Persistence of the Oral Probiotic Streptococcus salivarius M18 Is Dose Dependent and Megaplasmid Transfer Can Augment Their Bacteriocin Production and Adhesion Characteristics

Bacteriocin-producing probiotic Streptococcus salivarius M18 offers beneficial modulatory capabilities within the oral microbiome, apparently through potent inhibitory activity against potentially deleterious bacteria, such as Streptococcus pyogenes. The oral cavity persistence of S. salivarius M18 was investigated in 75 subjects receiving four different doses for 28 days. Sixty per cent of the subjects already had some inhibitor-producing S. salivarius in their saliva prior to probiotic intervention. Strain M18’s persistence was dependent upon the dose, but not the period of administration. Culture analysis indicated that in some individuals the introduced strain had almost entirely replaced the indigenous S. salivarius, though the total numbers of the species did not increase. Selected subjects showing either high or low probiotic persistence had their salivary populations profiled using Illumina sequencing of the V6 region of the 16S rRNA gene. Analysis indicated that while certain bacterial phenotypes were markedly modulated, the overall composition of the oral microbiome was not modified by the probiotic treatment. Megaplasmids encoding bacteriocins and adhesion factors were transferred in vitro to generate a transconjugant S. salivarius exhibiting enhanced antimicrobial production and binding capabilities to HEp-2 cells. Since no widespread perturbation of the existing indigenous microbiota was associated with oral instillation and given its antimicrobial activity against potentially pathogenic streptococci, it appears that application of probiotic strain M18 offers potential low impact alternative to classical antibiotic prophylaxis. For candidate probiotic strains having relatively poor antimicrobial or adhesive properties, unique derivatives displaying improved probiotic performance may be engineered in vitro by megaplasmid transfer.


Introduction
In the past, properties such as the impact of dose on probiotic persistence and modulation of the microbiota have been less frequently studied, as the major focus of probiotic research has been on achieving efficacious outcomes, often with the largest costeffective dosage regimen. Probiotics are often touted as being like small factories producing biologically active substances that benefit the host, but as most probiotics rapidly transit through the oral and digestive tracts following their ingestion, the likelihood of persistence at their principal target site is low. Previous studies have found that probiotic bacteria do not generally persist for more than a few hours or days in the intestinal tract [1] or oral cavity [2], though there are some exceptions reported for the oral cavity and vagina [3,4].
Bacterial pathogens tend to have specific virulence traits that facilitate their attachment and subsequent invasion, particularly of oral and intestinal tissue, even in the presence of a protective layer of commensal bacteria, which themselves have adapted for attachment and survival, yet, seldom do the same commensal species become established in detectable numbers when administered in probiotic formulations. This has led to the suggestion that host factors influence the persistence of a microorganism newly introduced to an already established microbial ecosystem. Studies in which probiotic strains do not persist have largely failed to elucidate whether this is because of colonisation resistance, damage to the probiotic strains during their preparation, host specific attachment incompatibilities, dosage deficiencies or other mechanisms [5,6]. Interestingly, some probiotic bacteria do not appear to perform consistently well in clinical studies and there appear to be often ill-defined factors relating to the manufacturing process that can influence the subsequent performance of a microorganism in vitro or in vivo [5,6,7]. In practice most probiotic strains have been propagated in vitro many times prior to being used in scaled commercial fermentations. During the course of serial passaging, bacteria can slowly accumulate random mutations within their chromosomes or jettison rarely translated DNA [8,9]. 'Muller's rachet' theory holds that asexual organisms (in this case bacteria kept in pure culture that are not transferring or receiving DNA from their surrounding community), display genetic drift resulting in a loss of functions as a consequence of the gradual genetic decline effected by random mutation [9,10].
As a result of the oral microbiota being implicated in a variety of systemic conditions [11], attempts are being made to address these through treating the oral cavity. The proximal location of the oral cavity provides comparatively uncomplicated access for probiotics and for sampling to assess their impact. Studies have shown that the microbial composition of saliva is derived from a variety of oral ecosystems, such as the dorsum of the tongue [12]. Streptococcus salivarius has an innate capability of binding to and persisting on the tongue dorsum and some strains release into saliva copious quantities of bacteriocins that could provide a targeted way of removing deleterious bacteria [4,13,14]. In vivo bacteriocin production is often cited as the principal means by which health benefits are realised following the consumption of probiotic products [15,16]. The passage of these organisms through the oral cavity and gut up-regulates bacteriocin production by members of the host's microbiota [17]. In order to achieve sustained bacteriocin release, persistence of the probiotic bacteria is likely necessary [18].
To date, almost all of the bacteriocin-producing capability of S. salivarius appears attributable to genetic determinants localized on a megaplasmid (size 160 to 220 kbp). Naturally occurring transmission of these bacteriocin-encoding megaplasmids has been shown to occur both in vitro and in vivo. Their highly-flexible propensity for acquisition, expression and de-commissioning of a wide variety of bacteriocin loci may help account for the numerical prominence of S. salivarius in the oral cavity and be a mechanistic basis for S. salivarius having a major role in the maintenance of a balanced oral ecosystem. It has been suggested that S. salivarius megaplasmids may function as repositories for bacteriocin determinants acquired from a variety of oral species via transposition of IS elements [13]. Preliminary molecular analyses have indicated that these megaplasmids can also encode molecules aiding host cell adhesion, without induction of antibiotic resistance [19].
The goal of the present study was to investigate the effect of different dosage levels of Streptococcus salivarius M18 on the persistence of these bacteriocin-producing cells. Bacteriocin and adhesion determinants have previously been identified on certain S. salivarius megaplasmids and the influence of inter-strain transfer of these on host cell adhesion and antimicrobial activity was evaluated using in vitro models for the purpose of tailoring probiotic strains.

Subjects
The protocol and consent documents were reviewed and approved by the New Zealand Lower South Regional Ethics Committee. Seventy-five subjects (average age of 19 years) were recruited from a University class of approximately 200 students. All subjects were older than 18 years of age and gave informed written consent. The subjects were randomized and blinded to one of four identical looking dosage groups (taking lozenges containing 1610 6 (n = 19), 1610 7 (n = 20), 1610 8 (n = 17) or 1610 9 (n = 19) colony-forming units [CFU] of Streptococcus salivarius M18 per dose). The subjects used one lozenge per day for 28 days after tooth brushing. Samples (ca. 1 ml) of unstimulated saliva were obtained upon entry to the study and then each week at least 12 hours after lozenge treatment.
The saliva samples were serially diluted in phosphate buffered saline (PBS) and 50 ml was spiral plated onto various agar media including Columbia blood agar base, supplemented with 0.1% (w/ v) calcium carbonate and 5% (v/v) human blood [Fort Richards Laboratories, Ltd., Auckland, NZ] (BACa). Streptomycin (100 mg/ml) was added to facilitate detection of the marked streptomycin resistant S. salivarius strain M18, and Mitis Salivarius agar (MS, [Fort Richards Laboratories, Ltd.]) was used for the general selection and enumeration of S. salivarius. Agar cultures were incubated for 18 hours in air supplemented with 5% CO 2 at 37uC. Forty colonies having the characteristic S. salivarius phenotype on the MS plate from each subject were picked into a lawn culture of Micrococcus luteus I1 that had just been freshly inoculated on BACa. This indicator bacterium is known to be sensitive to a wide variety of streptococcal bacteriocin-like inhibitory substances (BLIS). Streptococcus-like colonies that grew on BACa streptomycin (i.e. presumptive S. salivarius M18) were picked into a freshly seeded BACa lawn culture of Streptococcus mutans OMZ 175, an indicator strain uniquely sensitive to BLIS activities of strain M18 (Burton unpublished). Total S. salivarius populations from MS plates were also tested for bacteriocin like inhibitory substances by deferred antagonism testing [20].

DNA Purification and Sequence Analysis
Total DNA was extracted from 500 ml of sample saliva preincubated for 10 min at 37uC with 50 ml of 8.8 mmol/l dithiothreitol using the PureLink TM genomic DNA kit (Invitrogen, Auckland, NZ) as per the manufacturer's instructions for Grampositive bacteria. DNA was eluted from the column in 100 ml of elution buffer. The primers L-V6 (59-CAACGCGARGAACCT-TACC-39) and R-V6 (59-ACAACACGAGCTGACGAC- 39) were chosen to amplify the V6 hyper variable region of the 16S rRNA gene [27]. The following PCR reaction conditions were used: 5 units Taq platinum: 1.7 mM MgCl, 210 mM dNTPs and 640 nM of each primer. A touchdown protocol was employed with: initial denaturation 94uC for 2 min; denaturation 94uC; annealing starting at 61uC and dropping with 1uC over 10 cycles with the remaining 15 cycles at 51uC; extension at 72uC; all for 45 seconds and a final elongation step for 2 min. A negative control including all ingredients but with water instead of DNA template, and a positive control with a lower limit of detectable DNA, were performed alongside all test reactions. PCR-products were used when the negative control was free of PCR product and the positive control amplified. A constant volume aliquot of each amplification product was run on a 1.5% (w/v) agarose gel to determine the approximate amount of product and sent for sequencing at The Next-Generation Sequencing Facility (Illumina) in The Centre for Applied Genomics at the Hospital for Sick Children, Toronto, Canada. Data analysis was performed as previously described [28]. 16S rRNA Operational Taxonomic Units seed sequences were deposited in NCBI Short Read Archive with the BioSample accession SAMN02055382.

In vitro Intra-species Transfer of Megaplasmids
Using differences in sensitivity to antibiotics and bacteriocin of both megaplasmid donor and recipient strains, the transcongugants were generated and confirmed by ERIC-PCR [8]. In brief, the plasmid donor strain (M18 or K12) was grown on CNA-P agar (a medium that represses salivaricin A [SalA] and salivaricin B [SalB] production) [29] while the recipient strain (M18 2/2 or K12 2/2 , plasmid-negative derivatives resistant to 500 mg/ml streptomycin or spectinomycin) in brain heart infusion (BHI) (Difco, MD) for 18 h at 42uC in a candle jar. Plasmid donor cells were collected on a sterile cotton swab and resuspended in 5 ml fresh BHI. One milliliter of the plasmid recipient strain was added to the mixture and then incubated in a candle jar for 18 h at 42uC (incubation above 37uC represses SalA and SalB production). Positive controls were included in all experiments and consisted only of donor cells or recipient cells in 5 ml BHI. Samples of the control and test mixtures were taken at 0, 2, 4 and 18 h using a cotton swab that was dipped into the BHI culture and then used to swab the surface of a BACa +Str plate. The plate was incubated for 18 h at 37uC in 5% CO 2 in air.
Subsequent selection for SalA-and spectinomycin-resistant colonies was carried out as follows; Streptococcus salivarius A23-4, a SalA producer (and therefore possesses the SalA immunity genes) was inoculated diametrically across a BACa plate and incubated for 18 h at 37uC in 5% CO 2 in air. Visible bacterial growth was removed the next day using the edge of a microscope slide, following which the plate surface was exposed to chloroform vapors for 30 min to kill residual producer cells. The plate was left to air for 1 h leaving only the deposited bacteriocin. A cotton swab charged with growth from each BACa +Str plate was then used to streak across the sterilized BACa plates and the plate incubated at 37uC in 5% CO 2 in air for 18 h. Thirty SalA-resistant colonies in this region were picked off and stab-inoculated onto a BACa plate pre-seeded with an Micrococcus luteus I1 lawn, which is highly sensitive to antimicrobial activities. The lawn was seeded by charging a cotton swab with a I1 THB culture (18 h, 37uC, 5% CO 2 in air) and creating a confluent lawn over the entire BACa plate. This was then incubated for 18 h at 37uC in 5% CO 2 in air. Donor and recipient cells were stab-inoculated onto each plate as positive and negative controls. SalA-resistant colonies which produced a detectable inhibition zone against I1 were then struck onto a BACa +Spec plate (100 mg/ml) and incubated for 18 h at 37uC in 5% CO 2 in air. Colonies which grew on the BACa +Spec plate were considered to be recipient cells that had successfully acquired SalA-encoding DNA from the plasmid-containing donor strain (i.e. they were putative plasmid recipients). These were also checked for their bacteriocin-producing abilities (as described).
The simultaneous antagonism method was used to detect the ability of a producer strain (S. salivarius) to inhibit the growth of coseeded indicator strains. A lawn of P. intermedia, P. gingivalis or P. canoris on either BACaHV or BACa from cells suspended in 3 ml THB to a 0.5 McFarland Standard onto a fresh agar plate. Then a pure-producer colony, which was 'picked' and then 'stabbed' into the agar plate, already inoculated with the lawn. These were incubated anaerobically, at 37uC, for 2-4 days or until the lawn of the indicator strain was confluent. The resulting zone diameter was recorded.

Attachment of S. Salivarius Strains to HEp-2 Cells
Streptococcus salivarius strains were tested for their ability to bind to the mammalian HEp-2 cells; M18, M18 2/2 (megaplasmid deficient), M18 K12p and K12 M18P (as derived above with exchanged megaplasmids), K12, K12 2/2 (megaplasmid deficient), JIM8777, ATCC7073, Min5, Tove R, Tove S and JH. To assess adherence, HEp-2 cells (ATCC CCL23) were grown into a monolayer in flat-bottomed 96-well plates while S. salivarius strains of interest inoculated in 3 ml THB and incubated overnight, at 37uC in 5% CO 2 in air. Overnight bacterial cultures were spun in micro-centrifuge tubes for 5 minutes. The supernatant was discarded and the cells were suspended in PBS before being centrifuged again for 5 minutes. Cells were resuspended in Dulbecco's Modified Eagle's Medium (DMEM) [Invitrogen, Auckland, NZ] and diluted to 5610 5 -5610 6 cells/ml. To obtain an initial cell count, the samples were further diluted up to 10 25 and plated out using the Miles-Misra method technique (10 ml in triplicate per dilution per strain) onto 2YT agar plates [30]. The monolayers of HEp-2 cells were washed three times in PBS and then mixed with 100 ml of strain of interest into each well containing HEp-2 cells. All tests were carried out in triplicate and incubated for 3 hours at 37uC in 5% CO 2 in air. The wells were then washed three times in PBS to remove non-adherent bacteria. To dissociate the HEp-2 cells from the wells, 30 ml of 0.5 g/l trypsin and 0.2 g/l EDTA solution was added to each well and incubated for 30 minutes at 37uC in 5% CO 2 in air. Using an inverted microscope, cells were checked to ensure that most HEp-2 cells have been dissociated from the wells. Then 70 ml of THB was added to the wells and pipette used to dislodge the HEp-2 cells from the walls. These were then plated using the Miles-Misra technique onto 2YT agar plates and incubated at 37uC in 5% CO 2 in air, overnight. For enumerating S. salivarius in cellassociation assays, 2YT (yeast/tryptone) agar (2% (w/v) tryptone [Bacto, BD], 1% (w/v) yeast extract [Bacto, BD], 1% (w/v) sodium chloride, 1.5% (w/v) bacteriological agar) was used. Percentage adherence for each strain was calculated by determining the average number of CFU per dilution (as CFU/ml), this value was divided by the initial number of colonies/ml added to the corresponding well and then multiplied by 100.

Detection of Adhesion Factors
Twenty colonies from a fresh culture on solid media were individually resuspended in 300 ml of 0.85% (w/v) NaCl in 1.5 ml microfuge tubes and were DNA extracted. The genes encoding cspA, cspB, orf 166 and orf 176 were investigated through PCR. The cspA and cspB products were amplified with primers (59-39), cspA forward; GCC TAA CGC TAC GGA TAC TGC TAA T, cspA reverse; ACT GCT CCT CCT GCC TGT GAA G, cspB forward; CCA ACA TAA AGG GAC ACC AAC TAC GAG, cspB reverse; CCC ATC CGG ATT AAC GCT ACC A. Amplification utilised an initial denaturation step at 92uC for 2 minutes, annealing at 55uC for 2 minutes, followed by elongation at 65uC for 5 minutes. This was followed by 32 cycles of denaturation at 92uC for 30 seconds, annealing at 55uC for 30 seconds and elongation at 65uC for 3 minutes. The PCR conditions for the amplification of orf 166 and orf 176 used primers Orf 166 forward; CGA GAG TTT GCT GCC ATA CA, Orf 166 reverse; GGC AAC ACC AGC GTT TTT AC, Orf 176 forward; CTT TCT CGA CAG TAA GGC GG and Orf 176 reverse; TGA AAT TCC AAC TCC TTG CC and included an initial denaturation step at 94uC for 2 minutes, followed by 30 cycles of denaturation at 94uC for 30 seconds, then annealing at 55uC for 30 seconds and elongation at 72uC for 1 minute. This was followed by a final elongation step at 72uC for 5 minutes.

Co-aggregation Assay
To determine the ability of S. salivarius to co-aggregate with strains Porphyromonas gingivalis ATCC 33277, Aggregatibacter actinomycetemcomitans V29523 and Fusobacterium nucleatum FH2, the S. salivarius strains were inoculated into 10 ml THB and incubated at 37uC, in 5% CO 2 in air, overnight. The periodontal pathogen strains were inoculated into THB and incubated at 37uC anaerobically for 2-3 days. The bacteria were harvested by centrifugation at 30006g for 10 minutes and washed three times in 1 ml volumes of aggregation buffer (0.121 g Tris, 0.022 g calcium chloride, 0.031 g hydrated magnesium chloride, 8.766 g sodium chloride/L) to ensure complete removal of culture medium. The cell suspensions were diluted 100-fold with aggregation buffer and 500 ml aliquots of each periodontal pathogen was individually mixed with 500 ml of each of the test S. salivarius strains. The turbidity of the mixtures were recorded at 15 minutes, 40 minutes and 8 hours, and given a score depending on the aggregation.

Indigenous S. salivarius Strains Commonly Produce Bacteriocins
Sixty two per cent of subjects had BLIS-producing S. salivarius detected in their saliva prior to dosing and in 21% of these individuals the BLIS producers represented .80% of the total S. salivarius population (Figure 1). Further analysis of this BLIS activity using the P-typing method [20] showed that 26% of subjects had Streptococcus salivarius which produced antimicrobial substances, which inhibited at least 3/9 indicator organisms, tested (data not shown). None of the subjects carried S. salivarius strains having inhibitory activity against Streptococcus mutans OMZ 175, an organism sensitive to the bacteriocin produced by strain M18, as tested against 40 isolated S. salivarius colonies from each subject (data not shown).

Streptococcus Salivarius M18 Oral Persistence Appears Partially Dose Dependent
No differences in the total mean salivary S. salivarius (CFU/ml) for each subject was found from baseline during the course of dosing with strain M18 (Figure 2a). To enumerate strain M18, saliva was plated on CABCa agar supplemented with 100 mg/ml streptomycin. An average background level of approximately 1610 3 CFU/ml of naturally resistant S. salivarius was detected in the baseline samples (Figure 2b). However, after dosing with M18, the levels of streptomycin resistant S. salivarius increased substantially, indicative of the strain's presence within the salivary microbiota. The exception was the lowest dosage group, which only showed an increase in streptomycin resistant S. salivarius in the 4-week sample. Progressive increases in strain M18 dosage resulted in correspondingly greater proportions of streptomycin resistant S. salivarius in the salivary population. In the highest dosage group (1610 9 CFU/dose/day), the putative M18 colonisation levels were one log higher than in the other dosage groups.
The log counts showed that subjects who received higher probiotic dose retained higher numbers of M18 (Figure 3a). The percentage of subjects having the M18 strain detected in their saliva increased with the dose quantity ( Figure 3b). However, after day seven, the salivary probiotic numbers did not appear to increase, despite further dosing. The trend was for the cell numbers to slowly track downwards.

V6 Region 16S rRNA Sequence Analysis of the Salivary Microbiota does not Indicate Major Perturbations by Probiotic Instillation
To determine whether instillation of an exogenous organism influenced the composition of the salivary microbiota when persisting at different levels, six subjects that exhibited the most consistently low or high levels of M18 salivary levels had their microbial composition examined in greater detail. The average strain M18 salivary persistence levels were log 5.34 for the subjects in the high persistence group compared to 4.15 for the those in the low persistence group (P = 0.0026). 16S rRNA gene V6 region amplification and sequencing yielded an average of 258, 210 sequences per sample. These were consigned to 197 operational taxonomic units (OTU) based upon grouping sequences using a 95% DNA sequence identity cutoff. The number of sequence reads per OTU was converted to the total proportion per sample. The first five OTU groups accounted for more than 50% of the total sequences and the top ten accounted for over 70% of sequence types. Analysis of the composition of all samples from the subjects indicated no significant ecological shifts in the microbiota following the probiotic dosing by weighted UniFrac analysis (Figure 4 & 5). The relationships between the samples predominantly appeared to cluster by participant; that is, a subject's microbiota was most similar to itself at all time points than it was compared to any other subjects whether in either the 'high' or 'low' groups ( Figure 5).

Megaplasmid Transfer Results in a Recipient Strain with Hybrid Characteristics
All the S. salivarius strains tested, except strain NR, had inhibitory activity against P. intermedia. The plasmids of strains K12 and M18 must harbour the BLIS activities against strains ATCC25611 and BGBL, as their plasmid-negative derivatives lost the ability to inhibit these same strains ( Table 1). Inhibition of P. gingivalis strains JK45, W50, and P. canoris strain P21 was media-  dependent with inhibition on either BaCa or BaCaHV agar, depending on the indicator and producer strain (Table 1). After prolonged incubation of the indicators (up to seven days), the zones of inhibition were no longer visible, indicating that the initially observed inhibitory activity may have been bacteriostatic. Some synergy between chromosomal and extra chromosomally encoded elements exist as after megaplasmid transfer some of the antimicrobial activities were different in the transconjugants, for example M18 and K12 wild type strains both inhibited strain JK45 on BACa, but neither of the transconjugants displayed this activity on the same media.

Multiple Factors are Involved in the Attachment of M18 to Host and Bacterial Cells
Neither M18 nor its megaplamid-cured derivative could adhere to HEp-2 cells. However, strain K12 (and its variant strains including the megaplasmid deficient K12 2/2 and the transconjugant containing the M18 plasmid K12 M18p ), adhered well in vitro ( Figure 6). Strain M18 K12p had a relative adherence in excess of 400%, whereas, both the wild type and cured derivatives of strain M18 had adherence scores of fewer than 10%. Previous studies have shown that it is not possible for the HEp-2 cells to internalise M18 (Burton unpublished) and therefore all values were considered adherence. Previous studies have looked at S. salivarius adhesion to other bacteria and the host epithelium at the cellular and molecular level [31,32]. To elucidate why the change of adhesion characteristics of the wild type strains and subsequent change upon megaplasmid loss or acquisition, strains of S. salivarius were evaluated by PCR for their presence or absence of genes associated with these properties, namely cspA, cspB, orf 166 and orf 176 ( Table 2). Strains positive for cspA also have cspB, but did not necessarily possess orf 166 and orf 176. Strangely, cspA and cspB were not detected for the K12 M18p strain which may indicate modification or deletion post transfer of the megaplasmid in this case.
All S. salivarius strains co-aggregated to some extent with P. gingivalis ATCC 33277 (Table 3). Strain M18, containing the pSsalK12 plasmid had higher co-aggregation capability than either the parent or plasmid-free variant. None of the S. salivarius co-aggregated with A. aggregatibacter V29523, and only ATCC 7073 (a fibrillated S. salivarius strain used as a positive control) coaggregated with F. nucleatum FH2.

Discussion
In the present study, BLIS-producing S. salivarius were shown to be a very common component of the indigenous salivary microbiota and indeed the majority of individuals yielded isolates displaying some inhibitory BLIS activity. Previous studies on pharyngitis, tonsillitis, dental caries and cystic fibrosis have shown a correlation between a reduction in the levels of potential bacterial pathogens and the presence of these ''antagonistic'' streptococcal commensals within the upper respiratory tract microbiota [33,34,35,36,37,38,39,40]. However, probiotic application of BLIS-producing bacteria for prevention of infection has been limited [14]. There is some evidence that these probiotic organisms can confer the same level of protection as naturally harboured strains [40,41,42]. While antimicrobial activities may not be the only mechanism involved in host protection by indigenous S. salivarius, achieving persistence of probiotics likely requires mimicking the natural conditions, which is likely to be important in conferring protection.
Persistent low-level dosing did not appear to lead to cumulative increases in the proportion of the probiotic within the indigenous salivary population. However, the levels of persistence increased with higher doses. This has implications for the clinical applications of probiotics, which are often recommended to be given daily, because most do not colonize the host. Plus, their concentrations are invariably not high (apart from VSL#3 for inflammatory bowel disease), so they are often administered over the course of a day. The ability of a probiotic to persist at the target site likely allows the organism to have a greater impact on the host.
The total number of S. salivarius detected in the saliva of an individual did not appear to be increased above the baseline level, even when doses of up of 100 times the background levels of  indigenous S. salivarius were administered. This implies that there is a limited capacity for the host to increase its microbiota species load. It also raises the question of what mechanisms pathogens use, and that are not present in probiotic or commensal strains, to integrate or by-pass the indigenous microbiota.
In some subjects there clearly was a high level of persistence of the bacteriocin-producing probiotic strain. Indeed, in some subjects a substantial proportion of the original S. salivarius population was replaced. 16S rRNA gene sequence analysis of subject's samples with either high or low probiotic persistence did not show remarkable change in the predominant microbial composition. This echoes the findings of McNulty et al. (2011) who showed that probiotic yogurt did not alter the gut microbiota [43]. The microbiota of the young adults tested here seemed to be relatively robust with regards to perturbation by probiotic instillation (Figure 4). While there were some bacterial types that were significantly different in their proportions both compared to pre samples and between the two persistence groups, these changes did not represent major perturbations of the microbiota.
The absence of such disturbances actually supports the safety of strain M18, as regulatory agencies view disruption of the indigenous microbiota of healthy subjects as undesirable.
Our group [8] has demonstrated the transfer of bacteriocinencoding plasmids into indigenous oral S. salivarius strains. The present study has demonstrated the first in vitro transfer of bacteriocin-encoding megaplasmids between two strains of S. salivarius. This opens up the possibility of creating tailor-made probiotic strains through the transmission of megaplasmids from poorly-persisting, antimicrobial producing S. salivarius strains into poor bacteriocin-producing but strongly persisting indigenous S. salivarius, strains potentially conferring better protection to the host.
Adhesion is an essential first step for any colonisation of the oral cavity since non-adherent bacteria will be rapidly washed away in the salivary flow. The binding of bacteria to tissue cells involves specific adhesins [44,45]. These include proteins that contain a  LPXTG motif (membrane anchor) near the C terminus [46], such as cspA and cspB (cell-surface protein A & B) [32]. Lévesque et al. [32] suggested that cspB may be associated with fimbriae, and thus could contribute to adhesion specificity, as fimbriae have been noted to be key components of the cell-to-surface and cell-to-cell adhesion of oral bacteria [47]. Another surface component, the fibrils are also important for adhesion of streptococci to oral surfaces and co-aggregation with other oral bacteria [31,48].
In the present study, CspA and CspB proteins were found to be plasmid-encoded on M18 and K12, but was also detected the poorly adhering Tove-S, ATCC25975 and JH strains. Levesque et al. [32] suggested that the Orf176 protein might also be a transcriptional regulator for cspA and cspB and also orf166, which encodes cell-surface protein [49]. Only strains M18 K12p (with the pSsalK12 megaplasmid) and wild-type K12 possessed all four genes. Orf166 may have a role in the adhesion of S. salivarius to HEp-2 cells since strain M18 K12p had a high relative adherence to HEp-2 cells but strain M18 showed poor adhesion. This suggests that Orf166 helps to mediate the adhesion process and in future studies deficient derivatives may assist in determining this. However, as strain K12 2/2 also adhered strongly to HEp-2 cells, so other factors must also promote K12 adhesion.
Co-aggregation has been documented for strains of the oral cavity [50,51], including S. salivarius and periodontal pathogens [52]. It has been suggested that S. salivarius co-aggregation with potential pathogens may be a means of eradicating the reservoir of the salivary anaerobes [53]. Of the S. salivarius strains tested here, only ATCC 7073 (a fimbriated K + strain, carrying fibrils) was able to co-aggregate with F. nucleatum. This is interesting as it implies a degree of specificity between strains not due to fimbriae, as the fimbriated Kstrain, ATCC 25975 was unable to bind to F. nucleatum This finding was similar to that of Levesque et al. [52] except in their case, a F. nucleatum strain co-aggregated with strain ATCC 25975, as well as a fimbriae-negative mutant of strain ATCC 25975 called strain D37. This further rules out a role for fimbriae in coaggregation.
In conclusion, this study has shown that the persistence of a probiotic S. salivarius strain in the mouth was dose dependent. This ability to persist could allow the probiotic the opportunity to more effectively counter pathogens as well as inducing host gene expression pathways of homeostasis and cytoskeletal repair in the epithelial lining [54,55]. The use of high dose M18 to treat and prevent oral diseases, in comparison to antibiotics, warrants further clinical testing with the hope of providing alternative option in dental practice.