Introduction

Bacteriocins are antibacterial ribosome-synthesized peptides or proteins produced by bacteria (Jack et al. 1995; Nishie et al. 2012). These substances are attractive for the growth control of bacteria to prevent foodborne diseases and for improving food processing technology such as extending the shelf life and as a replacement for synthetic preservatives (Gálvez et al. 2007). Numerous studies on bacteriocins were previously performed from various scientific and industrial viewpoints (Cleveland et al. 2001; Gálvez et al. 2007; Balciunas et al. 2013). In previous studies (Cotter et al. 2005; Nissen-Meyer et al. 2009; Nishie et al. 2012), bacteriocins were characterized on the basis of their structural properties and were mainly classified into two groups, Class I, the lantibiotics, and Class II, the non-lantibiotics, which were further divided into four subgroups. Among the subgroups, Class IId is composed of one-peptide non-pediocin-like linear bacteriocins and was found to be produced by several types of bacterium. From a practical perspective, a Class I bacteriocin, nisin A, has been studied intensively and it is now practically applied in food industries. It has been added to the European food additive list and approved by the Food and Drug Administration (FDA) (Cotter et al. 2005).

Terasi shrimp paste is a well-known traditional ingredient in Indonesian cuisine. This product is made from salted small shrimps through a complicated process of grounding, drying, and solidifying (Putro 1993; Kobayashi et al. 2003; Tanasupawat and Visessanguan 2014). Various types of traditional fermented food are widely distributed in Southeast Asia (Tanasupawat and Visessanguan 2014) and have been attractive sources for the screening of orally ingested useful bacteria such as bacteriocin-producing bacteria, because these types of fermented food and the bacteria isolated from them are widely known to be safe for human consumption for their safety to human health. In previous reports on fermented shrimp paste such as terasi, Bacillus amyloliquefaciens was isolated as a producer of a novel bacteriocin from the Thai shrimp paste kapi (Kaewklom et al. 2013). In addition, Bacillus subtilis and a Bacillus sp. were also isolated as producers of antibacterial substances from the Myanmar shrimp paste ngapi (Aung et al. 2004). These findings suggest that bacteriocin-producing bacteria are widely distributed in various types of fermented shrimp paste and these paste are expected to be attractive sources for the isolation of such bacteria. However, to the best of our knowledge, there have been no detailed reports available on the study of bacteriocin-producing bacteria isolated from terasi shrimp paste. With this in mind, in the present work, we examined bacteriocin-producing bacteria isolated from this product.

Materials and methods

Isolation of bacteriocin-producing bacteria

Thirteen terasi samples were purchased from a local factory near Semarang City, Central Java, in 2002 and 2003. They were homogenized thoroughly and then spread on MRS agar plates (Merck, Darmstadt, Germany) supplemented with 10 % NaCl. After aerobic incubation at 27 °C for 3 weeks, an agar medium [2 % mycin assay agar arei (Mikuni Chemical Industry Co., Ltd., Tokyo, Japan) with 1 % Bacto yeast extract (Becton, Dickinson and Company, Franklin Lakes, NJ, USA), pH 7.0] containing an overnight culture of Listeria monocytogenes IID 577 (IID: Laboratory Culture Collection, Institute of Medical Science, University of Tokyo, Tokyo, Japan) at the final concentration of approximately 107 cells ml−1 was overlaid on the upper layer, which was followed by incubation at 27 °C. After 2 days of incubation, a total of 27 putative bacteriocin-producing colonies that showed growth inhibition zones were isolated from six terasi samples, and then strain P2 was chosen as a representative strain in this study to evaluate the characteristics of the antibacterial substance produced.

Taxonomical identification of bacteriocin-producing bacterium

Strain P2 was taxonomically identified in accordance with previous reports (Kobayashi et al. 2003; Moe et al. 2015). Phenotypic characterization was performed using trypticase soy broth (TSB) and tryptic soy agar (Becton, Dickinson and Company) supplemented with 10 % NaCl at 27 °C as the basal culture condition. 16S rRNA gene sequencing analysis was carried out as described previously (Kobayashi et al. 2003). The 16S rRNA gene was amplified using eubacterial primers and a thermal cycler (PC707, ASTEC, Fukuoka, Japan). Then, amplified products were labeled using a BigDye Terminator Cycle Sequencing kit (Applied Biosystems, Foster City, CA, USA), and then sequence analysis was carried out using a capillary DNA sequencer (Model 310, Applied Biosystems).

Measurement of antibacterial activity

Antibacterial activity was determined by agar well diffusion assay using a double-layer agar plate as described previously (Moe et al. 2015). L. monocytogenes IID 577 was used as an indicator strain for routine experiments other than the determination of the bacteriocin spectrum. One-hundred-microliter samples were added to 8-mm-diameter wells in a double-layer agar plate, followed by incubation at 27 °C for 1 day. One arbitrary unit (AU) was defined as the reciprocal of the highest dilution that resulted in the observation of a growth inhibition zone of L. monocytogenes IID 577 around a well.

Bacteriocin purification

All of the bacteriocin purification steps were carried out at 4 °C. Strain P2 was cultured under shaking conditions (120 rpm) in 100 ml of TSB medium supplemented with 10 % NaCl in a 500 ml baffled flask. Culture in a total of 4000 ml of TSB was performed for 1 day at 27 °C, and then the supernatant obtained by centrifugation at 8000 rpm (10,600×g) was precipitated with ammonium sulfate at 80 % saturation. The precipitate was solubilized in 300 ml of 0.01 M Tris–HCl (pH 7.0). After dialysis against 0.01 M Tris–HCl (pH 7.0), internal fluids were subjected to CM-cellufine C-500 m cation exchange chromatography (2.6 × 50 cm, JNC, Tokyo, Japan) equilibrated in 0.1 M sodium acetate (pH 4.0) and eluted with a linear gradient from 0 to 1 M NaCl. The active fractions were concentrated by ultrafiltration, which was followed by further steps of gel filtration chromatography using a Sephacryl S-200 HR column (1.6 × 97.5 cm, GE Healthcare, Buckinghamshire, UK) with elution using 0.05 M NaCl–0.01 M Tris–HCl (pH 7.0). Finally, the active fractions were subjected to Mono S HR 5/5 cation exchange chromatography (0.5 × 5 cm, Pharmacia Biotech, Uppsala, Sweden) with 0.05 M NaCl–0.01 M Tris–HCl (pH 7.4) and eluted with a linear gradient from 0.05 to 0.5 M NaCl. The antibacterial activity of each fraction was measured by agar well diffusion assay using a double-layer agar plate as described above. In addition, absorbance at 280 nm was also measured using a spectrophotometer (UV-2400 PC; Shimadzu, Kyoto, Japan). The protein concentrations of the samples in all the purification steps were measured using a commercially available DC Protein Assay kit (Bio-Rad).

Effects of enzyme treatments on bacteriocin activity

Six types of enzyme, namely, proteinase K (Tritirachium album, Merck, Darmstadt, Germany), pepsin (porcine gastric mucosa, Wako, Tokyo, Japan), lysozyme (egg white, Wako), protease (Streptomyces griseus, Wako), papain (Carica papaya, Wako), catalase (bovine liver, Wako), and trypsin (porcine pancreas, Wako) were used. Each enzyme was mixed well with the purified bacteriocin (100 AU ml−1) at the final enzyme concentration of 1 mg ml−1 at 37 °C for 2 h. The antibacterial activity of the reactants was measured by agar well diffusion assay using a double-layer agar plate as described above.

Effects of heat and pH treatments on bacteriocin activity

The purified bacteriocin (50 AU ml−1) was subjected to various treatments. Heat treatment was performed at 60 °C for 30 min, 100 °C for 15 and 30 min, and 121 °C for 15 min. pH treatment was performed at pHs 3.0 and 11.0 at 4 °C overnight. Then, the samples were neutralized using small amounts of NaOH and HCl solutions and the remaining activity was determined by agar well diffusion assay using a double-layer agar plate as described above.

Bacteriocin spectrum

Nineteen types of indicator strain obtained from six culture collections were used. These strains were cultured in a double-layer medium with wells containing 100 µl (100 AU ml−1) of the purified bacteriocin. To examine their minimum inhibitory concentrations (MICs), the spot-on-lawn method was performed. Ten-microliter samples [400 AU ml−1 (228 μg ml−1)] diluted serially were spotted onto a double-layer agar plate as described above, followed by incubation at 27 °C for 1 day. The highest dilution that resulted in the observation of a growth inhibition spot was determined.

Sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE)

The purified bacteriocin was subjected to tricine-SDS-PAGE (Schägger and Von Jagow 1987) in 15 % gel with molecular weight markers (Low-Range Rainbow Marker, GE Healthcare), and Coomassie Brilliant Blue (CBB) staining (Ez stain aqua, ATTO, Tokyo, Japan) was performed after electrophoresis. After electrophoresis, half of the electrophoresis gel was washed well in sterilized distilled water and then placed on the double-layer medium containing L. monocytogenes IID 577.

Analysis of bacteriocin using mass spectrometer and protein sequencer

MALDI-TOF MS analysis was carried out using AXIMA-CFR Plus (Shimadzu, Kyoto, Japan) to determine accurately the molecular weight of the purified bacteriocin. Sequence analysis of the amino acid residues was performed using two analytical methods, namely, N-amino acid sequence analysis based on the Edman degradation method using a PPSQ-21A protein sequencer (Shimadzu) and MALDI-TOF MS analysis using an AXIMA Performance MALDI TOF mass spectrometer (Shimadzu) according to the matrix for the in-source decay (ISD) technique.

Effect of bacteriocin on the growth behavior of Listeria monocytogenes

To evaluate the mechanism underlying the antibacterial activity of the purified bacteriocin against L. monocytogenes IID 577, 96-well plastic plates and glass test tubes were used as described previously (Moe et al. 2015). The growth turbidity and viable cell count in TSB media, which were incubated at 27 °C, were determined using both optical density (O.D.) measurement at 630 nm and the spread plate technique.

For the monitoring of optical density, L. monocytogenes IID 577 cells were precultured in TSB medium for 1 day at 27 °C. After that, 45 μl of this preculture was transferred into 4.5 ml of TSB medium; then, 180 μl of this diluted culture was added immediately into the wells of a 96-well plate (=0 h in Fig. 4). The O.D. of the broth medium was measured over a period of 26 h using a plate reader (SpectraMax M3, Molecular Devices, California, USA). After 11 h of incubation, 20 µl of the purified bacteriocin (500 AU ml−1) or 0.05 M NaCl-0.01 M Tris–HCl buffer (pH 7.0) was added to these wells.

In addition, for the monitoring of viable cell count, L. monocytogenes IID 577 cells were precultured in TSB medium for 1 day at 27 °C. After that, a 1000-fold dilution of this preculture was prepared using physiological saline solution; then, 50 μl of this dilute preculture was transferred into 4.45 ml of TSB medium. Immediately, 4.5 ml of this TSB medium was mixed well with 500 μl of the purified bacteriocin (500 AU ml−1) or 0.05 M NaCl-0.01 M Tris–HCl buffer (pH 7.0) in glass test tubes (=0 day in Fig. 5) and then incubated for 7 days. After incubation for 1, 2, 4, and 7 days, the number of viable cells was determined using Listeria Selective Agar (Oxford Formulation) (Oxoid Limited, Basingstoke, UK).

Results

Taxonomic identification of strain P2

Strain P2 was identified as a gram-positive, spore-forming, and rod-shaped bacterium. It can grow only aerobically on plate media, and motility was observed in liquid media under a light microscope. It was found to be catalase-positive. The 16S rRNA gene sequence of this strain showed high similarity to those of members of the genus Virgibacillus registered previously on BLAST searches of the DNA Data Bank of Japan (DDBJ). Among them, it showed the highest similarity of 100 % (1489/1489) to Virgibacillus salexigens JCM 30552T (LC016572) and the lowest similarity of 95.41 % (1288/1350) to Virgibacillus arcticus JCM 14839T (EF675742).

Isolation and purification of bacteriocin

In the case of gel filtration chromatography, the substance with antibacterial activity was eluted in one portion. Additionally, in the case of two types of cation exchange chromatography, the substance with antibacterial activity was either retained or eluted in one portion by a salt gradient. The behaviors of the antibacterial substance in the final chromatography step, which was by Mono S HR 5/5 cation exchange chromatography, are shown in Fig. 1. In comparison with the activity of the culture supernatant, about 500-fold purification was achieved at the final purification step. In addition, the final recovery yield of the substance with antibacterial activity was estimated to be approximately 10 %.

Fig. 1
figure 1

Cation exchange chromatography of active fraction from gel filtration chromatography. Column, Mono S HR 5/5 (ϕ 0.5 × 5 cm); buffer, 0.01 M Tris–HCl-0.05 M NaCl (pH 7.4); gradient, 0.05–0.5 M sodium chloride (10–90 min); flow rate, 0.5 ml min−1; fraction volume, 0.5 ml; absorbance of each fraction at 280 nm (open diamonds); diameter of inhibition zone of each fraction against L. monocytogenes, excluding diameter of well (closed squares)

Full size image

Effects of enzymes, heat, and pH treatments on bacteriocin activity

After treatments with proteinase K, protease, and trypsin, no growth inhibition due to bacteriocin around the wells on the double-layer plate was observed. Its activity was partially affected by treatment with papain. However, treatments using catalase and lysozyme did not affect the diameter of the growth inhibition zone (data not shown). As shown in Table 1, it was stable after heat and pH treatments, and it retained at least approximately 60 % of its initial activity after autoclaving and high-pH treatment.

Table 1 Effect of heat and pH treatments on purified bacteriocin from strain P2
Full size table

Spectrum of antibacterial substance

The antibacterial effect of the bacteriocin on the growth of various bacteria is shown in Table 2. This substance was effective for only gram-positive bacteria including genus Bacillus, Enterococcus, Lactobacillus, Leuconostoc, Listeria, and Virgibacillus.

Table 2 Antibacterial activity spectrum of purified bacteriocin from strain P2
Full size table

SDS-PAGE

Photographs revealing the antibacterial activity of the bacteriocin on an SDS-PAGE gel are shown in Fig. 2. The purified bacteriocin showed single blue bands on a 15 % tricine-SDS-PAGE gel with the same molecular mass of approximately 3.5 kDa under both reducing and nonreducing conditions of sample preparation. Antibacterial activity on the washed gel was also observed at the locations corresponding to the single bands in the CBB-stained gel.

Fig. 2
figure 2

Tricine-SDS-PAGE analysis of the purified bacteriocin from strain P2. a Electrophoresis gel stained with CBB. b Growth inhibition zone of L. monocytogenes under the gel after electrophoresis. The photograph was taken after an overnight incubation at 27 °C. M molecular weight marker. Lane 1 The sample was electrophoresed after treatment under reducing condition. Lane 2 The sample was electrophoresed directly

Full size image

Molecular weight and amino acid sequence analysis

The mass spectra of the purified bacteriocin are shown in Fig. 3. The monoisotopic mass weight ([M+H]+) of this substance was determined to be 5318.4 Da. Although the amino acid residues of this bacteriocin could not be determined in an initial experiment using a protein sequencer, surprisingly, it was determined after idling for thirteen Edman degradation reaction cycles, and 14 residues were unexpectedly determined to be TVINWIKNGATFSW. On the other hand, 30 amino acid residues were determined to be A(Q/K)(I/L)GT(Q/K)AA(Q/K)WAWAN(Q/K)GTV(I/L)NW(I/L)(Q/K)NGATFSW by MALDI-TOF MS analysis.

Fig. 3
figure 3

Mass spectra of the purified bacteriocin from strain P2. Scan range: a mass-to-charge ratio, 1000–6000; b mass-to-charge ratio, 5200–5400

Full size image

Effect of bacteriocin on growth of L. monocytogenes

Changes in O.D. and the viable cell count of L. monocytogenes in liquid culture are shown in Figs. 4 and 5, respectively. The addition of bacteriocin inhibited the increase in the O.D. of the medium, and as compared with the reference medium, the difference in O.D. reached 0.21 at the final measurement time after 26 h of incubation. On the other hand, the viable cell count of L. monocytogenes in the medium with bacteriocin was markedly different from that without bacteriocin during 2 days of incubation. In particular, the difference reached approximately 100,000 times after 1 day of incubation.

Fig. 4
figure 4

Effect of purified bacteriocin from strain P2 on O.D. of broth culture of L. monocytogenes. Data are expressed as mean ± standard deviation (SD) from three independent experiments. Tris–HCl buffer (open circles) and bacteriocin (closed squares) were added at 10.5 h

Full size image
Fig. 5
figure 5

Effect of purified bacteriocin from strain P2 on viable cell count of L. monocytogenes in broth culture. Data are expressed as mean ± standard deviation (SD) from three independent experiments. In the case of data of only 2-day broth culture with bacteriocin, data are expressed as mean ± minimum and maximum counts from two independent experiments and an outlier was 3.8 × 102 CFU ml−1. Tris–HCl buffer added (open circles), bacteriocin added (closed squares)

Full size image

Discussion

The bacteriocin-producing bacterium strain P2 was identified phylogenetically as V. salexigens owing to its 100 % similarity in the 16S rRNA gene sequence. Furthermore, this phylogenetic identification result does not contradict the phenotypic characteristics of strain P2, because the members of the genus Virgibacillus were reported to include gram-positive motile rods and catalase-positive aerobic or facultative anaerobes, which have common characteristics with some exceptions (Sánchez-Porro et al. 2014).

Antibacterial activity was observed at the locations corresponding to the single bands on the washed CBB-stained gel. Therefore, these bands, which could be stained with CBB, were identified as having an antibacterial substance. In addition, the same migration distance of the CBB-stained band of the purified bacteriocin was determined after treatment under both reducing and nonreducing conditions. The remaining activity of the antibacterial substance after treatments with catabolic enzymes and the results of electrophoresis suggest that this substance could be regarded as proteinaceous without considering the presence of disulfide linkages. Consequently, although the molecular size of this bacteriocin was determined to be approximately 3.5 kDa by SDS-PAGE, it was rigorously estimated to be 5318.4 Da ([M+H]+) by MALDI-TOF MS analysis.

The purified bacteriocin from strain P2 showed antibacterial activity against a typical foodborne bacterium, namely, L. monocytogenes. Therefore, this finding suggests that this substance might be an attractive antibacterial food additive to control this foodborne bacterium. In addition, the O.D. of the culture medium did not decrease markedly after adding this bacteriocin, which suggests that this substance does not have bacteriolytic activity against L. monocytogenes. In this study, Listeria Selective Agar was used for counting the cells of L. monocytogenes; therefore, viable but nonculturable (VBNC) cells will not grow and the bacteriocin activity will be overestimated. Although this point must be considered, the viability of L. monocytogenes was markedly inhibited in a 1-day broth culture, and this finding suggests that this antibacterial substance did not maintain a sustainable antibacterial effect on L. monocytogenes. The reason for the lack of effectivity of the bacteriocin after long periods of incubation is not understood yet; the selection of resistant variants of L. monocytogenes was considered as one of the reasons.

To evaluate the detailed characteristics of this bacteriocin, its amino acid sequence was determined using both a MALDI-TOF mass spectrometer and an N-terminal protein sequencer based on the Edman degradation technique. Although MALDI-TOF MS analysis cannot determine the direction of the amino acid sequence of proteinaceous samples, the result obtained from N-terminal protein sequencing overlapped with the sequence result obtained from the MALDI-TOF MS analysis. Eventually, despite the fact that neither end of the peptide sequence could be determined, both analytical results led to the conclusion that this bacteriocin has a partial consensus sequence of A(Q/K)(I/L)GT(Q/K)AA(Q/K)WAWAN(Q/K)GTVINWIKNGATFSW. Although, in this study, the inner amino acid sequence was determined from the results of Edman degradation, the sequences of the N-terminal region could not be determined for some reason. As compared with the following information obtained from alignment analysis, the N-terminal of this bacteriocin might be blocked by formylmethionine as well as related bacteriocins.

To evaluate the mutual relationship between the 30 amino acid residues of our consensus bacteriocin sequence determined in this study and the sequences of other bacteriocins and related substances that have been registered in a protein sequence database, we performed alignment analysis using BLAST searches of DDBJ databases. As shown in Fig. 6, the amino acid sequence of this bacteriocin showed similarity to those of the representative members of Class IId bacteriocins. Lacticin Z (BAF75975), which has already been studied in detail by both chemical and genetic engineering analysis techniques (Iwatani et al. 2007), shows sequence similarity only in its central region. At least thirteen different residues between our sequence and that of lacticin Z were observed. Moreover, as compared with those of aureocin A53 (WP_032072954) from S. aureus (Netz et al. 2002), our sequence shares lower similarities and sixteen different residues were observed. In the previous studies on lacticin Z and aureocin A53, their N-terminals were found to be blocked by Edman degradation analysis, and the presence of an N-terminal formylmethionine residue was also reported (Netz et al. 2002; Iwatani et al. 2007). In contrast, the sequence of the bacteriocin we obtained and those of bacteriocins from V. massiliensis Vm-5 (CDQ37939) and a Bacillus sp. (WP_035390639) matched extremely well. In accordance with the recent development of comprehensive genomic analysis, various hypothetical bacteriocin sequences have been registered, and the above bacteriocin sequences (CDQ37939 and WP_035390639) were also considered to be hypothetical sequences. Therefore, these results indicate that the bacteriocin from strain P2 could not be classified as any of the previously reported substances except for the putative Class IId bacteriocin predicted from the draft genome sequence of spore-forming gram-positive bacteria.

Fig. 6
figure 6

Alignment of the amino acid sequences of purified bacteriocin from strain P2 and related substances. Upper sequence reference sequence; lower sequence partial sequence of bacteriocin from strain P2. a Lacticin Z from Lactococcus lactis (BAF75975). b Aureocin A53 from Staphylococcus aureus (WP_032072954). c Putative bacteriocin from Virgibacillus sp. Vm-5 (CDQ37939). d Putative bacteriocin from Bacillus sp. (WP_035390639)

Full size image

There have been several reports on antibacterial-substance-producing bacteria isolated from shrimp paste distributed in Southeast Asian countries. Although detailed information such as the amino acid sequences of the antibacterial substances has not been found, Bacillus amyloliquefaciens strain SP-1-13LM was reported as a bacteriocin producer in a study on kapi, a Thai shrimp paste (Kaewklom et al. 2013). Its bacteriocin is called amysin, which was estimated to have a molecular mass of 5.2 kDa. Additionally, the antibacterial activity of an ethyl acetate extract from a liquid culture of Bacillus species isolated from ngapi, a Myanmar shrimp paste, was also reported (Aung et al. 2004). The properties of the bacteriocin from strain P2 were different from those of these antibacterial substances in terms of the broad spectrum of its antibacterial activity, which includes gram-negative bacteria.

In conclusion, various studies on antibacterial substances such as bacteriocins produced by isolates from traditional fermented shrimp paste products in Southeast Asian countries have already been reported; however, detailed information is limited at present. Therefore, to the best of our knowledge, this is the first detailed study showing a bacteriocin produced by a bacterial isolate from not only terasi shrimp paste but also other fermented shrimp pastes distributed in Southeast Asia.