Periodontal disease is a chronic inflammatory infection that affects the tissues surrounding and supporting teeth [1–3]. It is highly prevalent in adult populations around the world, and is the primary cause of tooth loss after the age of 35 [2–4]. The term ‘periodontal disease’ encompasses a spectrum of related clinical conditions ranging from the relatively mild gingivitis (gum inflammation) to chronic and aggressive forms of periodontitis; where inflammation is accompanied by the progressive destruction of the gingival epithelial and connective tissues, and the resorption of the underlying alveolar bone. It has a highly complex, multispecies microbial etiology; typified by elevated populations of proteolytic and anaerobic bacterial species [5]. Oral spirochete bacteria, all of which belong to the genus Treponema, have long been implicated in the pathogenesis of periodontitis and other periodontal diseases [6]. One species in particular: Treponema denticola has been consistently associated with both the incidence and severity of periodontal disease [6–11].

Over the past few decades, a significant number of T. denticola strains have been isolated from periodontal sites in patients suffering from periodontal disease; predominantly from deep ‘periodontal pockets’ of infection that surround the roots of affected teeth. Clinical isolates of T. denticola have previously been identified and differentiated by a combination of cell morphological features; biochemical activities (e.g. proteolytic substrate preferences), immunogenic properties (e.g. serotyping, or reactivity towards monoclonal or polyclonal antibodies) as well as multilocus enzyme electrophoresis [12–17]. However, these approaches are generally tedious and technically demanding, and often yield inconsistent or ambiguous results.

To date, only two complete genome sequences are available for oral spirochete bacteria; those of T. denticola ATCC 35405 (type strain) [18] and Treponema vincentii LA-1 (ATCC 35580), which has been sequenced by researchers at the J. Craig Venter Institute as part of the Human Microbiome Project [19], but is as yet unpublished. The 2.84 Mbp single circular chromosome of T. denticola ATCC 35405 contains ca. 2,770 predicted protein-encoding genes, whilst the 2.51 Mbp T. vincentii genome is predicted to have ca. 2,600 protein encoding genes (NCBI GenBank accession number NZ_ACYH00000000). The syphilis spirochete Treponema pallidum is closely-related to T. denticola at the genetic level, but contains a much smaller ‘host-adapted’ genome ca. 1.14 Mbp in size [20].

Over recent years, multilocus sequence analysis (MLSA) has proven to be a powerful method for the discrimination, taxonomic classification and phylogenetic analysis of closely related microbial species, subspecies and strains [21–29]. MLSA involves the systematic comparison of the DNA sequences of sets of (conserved) genes, usually 2 to 10 in number, within a given set of strains or species. Commonly, the total gene sequence data for a single isolate is concatenated prior to analysis using a variety of distance-based or criterion-based computational methods. MLSA offers many advantages over ‘single gene’ approaches; most notably its greater sensitivity and resolving power, and its ability to identify or overcome conflicting signals, such as those arising from horizontal gene transfer [22, 23, 29].

Although studies have consistently associated T. denticola with periodontal disease, its precise pathogenic roles remain to be fully established. This issue has been complicated by the use of a variety of different T. denticola strains in previously reported biophysical analyses, cell culture-based investigations or animal infection models. Very little is presently known about how similar or disparate these isolates may be at the genetic level. This prompted us to utilize an MLSA-approach to systematically analyze the genetic composition of 20 of the most commonly used strains of T. denticola; originally isolated from patients with periodontal diseases who were living in Asia, Europe or North America. Our results reveal that there is considerable genetic diversity within this species. Phylogenetic analyses of multi-gene datasets indicate that the T. denticola strains studied share a common genetic origin, which is distinct from that of T. vincentii or T. pallidum and appear to have a clonal structure.

The oral spirochete bacterium Treponema denticola is postulated to play an important role in the pathogenesis of periodontal disease; in particular chronic periodontitis, which is estimated to affect ca. 10-15% of the global population [3, 4, 6–9]. It is also implicated in the etiology of acute necrotizing ulcerative gingivitis (ANUG) [42] and orofacial noma [43], two other tissue-destructive diseases of the orofacial region. However, T. denticola is commonly detected in the oral microbiota in dentulous adults; albeit at relatively low levels, and its precise etiopathogenic mechanisms remain to be established. The elucidation of more specific disease associations is presently hampered by the lack of a reliable method for strain identification, and a very poor understanding of how strains differ at the genetic level.

Here, we utilized a seven protein-coding gene multilocus sequence analysis (MLSA) approach, to characterize genomic diversity and evolutionary relationships in a small, but carefully-selected collection of T. denticola isolates of diverse geographical origin. Our results revealed that there are relatively high levels of genetic diversity amongst T. denticola strains; with gene sequence similarities ranging between ca. 84 − 100% between the strains. These levels are considerably higher than in T. pallidum; where strains of the pallidum and pertenue subspecies share ca. 100-99.6% genome sequence identity, and genetic differences are largely confined to recombination ‘hotspots’ or other areas of acquired DNA sequence [20]. Whilst there were variations in the relative proportions of polymorphic sites present in the seven protein-encoding genes selected for analysis, all were under a strong (purifying) evolutionary pressure to conserve function. We found no evidence of genetic recombination in any gene sequence analyzed, indicating that genes had evolved as intact units in each strain. It is interesting to note that the flaA gene, which encodes an endoflagellar sheath protein that is a known a cell surface-exposed epitope [44], appeared to follow a similar evolutionary pathway as the pyrH and recA ‘housekeeping’ genes analyzed. Although we also sequenced the 16S rRNA (rrsA/rrsB) gene(s) from each strain, we did not add this to the concatenated multi-gene sequence for phylogenetic analysis. This was because it is present in two identical copies on the T. denticola genome [18], and may be under distinct evolutionary pressures, due to the fact that is not translated into a protein; e.g. it may have increased levels of nucleotide insertions or deletions (indels), or may have selection biases relating to its secondary structure [24].

Based on the concatenated 7-gene (flaA, recA, pyrH, ppnK, dnaN, era and radC) datasets, both the Bayesian (BA) and maximum likelihood (ML) topologies clearly indicated that all 20 T. denticola strains are monophyletic; i.e. they originated from a single common ancestor that was genetically distinct from T. vincentii and T. pallidum (see Figure 3). Our data also indicates that at the genetic level, T. denticola is more closely related to the oral treponeme T. vincentii, than the syphilis spirochete. Six well-defined clades (I-VI) were formed in both the BA and ML trees, which comprised 18 of the 20 strains analyzed. The OTK strain from the USA does not fall within any of the defined clades, possibly due to the relatively low sample size. The early-branching ATCC 700768 strain from China appears to be highly divergent from the other T. denticola taxa (discussed further below). The overall concordances in tree topologies obtained for the 7 individual genes, which are well-distributed around the ca. 2.8 Mbp chromosome, are consistent with T. denticola being predominantly clonal in nature. We did not attempt to estimate evolutionary timescales, as the precise dates of isolation are not known for these strains. Due to the high levels of sequence divergence and putatively clonal strain distributions, we speculate that T. denticola has been co-evolving in humans and animal hosts for a considerable period of time. However, genome sequence data from additional strains of known isolation date will be required to validate this proposition.

It should be noted that the majority of previous biophysical or culture-based investigations involving T. denticola have primarily utilized only three different (ATCC) strains: 35405^T (Clade III), 35404 (Clade I) and 33520 (Clade II); which are all of North American origin [30, 31]. Our data suggests that these three strains (lineages) may not be wholly representative of the T. denticola strains distributed within global populations. Whilst our sample size is modest, the scope of our MLSA analysis was limited by the relative paucity of T. denticola strains presently available. Oral treponemes such as T. denticola are fastidious, capricious and notoriously difficult to isolate; and there are very few laboratories in the world that actively maintain strain collections.

The ATCC 700768 (OMZ 830, China), ATCC 700771 (OMZ 834, China), OMZ 853 (China) and OTK (USA) strains, located in basal positions in the phylogenetic trees, appear to be the most genetically distant from the genome-sequenced ATCC 35405 type strain (Canada). This genetic divergence is consistent with literature reports, which have stated that these strains have notable phenotypic differences. For example, the primary sequence, domain structure and immunogenic properties of the major surface protein (Msp) in the OTK strain, were shown to be quite distinct from those of the ATCC 35405 or 33520 strains [14, 45, 46]. In another study, Wyss et al. reported that the FlaA proteins from the ATCC 700768 and ATCC 700771 strains reacted positively towards the ‘pathogen-related oral spirochete’ (PROS) H9-2 antibody (raised against T. pallidum); whilst the ATCC 35405, 35404, 33521, 33520 and ST10 strains were unreactive [15].

It is highly notable that several sets of T. denticola strains with similar genetic compositions were isolated from subjects living on different continents; i.e. the MS25 (USA), GM-1 (USA), S2 (Japan) and OKA3 (Japan) strains in Clade V; the ATCC 33520 (USA) and NY545 (Netherlands) strains in Clade II; the ATCC 33521 (USA), ST10 (USA) and OMZ 852 (China) strains in Clade IV; and the ATCC 35404 (Canada), OT2B (USA), NY531 (Netherlands), NY535 (Netherlands) and NY553 (Netherlands) strains in Clade I. We tentatively propose that this indicates that there may be a number of T. denticola clonal lineages, or closely-related clusters of strains, which have global distributions. We also identified closely-related strains that had been isolated from different subjects residing in the same geographical location: e.g. the ATCC 700771 and OMZ 853 strains from China (Clade VI).

This study represents the first in-depth multilocus sequencing approach that has been used to analyze strains belonging to a species of oral spirochete bacteria. However, it is important to note that alternative MLSA schemes have previously been used to characterize intra-species variation in other (pathogenic) spirochetes. A 21 gene MLSA approach was notably used to probe the origins, evolutionary history and possible migratory routes of T. pallidum, the causative agent of syphilis [28]. Genetic diversity within Borellia burgdorferi sensu lato, was similarly investigated using a seven gene MLSA system [27], enabling taxonomic relationships to be defined within this complex group of related (sub)-species. As far as other putative periodontal pathogens are concerned, Koehler and coworkers used a 10 gene MLSA system to investigate genetic relationships between 18 Porphyromonas gingivalis strains isolated from patients with periodontitis in Germany, and one isolate from the USA [47]. This revealed the presence of high levels of horizonal gene transfer, i.e. a panmictic population structure; quite unlike what we have found for T. denticola here. Subsequent studies have revealed that both P. gingivalis and another another ‘periodontopathogen’: Aggregatibacter actinomycetemcommitans both had specific lineages with increased association with periodontal disease; with apparently differing levels of carriage in certain ethnic groups or geographical populations [48–50]. It remains to be established whether T. denticola also possesses lineages with increased association with periodontal disease.

As the seven selected genes appear to be well-conserved in treponeme species, we envisage our MLSA framework as being readily adaptable for strain typing, as well as establishing intra- and inter-species phylogenetic relationships within diverse treponeme populations. For example, one interesting application would be to explore similarities and evolutionary relationships between closely-related strains and species of treponeme bacteria found in the human oral cavity, versus those present in animal reservoirs; especially those associated with polymicrobial tissue-destructive infections [51, 52].

Our sequencing data clearly reveals that clinical isolates of the periodontal pathogen T. denticola have highly diverse genotypes. We define 6 distinct clonal lineages present within strains isolated from subjects living in Asia, Europe and North America. Several T. denticola lineages are present on different continents, which is consistent with the existence of strains with widespread, possibly global, distributions. Our results lay the foundations for future systematic molecular investigations aimed at establishing the ecological distributions, disease associations or phylogeny of treponemes belonging to this and other species.