Isolation and Identification of Bifidobacteriaceae from Human Saliva

ABSTRACT Bifidobacteriaceae were isolated from saliva and infected dentine by using a mupirocin-based selective medium. Of the saliva samples, 94% harbored bifids. The mean concentration (± the standard error) was 4.46 (±0.12) log10(CFU per ml + 1), and the predominant isolates were Bifidobacterium dentium, B. longum, Scardovia inopinata, Parascardovia denticolens, and Alloscardovia omnicolens.

The Bifidobacteriaceae constitute a significant proportion of the microbiota in the human gut (12), but the bifid populations of the oral cavity have been much less well studied. Little is known of their intraoral distribution or of the environmental factors, including dietary components, which may influence their numbers or the proportions of individual species isolated from the mouth. The cultivable commensal oral bifids are Bifidobacterium dentium, Scardovia inopinata, and Parascardovia denticolens (9,20,31), and Scardovia genomospecies strain C1 and Scardovia sp. strain T01-04 have also been described previously (16,25). Oral bifids have been at best sporadically isolated, but when isolated or identified in 16S rRNA cloning and sequencing studies, they are usually found in carious dental tissues (1,4,7,8,11,15,17,(22)(23)(24)(25)33) but are not detected in healthy mouth tissues or in root caries lesion specimens (2,26). There is inconsistency in the results of these previous studies. The aim of this study was to use conventional cultural methods to determine the prevalences and identities of oral bifids and, consequently, to investigate the genotypic diversity of B. dentium and B. longum.
Compared to MTPY medium, the MMTPY had a reduced concentration of mupirocin, from 100 to 5 g/ml, and modifications to the peptone source and raffinose to promote the isolation of non-glucose-fermenting species (5,32). These modifications did not significantly affect the utility of the medium, as the growth of nonbifid organisms was inhibited and the counts of viable cells of type and reference strains were not significantly different from those on MTPY medium (data not shown). The numbers of mutans streptococci, lactobacilli, and yeasts (primarily Candida spp.) in the saliva samples were determined (39); the detection limit was 10 CFU per ml. The total number of presumptive bifid colonies on the MMTPY was established, a sample of each presumptive bifid colony type from 129 subjects was subcultured, and partial 16S rRNA gene sequences were obtained using the primers Bif164mod (5Ј-G GGTGGTAATRCCSRATG-3Ј) and Bif662mod (5Ј-CCACC GTTACACCGRGAA-3Ј), modified from those described previously (34), or the universal primers (21) 63F and 1387F. Sequences were determined with the ABI Prism BigDye Terminator version 3.0 cycle sequencing kit (Applied Biosystems) and analyzed by using information from the Ribosomal Database Project (http://rdp.cme.msu.edu/) or from BLAST searching. Bifids were isolated from 94% of participants, and the concentrations of bifids in saliva samples were not significantly different from those of mutans streptococci but were significantly greater than those of the other taxa (Table 1); the relative proportions of bifids and lactobacilli mirrored those in human adult feces (10). The improved detection of bifids was a consequence of the use of a selective medium and the observation that the 27F primer used in most 16S rRNA sequencing studies has three mismatches with the 27F sequence of all Bifidobacteriaceae (14), resulting in significantly reduced detection of this taxon. Approximately 50% of subjects had salivary bifid and mutans streptococcus counts of Ͼ10 5 (Fig. 1). The concentrations of all taxa in saliva samples correlated negatively with the salivary flow rate, and those of all taxa correlated significantly with one another ( Table 1), suggesting that similar intraoral factors, including the salivary flow rate, may influence the concentrations of these organisms in saliva.
Of 354 presumptive bifid colonies identified, 341 (96.3%) were Bifidobacteriaceae. B. dentium was isolated from 106 subjects (82.2%). S. inopinata and P. denticolens were isolated from six (4.7%) and four (3.2%) subjects, respectively, while B. scardovii and Alloscardovia omnicolens were each isolated from two subjects (1.6%). The identities of the predominant bifids were in accord with those in previous studies (9,23,30), but this is the first report of the isolation of A. omnicolens (19) and B. scardovii (18) from nonclinical specimens, suggesting that the mouth may be their normal habitat. B. longum was isolated from 14 subjects (10.8%), and from 8 subjects, only B. longum was isolated, with concentrations of up to Ͼ10 7 CFU per ml of saliva and a median of 4.4 ϫ 10 5 CFU per ml. The salivary bifid concentrations in the samples from which B. longum was isolated were significantly greater than those in the other samples (mean Ϯ standard error, 5.72 Ϯ 0.22 versus 4.60 Ϯ 0.12 log 10 [CFU per ml ϩ 1]; P Ͻ 0.001). B. longum isolates fermented (6) arabinose, glucose, and xylose, suggesting that they belonged to biovar longum (29).
To investigate the genotypic diversity of B. dentium, we examined 51 randomly selected independent isolates by repetitive extragenic palindromic PCR (REP-PCR) (3), and each isolate was distinct (similarity, Ͻ90%) ( Fig. 2A), similar to those of other oral commensal bacteria in previous studies (3,13,37). Individual root caries lesion samples harbored unique genotypes, and multiple genotypes were detected in each root caries lesion specimen (Fig. 2B). The isolation of B. longum from 10% of subjects indicates that there is a potential for B. longum derived from probiotic foods to establish in the mouth, although bifids consumed as probiotics do not persist in the normal fecal flora (28,32). To ascertain whether the B. longum strains were the same and therefore likely to be from the same source, 12 independent isolates were genotyped using REP-PCR (3) and the BOX1AR primer, 5Ј-CTACGGCAAGGCG ACGCTGACG-3Ј (23). Neither genotyping method differentiated all isolates, but the combination of patterns demonstrated that all isolates were distinct (Fig. 3), suggesting independent origins for these strains. Thus, isolates Mu55-1 and Mu91-2 were similar by REP-PCR but were different by analysis with the BOX1AR primer. Repetitive-sequence PCRs have been used previously to identify individual species but not to assess intraspecies diversity (23,38). In previous studies, three B. dentium isolates had apparently identical sequences as determined with the BOXA1R primer (23), while enterobacterial repetitive intergenic consensus-like primers gave similar band patterns for closely related species (35); here, REP-PCR primers (36) resulted in unique patterns for independent isolates, suggesting that REP- a The concentrations of bifids in saliva samples were not significantly different from the levels of mutans streptococci but were significantly greater than the levels of lactobacilli (P Ͻ 0.001) and yeasts (P Ͻ 0.001) in the samples. b 95% CI, 95% confidence interval. c Coefficients of correlation between levels of bifids and levels of caries-associated microorganisms in saliva samples and between these levels and the salivary flow rate (measured in milliliters per min). All correlation coefficients correspond to P of Ͻ0.001.  PCR may be useful for tracking isolates in epidemiological studies.
In conclusion, Bifidobacteriaceae were present at high levels in the saliva of adults and their numbers were significantly correlated with those of caries-associated organisms. Studies are now required to investigate the oral ecology of this taxon and to determine the utility of these organisms as markers of caries risk.