Divergence without difference: phylogenetics and taxonomy of Chlamydia resolved


  • Editor: Gunna Christiansen

Correspondence: Richard S. Stephens, 16 Barker Hall, Berkeley, CA 94720, USA. Tel.: +1 510 642 5940; fax: +1 510 643 1537; e-mail: rss@berkeley.edu


Members of Chlamydiaceae have been extensively characterized by complete genome sequencing. This information provides new understanding concerning their natural evolutionary history. Comparative genome analysis is consistent with the conclusion that host-divergent strains of Chlamydiae are closely related biologically and ecologically. The previous taxonomic separation of the genus based on ribosomal sequences is neither consistent with the natural history of the organism revealed by genome comparisons, nor widely used by the Chlamydia research community 8 years after its introduction; thus, it is proposed to reunite the Chlamydiaceae into a single genus, Chlamydia.


The Chlamydia field operates deep within a research framework immersed in genomics. There are complete genome sequences representing the entire range of chlamydial organisms providing unprecedented depth and breadth for comparison. To date there are 16 publicly available genomes that can be accessed and analyzed with additional genomes forthcoming (Table 1). This is a rich resource that has defined broad new visions of chlamydial biology and engaged an influx of new researchers who continue to contribute to the chlamydial research arena. Evaluating the set of chlamydial genomes gives a global picture of the genome organization and biological capacity of each species, and reveals new insights into how they have become shaped during their evolution. Chlamydia sp. have been placed in their own bacterial division because they have an evolutionary pathway deeply separated from other bacteria (Pace, 1997). Significantly, most of the different species within Chlamydiaceae similarly separated from each other early in their evolution (Stephens, 2002) resulting in a very broad spectrum of pathoadaptive species (Pallen & Wren, 2007).

Table 1.   Completed Chlamydia genomes
trachomatisA/Har-132005RML NIAID
trachomatisLGV L2/4342008Sanger Institute
trachomatisLGV L2/UCH-12008Sanger Institute
pneumoniaeJ1382000Yamaguchi University
pneumoniaeKoala2008University of Maryland
abortusS26/32005Sanger Institute
felisFe/C-562006Yamaguchi University
psittaci6BC2008University of Maryland
pecorum 2008University of Maryland
ProtochlamydiaUWE252004University of Vienna
Simkania 2008University of Maryland

One inescapable perspective is that this wealth of genome-based knowledge reframes the reference for interpreting phylogenetic history and, moreover, recalibrates the set-point for a 16S rRNA gene-based molecular clock and consequent taxonomic interpretation (Coenye et al., 2005). In 1999, it was proposed to assign chlamydial strains in the single genus, Chlamydia, to two genera, Chlamydia and Chlamydophila, based on apparent differential clustering of 16S rRNA gene (Everett et al., 1999). Because the separation of strains was not robust and lacked other consistent biological markers, the proposal was rejected by many in the scientific community (Schachter et al., 2001). Today the field suffers from a lack of a cohesive taxonomy that continues to plague the Chlamydia research community. Unfortunately this has been maintained as dictated by a singular faith in 16S rRNA gene phylogenetics without respect for the biological system; especially, a biology so unusual that it is unlike the bacterial examples that served to validate the original concepts. The classic bacterial concepts for genus and species criteria are based on data derived from a small number of fast growing organisms that enjoy much genetic exchange, including DNA exchange between different species. Chlamydiae are not meaningfully fast growing and only rarely may they exchange DNA with other highly related Chlamydiae and only within the same intracellular vacuole (Demars et al., 2007).

Species and genus

A bacterial ‘species’ is defined as a distinct set of ‘strains that have a close resemblance to one another’ where ‘each species differs considerably’ (Brenner et al., 2005). For 16S rRNA gene, a sequence similarity of <97% often marks different species. At the genus level, separate species that cluster based on 16S rRNA gene sequence similarity and/or share the ‘same general phenotypic characteristics’ are grouped together (Brenner et al., 2005). Thus phenotype continues to serve a decisive role in decisions about cutoff points for 16S rRNA gene similarity at the species and genus levels (Stackebrandt et al., 2002).

The reason for using 16S rRNA gene similarity is that it serves as a common and convenient marker among bacteria from which one can infer relationships that ideally reflect biological and ecological similarities and differences. Thus when two strains differ with <97% similarity, it can be meaningful to infer separate species because there is the expectation that this will reflect concomitant biological divergence (Palys et al., 1997; Stackebrandt et al., 2002; Gevers et al., 2005). This is reasoned because sequence similarity and clustering is usually coincident with ecologically distinct populations as any adaptive advantage of one population would result in replacement of another population that resides in the same niche (Palys et al., 1997). The evolutionary processes that drive evolution of bacterial lineages are gene gain from the microbial gene pool, gene loss, gene order, and gene mutation (Pallen & Wren, 2007).

Everett et al. (1999) cited Palys et al. (1997) as the theoretical foundation to support the taxonomic splitting of Chlamydia based on 16S rRNA gene similarity clustering; however, this basis was misunderstood as Palys et al. (1997) emphasized that there are exceptions. One notable exception is the unrestrained opportunity for neutral DNA sequence divergence for populations that live in the same environmental habitat, but are isolated geographically (Palys et al., 1997). In this case DNA sequences diverge because each isolated subpopulation cannot compete with each other nor exchange genes (Palys et al., 1997). Chlamydiae live in host-specific geographical islands, but otherwise live in the same ecological niche – that of the eukaryotic host, the eukaryotic host cell, and the inclusion vacuole. Living within the host cell, and especially within an inclusion vacuole, prevents Chlamydiae from having access to the exogenous microbial gene pool. Thus, speciation of chlamydial strains within a single genus Chlamydia is appropriate as each of the previously proposed Chlamydophila species is consistent with these fundamental principles –Chlamydia and Chlamydophila strains (1) cluster together, (2) often have <97% 16S rRNA gene sequence similarity, and (3) share all the fundamental and classically defined phenotypic characteristics.

How did this become distorted?

One can anticipate the origin of this confusion if two populations of strains become geographically isolated; their 16S rRNA gene sequences would then drift independently collecting unique mutations during geologic time. However, because each strain lives in an identical environmental niche with common selective pressures, there will be little biological change despite obtaining many nucleotide differences. In such a situation, inferences by percent similarity of 16S rRNA gene have very different levels of scale to be consistent with a taxonomy reflecting natural evolutionary relationships and cohesive differences in ecological niche. It may be convenient and even meaningful to link host adaptation to a species definition (Gevers et al., 2005), even if it will fail in some cases where Chlamydiae are found in multiple hosts (e.g. Chlamydia pneumoniae). Nevertheless, to establish a genus-level split at <95% 16S rRNA gene similarity (Everett et al., 1999) is arbitrary and scientifically untenable – independent of the previously elaborated inconsistencies (Schachter et al., 2001).

Gene print of evolution

How can genomic perspectives be used to resolve this mistake in nomenclature? Analyses of entire genomes are the ultimate measure of the similarities and differences in the biology and evolution of organisms as a genome documents the full natural history of the organism (Coenye et al., 2005). Two pertinent and fundamental observations can be immediately appreciated when comparing chlamydial genomes. The first observation is the reduction in genome size as a result of gene loss. Indeed, point mutation and gene loss may be the only mechanisms of divergence for Chlamydiae. This is apparent in 16S rRNA genes (Mira et al., 2001), and for Chlamydiae there has been the loss of multiple copies of 16S rRNA genes even to a single copy for some strains. Thus, adaptation to specific hosts has been dependent on minor mutation and gene loss rather than by the acquisition of new genes. The second observation is that the gene content and gene order (synteny) are extremely high. For almost every other bacteria, loss of gene synteny is one of the first changes to occur as strains diverge (Pallen & Wren, 2007). It is common that even between strains of the same species, there is virtually no discernable conservation of gene order beyond the operon (Coenye et al., 2005). The fact that there is nearly 80% conservation of genes and gene order between, for example, Chlamydia trachomatis, C. pneumoniae, and Chlamydia abortus is remarkable and reflects the common biological foundations of their intracellular niche, the loss of full systems for recombination (e.g. lack of RecG), and the lack of access to the microbial gene pool.

Developmental cycle limits evolution

Another essential biological process that profoundly affects the population structure and the evolution of Chlamydiae is their developmental cycle. The requirement to complete the cycle and produce infectious elementary bodies (EB) for strain propagation places a profound constraint on divergence of functional properties. Notably, the developmental cycle does not constrain neutral or even mildly deleterious mutations. In this case a biologically successful EB will often ‘trap’ neutral mutations that then become fixed in the population at higher rates than expected, analogous in some respects to the host bottleneck for endosymbionts (Moran, 1996).

Thus for Chlamydiae there are three important and fundamentally unique biological attributes that affect the appropriate understanding of simple counting of mutations in a single gene: (1) the bottleneck of the developmental cycle, (2) the geographic isolation within host cells and in different hosts, and (3) the paucity of horizontal gene transfer. All of these strongly bias the scale and, hence, interpretation of 16S rRNA gene differences between chlamydial strains.

Taxonomy and genomics

16S rRNA gene phylogenetic analyses demonstrating a split between the C. trachomatis group and all other Chlamydiae was always uncertain as it is supported by bootstrap analysis only 68% of the time (Everett et al., 1999). In independent studies that evaluated protein distribution based on complete chlamydial genome sequences, the distribution of proteins was not consistent with this early separation (Griffiths et al., 2006; Gupta & Griffiths, 2006). Similar inconsistency was observed when evaluating chlamydia-specific indels (Gupta & Griffiths, 2006). It is also noted that the chlamydial plasmid found in both proposed genera are very similar and the plasmid was likely acquired before their divergence (Griffiths et al., 2006). Analysis of the distribution of all encoded ORFs for C. trachomatis, C. pneumoniae, and Chlamydia caviae revealed one protein that is present in C. trachomatis and C. pneumoniae, but not in any of the other members of the ‘Chlamydophila cluster’ (Griffiths et al., 2006). This fact is inconsistent with two independent monophyletic branches. Moreover, only 20 proteins were unique for each of the two proposed genera and all of these were uncharacterized or hypothetical proteins (Inc proteins can be included in this category because we do not know their function). It can be seen that there are 89 protein differences among strains of Helicobacter pylori and over 200 differences for strains of Neisseria meningitidis (Lan & Reeves, 2000) – these are differences among strains of the same species. Given the ‘sporadic distribution’ of proteins across the chlamydial genera and unknown functions of these proteins, the biological and evolutionary interpretation of a meaningful difference is equivocal (Griffiths et al., 2006).

These findings challenge and undermine the validity of the proposal for separate genera. There is agreement that these chlamydial strains may be defined as different species and that there has been early divergence and separation of the chlamydial species during their evolution. This separation is often associated with isolation to a specific host. The most parsimonious and scientifically robust decision is that these are relationships consistent with species-level, not genus-level, affiliation.

Taxonomy resolved

The proposal by Everett et al. (1999) set the boundary arbitrarily at <95% similarity in 16S rRNA gene as the threshold for deciding which of two genera chlamydial strains should be placed. It is inconsistent that this criterion is not met by most species (Fig. 1). Nevertheless, with several exceptions, they are all >97% similar, which is typically used for bacterial speciation, not genus-level demarcation. The deception is the apparent bifurcated clustering of chlamydial species by 16S rRNA gene sequence comparison; however, this is merely a biased outcome of their early separation in geologic time, their isolation in geographic host islands, and the evolutionary constraints imposed by the chlamydial developmental cycle.

Figure 1.

Chlamydia species names showing % 16S rRNA gene similarity between species. Data from Everett et al. (1999). Note: none of the relationships meet the <95% similarity level proposed for the family separation into two genera especially between Chlamydia caviae and Chlamydia muridarum at 96% (Everett et al., 1999).

A phylogenetic analysis based on 110 concatenated genes conserved in all Chlamydia genomes confirms and strengthens the close and linked evolutionary relationship among Chlamydiae (Fig. 2). In this tree, species trachomatis, muridarum, pneumoniae and pecorum cluster away from species felis, caviae, psittaci and abortus, further depreciating the value of 16S rRNA gene-based trees for the purpose of taxonomic classification in the Chlamydiaceae.

Figure 2.

 One hundred and ten conserved proteins from the genomes of Chlamydia trachomatis, Chlamydia pneumoniae, Chlamydia muridarum, Chlamydia pecorum, Chlamydia felis, Chlamydia psittaci, Chlamydia abortus and Chlamydia caviae were identified using the blast score ratio (BSR) approach [BSR ≥0.8; Rasko et al. (2005)]. Protein clusters were aligned using clustalx (Thompson et al., 1997) and back-translated into nucleotide alignments using transalign, part of the emboss (Rice et al., 2000) software package. Concatenated aligned genes, spanning a total of 121 674 positions with a sequence similarity of 82.2% and identity of 58.8% were further analyzed by the HKY85 method (Hasegawa et al., 1985) with 200 bootstrap replicates, and the results used to generate a phylogenetic tree by application of the PhyLM algorithms (Guindon et al., 2005).

With genomics, we now understand the true evolution of Chlamydiae as they began their separation into different hosts 60–100 million years ago (Stephens, 2002), but remained true to the common constraints of their intracellular and intravacuolar niche. Today this natural history is reflected in the accumulation of mutations and gene loss while retaining their unifying biology. It can be confidently concluded that the spectrum of divergence among Chlamydiae is without generic difference.

It is of paramount importance to have a taxonomic nomenclature that is meaningful and useful for the research and medical communities, and not limited to the view promoted by rRNA systematics that disregards the unique biology of this system. The split of the genus was proposed over 8 years ago and still has not been accepted by the scientific community. This is easily and objectively established by the overwhelming majority of the publications that use only ‘Chlamydia pneumoniae’ (e.g. 81% in 2006) as the name vs. those that use only ‘Chlamydophila pneumoniae’ (16% in 2006). Such analysis demonstrates a tacit rejection of the two-genus nomenclature by researchers in the field, in spite of its inherent bias in favor of the two-genus system by neither including the quantitatively predominant research articles that include C. trachomatis in the title, nor accounting for reluctant authors who are required to use the ‘official’ two-genus system by editors. Unlike phylogeny, taxonomy is primarily a tool meant to facilitate analysis. The genus Chlamydophila is apparently still an unpractical tool for the vast majority of its users 8 years after it was first introduced. It is time now to end the taxonomic confusion by adopting the one-genus, multiple-species nomenclature (Fig. 1) that is both practical and scientifically justified.


The authors sincerely thank S. Dela Cuesta, K. Hybiske, S. Abromaitis, S. Ruiz, and B. Hoft for their constructive and insightful criticisms and discussions concerning the content of this manuscript. This work was supported in part by NIH RO1 AI51472 (G.M. and P.M.B.).


Adapted with permission from Proceedings of the 6th Meeting of the European Society for Chlamydia Research, Aarhus, Denmark, July 1–4, 2008, Gunna Christiansen, Editor.