hemolytic potential. Furthermore, the appearance of a greenish halo around both strains, 8p-a3 and DMC-S1, has been observed on sheep blood agar (Fig. S2), confirming α-hemolytic activity, i.e., the ability to cause partial degradation of erythrocytes, similarly to S. sobrinus, which served as a reference.

The potential of Lactobacilli to produce metabolites that are potentially toxic for humans was evaluated via BLASTn search for the respective genes. While the genes of biogenic amine synthesis were not detected in any strain, both genomes contained the key genes for D-lactate synthesis and genes associated with nitro compound formation (Table S10). Moreover, while the qRT-PCR data indicated the constitutive expression of genes for D-lactate synthesis in cells of both strains, their expression differs between the strains and depends on the temperature, significantly increasing at 37°C in the probiotic strain. The data obtained indicate that the strains studied are similar in terms of “potential genomic liabilities” requiring further validation.

Determinants of antibiotic resistance

Using RGI and CARD, the dltABCD and mprF genes responsible for resistance to cationic antimicrobial peptides have been identified in both genomes. While other genes associated with AMR were not detected in both strains at a high probability threshold (Table S11), both strains were resistant to streptomycin, aminoglycosides, tetracycline, cephalosporins (except ceftazidime for L. plantarum 8p-a3), vancomycin, and fluoroquinolones, and were sensitive to amoxicillin, ampicillin, carbapenems, and macrolides (Table S12).

Virulence and persistence factors

While VirulenceFinder did not identify any virulence genes in both genomes, BLASTn and VFDB revealed a number of virulence-associated genes (Table S13), although with low identity. Some marker genes found in probiotic strains encoding adhesins and immunomodulators may also ensure the bacterial survival and persistence (are associated with virulence in pathogens), as well as expansion under stressful conditions (Tables S14 and S15).

While the strains differ only slightly in genes required for persistence, significant differences are found in the profiles of adhesin genes cwaA, msa, mub, and LPXTG. Of note, proteins containing LPXTG motif are considered essential adhesins, and the msa is a marker of probiotic strains. L. plantarum 8p-a3 has a genotype of cwaA⁺ msa⁻ mub⁺ LPXTG⁵⁺, and L. plantarum DMC-S1 is cwaA⁻ msa⁺ mub⁻ LPXTG⁶⁺, thus being expected to differ in adhesion and autoaggregation abilities. These data matched the results of in vitro tests (Fig. S3 and S4): the DMC-S1 strain exhibited a higher rate of autoaggregation compared to 8p-a3 (78% vs 62%) and twofold higher hydrophobicity.

Significant differences between the strains are found in the set of genes for tannin resistance. In addition to the tanB gene, the probiotic strain 8p-a3 contains the tanA gene encoding a rare subtype of tannase, which may provide bacteria with unique opportunities for stress survival. The gene has a constitutive expression while significantly increased at 37°C (Fig. S1).

The elucidation of the role of AckA and MprF in immunomodulation, which is critical for the outcome of the interaction of L. plantarum with a model organism (D. melanogaster), requires the need for attention to the gene sequences of the corresponding proteins. The strains studied proved to carry different variants of the ackA and mprF genes (Fig. S8 and S9), and the expression patterns of these genes differ between the strains, and depend on the temperature, significantly increasing at 37°C in probiotic strain (Fig. S1).

Thus, strains differ by the pattern of genes required for bacterial persistence and in the outcome of their interaction with the host, which is also consistent with in vitro data. While the resident strain is superior in adhesion capacity, the presence of genes of two subtypes of tannase increases the survival potential of the probiotic strain under unfavorable conditions.

Bacteriocin-encoding genes

Using the BAGEL4, clusters of bacteriocin-encoding genes have been identified in genomes of both strains (Fig. 4; Table S16).

Fig 4.

The organization of a bacteriocin synthesis cluster in the genomes of L. plantarum 8p-a3 (А) and DMC-S1 (В), predicted by BAGEL4. The genes in blue were determined using BLASTp.

Nevertheless, the clusters vary between strains. L. plantarum DMC-S1 has a cluster of genes belonging to the plantaricin J class, which contains the core peptides PlnEF, PlnA, PlnN, and PlnJK. In the L. plantarum 8p-a3 genome, another variant of the plantaricin cluster was found, containing genes for the core peptides PlnNC8αβ, PlnJ, PlnA, PlnEF, and Enterocin-x-β. The BLASTp analysis revealed that the genes encoding (i) bacteriocin immunity proteins PlnI and L (in both strains), PlnM and P (in DMC-S1); (ii) transcription regulators PlnC and D, histidine protein kinase PlnB, membrane proteases containing CAAX signature PlnT, U, V, and W (in both strains), protein PlnR with unknown function (in DMC-S1), protein of unknown function PlnQ, and lactococcin/core bacteriocin ComC/BlpC family leader-containing protein (in 8p-a3).

Plantaricin genes apparently provide antagonistic activity of the strains observed in in vitro tests against various pathogens in spot-on-lawn tests, as well as the repression of the biofilm formation by the cell-free culture liquid of these bacteria (Fig. S5 and S6). Interestingly, the antagonistic activity between these two strains was not observed in the compatibility streak-seeding test, suggesting that the strains are tolerant of the antimicrobial compounds produced (Table S16). Thus, despite some differences in the profiles of the bacteriocin cluster genes, the strains seem to be biocompatible and demonstrate similar antagonistic activity.

Genes-based phylogenetic reconstruction

The analysis of average nucleotide identity (ANI), phylogenetic reconstruction based on core genes, and conserved genes encoding the crucial symbiosis factors AckA and MprF (regulating the level of immunotolerance relevant to the bacterium and the host) was used to determine the phylogenetic position of L. plantarum 8p-a3 and L. plantarum DMC-S1 among other L. plantarum strains isolated from sources with similar or different conditions and carrying similar or divergent gene variants (Fig. 5; Fig. S7 to S9).

Fig 5.

The phylogenomic tree of L. plantarum strains isolated from disparate sources is based on genome-wide data. The numbers adjacent to the branches are GBDP pseudo-load support values exceeding 60% in 100 replications, with an average branch support value of 14.6%. The sources of bacterial excretion are as follows: red, D. melanogaster; brown, human feces; green, fermented foods; blue, dairy products; and gray, other sources. The numbers in the boxes indicate different gene variants; ps, pseudogene. A phylogenomic tree was reconstructed using the GGDC web server (http://ggdc.dsmz.de/).

L. plantarum DMC-S1, isolated from the intestine of Drosophila, has the highest homology with strains isolated from similar sources (WJL, dm, and BDGP2). The genome of L. plantarum 8p-a3 is more similar to other strains obtained from fermented sources, as well as the gastrointestinal tract of flies (EM, DietG20.1.2, and JCM8341). The strains closest to L. plantarum DMC-S1 by ANI are dm (ANI 99.95%), WJL (ANI 99.84%), and BDGP2 (ANI 99.80%). For L. plantarum 8p-a3 (as well as 8P-A3), the closest strains are EM (ANI 99.45%) and HOM3201 (ANI 99.31%). ANI between L. plantarum 8p-a3 and L. plantarum DMC-S1 is 98.99%. According to the phylogenetic data, a strict pattern of the relationship between the ackA and mprF gene variants or the degree of genomic homology and the sources of strains is not evident. For example, the ackA sequence variant, “the moderate Drosophila-growth-promoting ability variant,” is present in L. plantarum strains with varying degrees of genetic proximity that were isolated from different sources (including the commercial probiotic Lactobacterin), and vice versa (Fig. 5). Thus, the results of phylogenetic reconstruction indicate that L. plantarum 8p-a3 and L. plantarum DMC-S1 are phylogenetically close—the probiotic strain 8p-a3 turned out to be close to the Drosophila-associated strain line. However, the versions of the immunomodulator genes, crucial for the outcome of the interaction between L. plantarum and D. melanogaster differ between the strains.

Extracellular vesicles: physicochemical properties and bioactivity

An important part of the bacterial secretome, relevant for intercellular communication and targeted action on recipient cells, is associated with extracellular bacterial vesicles (3). Using TEM and NTA, we found that L. plantarum 8p-a3 and DMC-S1 produce extracellular vesicles (EVs), membrane-surrounded nanostructures with average diameters of 148 ± 1.9 nm and 157 ± 2 nm, respectively. These EVs represent a heterogeneous pool consisting of vesicle subpopulations that vary in size and exhibit train-specific profiles (Fig. S10A).

According to infrared spectroscopy (IR) data, lipids, nucleic acids, polysaccharides, and proteins are present in EVs of both strains (Fig. S10B). The specific features of the IR spectra in a number of frequency ranges indicate differences in the vesicular composition of the strains in terms of the qualitative and quantitative characteristics of the corresponding compounds, including proteins (1,550 cm⁻¹). The LC-MS/MS proteomic analysis of EVs revealed 37 proteins in 8p-a3 vesicles and 39 proteins in DMC-S1 vesicles, among which 23 proteins were common, while 14 and 16 were strain-specific, respectively (13).

A significant part of the proteins in the vesicles of both strains belongs to those with the potential to affect the biofilms, exhibiting pro- or antibiofilm activity (Fig. S11). The spectrum of such proteins in the strains turned out to be similar, suggesting similar effects on bacterial biofilms. Indeed, in vitro both strains could affect the density of biofilms of some ESKAPE group pathogens. Significant decreases in the biofilms of E. faecium, S. aureus, and P. aeruginosa have been observed, but not E. coli (Fig. S12).

The ex vivo data indicate that vesicles of both strains can modulate immunoreactivity in the gut of D. melanogaster (Fig. S13), although the expression patterns of D. melanogaster genes associated with the immune response to vesicles of L. plantarum 8p-a3 and DMC-S1 differed significantly. In the case of vesicles from 8p-a3, predominant expression of genes associated with the activation of the immune response was observed, but not protection from oxidative stress (which obligately accompanies activation of immunoreactivity), whereas in the case of vesicles from DMC-S1, the opposite effect was found. A significant increase in the expression of the Drs, Mtk, CecA1, PRDX5, and Reaper genes recorded in the intestinal cells of flies treated with 8p-a3 vesicles (but Drs, cat, dFoxo, and Reaper in the cells treated with DMC-S1 vesicles) indicates activation (in the case of the probiotic strain) of the immune pathways associated with the induction of host tissue damage.

The in vivo effects of the vesicles derived from 8p-a3 and DMC-S1 proved to be in line with those of the strains. Pronounced negative effects were registered in the case of the probiotic strain (Fig. S14 and S15). Intrathoracic inoculation of adult flies with vesicles obtained from the L. plantarum 8p-a3 strain (but not DMC-S1) negatively affected the reproductive status of insects, reducing the number of eggs laid by females 3.5 times compared to the control group (Fig. S14A) and 3.3 times compared with flies inoculated with vesicles from DMC-S1. Inoculation of imago with vesicles from both 8p-a3 and DMC-S1 led to a significant increase in the fetal death frequency by 1.6 and 1.47 times, respectively, compared to the control group (Fig. S14B). Treatment of third-instar larvae for 24 h with vesicles isolated from the L. plantarum 8p-a3, but not DMC-S1, led to an increase in DNA damage in intestinal enterocytes, increasing the DNA comet index by 1.43 times compared to the control (Fig. S14E). The shift in the distribution of DNA comet types (from n0 and n1 toward n3 and n4) and histological data (Fig. S14F) also indicates the (geno)toxic effects of the probiotic strain vesicles on intestinal tissues in fruit flies.

Proteins associated with virulence, which determine the genotoxicity, are not found in vesicles of the probiotic strain, whereas the product of the hupB gene (HU family protein, a histone-like protein), which is associated with bacterial virulence and is absent in the vesicles of the DMC-S1 strain, has been detected (Fig. S13). It can be secreted through bacterial vesicles and enter host cells, bind DNA, specifically modulate gene expression, potentially contributing to the host DNA damage and the development of pathological processes. However, so far it has been shown for pathogenic bacteria [14, 15].

Thus, these data show pronounced differences in the proteomic composition and biological effects of vesicles derived from the strains studied. The negative effects of L. plantarum 8p-a3 vesicles found in the experimental model used (D. melanogaster) determine the need for further relevant studies in higher-order model systems to confirm the safety of the probiotic strain.

DISCUSSION

The molecular basis of the beneficial effects and the causes of the negative effects of probiotics are not entirely clear (1, 4). Clarifying these issues is important for understanding the biology of these microbes, the mechanisms of their interaction with the host and other microbes, and for assessing the safety of probiotic strains. Omics technologies have opened up new resources for obtaining relevant knowledge. New knowledge opens up new facets of probiotics and defines new directions for their research, advancing probiogenomics (9). The establishment of the fact that extracellular vesicles are key intermediaries of bacterial interaction with cells of macro- and microorganisms determines the relevance of analyzing the properties of these nanostructures for understanding the biology of intestinal commensals/symbionts, as well as for evaluating the functionality and safety of probiotic strains, especially if their producers have been found to have negative properties (3, 5).

We have previously shown that the approved probiotic strain L. plantarum 8p-a3 (isolated from the commercial probiotic drug Lactobacillus and used in clinical practice in Russia since 1973) exhibits negative effects on the model host organism D. melanogaster (Canton-S). On the contrary, the L. plantarum strain DMC-S1, isolated by us from the intestine of D. melanogaster in 2022, has an exceptionally positive effect on the host organism (12). Here, for the first time, we present the results of a comparative analysis of the functional potential and safety of these two strains to identify candidate factors associated with the negative effects of the probiotic strain in the D. melanogaster model. Through genomic analysis, extracellular vesicle studies, and in vitro and in vivo assays, we have identified the common and specific characteristics of the strains. We performed a phylogenetic reconstruction of the strains. The strains turned out to be phylogenetically similar (the probiotic turned out to be close to the “Drosophila lineage”), including in terms of their functional potential and safety. The strains proved to be similar in a set of genes that determine benefits to the host, as well as in the presence of some unwanted sequences (common among L. plantarum strains and requiring further safety studies).

Significant differences between the strains were found in the sets of genes responsible for adhesion, antimicrobial peptides, sialic acid metabolism, mucin degradation, and tannin resistance. In addition, important differences have been identified in the genes of immunomodulators, which are crucial for the microbe-host outcome, including their primary structures and expression. These genetic factors essential for the persistence of bacteria in the host can determine the specific features of strains.

The adhesion is an essential aspect for the survival of a probiotic in the gastrointestinal tract (16). Therefore, the hydrophobicity and autoaggregation are assessed as indicators of microbial potential for adhesion and attachment to the gut. The genetic potential of the strains for adhesion turned out to be different, and in vitro analysis indicated a correlation between genotype and phenotype. In the genomes of both strains, genes encoding proteins with LPXTG motifs that anchor the cell wall and promote bacterial colonization in the gastrointestinal tract are found (17) (Fig. S14). However, the msa gene, considered as a marker of reliable adhesion of probiotic L. plantarum strains (18), is present in DMC-S1, while it is absent in the 8p-a3 strain. This fact could explain the observed lower hydrophobicity and autoaggregation capacity of the 8p-a3 strain compared to DMC-S1 (Fig. S3 and S4). In turn, autoaggregation is believed to promote competitive exclusion of microbes from the econiche and displacement of pathogens of intestinal commensals and provides resistance to antimicrobials, survival, and persistence in certain niches (19). Interestingly, the antagonistic activities of both strains were close and relatively high against various pathogenic bacteria (Fig. S4), including the biofilms (Fig. S5 and S6). This phenomenon seems surprising against the background of significant differences between strains in relation to bacteriocin genes.

L. plantarum strains differ significantly in bacteriocin synthesis, and some isolates do not produce the required range of these compounds (20). Meanwhile, according to in silico data, L. plantarum 8p-a3 and DMC-S1 have significant bacteriocinogenic potential, with 29 and 31 genes involved in bacteriocin production found in the genomes of the strains, respectively (Fig. 4; Table S16). These genes may provide expression of peptides suppressing Gram-positive and Gram-negative bacteria, fungi (21), and viruses, including the Flaviviridae family (22, 23). These findings match the pronounced ability of these strains to suppress the growth of pathogens in vitro (Fig. S5 and S6). On the other hand, different patterns in the bacteriocin clusters indicate putative different capabilities of the strains to hit targets. The presence of genes for NC8αβ in the probiotic strain allows suggesting its possible contribution to the modulation of cell proliferation (24) and the structure of the host intestinal microbiota (25), also observed in the case of L. plantarum 8p-a3 in our studies (12). Elucidation of the effects of various plantaricins on molecular processes in host cells is a task of future research relevant to assessing the safety of probiotics.

Adhesion of probiotics to intestinal cells leads to their successful competition for binding sites with other bacteria in the community, including pathogens. Bacteria must overcome the gut mucin layer: several glycosyl hydrolases (GHs) work in cooperation in its degradation (1, 26–28). Bacteria of the genus Lactobacillus have a small spectrum of GHs associated with mucin degradation. According to genomic analysis, they carry genes for only 1–4 types of GHs, but not of GH33. The absence of a gene for GH33 does not allow considering the strains as involved in significant degradation of mucin. However, it should be noted that the potential of the L. plantarum 8p-a3 strain (GH2, GH20, GH42, and GH85) in relation to mucin-lytic abilities is still higher than that of DMC-S1 (GH2, GH20, and GH42) (Table S15).

Resistance to tannins is an essential factor for the persistence of bacteria in the intestine. The presence of tannase genes in bacteria determines the possibility of tannin degradation. Regarding the genetic determinants of tannin resistance, L. plantarum 8p-a3 and DMC-S1 strains differ significantly: the potential of the probiotic strain exceeds that of the DMC-S1. Genes for two tannase subtypes, A and B, were found in the genome of L. plantarum 8p-a3, while only a gene for subtype B was found in DMC-S1. Generally, L. plantarum strains have only the tannase B gene, and the presence of genes for two tannase subtypes is extremely rare (Fig. 5). The presence of subtype A tannase can significantly increase the potential of bacterial stress resistance, and the presence of two subtypes of tannases can cause unexpected consequences for the host. The ability of S. gallolyticus and F. nucleaticus to actively degrade tannins due to the presence of two subtypes of tannases (A and B) is considered a possible reason for the successful survival of malignant cells in the presence of these bacteria (29). It remains to be revealed whether L. plantarum 8p-a3, as the owner of two tannase subtypes at once, can exhibit similar properties. According to in vitro analysis (Fig. S1), at temperatures relevant to poikilothermic organisms, the tannase A gene is constitutively expressed in L. plantarum 8p-a3 cells, and at temperatures relevant to the human body, the expression level of this gene increases significantly. Whether this is the case in vivo, and what the consequences of this strain feature are for the host, are questions for future research relevant to assessing the safety of probiotics.

A significant contribution to bacterial persistence in higher eukaryotes is made by proteins with the potential as immunomodulators that ensure host immunotolerance against the corresponding microbes. Some mutations in such genes can dramatically affect the outcome of an interaction (symbiosis/pathogenesis). In the case of the L. plantarum–D. melanogaster association, such genes include ackA and mprF. The products of these genes involved in specific metabolic processes have proven to be important regulators of immunotolerance (30, 31). Mutation related to the “loss of function” in the case of mprF causes sensitivity to cationic peptides (12, 30), as well as a change in the LTA structure in the bacterium, and leads to disruption of immunotolerance in the host (the development of inflammatory processes in response to the persistence of this bacterium) (12, 30). Mutation related to the “loss of function” in the case of ackA causes an imbalance of acetyl phosphate (32), associated with the development of virulence in bacteria, as well as serious pathologies in humans (33, 34). In L. plantarum, this mutation was shown to be accompanied by increased formation of acetylated amino acids, including N-acetyl glutamine, which induces host immunotolerance (31). In this regard, a comparative analysis of the sequences and expression of the corresponding genes is very relevant for assessing the functionality and safety of probiotic strains. The L. plantarum 8p-a3 and DMC-S1 strains differ in terms of variants of primary sequences for ackA as well as mprF (Fig. S8 and S9). The sequences of these genes in the strains are “working”; their differences are associated with single non-synonymous substitutions without the effect of shifting the reading frame and the formation of stop codons. The pronounced difference in the levels of the corresponding transcripts in L. plantarum 8p-a3 and DMC-S1 strains (Fig. S1) indicates the possibility of significant differences between the strains in terms of regulation of ackA and mprF expression. Whether that is the case remains to be seen. It is obvious that the genes of immunomodulators, which are critical for the outcome of the interaction of probiotic bacteria with the host, should be the focus of special attention in future studies. Subtle differences in the sequences and/or expression of these genes can be very significant for the metabolism of the bacterium, as well as its interaction with other microbes and host cells, and are accordingly relevant for assessing the safety of probiotics.

Pronounced differences between the strains were found in terms of composition and biological effects of their vesicles (Fig. S12 to S14). The results of in vivo studies indicated (geno)toxicity of the probiotic strain vesicles against intestinal tissue of the model organism used (Fig. S14), and the data correlated with in vivo data for the strains (12). Identifying the trigger of this effect is extremely important for understanding the molecular base of the phenomenon and evaluating the safety of probiotics.

According to the data of bioinformatic resources recommended for the search for virulence factors, proteins associated with virulence that determine the genotoxic potential are not found in probiotic vesicles of L. plantarum 8p-a3. Nucleases can potentially cause genotoxic effects, but the nuclease profiles of vesicles in the strains are similar, and the search tools do not identify them as the corresponding agents. We hypothesize that the DNA-binding protein of the HU family (a product of the hupB gene) may be a candidate factor associated with the negative effects of L. plantarum 8p-a3 vesicles in the D. melanogaster model. This protein is stably detected in vesicles of L. plantarum 8p-a3, but not L. plantarum DMC-S1. Histone-like protein HU, mediating the modulation of DNA topology and gene expression, is crucial for the survival of pathogenic bacteria (14, 35). The dual role of the HU protein (“internal” — protection of the bacterial genome and “external” — interference in the genome of host cells) makes it a key factor in persistence and virulence of pathogenic bacteria. The contribution of this protein to the interaction of commensals with host cells is not yet studied. The possibility of its participation in the negative effects of the L. plantarum 8p-a3 observed in the fruit fly model system remains to be investigated. Clarifying these issues is extremely important for evaluating the safety of probiotics. At the same time, taking into account the differences between the strains with respect to the infrared spectra of vesicles, it cannot be excluded that other classes of molecules may (also) be involved in the negative effects of the probiotic strain. sRNAs patterns of vesicles will require special attention (36). The features of extracellular vesicles secreted by the bacterium with GRAS status revealed through our research using the fruit fly model, open up new facets in assessing the functionality and safety of probiotics.

It is important to note that in silico data of our study generally correlated with in vitro and in vivo data, testifying to the validation of predictions important for evaluating the functionality and safety of probiotics. The exception was the results of the analysis of antibiotic sensitivity. The genomes of both strains contain genes dltABCD and mprF, encoding resistance to cationic antimicrobial peptides, which are crucial for survival in the gastrointestinal tract (Table S11). They belong to the intrinsic antimicrobial resistance of bacteria and are not considered to pose a significant risk in regard to the transmission of resistance to other microbes. The genes determining resistance to other groups of antibiotics are not detected in the genomes of the strains at high confidence, while some genes responsible for target protection from antibiotics, antibiotic efflux, antibiotic inactivation, and reduced permeability to antibiotics, with identity ranging from 20% to 71%, are observed (Table S11). However, in vitro, both strains exhibited phenotypic resistance to antibiotics of different groups (in addition to cationic antimicrobials) and differed slightly in their antibiotic sensitivity profiles (Table S12).

According to previously published reports (36–41), the phenotypic profiles of lactobacilli were determined using the standard agar disk diffusion method. Certainly, to achieve reliable results and consistency between genomic predictions and phenotypic data, it is necessary to avoid differences in experimental methods and conduct antimicrobial sensitivity testing in accordance with the recommendations of EUCAST and/or CLSI (42, 43). However, current guidelines do not establish a sufficient number of control points for testing lactobacilli for antibiotic sensitivity, probably because lactobacilli are commensals rather than clinically significant microorganisms. Future studies should adjust the relevant recommendations.

Differences in phenotypic/genotypic resistance in Lactobacillus isolates of various origins have also been noted in other studies (41, 44, 45). The discrepancy between genomic predictions and experimental antibiotic resistance in L. plantarum is biologically plausible and expected. This species has innate tolerance mechanisms such as pumps for multiple drug excretion, thickened peptidoglycan layers, and low membrane permeability, which can lead to phenotypic resistance without appropriate antibiotic resistance genes (41, 44, 46). In addition, condition-dependent gene expression, incomplete genome assembly, and database distortions can lead to underestimation or overestimation of the prediction of resistance genes. Differences in the analysis conditions (pH, seed density, and diffusion medium) can also affect the inhibition zones. Finally, the involvement of yet unknown resistance mechanisms and related genetic determinants cannot be ruled out (47–50), especially since not all genes in the genomes of even well-studied bacteria are annotated.

Despite a significant breakthrough in defining the genetic landscape of bacteria, part of their genome remains in the shadows (51, 52). A number of genes of L. plantarum 8p-a3 and L. plantarum DMC-S1, as well as many other probiotic strains, are genes for proteins with an unknown function (Table S16). This limitation seems to be a serious obstacle to the correct assessment of the safety of bacteria used in medicine, agriculture, food industry, and biotechnology (53). Obviously, for practical use and safety, it is advisable to use only strains with fully annotated genomes that do not contain undesirable sequences and/or to develop “precision postbiotics.”

Study limitations

Our study has some methodological and biological limitations. In particular, the genomic data for the 8p-a3 strain is based on rough assembly, which limits the validity of SNPs in a genomic context. The analysis of the functionality and the expression of a number of critical genes was carried out only in vitro, without transcriptomic or proteomic/metabolomic validation, and the activity of genes of L. plantarum in the host intestine remains unknown. The vesicles were obtained from bacteria grown under limited environmental conditions, and the identification of their components was performed only with respect to proteins, but not metabolites, RNA, and other compounds. Finally, in vivo effects have been tested only in the D. melanogaster model. All these points need to be taken into account in further research for a deeper understanding of the interaction of the bacterium with the host, which is relevant for assessing the functionality and safety of probiotics.

Conclusion

For the first time, a probiogenomic approach was used to identify common and specific features related to functionality and safety in phylogenetically closely related strains of the bacterium with GRAS status—the approved probiotic strain L. plantarum 8p-a3 and the intestinal resident Drosophila L. plantarum DMC-S1—which have the opposite effects on the model host organism (D. melanogaster). The genomic analysis of the studied strains was supplemented by the analysis of their extracellular vesicles. Integrative data analysis in silico, in vitro, and in vivo revealed the unique abilities of the strains associated with the potential of their beneficial and undesirable properties against the host. It was found that L. plantarum 8p-a3 has special properties, including those related to the effects of the bacterial vesicles in vivo on the fruit fly model, which determine the need for further appropriate studies in higher-order model systems to assess the safety of the probiotic strain. Emerging evidence indicates that, for assessing the safety of probiotics, not only the abundance of genes, but also variants of the primary structure of genes matter. The findings highlight the challenges associated with gaps in our knowledge of probiotic biology, as well as the standardization of probiotic strain research. It is obvious that the power of modern high-resolution methods should be directed to solving these problems in order to develop an effective system for evaluating the safety of probiotics.

MATERIALS AND METHODS

Bacterial strains and growth conditions

Lactiplantibacillus plantarum DMC-S1 has been previously isolated from the gut of D. melanogaster (12). L. plantarum 8p-a3 strain is approved as a probiotic strain (Biomed, Russia) and is effective against various dysbiosis-associated disorders (54). LAB strains were grown in MRS broth (Difco, USA), and Staphylococcus aureus subsp. aureus ATCC 29213, Enterococcus faecium ATCC 19434, Escherichia coli ATCC 25922, and Pseudomonas aeruginosa ATCC 27853 were used as test bacteria and grown in Luria-Bertani (LB) broth. For the biofilm assay, the BM broth (glucose 5.0 g, peptone 7.0 g, MgSO₄·7H₂O 2.0 g, and CaCl₂·2H₂O 0.05 g in 1.0 L tap water) was used (55). The LB agar plates supplemented with 5% sheep blood were used for the hemolytic activity tests. For E. faecium, media were supplemented with 5% fetal bovine serum (FBS).

Bacterial antibiotic susceptibility test

Antibiotic susceptibility of the LAB strains was determined by disc diffusion assay as described in reference 50. Cultivation was carried out at 37°C for 18 h on MRS agar, as recommended in references 51–53. Bacteria were classified as resistant (R), intermediate (I), or susceptible (S), according to the EUCAST recommendations.

Antagonistic activity

Antagonistic activity was examined by agar spot test described in (56). Briefly, overnight cultures of individual strains were spotted (2 μL) on the surface of MRS agar and incubated anaerobically (Anaerogas gaspack, NIKI MLT, Russia) for 24 h at 37°C to develop the spots. A 100 μL volume of an overnight culture of test bacteria was mixed with 7 mL of soft Luria-Bertani (LB) agar (0.7%), poured over the plate, and plates were incubated aerobically at 37°C. After 24 h of incubation, diameters of bacterial growth inhibition zones were measured. Each test was performed in triplicate.

Cell surface hydrophobicity

The bacterial adhesion to hydrocarbons was measured as described in reference 57. The bacterial cells grown in the MRS broth at 37°C for 18 h were harvested by centrifugation, and the cell pellet was washed twice with 0.1 M KNO₃ (pH 6.2) and resuspended in the same solution to an OD₄₀₀ = 0.4 (A0). Bacterial cell suspension (2.4 mL) was intensively mixed with n-hexadecane (0.4 mL) and incubated at 37°C for 15 min until a complete phase separation in the mixture. The aqueous phase was gently taken out, and absorbance at 400 nm was measured (A1). The surface hydrophobicity (%) was calculated as (1 − A1/A0) × 100. Strains were classified as low (L) or medium (M) according to their hydrophobicity capacities (58).

Autoaggregation

The autoaggregation ability of isolates was tested as described in reference 59. Briefly, the bacterial cells grown in MRS broth at 37°C for 18 h were harvested by centrifugation, washed twice with PBS, and resuspended in PBS to OD₆₀₀ = 0.5 (A0). Bacterial cell suspensions (4 mL) were incubated at 37°C in tubes for 4 or 24 h under static conditions, and absorbance of the upper phase was measured at 600 nm (A1). Autoaggregation percentage was calculated as (1 – A1/A0) × 100.

Isolation and characterization of the extracellular vesicles of L. plantarum strains

Extracellular vesicles were isolated from L. plantarum 8p-a3 and DMC-S1 cultures as described in reference 60 with modifications. Isolated vesicles were treated with DNase and RNase and suspended in PBS. Transmission electron microscopy (TEM) of isolated extracellular vesicles was carried out on a Hitachi HT 7800 electron microscope as described previously (61). The size distribution and concentration of extracellular vesicles in samples were determined by nanoparticle tracking analysis (NTA) using a NanoSight LM-10 analyzer (Malvern Instruments) (62). IR spectra of vesicles were recorded using an Invenio instrument (Bruker) with an MCT detector and a BioATR II ATR unit with a ZnSe crystal and a Si external coating. Spectra were recorded at a spectral resolution of 4.0 cm^–1 with 128 scans. Band assignment was carried out as in reference 63.

Protein identification

Vesicular proteins were identified by LC-MS using an UltiMate 3000 nanoflow HPLC instrument (Thermo, United States) as described previously (61). A C18 PepMap100 precolumn (Thermo, United States) was used with a Peaky75-30 capillary reversed-phase column (Molekta, Russia). For each spectrum, identification was performed using the MSFragger search engine of the software package FragPipe 19.1. Samples were examined by targeted mass spectrometry using a Sciex QTRAP 6500 + mass spectrometer (Sciex, United States) and an ExionLC HPLC system (Sciex, United States) to validate the proteome data. Peptides were separated on an Ultra Aqueous C18 HPLC column (3 μm, 150 × 3.0 mm) (Restek) with an Ultra Aqueous C18 HPLC guard cartridge (Restek). Identification with the MSFragger search engine of the FragPipe 19.1 package was performed for each spectrum.

Biofilm assays

To test the antagonistic activity of L. plantarum strains against other bacteria in growing biofilms, L. plantarum and opportunistic bacteria were grown together under static conditions for 48 h in BM broth with a minimal amount of glucose (0.2%) to minimize acidification. Then, CFUs were differentially counted on Endo agar (for E. coli), mannitol salt agar (for S. aureus), and cetrimide agar (for P. aeruginosa) using the drop-plate assay (64).

The ability of vesicles to eradicate the formed microbial biofilm was assessed as described in reference 16. The 24-hour-old biofilms of opportunistic bacteria were washed with sterile saline (0.9% NaCl), and wells were filled with BM broth supplemented with vesicles normalized according to NTA data and incubated for the next 24 h. Then, the wells were rinsed with water, and biofilms were stained with crystal violet (64, 65). Data from five independent experiments were shown as medians with IQR.

The effect of L. plantarum and vesicles on D. melanogaster

D. melanogaster Canton-S were cultured on a standard nutrient medium (agar-agar, 10.0 g; yeast, 10.0 g; semolina, 25.0 g; sugar, 25.0 g; propionic acid, 4.0 mL in 1.0 L tap water) at 25°C and a relative humidity of 65% (66). Infection of flies with L. plantarum 8p-a3 and L. plantarum DMC-S1 strains and assessment of the flies’ adaptability were performed as described earlier (12). The locomotor activity of Drosophila was measured using the Climbing assay; groups of 10–20 individual flies were tested in three repeats in each experiment (67).

An alkaline comet assay was performed to assess DNA damage in enterocytes of flies infected with vesicles (68). Briefly, a third-instar D. melanogaster larvae were infected via a nutrient substrate containing 25 µL of vesicles of L. plantarum 8p-a3 and DMC-S1 strains. Larvae grown in a pure medium (25 µL of PBS) served as controls. After 24-hour incubation with vesicles, intestines of larvae were extracted (n = 5) in Poel’s saline solution (15 mM NaCl, 6.4 mM NaH₂PO₄, 42 mM KCl, 7.9 mM CaCl₂, 1.8 mM KHCO₃, 20.8 mM MgSO₄; pH 6.95). Further, the samples were centrifuged for 5 min (5,000 rpm) at 4°C, and a supernatant was kept and applied to 80 µL of 0.65% low-melting agarose and loaded onto glass slides. After complete solidification of the agarose, lysis was performed for 1 h at 4°C in L buffer (2.5 M NaCl, 100 mM EDTA, 10 mM Tris, 1% Triton X-100; pH 10). Next, the slides were transferred to alkaline electrophoresis buffer (0.3 M NaOH, 1 mM EDTA, pH 13) and incubated for 15 min. Electrophoresis was performed in the same buffer for 25 min (25 V, 300 mA). The slides were then neutralized in 0.4 M Tris (pH 7.5). DNA was stained with DAPI for 20 min in the dark. The slides were visualized using a Carl Zeiss Axio Imager M2 fluorescence microscope (Germany). At least 100 randomly selected cells were analyzed for each sample. The degree of DNA damage was assessed by visual classification (N0–N4).

To analyze intestinal tissue damage caused by L. plantarum vesicles, intestines (n = 5–7) were taken from infected D. melanogaster larvae and fixed in 4% PFA for 30 min at 25°C. Next, the samples were washed three times in PBS for 5 min and incubated twice in a PBST solution (1 × PBS with 1% Tween-20) for 15 min. Ten microliters of dye solution (propidium iodide [PI] [69], Hoechst [70], and DAPI [4′,6-diamidino-2-phenylindole] [71]) was added to samples and incubated for 5 min in the dark at 25°C. Imaging was performed on a Carl Zeiss Axio Imager M2 fluorescent microscope (Carl Zeiss, Germany) at appropriate excitation/emission wavelengths. The resulting images were processed and analyzed using ImageJ software.

Quantitative RT-PCR

The expression of хhlA, ackA, mprF, tanA, and D-lactate dehydrogenase genes (ldh1, ldh2) in L. plantarum strains was evaluated by qRT-PCR using the rpoC as a reference gene. For that, L. plantarum 8p-a3 and DMC-S1 were grown in MRS broth at either 25°C or 37°C for 24 h, and total RNA was extracted from 1 mL of cultures using RNA Solo (Evrogen, Moscow, Russia).

To assess the effect of L. plantarum vesicles on the immune response and stress response of D. melanogaster, total RNA was isolated using a DNA/RNA isolation kit with a co-precipitant (Biolabmix, Novosibirsk, Russia) from intestinal homogenates (n = 15) of a third-instar larvae infected with vesicles as described above. The transcription of genes encoding antimicrobial peptides (Cecropin A1, Defensin, Drosocin, Drosomycin, and Metchnikowin) and stress response proteins (Catalase, dFoxo, Peroxiredoxin 5, Reaper, and ANT) was quantified by qRT-PCR using βTub56D and eEF1α1 as reference housekeeping genes.

The cDNA was synthesized using MMLV RT kit (Evrogen, Moscow, Russia) with random primers following the manufacturer’s instructions. The qPCR of хhlA, ackA, mprF, tanA, and D-lactate dehydrogenase genes (ldh1, ldh2) cDNA was performed using the 5× SYBR + low ROX mix (Evrogen, Moscow, Russia) and primers presented in Table S17. The D. melanogaster gene expression was evaluated in qPCR performed with BioMaster HS-qPCR HI-ROX SYBR (BiolabMix, Novosibirsk, Russia). The reactions were carried out in triplicate on the QuantStudio 5 (Thermo, USA). The 2^-ΔΔCt method (72) was used for calculations.

Genome sequencing

The complete nucleotide sequence of L. plantarum 8p-a3 genome has been determined using the MiSeq platform in 2 × 300 bp sequencing mode (Illumina, USA). The sequencing of L. plantarum DMC-S1 genome was performed using the NovaSeq platform in 2 × 100 bp (Illumina, USA), as well as the MinION platform (Oxford Nanopore Technologies, UK). The genome coverage was 175× for L. plantarum 8p-a3 and 180× (75× on the Illumina NovaSeq and 105× on the Oxford Nanopore MinION) for the L. plantarum DMC-S1 strain.

Genome mining

Genome annotation was performed using the NCBI Prokaryotic Genomes Annotation Pipeline (PGAP) version 6.9 (73). The genome maps of the L. plantarum DMC-S1 and 8p-a3 strains were visualized using Proksee (74). Gene annotation against COG, KEGG, Pfam, and TCDB databases was performed using eggNOG mapper 2 (e-value < 0.001, identity > 40, query > 20, emapper version 2.1.12, eggNOG DB version 5.0.2) (75), and gene annotation against SEED categories was performed using RAST (76). Average nucleotide identity (ANI) analysis was performed using FastANI (77).

Comparison of the genome sequences of L. plantarum 8p-a3 and L. plantarum 8P-A3 was performed using GSAlign v.1.0.22 (78) to align the whole genomes and Bowtie 2 (79) to align the raw reads of the 8p-a3 strain to the complete genome of L. plantarum 8P-A3. Variant calling was performed using samtools, bcftools (80), and SNPEff (81). The variant calling threshold was set at Q > 30 and coverage > 15. Only variants confirmed by both whole genome and raw reads alignment methods are provided.

To search for insertion elements in the L. plantarum genomes, ISFinder and BLASTn v2.2.31 were used with an E-value threshold of < 5e−25 (82). The sequences encoding prophage genes were detected using the PHASTEST (83). CRISPR-Cas elements were identified using CRISPRCasFinder (84).

Clusters of genes associated with bacteriocin synthesis were identified using BAGEL4 (85) and verified with BLASTp v2.2.26 (E-value < 1e-50) (86). Antibiotic resistance genes were identified using CARD and Resistance Gene Identifier with perfect, strict, and loose hits models.

The presence of genes encoding bile acid hydrolases, esterases hydrolyzing plant phenolic compounds, and genes encoding vitamins, biogenic amines, D-lactate, and nitro compounds was determined by aligning the amino acid sequences of previously characterized genes using BLASTp v2.2.26 (E-value < 1e-50) (86).

Virulence genes were identified using the VFDB virulence factor database (87) and BLASTx (E-value < 1e-5, identity > 80%).

Statistical analysis

Experiments were carried out in biological triplicates with three technical repeats in each, unless otherwise mentioned. Data are shown as averages with standard deviation. The statistical significance of the results was assessed using one-way analysis of variance (one-way ANOVA) with Tukey’s multiple-comparison test, with a significance threshold at P < 0.05, using Statistica 12.0 software. The fly survival data were analyzed using the log-rank (Mantel-Cox) test.

DATA AVAILABILITY

Genome data are available in the GenBank database (BioProject IDs PRJNA528387 and PRJNA956517, and BioSample IDs SAMN11180819 and SAMN34217994 for L. plantarum 8p-a3 and DMC-S1, correspondingly). Raw reads were deposited to NCBI Sequence Reads Archive under SRA IDs SRR36237018 (L. plantarum 8p-a3, short reads), SRR36236716 (L. plantarum DMC-S1, long reads), and SRR36236717 (L. plantarum DMC-S1, short reads). Genome assembly accessions are ASM440304v2 for 8p-a3 strain and ASM2985510v1 for DMC-S1 strain.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/spectrum.02314-25.

File S1. spectrum.02314-25-s0001.xlsx.

Islands.

Supplemental figures and tables. spectrum.02314-25-s0002.pdf.

Fig. S1 to S15 and Tables S1 to S17.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.