Taxonomic status of the intracellular bacterium Wolbachia pipientis

Introduction

Wolbachia pipientis (class Alphaproteobacteria, order Rickettsiales) and its taxonomically undescribed close relatives are intracellular, maternally transmitted symbiotic bacteria found in more than 20 % of all insect species, numerous other arthropods and filarial nematodes (Bandi et al., 1998; Jeyaprakash & Hoy, 2000; Plantard et al., 1999; Taylor & Hoerauf, 1999; Wenseleers et al., 1998; Werren et al., 1995; Werren & Windsor, 2000). They are well known as manipulators of arthropod host reproduction, causing cytoplasmic incompatibility, feminization, parthenogenesis and male-killing (Stouthamer et al., 1999; Werren, 1997), and have, more recently, been shown to act as obligate mutualists in filarial nematodes (Bandi et al., 2001; Taylor & Hoerauf, 1999).

W. pipientis was discovered in 1924 in the cells of the mosquito Culex pipiens (Hertig & Wolbach, 1924) and was formally described in 1936 (Hertig, 1936). In recent decades, the use of PCR and sequencing techniques has revealed a large number of bacteria with close phylogenetic affinity to W. pipientis. Although they form a clear monophyletic cluster with W. pipientis, many of these undescribed strains show significant differences in the 16S rRNA gene when compared to the type strain (i.e. more than 3 %). Due to some uncertainty about whether such bacteria might represent different species, most researchers in the field now refer to these strains simply as Wolbachia. This term, although convenient and well-accepted, is not taxonomically legitimate. Owing to the growing interest in the biology and applications of W. pipientis and its relatives, a uniform nomenclatural system for these bacteria is desirable. One way to resolve this issue is to recognize all closely related strains as W. pipientis; an alternative would be to split the group into two or more species. In this note, we argue for the former option, based on results from phylogenetic and recombination studies, as well as a taxonomic precedent in other symbiotic bacteria.

Molecular phylogenetic and recombination analyses of W. pipientis

No W. pipientis strains have yet been isolated in pure culture, largely because of their fastidious requirements. Traditional methods for bacterial species and strain determination, such as those involving DNA–DNA hybridization and biochemical parameters, have therefore not been feasible. Early phylogenetic analysis of 16S rRNA gene sequences assigned W. pipientis clearly to the Alphaproteobacteria, revealing a sister-group relationship with the genera Ehrlichia, Anaplasma and Neorickettsia (O'Neill et al., 1992). This study showed that W. pipientis sensu stricto formed a monophyletic clade with other insect-associated micro-organisms and suggested ‘the classification of these bacteria as members of the same species' (O'Neill et al., 1992).

Further 16S rRNA-based studies of intracellular bacteria from various arthropods and filarial nematodes showed that they formed a monophyletic group with W. pipientis (Rousset et al., 1992; Sironi et al., 1995; Stouthamer et al., 1993) (see also Dumler et al., 2001). A study of several different W. pipientis arthropod strains based on the more variable gene ftsZ showed that two divergent clades exist (A, B) (Werren et al., 1995). The type strain (from C. pipiens) was found within clade B. A subsequent analysis of intracellular bacteria from filarial nematodes found that they formed two additional clades (C, D) (Bandi et al., 1998). These clades have since been termed ‘supergroups’; a schematic summary of currently recognized supergroups is shown in Fig. 1⇓. A number of other supergroups have been proposed more recently, including supergroups E (Collembola) (Czarnetzki & Tebbe, 2004; Vandekerckhove et al., 1999), F (arthropods and filarial nematodes) (Campbell et al., 1992; Casiraghi et al., 2005; Lo et al., 2002; Rasgon & Scott, 2004), G (spiders) (Rowley et al., 2004) and H (termites) (Bordenstein & Rosengaus, 2005). A number of divergent lineages, including those from various flea species and the filarial nematode Dirofilaria repens, have yet to be designated supergroups (Casiraghi et al., 2005; Gorham et al., 2003).

Relationships among the various supergroups are not well understood, primarily because of the absence of a suitable outgroup for rooting the inferred trees (Lo et al., 2002). The closest known relatives of W. pipientis are members of a clade including the genera Anaplasma and Ehrlichia, which have corrected 16S rRNA gene sequence identities of ∼90 % with W. pipientis. The last common ancestor of W. pipientis and these outgroups is likely to have existed several hundred million years ago, which has led to the erosion of phylogenetic signal in many sequence characters. In a recent study (Casiraghi et al., 2005), three protein-coding genes were combined and analysed, including their homologues in Anaplasma and Ehrlichia as outgroups. Using a statistical test that distinguishes between competing hypotheses (Shimodaira & Hasegawa, 1999), the scores of a number of different trees with differing outgroup positions were compared. It was found that no outgroup placement was significantly more likely than another. Without a reliable root for the tree, it is not possible to assign supergroups confidently to different, monophyletic species. A more prudent approach is to consider all strains as members of W. pipientis.

A number of recent studies have shown that exchange of genetic information has occurred among supergroups. Evidence has been found for recombination of genes within and between supergroups A and B (Baldo et al., 2005, 2006a; J. C. Dunning Hotopp and J. H. Werren, unpublished data). Exchange of genes/gene fragments within bacteriophages that infect different W. pipientis supergroups has also been demonstrated (Bordenstein & Wernegreen, 2004; Masui et al., 2000). Genome sequencing studies show that W. pipientis has the genetic machinery for gene transfer (Wu et al., 2004) and a number of arthropods have been shown to be infected by both A and B representatives (Werren et al., 1995; Werren & Windsor, 2000), which would provide a means of gene flow. It has been clearly demonstrated that recombination in the wsp gene encoding an outer-surface protein is widespread among supergroup A and B bacteria (Baldo et al., 2005; Jiggins et al., 2001; Werren & Bartos, 2001). This gene should therefore be avoided when trying to understand relationships among W. pipientis strains. Due to the level of recombination in W. pipientis and close relatives, reliable strain identification requires a multilocus strain typing (MLST) approach (Maiden et al., 1998). A comprehensive MLST system has been developed for W. pipientis using five standard housekeeping genes (Baldo et al., 2006b; Paraskevopoulos et al., 2006). Results reveal frequent recombination and a remarkable diversity among W. pipientis: 35 of 37 haplotypes were unique (Baldo et al., 2006b). Such frequent recombination is consistent with W. pipientis and its closely related strains being members of one species.

Proposal to recognize all Wolbachia formally as members of the species W. pipientis

At present, the majority of strains that cluster with W. pipientis do not have a proper scientific name. We and other W. pipientis researchers frequently encounter dissatisfaction from the microbial community over this issue, particularly editors and reviewers who request the addition of ‘pipientis’ to the title of manuscripts. As interest in these bacteria continues to increase, it is important to minimize taxonomic confusion within the broader community.

In the case of the aphid obligate primary symbionts, only one species name, Buchnera aphidicola (Munson et al., 1991), is recognized and commonly used by researchers in the field. This is despite the fact that separation of the main lineages of these bacteria occurred a few hundred million years ago, and corrected genetic divergences in 16S rRNA typically exceed 8 % (Moran et al., 1993). Strains of B. aphidicola are simply known by the host aphid they infect.

On the basis of results from phylogenetic and recombination analyses, as well as the taxonomic precedent set by aphid-symbiotic bacteria, we propose that bacteria with a close phylogenetic affinity to W. pipientis be formally included in this species. This includes those bacteria currently within the recognized supergroups (A–H) and those that display a similar genetic distance from the type strain. This was the consensus reached at a recent international conference on W. pipientis (Heron Island, Australia, August 2004). We note that this formal recognition will not prevent researchers from continuing to use the common name ‘Wolbachia’ in their papers. Without formal recognition of all strains as W. pipientis, the research community runs the risk of prematurely naming novel species in the genus. Because W. pipientis is so abundant in insects, with at least 2–6 million infected species (Werren et al., 1995), naming species based on host associations would lead to an unmanageable proliferation of species names.