Lactobacillus reuteri AN417 cell-free culture supernatant as a novel antibacterial agent targeting oral pathogenic bacteria

Lactobacillus reuteri AN417 is a newly characterized probiotic strain. The activity of AN417 against oral pathogenic bacteria is unknown. We investigated the antibacterial activity of cell-free L. reuteri AN417 culture supernatant (LRS) against three oral pathogens: Porphyromonas gingivalis, Fusobacterium nucleatum, and Streptococcus mutans. P. gingivalis and F. nucleatum have been implicated in periodontal disease, whereas S. mutans causes dental caries. Exposing these oral pathogenic bacteria to LRS significantly reduced their growth rates, intracellular ATP levels, cell viability, and time-to-kill. The minimal inhibitory volume of LRS was 10% (v/v) against P. gingivalis, 20% (v/v) for F. nucleatum, and 30% (v/v) for S. mutans. LRS significantly reduced the integrity of biofilms and significantly suppressed the expression of various genes involved in P. gingivalis biofilm formation. The L. reuteri AN417 genome lacked genes encoding reuterin, reuteran, and reutericyclin, which are major antibacterial compounds produced in L. reuteri strains. LRS treated with lipase and α-amylase displayed decreased antibacterial activity against oral pathogens. These data suggest that the antibacterial substances in LRS are carbohydrates and/or fatty acid metabolites. Our results demonstrate that LRS has antimicrobial activity against dental pathogenic bacteria, highlighting its potential utility for the prevention and treatment of P. gingivalis periodontal disease.


Results
Isolation and characterization of L. reuteri strains. We isolated 135 L. reuteri strains from human infants and 6-month-old female swine under anaerobic conditions. Bacterial isolates were identified by 16S rRNA gene sequencing and a matrix-assisted laser desorption/ionization time of flight (MALDI-TOF) biotyper (Bruker, Table 1). Primary screening for antimicrobial activity against pathogens, including Escherichia coli (KCTC 2571), Pseudomonas aeruginosa (DSM 50071), and S. mutans (KCTC 3065), was performed to select L. reuteri strains that had exhibited antimicrobial activity against periodontopathic bacteria in a disk diffusion assay (Fig. 1A). The results showed that L. reuteri AN417 displayed the strongest antibacterial activity against pathogens.
In addition, the production of 1, 3-propanediol (1, 3-PDO) by the isolated L. reuteri strains was determined using a high-performance liquid chromatography (HPLC) system (Table 1 and Supplementary Fig. S1A). The 1, 3-PDO is valuable compound used as a replacement for petroleum-based glycols, including propylene glycol, butylene glycol, and glycerin. It has been reported that L. reuteri metabolizes glycerol to reuterin (3-hydroxypropionaldehyde, 3-HPA) and then converts reuterin to 1, 3-PDO. L. reuteri cannot grow on glycerol as the sole carbon source and the conversion to 1, 3-PDO from glycerol requires NADH produced by glucose metabolism. Therefore, we tested the production level of 1, 3-PDO under co-fermentation in the presence of both glycerol and glucose in the culture medium.
Potentially important inhibitory activity of L. reuteri AN417 supernatant (LRS) against oral bacterial pathogens. When the antimicrobial activity of the newly identified L. reuteri strains was assessed, we observed that L. reuteri AN417 culture supernatant (LRS) outperformed those of our other tested strains. We first observed whether isolated L. reuteri strains affected the growth of the oral pathogens S. mutans KCTC 3065 and P. gingivalis BAA-308 (Fig. 1A, B). The clear zone was the largest when S. mutans was treated with LRS compared to the supernatants from other strains. Moreover, the growth of P. gingivalis was more highly inhibited when 10% (v/v) LRS was added to the medium than when supernatants derived from other L. reuteri strains were treated. Furthermore, growth of P. gingivalis was most inhibited when 10% (v/v) LRS was added to the medium (Fig. 1B). The findings indicated that cell-free culture supernatant derived from L. reuteri AN417 exhibited the highest activity against P. gingivalis and potentially against other oral pathogenic bacteria.
Next, we determined that the antimicrobial bioactive substances are present in the culture supernatant, not inside the bacterial cells themselves. To establish whether the active substances were present in the culture supernatant or in the bacterial cells, bacterial cell extracts (BE) were prepared using ethyl acetate. BE and LRS, which were concentrated and diluted to the desired values, were treated with P. gingivalis for 24 h. Compared with the control treated only with HEPES or de Man Rogosa and Sharpe (MRS), LRS substantially inhibited the growth of P. gingivalis. However, no effect was observed with the BE treatment. Furthermore, compared with control treatments, LRS significantly reduced P. gingivalis intracellular ATP levels, whereas the BE treatment did not (Fig. 1C). Administration of 20% or 40% (v/v) LRS for 96 h significantly reduced ATP levels and growth of the pathogenic bacteria (Fig. 1D).
LRS inhibits the growth of oral pathogenic bacteria. The inhibitory effects of 10%, 20%, 30%, and 40% (v/v) LRS on the growth of selected oral pathogenic bacteria (P. gingivalis, F. nucleatum, and S. mutans) were assessed. As the results, growth inhibition against tested pathogens was dependent on LRS concentrations. Treat-  LRS effectively reduces the viability of oral pathogenic bacteria. In fluorescent cell-staining assays, LRS increased the number of dead pathogenic bacteria and decreased the number of live pathogenic bacteria. To evaluate the effect of LRS on the viability of the three pathogenic bacterial strains, a mixture of SYTO9 green fluorescence nucleic acid stain and propidium iodide was used for cell staining. The cells were observed by fluorescence microscopy. Compared with the negative control (MRS or Brain Heart Infusion [BHI] medium), the intensity of green fluorescence emitted by live bacteria decreased in P. gingivalis, F. nucleatum, and S. mutans cultures treated with LRS. In S. mutans, the intensity of red fluorescence, indicating dead bacteria, increased with LRS treatment (Fig. 3A). After treatment with LRS, bacterial death was observed over time. As a result, it was observed that P. gingivalis died rapidly starting 8 h after LRS treatment (Fig. 3B). These results were consistent with those of LIVE/DEAD BacLight analysis, with LRS treatment substantially reducing colony forming units compared with the MRS treatment.
Antimicrobial activity of LRS against oral pathogenic bacteria, especially P. gingivalis. Treatment of the three oral pathogenic bacteria with LRS revealed a minimum inhibitory volume (MIV) of approximately 10% (v/v) for P. gingivalis. The MIV was 20% and 40% (v/v) for F. nucleatum and S. mutans, respectively (Fig. 3C).

LRS impedes biofilm formation by the oral pathogenic bacteria.
To confirm the antibiofilm activity of LRS against the biofilm formation during the early stage of bacterial colonization, LRS was added to P. gingivalis and S. mutans cultures immediately after bacterial inoculations, so that the effects of LRS on biofilm formation during the initial attachment phase could be examined. LRS treatment substantially reduced the fluorescence intensity compared with the control treatment ( Fig. 4A), which was consistent with the quantitative results. To determine the concentration of LRS that eradicates established P. gingivalis biofilms, biofilms developed for 5 days were treated with LRS and stained with crystal violet. Compared with the control treatment,  Table 1. Isolation of Lactobacillus reuteri strains, and determination of 1,3-PDO and reuterin production. 1,3-PDO and reuterin production were determined in Lactobacillus reuteri isolates using HPLC and Colorimetric method, respectively. a 1,3-PDO production:+++; > 12 g/L,++; > 7 g/L,+; > 0 g/L, -; not detected. b Reuterin production:+++; > 0.5 at 560 nm,++; > 0.3 at 560 nm,+; > 0.1 at 560 nm, -; not detected.  Genomic analysis and characterization of L. reuteri AN417. Genome-genome relatedness of L.
reuteri AN417 was also analyzed by calculating the average nucleotide identity (ANI) and constructing a phylogenomic tree of 31 genome sequences with fewer than 30 scaffolds among the genomes of L. reuteri strains from the GenBank/EMBL/DDBJ database. L. reuteri genomes were an average size of 2.10 Mb and G + C ratio of 38.43-39.31 ( Table 2). The whole genome of L. reuteri AN417 showed ANI values ranging from 94.8 to 99.6% with L. reuteri strains. The highest ANI values were obtained for the pig strains (Fig. 5A). Additionally, whole genome phylogenetic analysis was performed with 27 L. reuteri genomes, excluding those with many contigs. A phylogenetic tree of L. reuteri strains was constructed using the amino acid alignments of 766 core genes using the maximum likelihood approach. The tree showed clear separation of L. reuteri strains into five host-defined phylogenetic lineages (Fig. 5B). This analysis also revealed that L. reuteri AN417 isolated from pigs clustered in lineage IV, which is contained in strains originating from pigs. The whole genome sequence of L. reuteri AN417 encodes the urease complex UreABCEFGD unlike general strains isolated from human and pigs, contributing to its viability under acidic conditions (Fig. 5C). The mucin binding protein, Muc2, contributes to host adaptation and adhesion to mucus. However, genes encoding reuteran, reutericyclin, and reuterin, which are important in the antimicrobial activity of L. reuteri strains, were absent in www.nature.com/scientificreports/ the genome (Fig. 5C). In addition, reuterin production was also measured in L. reuteri AN417 (Table 2), demonstrating that L. reuteri AN417 did not produce reuterin. This result was consistent with the genomic analysis results (Table 1).

Antibacterial activity of LRS is mediated by the presence of fatty acids and sugars.
To categorize the type of metabolites responsible for the activity of LRS against periodontopathogens, the antibacterial effect of LRS against P. gingivalis in the presence of proteinase K, lipase, and α-amylase was evaluated. Treatment of LRS with lipase or α-amylase eliminated the inhibitory effect of LRS on P. gingivalis growth (Fig. 6A-C). These findings suggested that the antibacterial effect of LRS against oral pathogenic bacteria could be attributed to the presence of a fatty acid and/or sugar.

Discussion
The present study reports the antibacterial effects of cell-free culture supernatant from L. reuteri strain AN417 (LRS), a strain that was isolated from the porcine small intestine, against selected oral pathogenic bacteria. Based on our results, we anticipate that L. reuteri AN417 can positively affect oral health. Recently, there has been growing interest in the potential and utilization of microbial metabolites, termed postbiotics 23 . The cell-free culture supernatant of L. reuteri AN417 exhibited greater antimicrobial activity than those of the known Lactobacillus reference strains KCTC 3594 and KCTC 3678, which also inhibited periodontopathic bacteria (Fig. 1A,B). In our study, multiple lines of scientific evidence validated the antimicrobial activity of LRS. Evaluation of the antimicrobial effects of naturally-derived agents has largely focused on their activity against P. gingivalis, F. nucleatum, and S. mutans because these bacteria have received the most attention in relation to oral diseases and are implicated in periodontal diseases, dental caries, and endocarditis [7][8][9][10] . To date, the antimicrobial effects of naturally derived agents against these bacteria have been evaluated in numerous studies, with www.nature.com/scientificreports/ a recent study reporting the antimicrobial activities of L. reuteri supernatant against P. gingivalis 21 . Hence, we also focused on these three oral pathogens in our study. The major antibacterial compounds produced by L. reuteri strains are reuterin, reuteran, and reutericyclin. Genomic analysis of L. reuteri AN417 revealed the absence of a pdu-cbi-cob-hem gene cluster for the biosynthesis of reuterin and cobalamin (vitamin B 12 ), and genes for the synthesis of reutericycin and reuteran. However, L. reuteri AN417 encoded an inulin-type fructansucrase. The strain also encoded the UreABCEFGD urease complex, which was not found in the other strains isolated from humans and pigs, contributing to its viability under acidic conditions.
We compared the activity of concentrated L. reuteri AN417 cell-free culture medium (LRS) to the activity of L. reuteri AN417 cell extracts using organic solvents. The results showed that L. reuteri AN417 cell extracts had no antimicrobial activity, whereas LRS did (Fig. 1C). From these data, secondary metabolites produced during growth are thought to play a crucial role in antimicrobial activities against oral pathogenic bacteria 24 . Bungenstock et al. reported that the antibacterial effect of probiotics against fermented foodborne pathogens is attributable to their production of lactic acid and the associated increase in acidity 25 . However, the antimicrobial activity of LRS observed in this study may not have been due to lactic acid alone. Probiotic Lactobacillus spp. are Table 2. Lactobacillus reuteri genomes used for phylogeny reconstruction and comparative genomics. Whole genome phylogenetic analysis was performed with 27 L. reuteri genomes except for them with large numbers of contigs.  www.nature.com/scientificreports/ also known to produce various metabolites that defend against S. mutans colonization 26 . The level of intracellular ATP, the energy source for viable oral pathogenic bacteria, was reduced significantly by LRS, leading to the growth inhibition of oral pathogenic bacteria (Fig. 1D, right). The inhibitory effect was maintained for 96 h, indicating that the bioactive substances in the LRS were stable over time. Growth of S. mutans (a Gram-positive bacterium), F. nucleatum (Gram-negative), and P. gingivalis (Gram-negative) was inhibited by adding 30% (v/v), 20% (v/v), and 10% (v/v) of LRS, respectively ( Fig. 2A), demonstrating that bacterial growth inhibition by LPS was the most potent and most specific for P. gingivalis. In addition, we speculate that higher concentrations of antimicrobial substances are required to inhibit the growth of Gram-positive bacteria, which have thick peptidoglycan cell walls. Endotoxins, such as lipopolysaccharides (LPS), are produced from oral bacterial biofilms, including oral plaques. Endotoxins destroy alveolar bone and induce a series of inflammatory reactions that ultimately lead to tooth loss 27 . To determine whether the effect originated from bacterial membrane damage, SYTO9 and propidium iodide staining was performed. LRS treatment resulted in a marked decrease in viable bacterial cells (Fig. 3). This result may imply that bacterial membrane integrity was weakened by treatment with LRS, which led to the inhibition of bacterial growth and a reduction in the continuous release of endotoxins.
Biofilms are the cause of dental bacterial infections 28,29 . Biofilms are comprised of extracellular polymeric substances secreted by bacteria during metabolic processes. The biofilm structure confers antibiotic resistance 30 . Biofilms in the oral cavity generally contribute to periodontitis. Within biofilms, resident microbes are resistant to external attacks that include antibacterial agents, and the bacteria dispersed from the biofilms can cause infection 31 . Biofilm dispersal agents might be the most suitable targets for the prevention of periodontitis and dental caries. In previous studies, supernatants from cultures of Lactobacillus sp. inhibited biofilm formation and reduced the expression of genes related to the production of exopolysaccharides 26 . In the current study, LRS significantly reduced biofilm formation (Fig. 4), which could contribute to the prevention of dental caries and periodontitis.
LRS has excellent inhibitory abilities against the growth and biofilm formation of the tested bacteria. In particular, P. gingivalis required the lowest amounts of LRS to inhibit its growth and biofilm formation among the bacteria tested in our study. Thus, we focused on P. gingivalis in our other analyses, as this was the pathogen most strongly affected by LRS.
The antimicrobial activity of L. reuteri has been attributed to its production of organic acids, hydrogen peroxide, and bacteriocin-like compounds 32 , such as reuterin, reuteran, and reutericyclin 33 . However, L. reuteri AN417 lacks the ability to produce them because of the absence of the required genes. Thus, to identify the antimicrobial substance in LRS, various enzymes such as α-amylase, lipase, and proteinase were added to LRS to catabolize and inactivate any sugars, lipids, or proteins that could confer antimicrobial activity. To confirm which enzymes cause loss of activity, each LRS treated with enzymes was added to P. gingivalis cultures. LRS treated with lipase and α-amylase did not inhibit P. gingivalis growth, suggesting that the active substance responsible for LRS antibacterial activity was either a fatty acid or a sugar. Although L. reuteri reportedly produces antimicrobial molecules, including lactic acid, acetic acid, ethanol, and reutericyclin 34 , fatty acid and sugar-based antimicrobial substances have not yet been reported.
According to a previous study 35 , soluble or immobilized PLNC8 αβ bacteriocins from L. plantarum strains NC8 and 44048 prevent P. gingivalis colonization and pathogenicity. Therefore, it is necessary to conduct additional experiments under various conditions to support the findings of this study. Furthermore, the bioactive substances in LRS should be identified and purified for practical use.
In this study, we show that supernatants derived from L. reuteri AN417 cultures are able to suppress the growth and biofilm formation of oral pathogenic bacteria. Interestingly, L. reuteri AN417 does not produce reuteran, reutericyclin, and reuterin, which are important in the antimicrobial activity of reported L. reuteri strains. Although further studies are required, the antibacterial substance in LRS is suspected to be a fatty acid or a sugar, which has not yet been reported. Thus, LRS has potential as a novel bioactive substance for the prevention and treatment of pathogens associated with periodontitis.

Methods
Strain isolation and identification. Novel Lactobacillus reuteri strains were isolated from seven infants aged 3-7 years and from the small intestines of 13 6-month-old female pigs in the Republic of Korea. The pigs were fed a mixed diet. The intestinal contents from each child or pig were resuspended and serially diluted in sterile 0.85% NaCl. Aliquots were cultured anaerobically on MRS agar in an atmosphere of CO 2 :H 2 :N 2 (7:7:86) atmosphere. After 2-3 days, single colonies were subcultured on fresh MRS agar. All colonies were selected irrespective of their shape and size. Genomic DNA (gDNA) was extracted and purified from cells grown on MRS agar as described previously 36 . The gDNA was used for 16S rRNA gene amplification and sequencing, and whole genome sequencing. The complete 16S rRNA gene sequence was amplified using universal primers: 27F (5′-AGA GTT TGATCMTGG CTC AG-3′) and 1492R (5′-TAC GGY TAC CTT GTT ACG ACTT-3′). The amplified genes were sequenced and compared with sequences obtained from the EzBioCloud 37 and GenBank/EMBL/ DDBJ (http://www.ncbi.nlm.nih.gov/blast ) databases.
Identification of isolates using a MALDI-TOF Biotyper. One colony of each bacterial isolate was subcultured for 24 h and used for MALDI-TOF Biotyper analysis. The colony was acquired using a toothpick and spotted onto a polished steel MALDI target plate. One microliter of formic acid (70% in water) was added to the spot and dried. Subsequently, 1 µL of MALDI matrix (10 mg/mL solution of α-cyano-4-hydroxycinnamic acid (HCCA) in 50% acetonitrile/2.5% trifluoroacetic acid) was added to the spot and dried. The MALDI target plate was placed in the MALDI-TOF/Microflex LT instrument (Bruker Daltonics, Billerica, MA, USA) for automated measurement and data interpretation. The MALDI Biotyper output is a log (score) between 0 and 3.0, which is calculated from a comparison of the peak list from an unknown isolate with the reference MSP in the database. A log (score) ≥ 1.7 was indicative of a close relationship at the genus level. A log (score) ≥ 2.0 was set as the threshold for a match at the species level. Isolates with a log (score) ≥ 2.0 were accepted as the correct identification.  Minimal inhibitory volume (MIV) assay. The MIV of LRS was determined using a modified method of a previously described procedure 39 . Pathogenic bacterial cultures generated in suitable liquid culture media as detailed above were diluted with nutrient broth to an OD 600 of 0.005 or 0.05. The diluted bacterial suspensions were then treated with either MRS medium (control) or 1 × LRS, dispensed into the first well of a 96-well plate, and serially diluted into consecutive wells. Plates were incubated at 37 °C for 24 h, and the absorbance at 600 nm was measured using a microplate reader. Time-kill assay. The time-to-kill P. gingivalis following LRS treatment was determined based on a previously described protocol 40 . A bacterial suspension (OD 600 = 1) was treated with LRS and incubated anaerobically at 37 °C. An aliquot (100 μL) of this bacterial culture was removed at 0, 4, 8, 24, 48, and 72 h following LRS treatment, and plated on TSB agar to quantify the number of colony forming units (CFUs) in the treated cultures. Pure MRS medium was used as a negative control.

LIVE/DEAD
Real-time quantitative polymerase chain reaction (RT-qPCR) analysis. The potential of LRS to prevent biofilm formation and/or destroy established biofilms was investigated. P. gingivalis (OD 600 = 0.6) was dispensed into a polystyrene-coated 6-well plate and treated with MRS medium (control) or LRS for 24 h in an anaerobic incubator. In addition, P. gingivalis biofilms were established in a polystyrene-coated 6-well plate for 5 days, after which these biofilms were treated with either MRS medium (control) or LRS for 48 h in an anaerobic incubator. Total RNA was extracted with TRIzol reagent, and cDNA was synthesized using a PrimeScript RT reagent Kit (TaKaRa Bio, Shiga, Japan). RT-qPCR analysis of cDNA was performed according to the manufac- Effect of enzymes on antibacterial activity. Lactobacillus reuteri AN417 cell-free supernatants were treated with enzymes to evaluate the effect of enzymes on antibacterial substances. The LRS were treated with proteinase K (1 mg/mL), lipase (700 units/mg), or α-amylase (150 units/mg). For lipase and α-amylase treatment, the pH of the LRS was adjusted to 6.5 with NaOH to facilitate enzymatic activity. Enzymes were activated by incubation of the enzyme-treated supernatant at 37 °C for 3 h, after which the enzymes were immediately  41 . To correct the sequencing errors that can occur at both ends of a contig, the SMRT resequencing protocol was performed with assembly in which the first half of the contig was switched with the second half. Protein-coding genes were predicted using Prodigal v.2.6.3. Ribosomal RNA, transfer RNA, and miscellaneous features were predicted using Rfam v12.0 42 . CRISPR loci were predicted using the CRISPR recognition tool. ANI values were calculated using an online ANI calculator 43 .
Statistical analyses. All statistical analyses were performed using Student's t-test. The results are expressed as the mean ± standard deviation for each group. Multiple group data were analyzed using one-way analysis of variance, followed by Dunnett's multiple range test. The threshold for significance was set at p < 0.05. Data shown are representative of three independent experiments, except for Fig. 1B, in which the data are from two independent experiments.

Data availability
Whole genome sequences were deposited with Bioproject PRJNA637956 and Biosample SAMN15162791, respectively. GenBank accession numbers are CP054657 for single chromosome and CP054658-CP054661 for the four plasmids, respectively.