New names for old strains? Wolbachia wSim is actually wRi

Short abstract

A response to Serendipitous discovery of Wolbachia genomes in multiple Drosophila species by SL Salzberg, JC Dunning Hotopp, AL Delcher, M Pop, DR Smith, MB Eisen and WC Nelson. Genome Biology 2005, 6:R23

Abstract

A response to Serendipitous discovery of Wolbachia genomes in multiple Drosophila species by SL Salzberg, JC Dunning Hotopp, AL Delcher, M Pop, DR Smith, MB Eisen and WC Nelson. Genome Biology 2005, 6:R23

A recent paper published by Salzberg et al. [1] reports the discovery, assembly and comparative analysis of three partial Wolbachia endosymbiont genomes. These data were retrieved from the Trace Archive [2] from sequencing projects that were focused on the endosymbiont hosts - Drosophila simulans, D. ananassae and D. mojavensis - using the fully sequenced wMel Wolbachia genome [3] as a probe. Salzberg et al. refer to these partial genomes as belonging to Wolbachia strains wSim, wAna and wMoj respectively [1]. These strain names are new constructions and it appears that the annotated wSim genome sequence is essentially identical to the previously described wRi strain [4] and should be named accordingly.

There is a large body of previous work on the biology of Wolbachia infections of D. simulans. To date, five Wolbachia strains have been described from D. simulans (for a review see [5]), three of them belonging to group A, wAu [6], wRi [7] and wHa [4], and two belonging to group B, wNo [8] and wMa [9]. When the partial genome sequence of wSim [1] is compared to previously published sequences of the different D. simulans Wolbachia strains, it is clear that wSim is most likely to be the wRi Wolbachia strain that has been extensively studied over the years. Blastn analysis of numerous wRi sequences available at GenBank (accession numbers X61770, 16S rRNA; AB002288, groES and groEL; AB036661, bacteriophage WO gene for capsid protein; AF348330, ubiA, rbfA, infB, nusA, and acrD genes; AJ012073, glnA and dnaA genes and two genes encoding hypothetical proteins; and AJ580923, wspB gene) reveals that the wRi sequences are 99-100% identical to the partially assembled wSim genome [1]. On the basis of the molecular data publicly available in National Center of Biotechnology Information (NCBI) databases it is apparent that the strain designated as wSim by Salzberg et al. [1] is actually wRi. This strain was first described phenotypically by Hoffmann et al. in 1986 [7] in D. simulans collected in Riverside, California. wRi is characterized by the ability to induce high levels of cytoplasmic incompatibility (CI) in its native D. simulans host [7] and has the ability to spread quickly through host populations by the induction of CI [10,11]. Biogeographic studies have revealed that wRi is currently the most abundant strain infecting continental populations of D. simulans [12].

Finally, the Trace Archive for D. simulans contains reads from various D. simulans lines [13] of different biogeographic origin: wsim501 and sim6, both North American and most likely infected by wRi, and simNC48S from New Caledonia and potentially infected with wNo and wHa [12]. Therefore, it would be helpful if the authors could clarify which Trace data were used for the assembly of the wSim genome, as it might be possible that the assembly reported is chimeric, containing predominantly sequences from wRi and possibly some sequence from other Wolbachia strains.

While the discovery of these partial genome sequences in the Trace Archive is an exciting development, it is important that the finding is connected to the large established literature in this field if the data is to be of most value to the scientific community.

Julie Dunning Hotopp, William C Nelson and Steven L Salzberg respond:

We are aware that our newly discovered Wolbachia strain from the ongoing D. simulans sequencing project, which we have designated wSim [1], might be the same as wRi, as Iturbe-Ormaetxe et al. claim. Unfortunately, the evidence to support this claim, which is entirely based on sequence similarity, fails to distinguish it from other hypotheses. Iturbe-Ormaetxe et al. searched wSim against fragments of several D. simulans Wolbachia strains and found that wRi was the best match; from this they conclude that wSim and wRi are the same. If one searches these same wRi fragments against wAna, however, one finds an even closer match to wAna.

The small number of wRi genomic fragments available in GenBank (representing less than 18 kilobases (kb), not 'numerous sequences' despite the contention of Iturbe-Ormaetxe et a/.) are diverging too slowly to be used for definitive strain identification; in some cases even the wRi and wMel sequences cannot be differentiated. The wsp gene is simply missing from our wSim assembly, but is 99.9% identical between wAna and wRi. The wRi sequence of wspB is 99.2% identical over 788 base-pairs (bp) to wAna and 98% identical over 226 bp to wSim. The two longest genome fragments of wRi, AF348330 (9,235 bp) and AJ012073 (4,838 bp), match wSim and wAna equally well. Clearly, wRi, wSim, and wAna are closely related, as discussed in Table 2 of our paper [1], but if one uses sequence identity to assign strain designations, then wRi looks more like wAna than wSim.

As should be apparent from this analysis, the assertion made by Iturbe-Ormaetxe et al. that wSim = wRi rests on a logical fallacy; that is, that if the best unidirectional BLAST matches of genome A (wSim) correspond to genome B (wRi), then A = B. This ignores that fact genome B might have a better match to genome C - in this case wAna. Even more critical is the fact that only a tiny fraction of wRi has been sequenced. The BLAST analysis shows only that wSim and wRi are highly similar across a few sequence fragments representing less than 1.5% of their genomes.

We are aware that D. simulans has been reported to carry the wRi strain as well as the strain we designate wSim, and that some of the sequenced D. simulans strains carry the white mutation [13,14]. It should be noted, however, that although the D. simulans sequencing project included a mixture of three Drosophila strains, virtually all (99.9%) of the wSim sequences came from just one strain, sim6; thus both wSim and wRi were found in the California population of D. simulans. Neither this nor the BLAST alignments are, however, sufficient evidence to collapse the strains into one: Wolbachia species from closely related insect species often retain different strain identifiers [15-17] despite sharing some identical gene sequences. This is important because sometimes these Wolbachia infections result in different host phenotypes [16]. Less commonly, Wolbachia species with identical wsp genes isolated from the same insect species (for example, D. simulans) retain different strain designations [15].

This nomenclature is also common in other prokaryotes. Organisms with identical multi-locus sequencing typing (MLST) profiles isolated from the same geographical area will be given different strain designations to preserve information about their origin. This may be important if they have genomic rearrangements and single-nucleotide polymorphisms (SNPs) that confer different phenotypes. In Wolbachia, genomic rearrangements appear common [1,3], which may support the maintenance of separate strain designations to differentiate ancestry. In the absence of complete genome sequences, definitive genotyping assays, or phenotypic characterization of wSim, resolving strain differences is clearly complicated and beyond the scope of our paper.