The oral cavity is a very complex ecosystem which harbors more than 700 bacterial species [1,2]. These bacteria form specific microbial flora in several different habitats, including the tooth surface, the gingival sulcus and the dorsum of the tongue. The onset and development of periodontitis are associated with various factors. The bacterial factor is specific Gram-negative anaerobic bacteria such as Porphyromonas gingivalis [3].

Periodontitis is a recurrent disease with acute and chronic phases. In general, non-surgical periodontal treatments such as scaling and root planning are used for the chronic phase of periodontitis. In the acute phase, systematic and topical antibiotic therapies are applied to reduce the causative bacteria and prevent recurrence. However, antibiotic therapy has some harmful side effects, including possible allergic reactions, and antibiotic-resistant bacteria may appear after the prolonged use of antibiotics [4-6]. To develop an alternative therapy, there has been a focus on the usefulness of Lactobacillus and Bifidobacteria as probiotics [7]. Probiotics are defined by Fuller as, "a live microbial feed supplement which beneficially affects the host by improving its intestinal microbial balance" [8]. The effects of probiotics in the prevention of gastrointestinal infections and the improvement of allergy symptoms have been reported [9-11]. Recently, in the dental field, there have been several attempts to examine the possibility that probiotics may act directly in the oral cavity to prevent oral diseases such as dental caries and periodontal diseases. A number of clinical studies have already reported promising findings [12-14]. However, the antibacterial mechanism of probiotic bacteria against periodontal pathogens has not been clearly elucidated. The purpose of this study was to identify new probiotic strains that exhibit antibacterial activity against the representative periodontal pathogen, P. gingivalis, and to identify the antibacterial substances produced by the strains that may be applied to the prevention or treatment of periodontal disease.

The purification procedure of the antibacterial constituents against a periodontal pathogen produced by lactobacilli was shown in a supplemental file.

The present study used 50 strains of lactic acid bacteria (Table 1). All of the lactic acid bacteria strains were cultured in Man-Rogosa-Sharpe (MRS) broth (Difco, Becton Dickinson and Company, Sparks, MD, USA) at 37°C for 24 h under anaerobic conditions (N₂: 80%, CO₂: 10%, H₂: 10%). P. gingivalis ATCC33277 was cultured according to the previous study [15].

Table 1: Antibacterial activity of lactobacilli expressed as highest inhibitory dilution.

Screening of probiotic strains by antibacterial activity

Each lactic acid bacterial culture broth was centrifuged to obtain a culture supernatant. After adjusting the pH to 7.0 with 10N NaOH, the supernatant as test samples were serially diluted (two-fold) to 1/256 and P. gingivalis culture (10⁷ CFU/ml) was inoculated. The growth inhibition was determined as the absorbance value below 0.1 at 620 nm. Antibacterial activity against P. gingivalis was detected by the dilution folds as Units.

Fractionation of the L. plantarum 122 and L. fermentum ALAL020 culture supernatants.

After shaking an equal amount of ethyl acetate or acetone with the culture supernatants of each strains, the water soluble phase was applied to a Sephadex G-25 gel filtration column (100 mmφ × 350 mm) and eluted with deionized water (flow rate 5 ml/min). The total carbohydrate level (420 nm) of the peak fraction was detected using the phenol–sulfuric acid assay method [16]. The acetone supernatant of L. fermentum ALAL020 Sephadex G-25 fraction was analyzed by HPLC (JASCO Corporation, Tokyo, Japan) in a reverse-phase preparative ODS (C18) column (Wakoshil 5C18 AR Prep 20.0 mmφ × 250 mm), using gradients of (B) 50% acetonitrile in (A) 10% acetonitrile, both containing 0.05% trifluoroacetic acid: 10-30% of B over 20 min, in a 50% step-wise manner over 10 min, at a flow rate of 5 ml/min.

Determination of the antibacterial activities in the chromatography fractions of L. fermentum ALAL020.

After adjusting the pH to 7.0, 20 µl of each fractionated sample were aliquoted into a 96-well plate. Subsequently, 160 µl of GAM broth (Nissui, Tokyo, Japan) supplemented with 5 µg/ml hemin, 1 µg/ml menadione and 20 µl of the P. gingivalis culture (1 × 10⁶-10⁷ CFU/ml) were added into the wells. The mixtures in the 96-well plates were anaerobically cultured at 37°C for 48 h. The cultures were measured at an absorbance of 620 nm with a microplate reader. The antibacterial activity was measured as the Units used in screening or by the sodium lactate equivalence (SLU). The SLU was determined by dividing (L) mg/ml by (C) mg/ml, where (L) was the minimum inhibitory concentration (MIC) of sodium lactate against P. gingivalis and (C) was the concentration of the fraction that inhibited P. gingivalis to below the detection limit. The detection limit was determined as the absorbance value below 0.1 at 620 nm. To determine the antibacterial activity, the reciprocal value of (C) was divided by the reciprocal value of (L), using the following formula: (1/C)/(1/L)=L/C=SLU.

Quantification of sodium lactate and an analysis of the amino acids

Sodium lactate was quantified using an HPLC NANOSPACE SI-1 system (Shiseido, Tokyo, Japan) on a CAPCELL PAC C18 TYPE MG column (1.5×250 mm; Shiseido, Tokyo, Japan) equilibrated with 50 mM NH₄H₂PO₄-H₃PO₄ solution (pH 4.2) at a flow rate of 100 µl/min. The absorbance at 210 nm was monitored.

For the amino acid analysis, the sample was heat-hydrolyzed at 110°C with 6 N HCl for 24 h, fluorescently derivatized by 4-fluoro-7-nitrobenzofurazan (NBD-F), and eluted by gradients of (B) CH₃CN/MeOH/50 mmol/l KH₂PO₄=1/2/2 (v/v/v) in (A) CH₃CN/ 75 mmol/l H₃PO₄=16/84 (v/v). The gradient condition was 0% of B over 5 min, 0-70% over 20 min, and 70-100% over 20 min at a flow rate of 100 µl/min. The eluates were subjected to fluorescence detection (λ ex=470 nm, λ em=540 nm).

The LC-MS analysis of the purified product

An HPLC-purified sample was analyzed using an Ultimate 3000 UHPLC System (Thermo Fisher Scientific, Inc., Waltham, MA, USA), and a Shim-pack VP-ODS column (150 mm × φ4.6 mm, Shimadzu, Tokyo, Japan). Gradient elution was performed using 10-50% acetonitrile with 0.05% formic acid. An MS analysis was performed with a Mass Spectrometry Q Exaction^TM Quadrupole/Orbitrap hybrid mass spectrometer (Thermo Fisher Scientific). Ionization was performed using the ESI method and detected in the positive mode.

Screening of the probiotic strains

As shown in Table 1, two strains of lactobacilli out of 50 strains of lactic acid bacteria exhibited strong antibacterial activity against P. gingivalis: L. plantarum 122 (8 Units) and L. fermentum ALAL020 (16 Units). Other strains of the same species, showed only 2 Units. Therefore, the following analysis of the antibacterial constituents was performed with these two strains.

Fractionation of the L. plantarum 122 and L. fermentum ALAL020 culture supernatants.

The culture supernatant of L. plantarum 122 was separated on a Sephadex G-25 column into four fractions by absorbance at 260 nm; a substantial amount of carbohydrates was contained in fractions 3 and 4 (Figure 1). The culture supernatant of L. fermentum ALAL020 was separated into five fractions, however, the carbohydrate level was very low in comparison to L. plantarum 122 (Figure 2).

Figure 1: The fractionation of the culture supernatant of L. plantarum 122 using a Sephadex G-25 column: (A) Gel filtration chromatography of the water soluble phase of the culture supernatant. The solid line shows the UV absorption at 260 nm; the dotted line shows the chromogenic reaction of carbohydrate detected at 492 nm. (B) Antibacterial activity and sodium lactate were detected in fractions 1 to 4. Each fraction was adjusted to pH 7.0 before these experiments. Antibacterial activity against P. gingivalis was defined as the highest dilution (Units) to exhibit growth inhibition in the measurement of absorbance at 620 nm. ND: not detected.

Figure 2: The fractionation of the culture supernatant of L. fermentum ALAL020 using a Sephadex G-25 column: (A) Gel filtration chromatography of the water soluble phase of the culture supernatant. The solid line shows the UV absorption at 260 nm; the dotted line shows the chromogenic reaction of carbohydrate detected at 492 nm. (B) Antibacterial activity and sodium lactate were detected in fractions 1 to 5. Each fraction was adjusted to pH 7.0 before these experiments. Activity units were based on the sodium lactate equivalent (SLU) antibacterial activity. ND: not detected. SLU=(1/C)/(1/L)=L/C. C: The concentration of the fraction that inhibited P. gingivalis to below the detection limit. L: MIC of sodium lactate against P. gingivalis . (C) The antibacterial activity of the acetone supernatant and precipitate of Fr. 3 are shown as SLU.

The antibacterial activity of the L. plantarum 122 fractions

An antibacterial assay was performed. At the maximum dilution, there were 4 Units each in fractions 3 and 4 (Figure 1B). The concentrations of sodium lactate in fractions 3 and 4 were 23.4% (w/w) and 41.7% (w/w), respectively (Figure 1B). When the antibacterial activity of sodium lactate against P. gingivalis was measured in advance, the MIC of sodium lactate was 2.0% (20 mg/ml). The sodium lactate concentrations of fractions 3 and 4 were higher than the MIC of sodium lactate.

The antibacterial activity of the L. fermentum ALAL020 fractions

As shown in Figure 2B, fraction 3 showed the highest antibacterial activity of 1.82 SLU. The concentration of sodium lactate in fraction 3 was 1.3% (w/w) (Figure 2B), which was less than the MIC of sodium lactate (2.0%). This result indicated that L. fermentum ALAL020 might have some active substances other than sodium lactate. Therefore, the fractions were purified and the following analyses were performed.

Four volumes of acetone were added to the fraction 3 solution and the antibacterial activity of the acetone supernatant and precipitate was determined, revealing that the antibacterial activity of the acetone supernatant was as strong as 8.0 SLU (Figure 2C). The culture supernatant was then subjected to acetone precipitation for further experiments and the supernatant was separated using a Sephadex G-25 column. One peak fraction was obtained by this procedure (Figure 3A). This peak fraction was subsequently separated in the ODS (C18) column by reverse-phase HPLC and divided into eight fractions: H1-H8 (Figure 3B). H8-1 showed the strongest antibacterial activity against P. gingivalis (5.0 SLU), followed by H8-2 (3.1 SLU) (Figure 3C).

Figure 3: The purification of L. fermentum ALAL020: (A) Sephadex G-25 gel filtration chromatography of the acetone supernatant: The eluted peak area was then subjected to HPLC. (B) Reverse-phase chromatography by HPLC on an ODS (C18) column. (C) Antibacterial activity was detected in H1 to H8. Activity units were based on the sodium lactate equivalent (SLU) antibacterial activity.

The LC-MS analysis of the antibacterial constituents of L. fermentum ALAL020

An amino acid analysis of H8-1 and H8-2 was performed, however, no apparent peak was detected (data not shown). We therefore performed an LC-MS analysis. H8-1 and H8-2 were subjected to precision mass spectrometry. The molecular weights of both were estimated to be 226.131 Da (Figures 4A and 4B), with a molecular formula of C₁₁H₁₈O₃N₂.

Figure 4: The LC-MS analysis of H8-1 and H8-2:
(A) The spectrum of H8-1 contained major ions of m/z 227.1380. (B) The spectrum of H8-2 contained major ions of m/z 227.1390. The molecular weights of H8-1 and H8-2 were estimated to be both 226.131 Da.

P. gingivalis is the most important periodontal pathogen [17-19] because it is frequently isolated from patients with chronic periodontal disease. Vivekananda et al. [13] reported that L. reuteri DSM 17938 and L. reuteri ATCC PTA 5289, which we also used in our screening study, reduced the number of P. gingivalis in the oral cavity. In addition, Ishikawa et al. [20] reported that L. salivarius LS1 reduced the number of P. gingivalis cells in vitro. In the present study, L. plantarum 122 and L. fermentum ALAL020 showed stronger antibacterial activities than these other lactobacilli at a neutral pH.

The sterilization or antibacterial substances produced by lactic acid bacteria include organic acids such as lactic and acetic acids, hydrogen peroxide and bacteriocins [21-23]. Kang et al. [24] reported that L. reuteri showed antibacterial activity against oral pathogens including P. gingivalis, but it was decreased when the pH was neutralized. Thus, they concluded that the antibacterial substance of L. reuteri was lactic acid. Furthermore, Takahashi et al. [25] reported that P. gingivalis proliferation was inhibited at a pH ≤ 6.5. However, it is possible that low pH conditions in the oral cavity may induce dental caries or hypersensitivity. In this study, the antibacterial test against P. gingivalis was performed after the pH neutralized. Despite the neutral pH of the L. plantarum 122 and L. fermentum ALAL020 culture supernatants, both strains strongly inhibited the growth of P. gingivalis. These results indicate that sodium lactate, a neutralized form of lactic acid, also has antibacterial properties. This result is in line with the findings of Matsuoka et al. [26]. The antibacterial fraction of L. plantarum 122 contained more than 2% sodium lactate (MIC of sodium lactate against P. gingivalis). The presence of sodium lactate strongly contributed to the antibacterial activity of L. plantarum 122.

As for L. fermentum ALAL020, strong antibacterial activity was observed in the ethyl acetate or acetone supernatant of fraction 3 (Figures 2B and 2C), but the concentration of sodium lactate was only 1.3% (Figure 2B). Therefore, fraction 3 was divided in HPLC (Figure 3B) to obtain purified active fractions and analyzed by LC-MS (Figure 4). In L. fermentum, bacteriocin-like substances have been reported [27-29]. However, the antibacterial substances obtained from H8-1 and H8-2 in the present study showed a different molecular weight (226.131 Da, Figure 4) from bacteriocin-like proteins and the following composition formula: C₁₁H₁₈O₃N₂. The substances from H8-1 and H8-2 showed different behaviors in the LC, and were divided into two proximal peaks; we speculate that the substance was a structural isomer. An amino acid analysis revealed that this substance did not contain amino acids, and it was confirmed with FT-IR (data not shown). Reuterin, an antibacterial substance in L. reuteri (also known as 3-hydroxypropionaldehyde; molecular weight, 74 Da; composition formula, C₃H₆O₂) also has a low molecular weight and does not contain amino acids [30,31]. Because the molecular weight and formula of our isolated substance were different from those of reuterin, the isolated substance may be a novel antibacterial substance. In our preliminary experiment, L. fermentum ALAL020 was less effective against Prevotella intermedia or Fusobacterium nucleatum (date not shown), but highly effective against P. gingivalis. These results indicate the possibility that it may specifically target P. gingivalis. According to these results, it is necessary to analyze the mechanisms that underlie its antibacterial activity.

In conclusion, an antibacterial substance which exhibited activity against the periodontal pathogen P. gingivalis was isolated and identified from L. plantarum 122 and L. fermentum ALAL020. It was hypothesized that the major antibacterial substance produced by L. plantarum 122 was lactic acid. The antibacterial substance produced by L. fermentum ALAL020 we purified was a novel low molecular substance active in neutral pH. This substance might be useful in dental treatment and for the prevention of periodontal diseases. A further structural analysis of this novel substance is necessary to elucidate its mechanisms before its use in the clinical setting.

This work was supported by the Central Institute for Health Science, A.L.A Corporation, and the Oral Microbiology and Periodontology department of Tsurumi University School of Dental Medicine. This work was funded by the Japan Society for the Promotion of Science: KAKENHI(C) (No.24593173 to T.O.). The authors declare no conflicts of interest associated with this manuscript.