We read with interest the recent paper by Ros and colleagues (8) reporting recombination in the citrate synthase gene (gltA) from Wolbachia endosymbionts of filarial nematodes that belong to Wolbachia supergroups C and D. We were intrigued by this observation, since unlike the mostly parasitic relationship between Wolbachia endosymbionts and arthropod hosts, where coinfection with two or more Wolbachia strains and genetic exchange via recombination are common (1, 9), neither phenomenon had been detected previously for Wolbachia strains which mutualistically colonize nematodes (2, 5, 6). Furthermore, the nematode hosts of the Wolbachia strains implicated in the recombination event infect different mammalian hosts and are transmitted by distinct insect vectors. Since the phylogeny of filarial nematodes appears highly congruent with that of their Wolbachia endosymbionts, indicative of long, stable coevolution (3), and horizontal transfers between nematode hosts are not known, it is difficult to understand how such recombination might have occurred.
We suspected that the apparent recombination was due to a hybrid sequence previously deposited in GenBank and used in the analysis. The software used to identify recombination would not be able to discriminate such hybrids. The gltA sequence in question (GenBank accession no. AJ609644) from the Wolbachia endosymbiont of Wuchereria bancrofti (supergroup D) was obtained as part of an earlier study involving amplification of Wolbachia genes from 40 invertebrate hosts, including Brugia malayi (also supergroup D Wolbachia) and Onchocerca gibsoni (supergroup C) (4). In this study, each gltA amplicon was sequenced bidirectionally with the primers used for PCR. We believe a tracking issue during amplification, sequencing, and assembly of forward and reverse reads of the gltA fragment from the Wolbachia endosymbiont of W. bancrofti resulted in the accidental generation of a hybrid sequence. Ros et al. subsequently observed that the 5′ end of the gltA gene fragment (~210 bp) amplified from W. bancrofti is almost identical to the corresponding fragment from O. gibsoni, while the 3′ end (~625 bp) is a close match to the sequence from B. malayi, giving the appearance of recombination between Wolbachia supergroups C and D (8).
The recent draft genome sequence of W. bancrofti and its endosymbiont (http://www.broadinstitute.org/annotation/genome/filarial_worms/MultiHome.html) indicates no such recombination in the gltA genes of Wolbachia from filarial nematodes. Figure Figure11 presents an alignment of the Wolbachia gltA fragments depicted by Ros et al. (8) but is extended to include the new gltA sequence from W. bancrofti. At the 5′ end of the alignment, where 33 nucleotides of the earlier W. bancrofti sequence (GenBank accession no. AJ609644) had an identical match to O. gibsoni rather than B. malayi (8), we find that 31 of 32 positions covered by the new W. bancrofti gltA fragment exactly match the Wolbachia sequence from B. malayi and not that from O. gibsoni. The 3′ ends of both sequences from W. bancrofti are almost identical to that from B. malayi (not shown). While we cannot formally rule out the possibility that different isolates of W. bancrofti might contain Wolbachia strains with different gltA sequences, all available evidence argues against recombination in this gene. This is consistent with earlier studies that failed to detect recombination in filarial nematode Wolbachia strains (2, 5, 6) and the notion that recombination does not occur in mutualistic Wolbachia strains that show no horizontal transmission between worm hosts (6).