OUP user menu

A new Sodalis lineage from bloodsucking fly Craterina melbae (Diptera, Hippoboscoidea) originated independently of the tsetse flies symbiont Sodalis glossinidius

Eva Nováková, Václav Hypša
DOI: http://dx.doi.org/10.1111/j.1574-6968.2006.00620.x 131-135 First published online: 1 April 2007


Symbiotic bacterium closely related to the secondary symbiont of tsetse flies, Sodalis glossinidius, has been described from the bloodsucking fly Craterina melbae. Phylogenetic analysis of two genes, 16S rRNA gene and component of type three secretion system, placed the bacterium closer to the Sitophilus-derived branch of Sodalis than to the tsetse symbionts. This indicates that the Craterina-derived lineage of Sodalis originated independent of the tsetse flies symbionts and documents the capability of Sodalis bacteria either to switch between different host groups or to establish the symbiosis by several independent events.

  • Sodalis
  • symbiont
  • molecular phylogeny
  • 16S rRNA gene
  • type three secretion system
  • Craterina


The symbiotic bacterium Sodalis glossinidius (Dale & Maudlin, 1999) was originally reported as one of three phylogenetically distant bacteria inhabiting tissues of tsetse flies (Aksoy et al., 1997). The other two bacteria are the mutualistic Wigglesworthia glossinidia (Aksoy, 1995) and the broadly distributed parasite of the genus Wolbachia (O'Neill et al., 1993). Since its discovery, Sodalis glossinidius has become an important subject of a symbiosis-centered research due to several favorable circumstances. Primarily, it attracted attention as a companion of an important bloodsucking insect and was recognized as a possible factor influencing the vector competence/capacity (Maudlin & Ellis, 1985). While this view was later opposed by several authors (Moloo & Shaw, 1989; Geiger et al., 2005b), the research of Sodalis glossinidius further accelerated after Dale (2001) demonstrated that it utilizes the machinery of the type three secretion system (TTSS) to enter the host cell. Owing to this finding and the fact that Sodalis could be maintained in in vitro laboratory cultures (Welburn et al., 1987), Sodalis glossinidius became a model organism, particularly in studying host cell invasion in an early stage of the symbiosis evolution. As a result of this attention, the complete genome of Sodalis glossinidius has recently been sequenced and analyzed (Toh et al., 2006). The preliminary analysis revealed a number of genomic changes associated with the adoption of a symbiotic lifestyle and further stressed the significance of Sodalis as a model of evolutionary transition from free-living to symbiotic bacteria. However, genetic characterization of a symbiont and the elucidation of its molecular machineries only represent one facet of the evolutionary picture. To understand fully the biology and the evolutionary potential of such a bacterium, it is important to know its distribution among various hosts, the routes of interspecific transmission and the modes of symbiosis it can adopt in different hosts. This can be well demonstrated by the finding that Sodalis-related bacteria, associated with the weevils of the genus Sitophilus, display features typical for mutualistic primary symbionts (Heddi et al., 1998; Nardon et al., 2002; Lefevre et al., 2004). Such a phylogenetic-distribution pattern indicates that Sodalis may be capable of both the horizontal transfers between different hosts, and the long-term coevolution associated with establishment of a mutualistic relation. This makes the bacterium a potentially excellent model for the study of the changes and trends accompanying the adaptation to different symbiotic modes. Unfortunately, despite the considerable number of symbiotic bacteria described annually from various insects, no other closely related bacterium has been identified so far. In this study, a new lineage is reported from a bloodsucking relative of tsetse flies: the hippoboscoid species Craterina melbae (Diptera; Hippoboscidae). Based on two different genes, it is shown that this symbiont has been acquired independent of the S. glossinidius inhabiting the tsetse flies.

Materials and methods

The sample of Craterina melbae was provided by Pierre Bize, University of Lausanne, Switzerland. Total DNA was isolated from tissue of three adult insects using the DNEasy Tissue Kit (QIAGEN) and used with three different sets of primers for the PCR amplification. Primer pair F40: 5′-GCGGCAAGCCTAACACAT-3′ and R1060: 5′-CTTAACCCAACATTTCTCAACACGAG-3′ was designed to amplify a 1200 bp of 16S rRNA gene; the primer SodF 5′-ACCGCATAACGTCGCAAGACC-3′ was designed to amplify together with R1060 c. 1000 bp long fragment of 16S rRNA gene from Sodalis and related bacteria, and the primers SpaPQRf: 5′-ATGATGATGATGAGCCCG-3′ and SpaPQRr: 5′-AGCCCATGCATAACCCAAAA-3′ were adopted from Dale (2001) and used to amplify components of the TTSS. The identity and preliminary phylogenetic position of the obtained sequences were checked by blast Search, NCBI (http://www.ncbi.nlm.nih.gov). The phylogenetic relationships of both genes, the 16S rRNA gene and the Spa component, were further analyzed in a broader taxonomic context. For 16S rRNA gene, an additional 33 sequences were retrieved from the GenBank (NCBI); they included all available Sodalis 16S rRNA gene sequences and 13 other symbiotic and free-living bacteria as outgroups (Fig 1a). Owing to a high similarity of the sequences, the alignment was unequivocal and did not require any testing for aligning parameters. The matrix was designed manually in bioedit program (Hall, 1999) and the few ambiguous positions were removed before the analysis. Phylogenetic analysis was performed in paup* (Swofford, 1998) by maximum parsimony (MP) and maximum likelihood (ML) methods using the TBR swapping procedure with 30 replicates of random sequence addition. The MP analysis was performed under the transition: transversion ratio set to 1 : 1, 2 : 1 and 3 : 1, and the strict consensus of all trees was computed. The bootstrap support was obtained by 1000 replicates of MP analysis with Ts/Tv set to 1 : 2. The evolutionary model for ML analysis, was determined in Modeltest 3.6 (Posada & Crandall, 1998). To analyze the Spa genes, the sequences were aligned with those from Sodalis glossinidius, Sitophilus zeamais-symbiont and two outgroups represented by Chromobacterium violaceum and Salmonella enterica. Two regions of the Spa fragment displayed homology sufficient for an alignment and meaningful phylogenetic analysis. They included partial SpaP (84 bp) and SpaR (231 bp) genes extracted from the following GenBank sequences: AE016825 (Chromobacterium violaceum), AF306650 (Sodalis glossinidius), AF426456 (symbiont of Sitophilus zeamais), and AL627276 (Salmonella enterica). The matrix was analyzed by an exhaustive search for the MP tree and examined by eye for the potential molecular synapomorphies. Within the Sodalis group, the genetic distances for all genes were calculated in paup* using the Jukes–Cantor model.

Figure 1

(a) One of the 38 MP trees obtained by the analysis of 16S rRNA gene matrix under different Ts/Tv ratios (CI of the trees varied from 0.6 to 0.63). The consensus of all trees and compatible with ML analysis is designated by bold lines. The names of host taxa (in brackets) or the bacteria are provided together with their sequence accession numbers. The dashed line indicates an alternative position of the Sitophilus rugicollis symbiont as obtained under the nonhomogeneous model. The Sitophilus-derived symbionts are labeled SOPE, SGPE, SZPE and SRPE for Sitophilus oryzae, Sitophilus granarius, Sitophilus zeamais and Sitophilus rugicollis, respecively; Sgl, Sodalis glossinidius; (b) the MP tree of SpaP+SpaQ genes retrieved by an exhaustive search (CI=0.99). In both trees, bootstrap values higher 50% are printed at the nodes.

The sequences amplified from Craterina melbae were deposited in GenBank under the accession numbers EF174495 (16S rRNA gene) and EF174496 (Spa genes).

Results and discussion

Phylogenetic analysis of the 16S rRNA gene, performed together with the two previously described Sodalis lineages (inhabiting tsetse flies and weevils), revealed that the Craterina-derived bacterium is more closely related to the Sitophilus symbionts than to Sodalis glossinidius (Fig. 1a). An identical arrangement, although within a taxonomically more restricted sample, was obtained when the SpaP and SpaR genes were analyzed, either as two independent data sets or within a single combined matrix (Fig. 1b). In all of the analyses, the position of Craterina-derived bacterium as a sister taxon of weevil symbionts has been unequivocal and no other arrangement was retrieved. Moreover, this position was also well supported by the tree parameters, the distances within the Sodalis group (Table 1) and several clear molecular synapomorphies within the Spa genes (including a unique and relatively long motif FSGSIV within the SpaR region; Fig. 2). This phylogenetic picture indicates that the Sodalis bacterium in Craterina has not been inherited from a common Glossinidae–Hippoboscidae ancestor, but was acquired by an independent event. Such a result is far from unexpected as the low 16S rRNA gene diversity within Sodalis glossinidius has been considered an evidence of multiple symbiont acquisition even within the tsetse flies (Aksoy et al., 1997). On the other hand, the monophyly of Sodalis glossinidius isolates is difficult to reconcile with such a multiple-acquisition scenario. Moreover, a recently performed amplified fragment length polymorphism (AFLP) analysis of Sodalis glossinidius from two different species of tsetse flies showed that they constitute two genetically distinct host-specific clades (Geiger et al., 2005a). Owing to this contrast between the Sodalis glossinidius monophyly and genetic structure on the one hand and low overall DNA diversity on the other, the process underlying current distribution of Sodalis glossinidius has never been satisfactorily elucidated. As a possible solution for this puzzle, Aksoy (1997) suggested a physiological constraint preventing Sodalis gossinidius from invading other insect groups once it adopted the transmission route via the ‘milk glands’. The finding, presented in this paper, of a clearly independent lineage of Sodalis in another hippoboscoid insect, demonstrates the general capability of this bacterium either to switch between different host groups or to establish the symbiosis by several independent events.

View this table:
Table 1

Distances calculated by Jukes–Cantor model for the sequences within and between the clades of Sodalis group (see Fig. 1 for list of sequences)

Figure 2

Aligned regions of SpaP and SpaR genes. The potential synapomorphies are designated by open boxes.

Previously, Lefévre et al. (2004) hypothesized that the Sodalis glossinidius had been acquired by horizontal transfer from Dryophthoridae weevils to tsetse flies. They based this suggestion on the intermediate position of Sodalis glossinidius between the symbiont of Sitophilus rugicollis and the cluster encompassing the symbionts from all other Sitophilus spp. When the relationships among the ‘Sodalis’ lineages were investigated in a different context of symbiotic and free-living taxa by standard MP and ML methods, a different topology was obtained, with the Sitophilus rugicollis symbiont forming a distant lineage independent of other Sitophilus symbionts (Fig. 1a). This arrangement can be easily integrated into the general view of several symbiont replacements within the Dryophthoridae (Lefevre et al., 2004). Clearly, the problem of the Sitophilus rugicollis symbiont rests in the characteristic of its 16S rRNA gene. Compared with the typical Sodalis 16S rRNA gene, the sequence of the Sitophilus rugicollis symbiont is highly aberrant. It rather resembles those known from most of the P-symbionts, including the Dryophthoridae R-clade reported by Lefevre (2004), i.e. the sequences typical with a considerable change of AT/GC ratio. As this phenomenon can strongly influence the phylogenetic picture, the application of algorithms dealing with such homogeneities has been suggested in several studies (Galtier & Gouy, 1995, 1998; Lefevre et al., 2004; Herbeck et al., 2005). Indeed, when the distance method was used based on a nonhomogeneous model implemented in the program PhyloWin (Galtier et al., 1996), the symbiont from Sitophilus rugicollis clustered at the base of all Sodalis sequences (Fig. 1a). Nevertheless, it is important to stress that the Sitophilus rugicollis issue is not crucial for solving the Craterina melbae-symbiont phylogeny and has not been thoroughly addressed in the present analysis (in fact, it is suspected that the available 16S rRNA gene information is not sufficient to solve the phylogenetic position of all Sodalis-related symbionts). The most important finding in this respect is that regardless of the exact position of the Sitophilus rugicollis symbiont, the sequence derived from the Craterina melbae invariantly clusters as a sister group of the bacteria isolated from the three related Sitophilus species (Sitophilus zeamais, Sitophilus oryzae and Sitophilus granarius; Fig. 1a). The finding of an independent Sodalis lineage in another bloodsucking hippoboscoid species indicates that this bacterial lineage may be more broadly distributed than evidenced by the available symbiont records. While such a possibility might solve the contradiction between the monophyly and low diversity of Sodalis glossinidius (by showing that the Sodalis glossinidius is not monophyletic but only represents a sample from more broadly distributed bacterium), it raises the question of why these bacteria are so overlooked. One typical difficulty with screening for the symbiotic bacteria is associated with the specificity of 16S rRNA gene primers. It is shown elsewhere (Hypsa & Krizek, 2007) that specifically designed primers can detect a lineage of symbiotic bacteria that are otherwise ‘invisible’ to regularly used universal primers. This may occur either due to the predominance of other bacterial symbionts in the same sample or due to the mutations within the critical region of the primer sequence. As the symbiotic bacteria are known to accumulate mutations within their genomes, including otherwise conservative regions, the danger of losing priming sites is likely to be higher than in free-living bacteria. To facilitate screening for other Sodalis lineages, a specific 16S rRNA gene primer was designed based on the known Sodalis sequences (see ‘Materials and methods’). We tested these primers using the Craterina-derived symbiont as a positive sample and another S symbiont, Arsenophonus triatominarum, as the negative sample, and their specificity was confirmed for Sodalis 16S rRNA gene. These markers, together with the Spa-specific primers, might provide a tool allowing for identification of other Sodalis lineages and a more complex investigation of the distribution and biology of these interesting symbiotic bacteria.


The authors are grateful to Pierre Bize for providing the Craterina samples. This work was supported by Grant 206/04/0520 (Grant Agency of the Czech Republic) and the grants LC06073 and MSM 60076605801 (Ministry of Education, Czech Republic).


  • Editor: Marco Soria


View Abstract