In this study we reexamined the population diversity and dynamics of S. mitis
and two other important members of the commensal microbiota of the human oral cavity and pharynx using a highly sensitive nonculture detection method. The study design made it possible to evaluate the hypothesis that these anatomical sites and their many distinct ecological niches house a large number of clones, the levels of most of which are below the level detectable by traditional cultivation (15
). We have proposed that the apparent acquisition and loss of clones observed in cultivation studies (11
) to some extent may be a result of fluctuations in the relative proportions of clones in the microbiota (15
). This study confirmed this hypothesis.
To fulfill the aim of the study, we needed a method with sensitivity and discriminatory power that exceeded the sensitivity and discriminatory power of traditional culture techniques. The PCR-based strategy used has been successfully employed in studies of the biodiversity of complex microbial ecosystems (1
). Instead of 16S rRNA sequences, which reveal species diversity, our target marker was the gdh
housekeeping gene. Due to the unusual sequence divergence in housekeeping genes of S. mitis
, S. oralis
, and S. infantis
, which exceeds the sequence divergence of corresponding genes in pathogens like S. pneumoniae
by a factor 10 (Kilian et al., submitted) (Fig. ), this method is capable of detecting individual clones of the three commensal species with a high degree of accuracy (Kilian et al., submitted). Sequencing of close to 10,000 clones from the gdh
libraries generated from the samples collected from the two subjects on two occasions separated by 2 years ensured detection of clones that constituted minor proportions (<0.1%) of the populations of the three species.
By nature, our method allows comparison of only one gene locus, which theoretically limits its discriminatory power. However, comprehensive phylogenetic analysis of a large collection of isolates supported by polyphasic taxonomic analysis demonstrated that gdh
sequences provided unambiguous and reliable information about species affiliation and that sharing of the same allele by more strains is rare (Kilian et al., submitted), as confirmed by the data shown in Fig. . In contrast, recent studies demonstrated that no known phenotypic trait can be used to differentiate the species in question (16
). As a consequence, previous conclusions concerning the diversity of S. mitis
and S. oralis
) conceivably were based on mixtures of species.
Artifacts in the form of chimeric sequences are a potential risk in PCR-based analyses of complex microbiota. In this context it may be important that the streptococci in question are naturally competent for transformation and that the mosaic structures of genes reflect the fact that homologous recombination contributed to their evolution (9
). Therefore, our attempt to eliminate artificial chimeric sequences by discarding the sequences showing discordant phylogenetic affiliations at the two ends may, theoretically, have resulted in an underestimate of the natural diversity among the sequences.
S. oralis, S. mitis, and S. infantis are all members of the mitis group, which currently includes 12 recognized species. We have previously demonstrated that the primer set used in this study amplifies the target gdh sequence exclusively in all strains of S. mitis, S. oralis, S. infantis, S. pneumoniae, and S. pseudopneumoniae and in some strains previously assigned to “S. mitis biovar 2” (Kilian et al., submitted). Phylogenetic analysis of gdh sequences revealed distinct clusters of S. oralis and S. infantis. However, S. pneumoniae and S. pseudopneumoniae each constitute single lineages within a distinct cluster of multiple lineages, the remainder of which currently belong to S. mitis (Kilian et al., submitted). These phylogenetic relationships are also reflected in the tree shown in Fig. . Although some gdh alleles obtained in the present study were closely related to the reference sequences for S. pneumoniae, we excluded the possibility that they originated from this species based on our previous observation that gdh sequences of S. pneumoniae form a monophyletic subcluster (Kilian et al., submitted). However, we cannot exclude the possibility that some of the gdh sequences represent the recently described organism S. pseudopneumoniae, for which no information on occurrence is available. As indicated by the separate clustering of the two reference strains of “S. mitis biovar 2,” none of the sequences amplified from the samples represented this taxon. In agreement with our previous observation based on comprehensive phylogenetic analysis, “S. mitis biovar 2” represents a distinct species that is more closely related to S. oralis than to S. mitis.
At the time of the first sampling, a total of 39 distinct gdh
alleles (21 alleles in subject A and 18 alleles in subject B) were detected in the two individuals. According to the total richness estimate, these numbers represent approximately 95.5 and 100% of the total number of distinct alleles in the libraries established for the two subjects. We equated these figures with the number of clones while recognizing that this may have caused a marginal underestimate of the actual number of clones due to potential sharing of gdh
alleles by distinct clones (Table ). Of the 39 clones present at the time of the baseline observation, 18 were redetected at the time of the second observation. Combined with the fact that 38 of 56 clones identified at the second sampling time were not detected at the initial sampling time, these observations suggest that acquisition and loss contribute to the strikingly unstable population dynamics of each of the three species, but we cannot exclude the possibility that some of these clones were present on both occasions but below the detection level. The suggestion that acquisition contributes to this phenomenon is in agreement with the previously observed transfers of S. oralis
clones among cohabiting couples (2
) and occasional sharing of S. mitis
clones by infants and their parents (11
Possible biases introduced by the PCR-based technique prevent the relative abundance of a genotype from being precisely determined from its proportional representation among amplified sequences. Yet previous studies demonstrated good agreement between the proportions of bacterial species detected by in situ hybridization and the relative proportions in clone libraries (7
). On this basis, Fig. clearly demonstrates that the majority of the clones carried by the two individuals studied accounted for minor proportions of the populations of the respective species. It is conceivable that such clones would have been missed by cultivation even if multiple isolates were examined. Furthermore, Fig. shows examples of fluctuations from minor proportions to predominance and vice versa. This finding is in agreement with our hypothesis that part of the apparent acquisition and loss of clones observed in cultivation studies may be explained by such fluctuations (11
). Furthermore, our longitudinal data demonstrate that such fluctuations also affect the balance between individual species. While S. oralis
was the predominant species among the three species harbored by the two individuals at the baseline sampling time, both subjects exhibited a shift toward predominance of S. mitis
2 years later (Table ). Due to the fact that this study was based on pooled samples from multiple surfaces in the oral cavity and pharynx, we could not discern if such fluctuations are confined to particular habitats.
In conclusion, this study revealed significant clonal diversity within populations of S. mitis, S. oralis, and S. infantis carried by individuals in the upper respiratory tract. The total number of clones, the demonstration of numerous clones in proportions that are not detectable by traditional culture methods, and the observed significant changes in proportions over time demonstrate that the populations of commensal streptococci in the oral cavity and pharynx are constantly changing due to significant fluctuations in the relative proportions of existing clones and species possibly combined with loss and acquisition from close contacts. One important biological implication of these findings is that, in contrast to individual clones of potential pathogens, like the closely related organism S. pneumoniae, commensal bacteria are not subject to rapid elimination by the adaptive mucosal immune system. Rather, the observed fluctuations in abundance of individual clones of commensal streptococci are due to interactions within the complex microbiota, the nature of which is largely unknown.