Comparative genomic analysis of hydrogen peroxide and nitric oxide metabolic pathways in Limosilactobacillus fermentum

Abstract

Limosilactobacillus fermentum strains influence human health through distinct metabolic pathways; however, the genetic basis of these strain-specific functions remains unclear. This study investigated phenotypic divergence between two L. fermentum strains isolated from the human oral microbiome. Despite 98.2% average nucleotide identity, comparative genomics revealed substantial strain-specific gene repertoires (485 and 542 unique genes in DM072 and DM075, respectively). Functional characterization demonstrated that DM072 synthesizes hydrogen peroxide via pyruvate oxidase (EC 1.2.3.3; K00158), conferring strong antimicrobial efficacy against the Streptococcus mutans. Conversely, DM075 lacks this oxidative pathway but exhibits six-fold elevated nitrate reductase activity during the stationary phase. Transcriptomic profiling revealed significant temporal upregulation of glutamate synthase (gltB, p < 0.05) and alkyl hydroperoxide reductase (ahpC, p < 0.05) in DM075, indicating coordinated nitrogen assimilation and oxidative stress responses. Bioinformatic analyses identified strain-specific enzymatic profiles, including differential distributions of glycosyl hydrolases and transferases, alongside disparate acid tolerance (pH 2.5 for DM075 vs. pH 3.0 for DM072). These findings demonstrate functional specialization at the strain level within closely related taxa, highlighting the potential of DM072 as an antimicrobial probiotic for dental caries prophylaxis and DM075 as a potential cardiovascular homeostasis modulator via the nitrate–nitrite–nitric oxide pathway.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-29368-6.

Introduction

The human oral cavity harbors a complex and dynamic microbiome integral to both oral and systemic health^1,2. Increasing evidence suggests a close association between the balance of oral microbial communities and periodontitis, dental caries, cardiovascular diseases, diabetes, and immune dysregulation, highlighting the need for detailed functional analysis of oral microorganisms³.

Probiotic bacteria of human origin have garnered considerable interest for their potential to improve oral and systemic health by inhibiting pathogenic bacteria, modulating host immune responses, and producing bioactive substances⁴. Among these, lactic acid bacteria isolated from the oral environment are particularly promising because of their high colonization ability, established safety profile, and functional efficacy⁵.

Limosilactobacillus fermentum, a species isolated from fermented foods and various human body sites, is recognized for producing antimicrobial substances, such as hydrogen peroxide and bacteriocins⁶. L. fermentum DM072 also exhibits potent inhibitory effects against oral pathogens, including Streptococcus mutans, attributed to its high hydrogen peroxide production capability⁷. Hydrogen peroxide synthesis in bacteria is mediated by oxidases, such as reduced nicotinamide adenine dinucleotide (NADH) oxidase (Nox), pyruvate oxidase (Pox), lactate oxidase (Lox), and NADH-dependent flavin reductase⁸. In DM072, hydrogen peroxide production primarily involves Pox, encoded by pox, which catalyzes the conversion of pyruvate to acetyl-phosphate, with concomitant hydrogen peroxide release. This antimicrobial activity contributes to maintaining oral microbial homeostasis, potentially reducing the incidence of dental caries and periodontitis.

Conversely, L. fermentum DM075, isolated from the tongue coating of healthy donors in South Korea, demonstrates unique physiological properties despite lacking detectable hydrogen peroxide production⁶. Phenotypic screening using Griess reagent (Supplementary Figure S3) revealed that DM075 demonstrates nitrate reduction capability, indicating its potential role in the oral nitrate–nitrite–nitric oxide pathway. Nitrate reductase (Nar) in oral bacteria is encoded by narX, narG, narJ, narH, narY, narI, and narW, whereas nitrite reductase (Nir) is encoded by nirK and nirS^9,10. The nitric oxide production capacity of DM075, completely absent in DM072, is associated with nitrogen metabolism-related genes such as gltB, encoding glutamate synthase, which catalyzes the synthesis of glutamate—a precursor for arginine that can be subsequently converted to nitric oxide⁶.

Nitric oxide functions as a gaseous signaling molecule with diverse biological activities, including vasodilation, inhibition of platelet aggregation, immune modulation, and anti-inflammatory effects^11,12. Emerging evidence suggests that microbial nitric oxide production in the oral cavity contributes to systemic blood pressure regulation and cardiovascular health¹³. The ‘enterosalivary circulation’ pathway involves dietary nitrate being concentrated in saliva, reduced to nitrite by oral microorganisms, and converted to nitric oxide non-enzymatically in acidic environments, such as the stomach. Despite the known importance of this pathway in cardiovascular health regulation, the specific microbial species responsible for efficient nitrate reduction in the oral cavity remain poorly characterized, creating a critical knowledge gap in understanding the effects of oral microbiome composition on systemic nitric oxide bioavailability¹⁴.

In the present study, we aimed to investigate the genetic basis for differential hydrogen peroxide and nitric oxide production in DM072 and DM075 strains through comparative genomic analysis. We performed functional annotation and pathway mapping using the Kyoto Encyclopedia of Genes and Genomes (KEGG) and Clusters of Orthologous Groups (COG) databases, analyzed genetic variants and their functional implications, and explored the phylogenetic relationships. Our findings provide insights into strain-specific functional differentiation within L. fermentum and suggest distinct probiotic applications: DM072 for caries prevention through hydrogen peroxide production and DM075 for cardiovascular health promotion through the nitrate–nitrite–nitric oxide pathway.

Results

Comparative genomic analysis of L. fermentum DM072 and DM075

Whole-genome sequencing using PacBio and Illumina sequencing platforms¹⁵ revealed that DM072 comprised a 2,060,072 bp chromosome (GenBank accession no. CP102714.1) and a plasmid of 44,665 bp (pLFDM072, CP102715.1). DM075 contained a single 2,204,022 bp chromosome (CP100352.1) (Table 1)^16–25.

Table 1.

Comparative analysis of L. fermentum DM072 and DM075 metabolic features.

Feature	L. fermentum DM072	L. fermentum DM075	Functional Significance	Reference
Primary ecological function	Hydrogen peroxide production with antimicrobial activity against S. mutans	Nitrate reduction capability (nitrate → nitrite) contributing to the oral nitrogen oxide cycle	Different ecological roles: hydrogen peroxide production vs. nitrate reduction capability in oral microbiome	16
Pyruvate metabolism	Contains pyruvate oxidase (K00158, WP_252772585.1)	Lacks pyruvate oxidase; contains additional genes (K13954, K03777, K01958)	K00158 enables hydrogen peroxide production in DM072	17
Nitrogen metabolism	8 genes, including nitrate reductase components (K00370-K00374)	10 genes, including additional glutamate synthase genes (K00265, K00266)	Enhanced nitrogen assimilation capacity in DM075	18
ABC transporters	26 genes	28 genes including K05845, K05846, K09817	Greater nutrient acquisition capability in DM075	19
Glutathione metabolism	7 genes; glutathione synthase (K01919) with 2 variants	7 genes; glutathione synthase (K01919) with 4 variants	Potentially enhanced oxidative stress response in DM075	20
TCA cycle	6 genes	7 genes; additional malate dehydrogenase (K01958)	Enhanced energy metabolism in DM075	21
Glyoxylate and dicarboxylate metabolism	7 genes; includes alcohol dehydrogenase (K00018)	6 genes; lacks K00018	Differences in carbon metabolism flexibility	22
pH tolerance	Survival at pH 3.0	Survival at pH 2.5	Enhanced acid tolerance in DM075	23
Colonization ability	Strong adherence to oral epithelial cells	Not determined	Enhanced ability to establish in the oral cavity	24
Genome features	Chromosome: 2,060,072 bp; GC 51.5% (CP102714.1); Plasmid pLFDM072: 44,665 bp; GC 40% (CP102715.1); No. of coding sequences: 2,035 (Chromosome), 53 (Plasmid); 2,088(Total)	Chromosome: 2,204,022 bp; GC 51.0% (CP100352); No. of coding sequences: 2,129(Chromosome), 0 (Plasmid); 2,129(Total)	Similar genome size and GC content despite functional differentiation	7,25

Circular genome maps of both strains (Supplementary Figure S1) illustrate the structural features of their complete chromosomes. Both strains showed similar GC contents (51.5% and 51.0%, respectively), with 2,088 protein-coding genes in DM072 (2,035 chromosomal, 53 in plasmid) and 2,129 in DM075.

The average nucleotide identity based on BLAST (ANIb) percentage identity heatmap revealed that DM072 and DM075 exhibited high sequence similarity (approximately 0.98) but showed distinct patterns of similarity with other L. fermentum strains (Fig. 1A). Figure 1B displays a phylogenetic tree illustrating their evolutionary relationships based on whole-genome sequences of 41 L. fermentum strains, with DM072 and DM075 positioned on separate branches of the tree. The ANIb dendrogram (Fig. 1C) further highlights these phylogenetic differences, placing DM072 and DM075 in separate clades, suggesting divergent evolutionary pathways despite species-level identity. DM072 clustered with strains like 2,760 and FAM19471 (right clade), whereas DM075 clustered with DR9 (left clade), indicating distinct genomic lineages.

Fig. 1.

Comparative genomic analysis of strains DM072 and DM075. (A) ANIb percent identity heatmap showing nucleotide similarity among 18 L. fermentum strains. (B) Phylogenetic tree of 41 L. fermentum strains. Strains were selected to represent phylogenetic diversity across major clades (including DM072-associated lineage: CUL67, DS19_7, SRCM_103290; and DM075-associated lineage: SD381, SL241, DR9), genome quality (complete or high-quality draft genomes), and geographic/ecological diversity (human, food, and environmental sources). Bootstrap support from 1,000 replicates is shown at the nodes. DM072 is highlighted in red, while DM075 is highlighted in blue. Scale bar: 0.01 substitutions/site. (C) ANIb-based hierarchical clustering dendrogram of 18 strains. (D) Venn diagram of gene distribution showing 1,613 shared genes between DM072 and DM075, with 485 genes unique to DM072 and 542 unique to DM075.

Gene distribution analysis (Fig. 1D) revealed that DM072 and DM075 shared 1,613 genes with 485 and 542 unique genes, respectively. This genetic divergence further supports their distinct functional characteristics despite belonging to the same species. Pan-genome analysis of 45 L. fermentum strains, including DM072 and DM075 (Supplementary Figure S2), highlighted the substantial genetic diversity within this species—only 3% of genes constituted the core genome (present in ≥ 99% of strains). Notably, pyruvate oxidase (pox) was identified as an accessory gene present in 60% (27/45) of strains, whereas nitrate reductase genes (narGHI) were core genes found in all strains, though with variable functional activity. In comparison, 84% were cloud genes (present in < 15% of strains), highlighting the high genetic plasticity of Limosilactobacillus that enables strain-specific functional adaptations to different environments.

DM072 contained a pyruvate oxidase gene (pox, K00158; locus tag: NHG87_RS04095), which catalyzes direct H₂O₂ production from pyruvate (Pyruvate + O₂ + Pi → Acetyl-phosphate + CO₂ + H₂O₂) and was completely absent in DM075 (Supplementary Table S1, Table 2). The causal relationship between pox and H₂O₂ production was validated through catalase inhibition experiments, where H₂O₂-mediated antimicrobial activity of DM072 against S. mutans was completely abolished by catalase treatment, with the zone of inhibition diameter decreasing in a dose-dependent manner⁶. Furthermore, genotype-phenotype correlation analysis of our 12 L. fermentum isolates revealed 100% concordance: all nine H₂O₂-producing strains possessed the pox gene, whereas all three non-producers (including DM075) lacked this gene⁶. In contrast, DM075 uniquely harbored genes associated with alternative pyruvate metabolic pathways, including pyruvate carboxylase (K01958) and D-lactate dehydrogenase (quinone) (K03777). DM075 also had more variants of L-lactate dehydrogenase (K00016) than DM072.

Table 2.

Key genomic features explaining differential H₂O₂ production and oxidative stress responses in DM072 and DM075.

Feature	DM072	DM075	Functional Implication
SOD genes	Absent	Absent	Alternative oxidative stress strategies are required
Pyruvate oxidase (K00158)	Present	Absent	Direct H₂O₂ production vs. none
Thioredoxin system	Enhanced	Standard	Compensatory antioxidant in DM072
Dyp-type peroxidase	Pseudogene	Functional	H₂O₂ accumulation vs. scavenging
NarGHI complex	Inactive	Active	No nitrate reduction vs. active

Although both strains harbored nitrate reductase (narGHI) genes, functional assays revealed that this enzyme was only active in DM075 (Supplementary Figure S3). The Griess assay confirmed nitrate reduction, with DM075 showing distinctive red coloration in MRS solid medium containing the Griess reagent.

Comparison of metabolic pathways in DM072 and DM075 through KEGG and COG analysis

To systematically understand the functional differences between DM072 and DM075, we performed a comparative analysis of metabolic pathways and functional categories using the KEGG and COG databases^26,27. As shown in Fig. 2A, both strains showed similar patterns in overall KEGG functional category distribution when analyzed as relative gene proportions, with notable differences in specific pathways.

Fig. 2.

Functional differences identified through metabolic pathway comparison of L. fermentum DM072 and DM075. (A) Relative gene proportions (%) by KEGG functional categories generated using GHOST KOALA (KEGG: Kyoto Encyclopedia of Genes and Genomes, https://www.kegg.jp/). Values are shown for DM072 (red) and DM075 (blue) across each category, standardized to account for different total gene counts (2,088 vs. 2,129 genes, respectively). (B) Percentage of genes in eight major KEGG metabolic categories reveals strain-specific metabolic capabilities. KEGG pathway maps, including Glutathione metabolism (map00480), Oxidative phosphorylation (map00190), and Pyruvate metabolism (map00620), were used with permission from Kanehisa Laboratories.

To contextualize these observations within species-level diversity, we performed comparative functional annotation across 46 L. fermentum genomes, comprising both complete and high-quality draft genomes. We found that both DM072 and DM075 fall within normal species variation for protein count (1,817.7 ± 76.4), COG categories (20.1 ± 0.3), and KEGG pathways (318.9 ± 7.9), confirming that the observed phenotypic differences reflect strain-specific adaptations within typical L. fermentum genomic diversity.

Both strains showed the highest proportion of genes associated with global and overview maps, carbohydrate metabolism, and amino acid metabolism, consistent with the general characteristics of Limosilactobacillus. Quantitative analysis revealed that DM072 contained higher proportions of genes involved in carbohydrate transport and metabolism (4.11% vs. 3.79%) and cell wall/membrane/envelope biogenesis (4.91% vs. 4.82%) than those in DM075. Conversely, DM075 showed higher proportions of genes related to replication, recombination, and repair (16.03% vs. 11.47%), transcription (7.33% vs. 6.93%), and defense mechanisms (1.43% vs. 1.07%) (Supplementary Figure S4).

Analysis of specific KEGG metabolic categories (Fig. 2B) revealed differential gene proportions across eight key pathways: pyruvate metabolism, oxidative phosphorylation, glutathione metabolism, peroxisome enzymes, ABC transporters, glyoxylate metabolism, tricarboxylic acid (TCA) cycle, and nitrogen metabolism. DM075 harbored more genes related to ABC transporters (28 vs. 26) and nitrogen metabolism (10 vs. 8), consistent with its enhanced nutrient acquisition capability and nitrogen metabolism capacity²⁸. DM075 also contained a higher number of oxidative phosphorylation genes (14 vs. 13), indicating a more diverse electron transport chain (Table 3)^29–34.

Table 3.

Comparative analysis of genes involved in the nitrate reduction pathway between DM072 and DM075 strains.

Gene/System	Function	DM072 (Gene annotation)	DM072 Count	DM075 (Gene annotation)	DM075 Count	Functional Significance	Reference
Nitrate reductase (narGHI)	Catalyzes nitrate → nitrite	NHG87_RS04930 (WP_104878748.1), NHG87_RS04935 (WP_003683514.1), NHG87_RS04945 (WP_057195147.1)	3	NHN16_RS05085 (WP_070955903.1), NHN16_RS05090 (WP_003683514.1), NHN16_RS05100 (WP_041812814.1)	3	Although the gene annotation for nitrate reductase (NarGHI) shows three copies in both DM072 and DM075, the Griess assay in Supplementary Fig. 3 demonstrates that nitrate reduction activity is functionally evident only in DM075.	29
Molybdenum cofactor synthesis	Required cofactor for nitrate reductase	NHG87_RS04835 (WP_003683477.1),NHG87_RS04905 (WP_231932212.1),NHG87_RS04830 (WP_003683474.1),NHG87_RS04925 (WP_252772622.1),NHG87_RS04915 (WP_252772621.1)	5	NHN16_RS04990 (WP_163586853.1),NHN16_RS05060 (WP_012391331.1),NHN16_RS04985 (WP_003683474.1),NHN16_RS05080 (WP_003683510.1),NHN16_RS05070 (WP_070955630.1),	5	Both DM072 and DM075 possess the full set of genes required for molybdenum cofactor synthesis, indicating equivalent genetic capacity to activate nitrate reductase.	30
Iron-sulfur cluster assembly	Electron transfer components	NHG87_RS04770 (WP_252772616.1), NHG87_RS05755 (WP_164879413.1), NHG87_RS02670 (WP_107759981.1), NHG87_RS04750 (WP_003683444.1), NHG87_RS04755 (WP_104878228.1), NHG87_RS04765 (WP_104878226.1), NHG87_RS05755 (WP_164879413.1), NHG87_RS06835 (WP_003700546.1), NHG87_RS06855 (WP_108458790.1)	9	NHN16_RS04925 (WP_014562468.1), NHN16_RS06055 (WP_015638938.1), NHN16_RS03170 (WP_070955730.1), NHN16_RS07400 (WP_163586791.1), NHN16_RS04905 (WP_049183842.1), NHN16_RS04910 (WP_046025943.1), NHN16_RS04920 (WP_070955641.1), NHN16_RS06055 (WP_015638938.1), NHN16_RS07400 (WP_163586791.1), NHN16_RS07405 (WP_163586790.1), NHN16_RS07445 (WP_163586786.1), NHN16_RS08625 (WP_003684108.1)	12	DM075 harbors a broader repertoire of iron-sulfur cluster assembly genes than DM072, potentially facilitating more efficient electron transfer to nitrate reductase and supporting its enhanced nitrate metabolism.	31
Electron transport chain	provides electrons for reduction	NHG87_RS01465 (WP_104878565.1), NHG87_RS05340 (WP_252772643.1)	2	NHN16_RS09405 (WP_004562908.1),NHN16_RS10835 (WP_003685576.1),NHN16_RS01645 (WP_004563331.1)	3	DM075 possesses more diverse electron transport chain-related genes (e.g., NAD(P)H-dependent oxidoreductase) than DM072, potentially providing a wider range of electron donors for nitrate reduction pathways. This diversity may enhance nitrate reduction efficiency and metabolic flexibility.	32
Transcriptional regulation	Controls the expression of nitrate reduction genes	NHG87_RS00635(WP_252772366.1), NHG87_RS01260(WP_086439969.1), NHG87_RS09815(WP_053069022.1), NHG87_RS00980(WP_086439167.1), NHG87_RS03135(WP_049183246.1), NHG87_RS04880(WP_012391333.1), NHG87_RS08290(WP_057194870.1), NHG87_RS10030(WP_035437552.1), NHG87_RS10485(WP_003685635.1)	8	NHN16_RS10160(WP_035429107.1), NHN16_RS00085(WP_003685635.1), NHN16_RS00620(WP_254183962.1), NHN16_RS02400(WP_023465572.1), NHN16_RS03565(WP_003683732.1), NHN16_RS04135(WP_035428905.1), NHN16_RS05035(WP_021349913.1), NHN16_RS06135(WP_163586828.1), NHN16_RS06140(WP_163586827.1), NHN16_RS09855(WP_021349574.1), NHN16_RS09860(WP_111523516.1), NHN16_RS00090(WP_015638411.1), NHN16_RS00625(WP_015638486.1), NHN16_RS02400(WP_023465572.1), NHN16_RS03570(WP_075667566.1), NHN16_RS05040(WP_076811412.1), NHN16_RS06140(WP_163586827.1)	17	DM075 has higher regulatory complexity and redundancy, potentially enabling more robust nitrate reduction gene expression.	33
Nitrate/nitrite transporters	Import of nitrate, export of nitrite	NHG87_RS04840	1	NHN16_RS04995	1	Both have canonical narK	34

Conversely, DM072 had slightly more genes related to glyoxylate metabolism (7 vs. 6) and TCA cycle (8 vs. 7), suggesting different energy generation strategies of the strains³⁵. In glutathione metabolism, DM075 had more variants of glutathione synthase (K01919) (4 vs. 2) and additional thiol-disulfide oxidoreductase (K01535), indicating a more efficient defense system against oxidative stress³⁶.

Analysis of nitrate reduction-related genes by COG functional categories (Supplementary Figure S5) showed that DM075 possessed more transcription-related genes (approximately 140 vs. 125), suggesting more robust regulatory mechanisms for nitrate reduction³².

Analysis of hydrogen peroxide production and management-related genes in DM072 and DM075

To further investigate the molecular basis underlying the distinct physiological characteristics of the two strains, we analyzed genes associated with hydrogen peroxide production and management according to COG functional categories. Figure 3 visually presents the distribution and functional differences of these genes across six key categories³⁷.

Fig. 3.

Clusters of Orthologous Groups (COG) functional categories related to H₂O₂ production and management in L. fermentum strains DM072 and DM075. Distribution of key metabolic functions categorized according to COG analysis using eggNOG v5.0. Color-coding indicates strain specificity: blue (DM075-enriched), red (DM072-enriched), and purple (shared by both strains).

Categories C (energy production and conversion) and G (carbohydrate metabolism and transport) were enriched explicitly in DM075, whereas category Q (secondary metabolite biosynthesis, transport, and catabolism) was specific to DM072³⁸. Categories H (coenzyme metabolism), O (post-translational modification), and P (inorganic ion transport) were common to both strains.

In DM075, genes in the energy production category (C) were closely associated with electron transport chain components essential for nitrate reduction¹⁷. DM075 harbored a more diverse array of energy production genes, with notably different compositions of F-type H⁺-H-transporting ATPase subunits (K02108–K02115), suggesting a more efficient energy generation system during nitrate reduction³⁹.

CAZyme analysis (Supplementary Figure S6) showed that DM075 was enriched with GT2, GH70, and GH13 enzymes associated with cell wall polysaccharide synthesis, strain-specific strategies for cell wall architecture, and biosynthesis⁴⁰.

In category Q, genes directly linked to hydrogen peroxide production were identified in DM072. To confirm our observation regarding pyruvate oxidase (K00158) and carbonyl reductase (K22373), we employed multiple complementary approaches. KEGG annotation mapping of our eggNOG v5.0 results identified pyruvate oxidase (K00158, protein accession WP_252772585.1) in DM072 while confirming its absence in DM075. Similarly, carbonyl reductase (K22373, WP_099032568.1) was exclusively present in DM072. Reciprocal BLAST searches of these protein sequences against both genomes confirmed their presence in DM072 and absence in DM075. Additionally, InterProScan domain analysis validated the presence of functional domains required for hydrogen peroxide production in DM072 pyruvate oxidase. These in silico predictions were consistent with our experimental data demonstrating strain-specific hydrogen peroxide production capabilities. The differential pyruvate metabolism gene analysis highlights the alternative pyruvate metabolism pathways in DM075, including pyruvate carboxylase, pyruvate formate-lyase activating enzyme, and malic enzyme (Supplementary Table S1).

In category H, DM075 showed enhanced molybdenum cofactor synthesis, suggesting improved capability for nitrate reductase activity⁴¹. DM072 encoded specialized redox-regulated chaperones for resistance to self-produced hydrogen peroxide, whereas DM075 had more variants of glutathione synthase (K01919) and additional thiol-disulfide oxidoreductase (K01535), representing a more robust general oxidative stress defense system.

In category P, DM075 showed enhanced nitrate/nitrite transporters, suggesting more efficient substrate availability for nitrate reduction reactions.

Nitrite production capacity and gene expression profiles of DM072 and DM075

To elucidate the differences in nitrate reduction capability between DM072 and DM075, we quantified nitrite production levels and expression profiles of related genes. Both strains were cultured in an MRS liquid medium supplemented with sodium nitrate to assess their ability to reduce nitrate to nitrite. DM075 exhibited a significant increase in nitrite concentration over time, with approximately 6-fold higher levels after 24 h of cultivation than at the initial level (p < 0.001; Fig. 4A). In contrast, DM072 consistently maintained low nitrite concentrations throughout the cultivation period, indicating minimal nitrate reduction capability or absence⁴².

Fig. 4.

Nitrite production capability and related gene expression in L. fermentum strains. (A) Nitrite concentration measured using the Griess assay in DM072 and DM075 over 24 h of cultivation (n = 3). Data were analyzed using two-way repeated measures ANOVA with a Bonferroni post-hoc test. (B) Expression levels of ahpC, narG, and narH in DM072 at 0, 6, and 24 h during nitrate exposure (n = 3). (C) Expression levels of gltB, ahpC, narG, and narH in DM075 at 0, 6, and 24 h during nitrate exposure (n = 3). Gene expression data were analyzed using one-way ANOVA with Tukey’s post-hoc test. Significance levels: *p < 0.05, ***p < 0.001.

Gene expression analysis revealed clear differences between the two strains (Fig. 4B, C). In DM072, the expression level of alkyl hydroperoxide reductase (ahpC), nitrate reductase large subunit (narG), and nitrate reductase small subunit (narH) genes showed no significant changes at any time point (Fig. 4B), suggesting a lack of activation of nitrate reduction-related genes. In contrast, DM075 exhibited notable expression patterns (Fig. 4C). The glutamate synthase large subunit (gltB) expression increased significantly over time, showing approximately 1.7-fold higher expression after 24 h compared to the initial level (p < 0.05). Although gltB is not directly responsible for reducing nitrite to ammonium ion, a process catalyzed by nitrite reductase (nir). These enzymes play critical roles in assimilating ammonium ions into glutamate via the glutamate–glutamine cycle. The exclusive presence and significant upregulation of gltB in DM075 suggests a more active and complete nitrogen assimilation pathway. This enhanced gltB expression indicates a broader activation of nitrogen metabolism, potentially supporting improved growth and stress adaptation under nitrogen-utilizing conditions⁴³. Table 3 summarizes the key nitrate reduction genes and associated systems in both strains, highlighting the molecular differences contributing to their distinct nitrate reduction capabilities. Additionally, ahpC in DM075 showed significantly increased expression over time, exhibiting more than 2-fold increased expression levels after 24 h (p < 0.05). As ahpC encodes an antioxidant enzyme that removes peroxides, its upregulation suggests that DM075 has developed adaptive strategies to respond to oxidative stress accompanying nitrate reduction.

Nevertheless, the expression of the narG and narH genes did not differ significantly over time among the strains, although narH showed a slightly increasing trend at 24 h in DM075. This finding suggests that differences in nitrate reduction capability between the strains may be attributed not primarily to changes in nitrate reductase expression, but rather to the activation of supporting systems, particularly the increased expression of gltB and ahpC in DM075⁴⁴.

Comparative analysis of nitrate reduction capability differences between DM072 and DM075

To comprehensively understand the molecular mechanisms underlying the significant differences in nitrate reduction capability between DM072 and DM075, we conducted a comparative analysis of the nitrate reduction components in both strains. Figure 5 schematically illustrates these differences, comparing the status of four key components NarGHI, Mo-cofactor, Fe–S cluster, E-transport) necessary for nitrate reduction⁴⁵.

Fig. 5.

Nitrate reduction pathway differences between L. fermentum strains DM072 and DM075. The DM075 strain (left) exhibits enhanced Fe–S clusters and complete MoCo, facilitating nitrate reduction (activation function). In contrast, the DM072 strain (right) displays defective cytochrome b and limited MoCo, resulting in impaired nitrate reduction (inactivation function). The diagram depicts electron transport during respiration and nitrate/nitrite movement via the NarK antiporter across the cytoplasmic membrane. Protein components: α (NarG): Nitrate reductase catalytic subunit containing molybdenum cofactor; β (NarH): Iron-sulfur cluster subunit for electron transfer; γ (NarI): Cytochrome b subunit for membrane anchoring; Cyt-b: Cytochrome b complex; Fe–S: Iron-sulfur clusters; MoCo: Molybdenum cofactor. Numbers 1–8 indicate the relative abundance or functional activity levels of the respective components, with higher numbers representing a greater abundance or enhanced functionality in the respective strain.

The nitrate reductase (narGHI) complex exists at the gene level in both strains but is functionally active only in DM075⁶. As shown in Table 3, both strains possess three copies of narGHI, which encodes the core enzyme catalyzing the reduction of nitrate to nitrite; however, this enzyme was inactive in DM072 but active in DM075, directly explaining the significant difference in nitrite production observed in Fig. 4A⁴⁶.

The molybdenum cofactor (Mo-cofactor) synthesis system, essential for nitrate reductase catalytic activity, is present in both strains⁴⁷. Both DM072 and DM075 harbor the full set of five genes required for molybdenum cofactor synthesis (Table 3), indicating an equivalent genetic capacity for activating nitrate reductase.

The iron-sulfur cluster assembly system, essential for electron transport processes, is more extensive in DM075 than in DM072 (dashed lines in Fig. 5). DM075 harbors a broader repertoire of iron-sulfur cluster assembly genes (12) than DM072 (9) (Table 3), potentially facilitating more efficient electron transfer to nitrate reductase and supporting its enhanced nitrate metabolism⁴⁸.

The electron transport system, essential for facilitating electrons necessary for nitrate reduction, shows greater diversity in DM075 than in DM072. DM075 possesses more diverse electron transport chain-related genes (3) compared to DM072 (2), including additional genes annotated as NAD(P)H-dependent oxidoreductase (Table 3), potentially providing a wider range of electron donors for nitrate reduction pathways. This diversity may enhance nitrate reduction efficiency and metabolic flexibility⁴⁹.

Additionally, the transcriptional regulation of nitrate reduction pathways appears more complex in DM075, which has 17 regulatory genes compared to eight in DM072. This higher regulatory complexity and redundancy may enable more robust nitrate reduction gene expression in DM075⁴⁵.

Our comparative analysis confirms that DM075 has a more comprehensive system necessary for nitrate reduction, whereas DM072 has limitations in key components. Although DM072 possesses the required molybdenum cofactor synthesis genes, its nitrate reduction capability appears to be limited by a less extensive iron-sulfur cluster assembly system and less diverse electron transport components⁵⁰.

The increased expression of gltB and ahpC genes observed in DM075 (Fig. 4C) supports the activation of nitrogen assimilation and oxidative stress response systems accompanying nitrate reduction. Although gltB is not directly involved in the reduction of nitrite to ammonium, a reaction catalyzed by nir, its exclusive presence and significant upregulation in DM075 suggest an enhanced capacity for incorporating ammonium ions into glutamate via the glutamine–glutamate cycle. This finding reflects a more complete nitrogen assimilation pathway and broader activation of nitrogen metabolism in DM075⁵¹.

Discussion

In this study, we elucidated the molecular basis of functional differentiation within the same species through comparative genomic analysis of two L. fermentum strains isolated from the human oral microbiome⁴⁴. Strains DM072 and DM075 exhibited contrasting physiological functions— hydrogen peroxide production and nitrate reduction, respectively—which were confirmed to originate from distinct genomic compositions.

The hydrogen peroxide production capability of DM072 is primarily attributed to the presence of pox (K00158)⁶. Pox catalyzes the conversion of pyruvate to acetyl-phosphate, generating hydrogen peroxide in the presence of oxygen, thereby mediating effective antimicrobial activity against catalase- and peroxidase-deficient bacteria⁵². This characteristic suggests the potential of DM072 as a probiotic agent for maintaining oral microbial homeostasis and contributing to caries prevention.

Conversely, the nitrate reduction capability of DM075 represents another significant physiological function. Here, we showed that both strains harbored similar nitrate reductase (narGHI) gene clusters; however, nitrate reduction was observed exclusively in DM075. Although narG and narH expression levels did not differ significantly over time in either strain, DM075 exhibited a slightly increasing trend in narG and particularly narH expression at 24 h. This finding is important considering that NarG (nitrate reductase subunit alpha) is responsible for catalytic activity and requires Mo-cofactor, whereas NarH (nitrate reductase subunit beta) contains iron-sulfur clusters essential for electron transfer⁵³.

Furthermore, our findings showed that DM075 possesses more comprehensive supporting systems necessary for nitrate reduction, particularly in iron-sulfur cluster assembly (12 genes vs. 9) and electron transport systems (3 genes vs. 2) than DM072. Both strains possess equivalent genetic capacity for Mo-cofactor synthesis with five genes each. We speculated that the superior nitrate reduction capability of DM075 is attributed not merely to the presence of core enzymes (NarGHI) but to the more comprehensive auxiliary systems (particularly iron-sulfur cluster assembly and electron transport diversity) required for their proper functionality²⁹.

The enhanced nitrate reduction capability of DM075 correlated with increased expression of gltB (1.7-fold, p < 0.05) and ahpC (2.0-fold, p < 0.05) during nitrate exposure (Fig. 4C), suggesting functional involvement of these genes in nitrogen metabolism and oxidative stress response. Although these correlations provide strong circumstantial evidence for genotype-phenotype linkages, we acknowledge that definitive causal relationships would require genetic manipulation studies such as knockout or complementation experiments.

We also showed that gltB is upregulated in DM075. Although nitrite reductase, which catalyzes the conversion of nitrite to ammonium ion, was not identified in either strain—precluding the presence of a complete nitrogen assimilation pathway—the increased gltB expression in DM075 suggests the enhanced capacity of DM075 for overall nitrogen metabolism. Additionally, the upregulation of ahpC indicates that DM075 has developed adaptive defense mechanisms against oxidative stress that may accompany nitrate reduction processes^9,54.

These findings collectively demonstrate that effective nitrate reduction depends not on a single gene or enzyme but on the coordinated function of multiple integrated systems: (1) the core nitrate reductase complex, (2) supporting cofactor synthesis pathways, (3) electron transport systems, (4) nitrogen assimilation machinery, and (5) oxidative stress response mechanisms. This constitutes a compelling illustration of how functional expression and integrated operation of support systems hold greater significance than the mere presence of genes.

To contextualize these strain-specific characteristics within the broader species diversity, we conducted pan-genome analysis of 45 L. fermentum genomes. This analysis revealed that pyruvate oxidase (pox), essential for H₂O₂ production in DM072, is an accessory gene present in 60% (27/45) of strains, whereas nitrate reductase genes (narGHI) are core genes found in all strains. Our phenotypic screening of 12 L. fermentum isolates from healthy donors (9 H₂O₂-producers, 3 non-producers) aligns with this genomic distribution, confirming that both metabolic phenotypes represent natural variants within the species’ normal diversity rather than anomalies. This finding underscores that the functional divergence between DM072 and DM075 reflects the inherent metabolic plasticity of L. fermentum, enabling strain-specific adaptations to different ecological niches.

Our results have important implications for probiotic development. Traditionally, probiotics have been selected and evaluated at the species level. Our findings demonstrate that strains within the same species can exhibit substantially different functional characteristics. The strain-specific characteristics of DM072 and DM075 identified in this study suggest their distinct ecological roles and potential probiotic applications. DM072 could be developed as a probiotic for caries prevention through its antimicrobial activity, whereas DM075 shows promise for cardiovascular health promotion through its contribution to the nitrate–nitrite–nitric oxide pathway^6,14. These findings suggest that strain-specific genomic and functional analyses beyond species-level classification are essential in probiotic selection and development⁵⁵.

The present study provides valuable insights into the distinct functional characteristics of the tested L. fermentum strains. Nevertheless, several limitations must be acknowledged. First, the results are based on in vitro experiments, which may not fully predict the behavior of the strain in the complex in vivo oral environment⁵⁶, owing to the challenges imposed by salivary flow, biofilm dynamics, and host immune responses⁵⁷. Second, although significant differences in nitrate reduction and hydrogen peroxide production were observed between DM072 and DM075, the underlying molecular mechanisms remain partly inferred. Additional functional studies involving gene knockouts or heterologous expression systems would strengthen our mechanistic conclusions⁵⁸. These limitations highlight the need for further in vivo studies, including animal models and human clinical trials, to validate our findings, evaluate the clinical efficacy of these strains as next-generation oral probiotics⁵⁹, and fully understand their potential health benefits⁶⁰.

In conclusion, this study demonstrates that significant functional diversity can exist between strains of the same species, as exemplified by the contrasting nitrate reduction and hydrogen peroxide production capabilities of L. fermentum DM075 and DM072. These findings underscore the importance of strain-specific characterization in probiotic development and highlight promising avenues for tailored probiotic applications targeting oral and systemic health. Future in vivo validation will be crucial to translate these insights into clinical practice.

Both DM072 and DM075 have been developed as commercial probiotic strains and are currently available as dietary supplement products in Korea. DM075 is under clinical investigation for cardiovascular health applications (registered at CRIS: KCT0010680), whereas both strains have undergone safety evaluations. This study provides the genomic and mechanistic basis for understanding the functional properties of these strains—specifically, the presence of the pox gene in DM072, enabling H₂O₂ production, and the enhanced electron transport systems in DM075 supporting nitrate reduction. These genomic insights provide scientific validation for the observed functional characteristics rather than speculative applications.

Limitations

Second, although significant differences in nitrate reduction and hydrogen peroxide production were observed between DM072 and DM075, the underlying molecular mechanisms remain partly inferred. Although our data show complete correlation between pox presence/absence and H₂O₂ phenotype across 12 isolates, and temporal correlation between gltB expression and nitrate reduction, these remain correlative rather than causal evidence. Additional functional studies involving gene knockouts or heterologous expression systems would strengthen our mechanistic conclusions.

Methods

Materials

All chemicals were of analytical grade and purchased from reputable suppliers. MRS broth and agar were purchased from Difco Laboratories (Detroit, MI, USA). Phosphate-buffered saline (pH 7.4) was prepared using NaCl (≥ 99.5% purity), KCl (≥ 99.0% purity), Na₂HPO₄ (≥ 99.0% purity), and KH₂PO₄ (≥ 99.0% purity), which were obtained from Sigma-Aldrich (St. Louis, MO, USA). 3,3’,5,5’-Tetramethylbenzidine (≥ 98% purity) and horseradish peroxidase (≥ 150 U/mg) were purchased from Sigma-Aldrich. Catalase (≥ 10,000 U/mg, bovine liver) was obtained from Merck KGaA (Darmstadt, Germany). Brain Heart Infusion (BHI) medium was purchased from BD Biosciences (Franklin Lakes, NJ, USA). Comparative genomic analyses were performed using different strain sets optimized for each specific analysis: (1) phylogenetic analysis involving 41 representative strains from major clades with complete genomes for accurate tree topology; (2) pan-genome analysis using 45 strains with complete genomes to accurately define core and accessory genes; (3) statistical comparison of functional annotations involving 46 strains comprising both complete and high-quality draft genomes to assess species-level variation. All genome sequences were obtained from the NCBI GenBank database.

Bacterial strains and growth conditions

L. fermentum strains DM072 and DM075 were isolated from the human oral cavity of healthy adult volunteers in South Korea. Samples were collected from 100 healthy adults (15 males, 85 females; mean age 28.23 years; range 21–51 years) with minimal supragingival plaque. The oral health status of participants was assessed by qualified dental professionals at Apple Tree Dental Hospital, and individuals with active oral diseases, including caries, periodontitis, or any mucosal lesions, were excluded from the study. Tongue coating biospecimens were collected using sterile swabs and immediately transferred to VMG2 transport medium. All procedures were conducted in accordance with relevant guidelines and regulations and were approved by the Ethics Committee of the Institutional Review Board of the Apple Tree Dental Hospital (approval number ATDH-2021–0001) and the Korea National Institute for Bioethics Policy (approval number P01-202111-31-002; http://public.irb.or.kr). All participants were informed of the purpose of the study and provided written informed consent before participation.

For routine culture, both strains were streaked on MRS agar plates and incubated at 37 °C for 48 h under anaerobic conditions using an anaerobic chamber (Coy Laboratory Products, Grass Lake, MI, USA) with an atmosphere of 85% N₂, 10% CO₂, and 5% H₂. Single colonies were inoculated into 10 mL MRS broth and cultured overnight (16–18 h) at 37 °C without shaking. S. mutans strain KCOM 1054, obtained from the Korean Collection for Oral Microbiology (KCOM, Gwangju, Korea), was cultured in BHI broth or on BHI agar plates at 37 °C under microaerobic conditions (5% O₂, 10% CO₂, 85% N₂).

RNA extraction and cDNA synthesis

Total RNA was extracted using the EZ™ Total RNA Miniprep Kit (EP301, Enzynomics, Korea) according to the manufacturer’s instructions. Bacterial cells were harvested by centrifugation (3,000 ×g, 10 min, 4 °C), and cell pellets were resuspended in extraction lysis buffer. After lysis and genomic DNA removal, RNA was purified using silica spin columns with on-column DNase I treatment. RNA purity and concentration were assessed using a NanoDrop spectrophotometer (A260/280 ≥ 1.8), and RNA samples were stored at − 70 °C until use.

First-strand cDNA synthesis was performed using the PrimeScript RT Reagent Kit (RR037A, Takara Bio, Japan) with a mixture of oligo dT and random primers. Reverse transcription reactions were carried out in a thermal cycler at 37 °C for 15 min, followed by 85 °C for 5 s. Each condition was processed in triplicate using independently extracted RNA.

Quantitative real-time PCR (qRT-PCR)

qRT-PCR was performed using TOPreal™ SYBR Green qPCR PreMIX (RT500, Enzynomics) following the manufacturer’s protocol. Each 20 µL reaction included 10 µL PreMIX, 1 µL cDNA, 1 µL each of forward and reverse primers (10 pmol/µL), and 7 µL RNase-free water. The following gene-specific primers were used in this study: triose phosphate isomerase A (tpiA; reference gene) forward 5’-CAAGGCCGGCTTCCCGATG-3’ and reverse 5’-CGCCGTTCAATGTGGCCTAAC-3’; ahpC forward 5’-GGCGGACGTCCTGGGTAAG-3’ and reverse 5’-GGCTGGGTCAGCCAACATTG-3’; gltB forward 5’-CGTTGCGGTCCAAAAAGGCG-3’ and reverse 5’-GTACCTCTACCGTCACGGTCG-3’; narG forward 5’-CGTTTGGATGAGCCCGCAAG-3’ and reverse 5’-GCCCCCCCGGTTCCCGGTG-3’; narH forward 5’-CAACGACCCGGAAATTGTCG-3’ and reverse 5’-CCGGGAAGATCATTTCCGGG-3’. Amplification was carried out on a QuantStudio Real-Time PCR System (Thermo Fisher Scientific) under the following conditions: 95 °C for 10 min, followed by 40 cycles of 95 °C for 10 s, 60 °C for 15 s, and 72 °C for 30 s. Melt curve analysis was conducted from 65 °C to 95 °C (0.5 °C/5 s) to confirm specificity. Relative expression was calculated using the 2^ΔΔCT method, with tpiA as the reference gene for normalization.

Nitric oxide production assay

Both L. fermentum strains were cultured in MRS liquid media supplemented with 10 mM sodium nitrate at 37 °C under anaerobic conditions for 0, 6, and 24 h to evaluate their nitrate reduction capabilities. Nitrite concentration, used as an indicator of nitric oxide production, was quantified using the PicoSens Total Nitric Oxide Assay Kit (BM-NIT-200; Biomax. Ltd., Gyeonggi-do, Korea). Culture supernatants were collected and filtered through Amicon Ultra centrifugal filters (10 kDa MWCO, Millipore). Following the protocol of the assay kit, 50 µL of each sample was transferred to a 96-well microplate and mixed with 50 µL of 1× assay buffer. Nitrite standard solutions (0–200 µM) were added to separate wells (60 µL per well) to generate a standard curve. Subsequently, Griess Reagent I (50 µL) was added to each well and incubated for 10 min at room temperature (20–25 °C). Then, Griess Reagent II (50 µL) was added to each well, and the plate was incubated for another 10 min at room temperature. Absorbance was measured at 540 nm using a SpectraMax iD3 microplate reader (Molecular Devices, San Jose, CA, USA). All measurements were performed in triplicate.

Genome sequencing and functional annotation

Complete genome sequences of DM072 and DM075 were obtained using PacBio Sequel and Illumina HiSeq platforms and deposited in GenBank under accession numbers CP102714.1 and CP102715.1 (DM072 chromosome and plasmid) and CP100352.1 (DM075 chromosome). Functional annotation of DM072 and DM075 genomes was conducted using eggNOG-mapper v2.1.12-3-g3666cb0 with the eggNOG 5.0 database. COG categories were assigned according to NCBI’s 25-letter functional classification system to categorize proteins based on their predicted functions. This analysis provided insights into the metabolic and functional capabilities of both L. fermentum strains, allowing comparative genomic assessment. KEGG orthology analysis was conducted using GHOSTKOALA, and genome features were cross-validated using the NCBI Prokaryotic Genome Annotation Pipeline (PGAP). Circular genome maps were generated using Proksee (https://proksee.ca/).

Comparative genomic analysis

Average Nucleotide Identity (ANI) was calculated using the OrthoANI algorithm. Genome sequences were fragmented into 1,020 bp segments, and BLASTN was performed between the genome segments. ANI values were calculated as the mean identity of all BLASTN matches.

Phylogenetic analysis was performed using the genome BLAST distance phylogeny (GBDP) approach. A total of 41 representative L. fermentum strains were selected for phylogenetic tree construction based on: (1) the inclusion of study strains DM072 and DM075; (2) phylogenetic diversity representing major clades, including DM072-associated lineage (e.g., CUL67, DS19_7, SRCM_103290) and DM075-associated lineages (e.g., SD381, SL241, DR9); (3) genome quality, prioritizing complete or high-quality draft genomes; and (4) geographic and ecological diversity across human, food, and environmental sources from diverse regions. Phylogenetic trees were constructed using FastME from the GBDP distances with 1,000 pseudo-bootstrap replicates. Pan-genome analysis was conducted using Roary with a 95% identity cutoff. Genes present in ≥ 99% of strains were identified as core genes; those present in ≥ 95% but < 99% were defined as soft-core genes; those present in ≥ 15% but < 95% of the strains were defined as shell genes, and those genes present in < 15% of strains were identified as cloud genes. ANIb percentage identity heatmaps were generated using the heatmap.2 function in the R package gplots with hierarchical clustering.

Metabolic pathway analysis

Metabolic pathway analysis was performed by mapping gene annotations to KEGG and COG databases. KEGG orthology analysis was conducted using GHOSTKOALA, and genes were assigned to specific metabolic pathways based on KEGG orthology identifiers. COG functional categories were assigned using eggNOG-mapper. The number of genes was counted and compared for each path between strains. This analysis provided insights into the metabolic and functional capabilities of both L. fermentum strains.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

J.-H.E. and H.S.K.: conceptualization, methodology; J.-H.E., M.-Y.C. and E.-M.C.: data curation, validation; J.-H.E.: formal analysis, investigation; M.-Y.C., J.-H.E., E.-M.C., S.-J.Y., Y.Y.K.: resources; M.-Y.C., J.-H.E., E.-M.C., S.-J.Y., D.-H.L. and H.S.K.: writing—original draft preparation J.-H.E., I.H. and H.S.K.: writing—review and editing; J.-H.E.: visualization; J.-H.E. and H.S.K.: supervision; J.-W.K.: laboratory analysis and technical support; J.-Y.H.: bacterial cultivation and strain maintenance; J.-H.E., H.B. and S.J.K.: project administration. All authors read and approved the final manuscript.

Data availability

The complete genome sequences of L. fermentum strains analyzed in this study are available in the GenBank database under accession numbers CP102714.1 (chromosome) and CP102715.1 (plasmid pLFDM072) for DM072 and CP100352.1 for DM075. The corresponding RefSeq accession numbers are NZ_CP102714.1, NZ_CP102715.1, and NZ_CP100352.1, respectively. Raw sequencing reads have been deposited in the Sequence Read Archive (SRA) under BioProject accession numbers PRJNA852394 (DM072) and PRJNA853106 (DM075).Sequencing was performed using hybrid assembly (PacBio + Illumina) with coverage depths of 518.3× (DM072) and 259.2× (DM075), achieving >90% bases with Phred scores >30. Assembly quality metrics include CheckM completeness of 93.18% (DM072) and 95.1% (DM075), with contamination estimates of 8.51% and 2.56%, respectively. The higher contamination estimate for DM072 likely reflects the presence of plasmid pLFDM072 (44,665 bp) rather than actual contamination. Both assemblies have been validated and accepted by NCBI quality control standards. Detailed sequencing and assembly methods have been previously described [7, 25].

Declarations

Competing interests

The authors declare no competing interests.