Isolation and characterization of bacteriocin-producing bacteria from the intestinal microbiota of elderly Irish subjects

Authors


Correspondence

Paul Ross, Teagasc Food Research Centre, Moorepark, Fermoy, Co. Cork, Ireland. E-mail: paul.ross@teagasc.ie

Abstract

Aims

To isolate and characterize bacteriocins produced by predominant species of lactic acid bacteria from faeces of elderly subjects.

Methods and Results

Screening over 70 000 colonies, from faecal samples collected from 266 subjects, using the indicator organisms Lactobacillus bulgaricus LMG 6901 and Listeria innocua DPC 3572, identified 55 antimicrobial-producing bacteria. Genomic fingerprinting following ApaI digestion revealed 15 distinct strains. The antimicrobial activities associated with 13 of the 15 strains were sensitive to protease treatment. The predominant antimicrobial-producing species were identified as Lactobacillus salivarius, Lactobacillus gasseri, Lactobacillus acidophilus, Lactobacillus crispatus and Enterococcus spp. A number of previously characterized bacteriocins, including ABP-118 and salivaricin B (from Lact. salivarius), enterocin B (Enterococcus faecium), lactacin B (Lact. acidophilus), gassericin T and a variant of gassericin A (Lact. gasseri), were identified. Interestingly, two antimicrobial-producing species, not generally associated with intestinally derived microorganisms were also isolated: Lactococcus lactis producing nisin Z and Streptococcus mutans producing mutacin II.

Conclusion

These data suggest that bacteriocin production by intestinal isolates against our chosen targets under the screening conditions used was not frequent (0·08%).

Significance and Impact of the Study

The results presented are important due to growing evidence indicating bacteriocin production as a potential probiotic trait by virtue of strain dominance and/or pathogen inhibition in the mammalian intestine.

Introduction

Bacteriocins produced by lactic acid bacteria (LAB) are ribosomally synthesized peptides, which generally have inhibitory action against closely related organisms and to which the producing strain is immune (Cotter et al. 2005; De Vuyst and Vandamme 1994). A number of bacteriocins produced by LAB have potential as novel antimicrobial agents in various practical applications from food to medicine (De Martinis et al. 2002; Joerger 2003; Klaenhammer et al. 2005; O'Sullivan et al. 2002; Todorov and Dicks 2007). Many LAB have been proven to function as probiotics, which are of benefit to human health when ingested in sufficient quantities. The isolation of potential bacteriocin-producing strains from the intestinal microbiota (comprising both Gram-negative and Gram-positive bacteria) has been previously reported (Gordon and O'Brien 2006; O'Shea et al. 2009). A diverse range of Gram-positive bacteriocin producers including lactobacilli, enterococci and streptococci were isolated in the latter study, with a high incidence (up to 70%) of repeated isolation for some bacteriocin producers. With respect to Gram-negative bacteria, the frequency of production of bacteriocins (such as colicins and microcins) by the Escherichia coli population component of the microbiota can vary from 10 to 80% (Gordon and O'Brien 2006).

Bacteriocin production may confer a significant advantage to the producing strain by allowing it to dominate complex microbial populations (O'Shea et al. 2009). The added advantage of using bacteriocin-producing probiotics over bacteriocins themselves is that in situ production potentially overcomes the adverse effect of proteolysis of the released peptide during gastric transit. A number of studies have demonstrated the potential use of bacteriocins in the control of important gastric pathogens including Salmonella spp. (Casey et al. 2004), Campylobacter jejuni (Stern et al. 2006; Svetoch et al. 2011), Listeria monocytogenes (Allende et al. 2007; Corr et al. 2007), E. coli O157:H7 (Brashears et al. 2003) and Clostridium difficile (Rea et al. 2007). With regard to the latter, Rea et al. demonstrated the potential of a narrow-spectrum bacteriocin to control Cl. difficile in a human colon model, without causing collateral damage to the surrounding microbiota (Rea et al. 2011).

The intestinal microbiota is considered a rich source of potential probiotic bacteria, and there is an increasing interest in the isolation of new bacteriocin-producing strains of human origin. Indeed, intestinal source has been reported as one of the primary criteria in selecting probiotic strains, as these strains may function better in an environment similar to its origin (Saarela et al. 2000). While there has been considerable research on LAB bacteriocins to date, there are few studies on the screening for bacteriocin-producing LAB from the human intestinal tract (Birri et al. 2010; Flynn et al. 2002; Verdenelli et al. 2009). The increasing interest in the isolation of bacteriocin-producing LAB of human intestinal origin that could be evaluated for probiotic effects was one of the motivating factors for this research.

The ELDERMET consortium was established in 2008 (http://eldermet.ucc.ie) and aims to determine the intestinal microbiota of 500 elderly (>65 year) Irish people. To date, this project has elucidated links between microbiota composition, diet and health (Claesson et al. 2011, 2012) and has generated a large collection of intestinal isolates (including bifidobacteria and lactobacilli), suitable for screening for use as probiotics for the elderly population. The aim of this study was to isolate and characterize bacteriocin-producing isolates from the ELDERMET biobank that may have potential applications for improving gut health in the elderly. Bacteriocin-like substances were characterized based on their spectrum of inhibition and identified by DNA sequencing and matrix-assisted laser desorption ionization time of flight (MALDI-TOF) mass spectrometry (MS) with a view to determining their potential for probiotic applications.

Materials and methods

Subject recruitment and sample collection

Subjects aged 65 years and older were recruited and examined at ELDERMET Clinics at two local hospitals (Cork University Hospital and St Finbarr's Hospital, Cork) as previously outlined (Claesson et al. 2011). The subjects included patients in long-term institutional care, rehabilitation centres, outpatient day hospitals, people being treated with antibiotics and a group of comparatively healthy elderly individuals. A total of 410 faecal samples from 266 elderly subjects were assessed. In addition to the initial sample (T0), faecal samples were also collected from some subjects at 3 (T3) and 6 months (T6). Stool samples were collected in a sterile container, stored at 5°C until transported on ice to the laboratory.

Bacterial strains, media and growth conditions

LAB strains were grown anaerobically at 37°C for 24–48 h on de Man, Rogosa, Sharpe (MRS; Difco Laboratories, Detroit, MI, USA) agar supplemented with 0·05% (w/v) L-cysteine hydrochloride (Sigma Chemical Co., St Louis, MO, USA) (denoted mMRS agar). All other bacterial strains were cultivated in brain–heart infusion (BHI) medium and were grown aerobically at 37°C. Where solid media were required, 1·5% agar (w/v) (Oxoid Ltd, Basingstoke, Hampshire, UK) was added, and for softer overlay medium, 0·8% (w/v) agar was added.

Enumeration of culturable bacteriocin-producing strains from faecal samples

One gram of the faecal sample was mixed with 9 ml maximum recovery diluent (MRD; Oxoid) in a Seward stomacher bag and subsequently diluted 10-fold as outlined previously (O'Sullivan et al. 2011). To enumerate total culturable LAB strains, the dilutions were spread plated (100 μl) onto mMRS agar. Agar plates were incubated anaerobically at 37°C for 48 h. Bacterial counts were recorded as colony forming units (CFU) per gram of faeces, and counts were expressed as log CFU g−1 faeces. These plates were overlaid with the indicator strains and were incubated at 37°C for 24 h. The indicator strains used were Lactobacillus delbrueckii subsp. bulgaricus LMG 6901 (for all samples) (BCCM/LMG Culture Collection, Gent, Belgium), Bifidobacterium breve NCIMB 8807 (for 228 samples) (National Collection of Industrial, Marine and Food Bacteria, Aberdeen, UK), Bifidobacterium lactis Bb12 (for 30 samples) (Chr. Hansen, Little Island, Cork, Ireland), Escherichia coli O157:H7 NCTC 12900 (for 11 samples) (National Collection of Type Cultures, Central Public Health Laboratory, London, UK) and Listeria innocua DPC 3572 (for 123 samples) (Teagasc Food Research Centre, Moorepark, Fermoy, Co. Cork, Ireland). Bacterial colonies generating a zone of inhibition against the indicator strain on the agar plates were selected and cultured in mMRS broth for 24 h at 37°C. These isolates were stocked in the ELDERMET Culture Collection for further characterization.

Antimicrobial activity assay

The antimicrobial activity of the cell-free supernatant (CFS) of the bacteriocin-producing strains was determined by the agar well diffusion method (Ryan et al. 1996). To eliminate the possibility of a zone of inhibition due to acid production, all culture supernatants were adjusted to pH 6·5 with 1 N NaOH and filter-sterilized (0·45 μm). Antimicrobial activity was assessed against a selection of Gram-positive and Gram-negative strains (Lact. bulgaricus LMG 6901, L. innocua DPC 3572, Listeria monocytogenes DPC 3785, Staphylococcus aureus DPC 5245, Salmonella typhi DPC 6452, Salm. typhi DPC 6046, E. coli DPC 6239 and Cronobacter sakazakii DPC 6440). All agar plates were examined for a zone of inhibition following overnight incubation.

Influence of heat treatment and enzymes on antimicrobial activity

To evaluate the heat stability of the bacteriocins, the neutralized supernatants from the bacteriocin-producing strains were heated to 100°C for 30 min. The untreated bacteriocin was used as a positive control. All culture supernatants were adjusted to pH 6·5 with 1 N NaOH and were mixed with an equal volume of each of the enzymes proteinase K, pepsin and catalase (25 mg ml−1; Sigma) and incubated at 37°C for 3 h. Following enzyme and heat treatment, the residual antimicrobial activity in the treated supernatant was assessed using the agar well diffusion assay with the indicator strain Lact. bulgaricus LMG 6901, in comparison with the untreated sample.

Genetic characterization of bacterial isolates

Genomic DNA was extracted as outlined previously (Coakley et al. 1996), and RNase (0·75 μl; Sigma) was added to the extracted DNA, incubated for 1 h at 37°C and then denatured at 65°C for 15 min. Bacterial isolates were identified by amplification of the 16S rRNA gene using the 16S eubacterial primers CO1 and CO2 (Table 1; Simpson et al. 2003), and the complete sequence of the 16S rRNA gene was determined by Sanger sequencing (Beckman Coulter, Essex, UK). The species was determined by nucleotide alignments (>98%) with deposited species in the NCBI database (http://blast.ncbi.nlm.nih.gov) and using the Ribosomal Database Project (RDP) (http://rdp.cme.msu.edu/). For Enterococcus isolates, further PCR analysis was used to distinguish the species present using previously published species-specific primers (Jackson et al. 2004). Primers specific to structural genes of known bacteriocins were also used (Table 1). All PCRs were performed in a G-Storm PCR machine (Gene Technologies, Essex, UK) and PCR products sequenced by Beckman Coulter Genomics. Sequences were aligned to the structural bacteriocin genes using the Lasergene 6 software program and subsequently analysed by BLASTP analysis (http://blast.ncbi.nlm.nih.gov/Blast.cgi).

Table 1. Primers used in this study
PrimersSequence (5′–3′)Annealing temp. (°C)Amplicon size (bp)References

CO1-Forward

CO2-Reverse

AGTTTGATCCTGGCTCAG

TACCTTGTTAC GACT

501500Simpson et al. (2003)

118α-Forward

118im-Reverse

ATGATGAAGGAATTTACA G

CCACGCTCTCACATAAC

50700Barrett et al. (2007)

Salivaricin B-Forward

Salivaricin B-Reverse

AGCGGACAGTTTGGTTATGA

TTCAGCACATAGAACACATCCTC

55900O'Shea et al. (2011)

Lactacin B-Forward

Lactacin B-Reverse

TTGAGTTATGTAATTGGTGGAG

CTGCTGAGCCTTTGATAACG

60181Tabasco et al. (2009)

Acidocin B-Forward

Acidocin B-Reverse

GTTGGAATAGTATGGTTTTCG

ATTCATATCCGGCACCAACG

58700This study

Gassericin A-Forward

Gassericin A-Reverse

CACTCGTAATGGATTTCCTA

GCCTTT TTTGAATACTCTATTC

55350Kawai et al. (2001)

Gassericin T-Forward

Gassericin T-Reverse

TGGATTTAAATTGCCTGAAAC

CATTCCCCCACTTGTTTCC

58645This study

Nisin Z-Forward

Nisin Z-Reverse

AACGGCTCTGATTAAATTC

CTATTTAGCAACCCTAAATAAC

50215This study

Enterocin B-Forward

Enterocin B-Reverse

GAAAATGATCACAGAATGCCTA

GTTGCATTTAGAGTATACATTTG

58159Toit et al. (2000)

Enterocin 62-6-Forward

Enterocin 62-6-Reverse

GTGGAAAGCTAGTATTTGCAAC

AGCGTTAAGCCGAATGTTTACAC

49511Dezwaan et al. (2007)

Mutacin II-Forward

Mutacin II-Reverse

AACGCAGTAGTTTCTTTGAA

TTCCGGTAAGTACATAGTGC

52444Kamiya et al. (2005)

Enterolysin A-Forward

Enterolysin A-Reverse

GGACAACAATTCGGGAACACT

GCCAAGTAAAGGTAGAATAAA

551007Nigutova et al. (2007)

Pulsed-field gel electrophoresis (PFGE) of bacterial isolates

Pulsed-field gel electrophoresis of isolates demonstrating antimicrobial activity in the CFS was performed as previously described (Simpson et al. 2002) using the ApaI restriction endonuclease. A low-range molecular weight DNA marker (9·42–194·0 Kb; New England Biolabs, Beverly, MA, USA) was used to determine the band size. Electrophoresis was performed using the Bio-Rad Chef-DR II instrument (Bio-Rad Laboratories, Richmond, CA, USA). For each PFGE pattern obtained, one strain was selected for further characterization.

Purification and identification of bacteriocins

Purification of antimicrobial peptides was performed by reverse-phase high-performance liquid chromatography (RP-HPLC). The bacteriocins were purified from an overnight culture of the producing strain in mMRS broth. Bacterial cells were harvested by centrifugation and resuspended in 70% (v/v) propan-2-ol and 0·1% (v/v) trifluoroacetic acid (TFA) and mixed at room temperature for 3–4 h. Following removal of the propan-2-ol by rotary evaporation, 20 ml of the sample was applied to a 5 g STRATA C18-E SPE column (Phenomenex, Cheshire, UK), which was pre-equilibrated with methanol and water. The column was washed with 20 ml 40% (v/v) ethanol, and the bacteriocin activity was eluted with 70% (w/v) propan-2-ol. Following removal of the propan-2-ol from the preparation, aliquots were applied to a semipreparative Proteo Jupiter (4 u Proteo, 90 Å, 250 × 10 mm) RP-HPLC column, previously equilibrated with 30% propan-2-ol containing 0·1% TFA. The column was subsequently developed in a gradient of 30% (v/v) propan-2-ol, containing 0·1% (v/v) TFA to 60% (v/v) propan-2-ol, containing 0·1% (v/v) TFA over 40 min. Bioactive fractions were collected at 1-min intervals, and bacteriocin activity was monitored by the well diffusion assay. For some strains, the supernatant of the overnight culture was used to determine the bacteriocin activity. MALDI-TOF mass spectrometry was performed on HPLC fractions as described previously (Cotter et al. 2006).

Results

Screening and isolation of bacteriocin-producing bacteria

Screening of over 70 000 colonies from 410 faecal samples (from 266 elderly subjects) resulted in the detection of 276 colonies exhibiting antimicrobial activity against Lactobacillus bulgaricus LMG 6901 (273 colonies) and Listeria innocua DPC 3572 (three colonies), representing an isolation frequency of 0·4% in the initial screening. Despite exhaustive parallel screening, colonies showing inhibitory activity against Bifidobacterium breve NCIMB 8807, Bif. lactis Bb12 and Escherichia coli strains were not detected in this study. The lack of activity against the indicator strain E. coli was not surprising due to the protective nature of the cell membrane in Gram-negative bacteria against the action of antimicrobial compounds (Gao et al. 1999). In addition, the culture conditions in this study focused mainly on the isolation of LAB, which generally kill or inhibit closely related Gram-positive strains (Klaenhammer 1988; Riley and Wertz 2002).

The 276 inhibitory isolates were subsequently reduced in number to 55 isolates (from 32 subjects) following neutralization of the CFS, representing a frequency of successful probable bacteriocin isolation of 0·08%. Further analysis of the 55 isolates by PFGE resulted in 15 individual strains (from 14 subjects) with distinct PFGE profiles (Fig. 1), corresponding to a probable bacteriocin detection frequency of 0·02%. Strains with similar PFGE patterns were isolated from multiple subjects, as shown in Table 2. A high incidence of repeated isolation was observed for Lactobacillus gasseri (12), Lactobacillus crispatus (15) and Enterococcus durans (6). Strains with identical PFGE patterns were confirmed to produce the same/similar bacteriocin by cross-sensitivity assays, as each isolate was cross-immune to the antimicrobial compound produced by the other (data not shown).

Table 2. Diversity of bacteriocin-producing strains among individual elderly subjects
StrainsNumber of isolatesNumber of subjectsNumber of pulsotypesPulsed-field gel electrophoresis profile
Lactobacillus gasseri 1263A, B & C
Lactobacillus salivarius 534D, E, F & G
Lactobacillus acidophilus 971H
Enterococcus faecium 422I & J
Lactococcus lactis 321K
Streptococcus mutans 111L
Lactobacillus crispatus 1572M & N
Enterococcus durans 641O
Figure 1.

Grouping of the macrorestriction patterns of the bacteriocin-producing strains generated 15 pulsed-field gel electrophoresis (PFGE) pulsotypes termed A to O, on digestion of chromosomal DNA with ApaI restriction enzyme [Lane 1 is the marker (Ma)].

Species identification of bacterial strains

The 16S rDNA gene sequences of the isolates were determined and compared with the available 16S rDNA sequences in the NCBI database. One of the most predominant species isolated in this study was Lact. gasseri. Other organisms identified included Lactobacillus salivarius, Lactobacillus acidophilus, Streptococcus mutans, Lactococcus lactis and Enterococcus spp. (Table 3). In addition to 16S rDNA gene sequencing, the identified enterococcal isolates were further typed using species-specific PCR primer sets for sequence variations in manganese-dependent superoxide dismutase genes of Enterococcus, as previously described (Jackson et al. 2004). Based on this method, two of the enterococcal strains (EM342-BC-1 and EM297-BC-T6-1) were determined to be Enterococcus faecium and EM133-BC-1 to be Ent. durans.

Table 3. Predicted bacteriocins and their enzyme sensitivity
Strains identified through 16S rDNA sequencingStrain IDPulsed-field gel electrophoresis profilePredicted bacteriocinAmino acid identity (%)aProteinase KPepsin
  1. ND, not determined; +: sensitive to enzyme; −: resistant to enzyme.

  2. a

    Amino acid alignments of the sequenced structural genes of strains identified in this study with structural genes from the characterized bacteriocins (NCBI).

Lactobacillus gasseri EM081-BC-T3-1AGassericin T99++
EM301-BC-T3-1BAcidocin B100++
EM315-BC-T6-1CAcidocin B100++
Lactobacillus salivarius EM100-BC-T3-1DABP-118 & Salivaricin B100++
EM253-BC-T6-2EABP-118 & Salivaricin B100++
EM347-BC-T3-1FABP-118 & Salivaricin B100++
EM351-BC-5GABP-118 & Salivaricin B100++
Lactobacillus acidophilus EM066-BC-T3-3HLactacin B98++
Enterococcus faecium EM342-BC-1IEnterocin B100++
EM297-BC-T6-1JEnterocin 62-6 & Enterocin M100++
Lactococcus lactis EM089-BC-T6-1KNisin Z100++
Streptococcus mutans EM315-BC-T3-2LMutacin II100++
Lactobacillus crispatus EM047-BC-T6-4MNDND
EM200-BC-T3-1NNDND
Enterococcus durans EM133-BC-1ONDND++

Inhibitory spectrum

The 15 antimicrobial-producing strains with distinct PFGE patterns were selected for further characterization. The inhibitory spectra of the strains using a range of indicator organisms were investigated and revealed a strain-dependent variation (Table 4). All the strains exhibited zones of inhibition against Lact. bulgaricus and other strains including Lact. salivarius, L. lactis and Enterococcus spp. also exhibited relatively wide spectra of activity, inhibiting L. innocua and Staphylococcus aureus. As expected, none of the strains tested exhibited antimicrobial activity against the Gram-negative indicator organisms, including strains of Salmonella typhi, Cronobacter sakazakii or E. coli.

Table 4. Inhibitory spectrum of ELDERMET isolates
ELDERMET isolatesIndicator organisms
Lactobacillus bulgaricus LMG 6901Listeria monocytogenes DPC 3785Listeria innocua DPC 3572Staphylo-coccus aureus DPC 5245
  1. +: strains exhibiting antimicrobial activity against indicator organism; −: strains not exhibiting antimicrobial activity against indicator organisms. No antimicrobial activity was observed against Salmonella typhi, Cronobacter sakazakii and Escherichia coli.

Lactobacillus gasseri +
Lactobacillus crispatus +
Lactobacillus acidophilus +
Lactobacillus salivarius ++++
Enterococcus durans +
Enterococcus faecium +++
Lactococcus lactis ++++
Streptococcus mutans +

Enzyme and heat sensitivity

The isolated crude bacteriocins exhibited stability at 100°C for 30 min, with the exception of bacteriocins from the Lact. crispatus strains. None of the inhibitory zones were affected by catalase, indicating that hydrogen peroxide was not involved in the inhibitory action. In contrast, the antimicrobial activity associated with 13 of the 15 strains was found to be sensitive to proteinase K, indicating the presence of antimicrobial compound of a protein nature in these cases (Table 3). The two Lact. crispatus strains were again exceptions, as their culture supernatants demonstrated antimicrobial activity even after proteinase K treatment.

Identification and purification of bacteriocins

Following species identification of the bacteriocin-producing isolates, primers specific to known bacteriocin structural genes from the isolated bacterial species were employed in test amplifications (Table 1). PCR analysis and subsequent DNA sequencing were used as an initial method to identify the putative bacteriocin structural genes present in each strain. This resulted in identification of ten different putative bacteriocins from the selected 15 isolates (Table 3). In parallel, the bacteriocins produced by these strains were purified by RP-HPLC, and the molecular mass of the peptides was determined using MALDI-TOF MS. Precise matching of these peptide masses to the theoretical masses was taken as evidence that the identified bacteriocin structural genes encoded the active bacteriocin. Certain bacteriocin producers were selected for further characterization based on either their unusual presence in the human intestinal microbiota (L. lactis EM089-BC-T6-1 and Strep. mutans EM315-BC-T3-2) and on their broad inhibitory spectra (Lact. salivarius EM100-BC-T3-1 and EM253-BC-T6-2) or due to the production of multiple bacteriocins (Ent. faecium EM297-BC-T6-1).

PCR and amplicon sequence analysis of Lact. salivarius EM100-BC-T3-1 and EM253-BC-T6-2 confirmed the presence of the three bacteriocin structural genes; these encoded salivaricin B and both peptides of the ABP-118 bacteriocin (α and β) in all strains. MALDI-TOF MS of the pure peptides, following RP-HPLC purification, confirmed molecular masses of 4096 and 4332 Da (Fig. 2a,b), which correspond to the previously characterized ABP-118α and ABP-118β peptides of the ABP-118 bacteriocin, respectively (Flynn et al. 2002). This was in accordance with the PCR and sequence analysis, which indicated the presence of the structural genes (100% identity) for this two-peptide bacteriocin within the strain. In addition, MS analysis also confirmed the presence of another previously characterized peptide, salivaricin B (Cataloluk 2001), showing a corresponding mass of 4433 Da (Fig. 2c). However, this peptide neither inhibited the indicator strain Lact. bulgaricus nor displayed antilisterial activity, which confirmed recent findings in our laboratory.

Figure 2.

Reverse-phase high-performance liquid chromatography (RP-HPLC) chromatograms and matrix-assisted laser desorption ionization time of flight (MALDI-TOF) mass spectrometry (MS) data (inserted panel) of (a) S1n1 produced by Lactobacillus salivarius EM100-BC-T3-1, (b) S1n2 produced by Lact. salivarius EM100-BC-T3-1, (c) salivaricin B produced by Lact. salivarius EM100-BC-T3-1, (d) nisin Z produced by Lactococcus lactis EM089-BC-T6-1, (e) mutacin II produced by Streptococcus mutans EM315-BC-T3-2, (f) enterocin M produced by Enterococcus faecium EM297-BC-T6-1, (g) enterocin 62-6 produced by Ent. faecium EM297-BC-T6-1 and (h) a variant of gassericin A produced by EM301-BC-T3-1. Arrows indicate location of the antimicrobial peptides.

PCR analysis and DNA sequencing data for Ent. faecium EM342-BC-1 confirmed the presence of the enterocin B gene, a previously characterized class IIc bacteriocin (Casaus et al. 1997). Interestingly, another Ent. faecium strain EM297-BC-T6-1 was confirmed to harbour the structural genes for a recently identified bacteriocin, enterocin M (Marekova et al. 2007), and also the genes for the two-component bacteriocin enterocin 62-6, which is closely related to enterocins L50A and L50B (Dezwaan et al. 2007). MS analysis of the bacteriocins from Ent. faecium EM297-BC-T6-1 identified proteins of two different masses corresponding to two putative bacteriocins. A peptide of 4628 Da was identified corresponding to enterocin M, which is an antilisterial class IIa pediocin-like bacteriocin (Fig. 2f), and a second bacteriocin consisted of two peptides (5206 and 5219 Da), corresponding to the two-peptide class IIc bacteriocins produced by a vaginal isolate of Ent. faecium 62-6 (Fig. 2g).

The sequence of the amplified PCR product from the L. lactis isolates revealed 100% amino acid identity to the translated structural gene of the well-characterized lantibiotic, nisin Z (Mulders et al. 1991), and was further investigated. MS analysis of bioactive RP-HPLC fractions purified from L. lactis EM089-BC-T6-1 revealed a peptide with a molecular mass of 3332 Da, which corresponds exactly to the peptide mass value expected for the nisin variant, nisin Z (Fig. 2d). 16S rDNA sequencing identified the ELDERMET isolate EM315-BC-T3-2 as Strepmutans. PCR and sequence analysis indicated the presence of the structural gene for the bacteriocin, mutacin II, whose expression was verified by MS analysis of the bacteriocin. The purified peptide (3244 Da) from this isolate corresponded to the mass of the lanthionine-containing mutacin peptide, mutacin II (Novak et al. 1994; Fig. 2e).

In this study, three Lact. gasseri strains were shown to produce bacteriocins: Lact. gasseri EM081-BC-T3-1, EM315-BC-T6-1 and EM301-BC-T3-1 (Table 3). Of the three strains, Lact. gasseri EM081-BC-T3-1 generated the expected-sized PCR product when amplified with gassericin T primers, which was confirmed by DNA sequencing. The other two strains were initially screened for the presence of the acidocin B and gassericin A structural genes using specific primers (Table 1). Due to the high similarity (>95% amino acid identity) between these two bacteriocins (Kawai et al. 1998b), both PCRs successfully generated a product in all strains. However, subsequent sequence analysis confirmed the presence of a structural gene with 98% amino acid identity to both acidocin B and gassericin A, indicating a new variant of this family of bacteriocins (Fig. 3). MS analysis of the purified antimicrobial peptides revealed molecular mass of 5683 Da, inferring the protein to be a variant of gassericin A, a previously characterized class IIc bacteriocin (Kawai et al. 1998b; Fig. 2h). PCR amplification of DNA from Lact. acidophilus EM066-BC-T3-3 confirmed the presence of the structural gene for lactacin B, and the protein alignments indicated 100% amino acid identity with the lactacin B structural gene product, identified previously in Lact. acidophilus La-5 (Tabasco et al. 2009). The strain isolated from this study, demonstrated a narrow inhibitory spectrum, inhibiting only Lact. bulgaricus, which is similar to the lactacin B bacteriocin (Barefoot and Klaenhammer 1983).

Figure 3.

Comparison of amino acid sequences of gassericin A (mature peptide) and variant produced by Lactobacillus gasseri EM301-BC-T3-1 and Lact. gasseri EM315-BC-T6-1. Amino acid differences are bold and underlined.

Discussion

As intestinal origin is generally considered a desirable attribute for probiotic strains for human consumption, this screening study was undertaken to isolate human intestinally derived bacteria with antimicrobial activity. The availability of a large number of faecal samples from older subjects recruited in the ELDERMET study enabled a comprehensive screening for such strains specifically associated with the intestinal niche of older subjects of different health states and resulted in the successful isolation of a broad range of bacteriocin producers of human intestinal origin. The initial activity screening of 70 000 colonies against Lactobacillus bulgaricus LMG 6901 resulted in the isolation of 273 isolates (representing an isolation frequency of 0·4%), which were subsequently reduced to 15 distinct strains through further characterization.

The bacteriocinogenic isolates were identified from the total bacterial count (on mMRS agar incubated anaerobically) ranging between log 106 to log 109 CFU per gram faeces. The isolation frequency of the bacteriocinogenic isolates was found to be low than expected (0·4% from total culturable bacteria) and did not represent a major part of bacteria culturable by the above methods. Due to the high number of colonies screened, the number of indicator organisms was restricted in this study. In addition, it does have to be noted that conventional culture methods can limit a screening study to the ‘easy to culture’ organisms from the intestine. However, it has also been widely documented that the elderly intestinal microbiota differs from younger adults (Biagi et al. 2011; Claesson et al. 2011) and that the compromised stability of the intestinal microbiota in the elderly (Biagi et al. 2012; Mueller et al. 2006) may have resulted in the lower isolation of bacteriocin-producing strains.

The majority of bacteriocin producers isolated in this study were Lactobacillus spp., with Lactobacillus gasseri being the predominant species isolated. This is in agreement with previous studies that consider Lact. gasseri to be one of the most commonly detected LAB in the human intestinal tract (Reuter 2001; Wall et al. 2007). Other bacteriocin producers identified included Lactobacillus crispatus, a common vaginal isolate (Pavlova et al. 2002), Lactobacillus salivarius, Enterococcus spp., Streptococcus mutans and Lactococcus lactis. All the isolated bacteriocins exhibited stability at 100°C for 30 min and were sensitive to proteases with the exception of bacteriocins from the Lact. crispatus strains. Further analysis is required to determine the nature of the antimicrobial compound produced by these strains.

Certain bacteriocin producers were selected for further characterization, including L. lactis EM089-BC-T6-1, Lact. salivarius EM100-BC-T3-1, Lact. salivarius EM253-BC-T6-2, Enterococcus faecium EM297-BC-T6-1 and Strep. mutans EM315-BC-T3-2. All four of the Lact. salivarius isolates (Table 3) from different ELDERMET subjects demonstrated inhibitory activity against Lact. bulgaricus LMG 6901, Listeria innocua DPC 3572 and the pathogenic Listeria monocytogenes DPC 3785 (Table 4). The presence of the three peptides ABP-118α, ABP-118β and salivaricin B were confirmed in these Lact. salivarius isolates. ABP-118 is a well-characterized two-component class II heat-stable bacteriocin, produced by Lact. salivarius subsp. salivarius UCC118 (Flynn et al. 2002) that inhibits the food-borne pathogen L. monocytogenes (Corr et al. 2007). In contrast to the ABP-118α and ABP-118β peptides, the salivaricin B did not exhibit antilisterial activity, which is in agreement with a recent study concerning another Lact. salivarius strain (O'Shea et al. 2011). These results demonstrate that the production of a two-component class II bacteriocin may be a common feature among intestinal Lact. salivarius strains.

Bacteriocin-producing species not commonly isolated from the human intestine, including L. lactis and Strep. mutans, were identified in this screening study. The bacteriocin produced by L. lactis EM089-BC-T6-1 (nisin Z) exhibited broad-spectrum inhibitory activity against the indicator strains Lact. bulgaricus LMG 6901, L. innocua DPC 3572, L. monocytogenes DPC 3785 and Staphylococcus aureus DPC 5245. Although it is well documented that nisin is antilisterial (Cai et al. 1997), the first human-derived nisin reported previously (Millette et al. 2007) did not show activity against Listeria spp. and was sensitive to temperatures >70°C. This outcome was explained by the fact that the nisin concentration produced by the strain was insufficient to inhibit the food pathogen (Millette et al. 2008). Further studies on this human-derived nisin Z from L. lactis EM089-BC-T6-1 isolate are necessary to determine its efficacy. Additionally, further characterization of bacteriocin production by the Lact. crispatus and Enterococcus durans isolated in this study is also required, as limited information is available on bacteriocin production by these strains (Hu et al. 2008; Tahara and Kanatani 1997; Yanagida et al. 2005).

Streptococcus mutans is one of the principal aetiological agents of dental caries (Kamiya et al. 2005), and although the presence of Strep. mutans in the human intestine has already been reported (Hamada and Slade 1980), to our knowledge, this is the first time a bacteriocin-producing Strep. mutans (producing mutacin II) has been identified from a human intestinal source. In this respect, it is noteworthy that mutacin production has been shown to be a colonization factor in the human oral cavity (Hillman et al. 1987).

Both the Ent. faecium bacterial isolates, EM342-BC-1 and EM297-BC-T6-1, exhibited inhibitory activity against Lact. bulgaricus LMG 6901, L. innocua DPC 3572 and L. monocytogenes DPC 3785. MS analysis of the purified antimicrobial peptides produced by Ent. faecium EM297-BC-T6-1 confirmed molecular masses for two putative bacteriocins. While it is characteristic of enterococci to produce different types of bacteriocins (Franz et al. 2007), little has been reported about the production of bacteriocins by enterococci of human intestinal origin (Moon et al. 2000). Enterocin producers are being characterized as novel probiotics and have been studied for their antagonistic nature against major food pathogens such as L. monocytogenes (Achemchem et al. 2005; Franz et al. 1999; Kang and Lee 2005; Renye et al. 2009). The present study reports the characterization of two active bacteriocins exhibiting antilisterial activity and suggests the need to assess such strains for their beneficial effect on the gastrointestinal microbiota.

The bacteriocins produced by the Lact. gasseri isolated in this study exhibited a narrow spectrum of activity, inhibiting just the Lact. bulgaricus assessed (Table 4). Sequence analysis identified gassericin bacteriocins, gassericin T and a new variant of gassericin A, in agreement with the MS data. In general, Lact. gasseri is considered one of the most predominant and commonly detected species of Lactobacillus in the human gastrointestinal tract (Reuter 2001) and is commonly associated with bacteriocin production (Kawai et al. 1998a). It was recently demonstrated that the combined effect of both gassericin A and gassericin T along with glycine could be used as a preservative in the food industry (Arakawa et al. 2009). Similarly, the gassericin-producing strains isolated in this study may be of benefit for the food industry.

Bacteriocin production can confer a competitive advantage to the producing strain, enhancing its dominance over other species in an ecological niche. To date, only a limited number of studies relating to bacteriocin production by strains of human intestinal origin have been performed. This study has highlighted the diverse range of bacteriocinogenic strains present in the human intestinal environment, which may help these strains to predominate in the bacterial population. Overall, following preliminary characterization, genotyping of all of the 55 bacteriocin-producing strains isolated revealed 15 different individual strains indicating a higher incidence of repeated isolation of similar strains. In addition to the known predominant species of the human intestinal environment, including Lact. gasseri and Enterococcus spp., this study has highlighted the isolation of unexpected species such as L. lactis and Strep. mutans. This screening has resulted in the isolation of a panel of bacteriocin-producing human strains with the potential to alter the gut microbiota when introduced as probiotic cultures.

Acknowledgements

This work was supported by the Government of Ireland National Development Plan by way of a Department of Agriculture, Food and the Marine, and Health Research Board FHRI award to the ELDERMET project, as well as by a Science Foundation Ireland award to the Alimentary Pharmabiotic Centre. We are grateful to all those people who participated in this study. We are also grateful to Mary Rea, Eileen O'Shea, Rebecca Wall, Nessa Gallwey, Ann O'Neill, Karen O'Donovan and Patricia Egan for technical and clinical help and to Siobhan Cusack and Eibhlis O'Connor for project management. This study is an output of the ELDERMET consortium (http://eldermet.ucc.ie), which has the following additional principal investigators: Colin Hill, Ted Dinan, Gerald Fitzgerald, Denis O'Mahony, Cillian Twomey, Douwe van Sinderen and Julian Marchesi.

Ancillary