Multilocus sequence analysis (MLSA) of ‘Rickettsiella costelytrae' and ‘Rickettsiella pyronotae’, intracellular bacterial entomopathogens from New Zealand

Authors


Correspondence

Andreas Leclerque, Julius Kühn-Institut (JKI), Institut für Biologischen Pflanzenschutz, Heinrichstraße 243, 64287 Darmstadt, Germany. E-mail: leclerque@hotmail.com

Abstract

Aims

Larvae of scarab beetles live in the soil and are frequently hosts for microbial pathogens. In New Zealand, larvae of the grass grub, Costelytrae zealandica (Coleoptera: Scarabaeidae), and manuka beetles, Pyronota spp. (Coleoptera: Scarabaeidae), have been collected from field populations showing loss of vigour and a whitened appearance. Diagnosis indicated an intracellular infection of fat body tissues by Rickettsiella-like micro-organisms. Rickettsiella bacteria are under evaluation as a possible new source of insect bio-control agents for important agricultural pests as, e.g. scarabaeid and elaterid larvae. The present study aimed at the unequivocal molecular taxonomic identification and comparison of the bacteria associated with Costelytra and Pyronota.

Methods and Results

Electron microscopy and phylogenetic reconstruction using a multilocus sequence analysis approach based on the 16S ribosomal RNA gene together with four protein-encoding markers (ftsY, gidA, rpsA, and sucB) demonstrated that both bacteria from New Zealand are phylogenetically closely related, but not identical, and belong to the taxonomic genus Rickettsiella.

Conclusions

The bacteria under study should be referred to as pathotypes ‘Rickettsiella costelytrae’ and ‘Rickettsiella pyronotae’, respectively. Moreover, on the basis of the currently accepted systematic organization of the genus Rickettsiella, both pathotypes should be considered synonyms of the nomenclatural type species, Rickettsiella popilliae.

Significance and Impact of the Study

The study demonstrates that Rickettsiella bacteria are geographically widespread pathogens of scarabaeid larvae. Implications of the phylogenetic findings presented for the stability of host adaptation by Rickettsiella bacteria are critically discussed.

Introduction

The genus Rickettsiella (Philip) comprises intracellular bacteria associated with and mostly pathogenic for a wide range of arthropods. Rickettsiella bacteria typically multiply in vacuolar structures within fat body cells and are frequently associated with protein crystals (Tanada and Kaya 1993). Due to their entomopathogenic potential, Rickettsiella are currently under evaluation as a possible source of new bio-insecticides, particularly for important agricultural pests with a long life cycle as, for instance, scarabaeids or elaterids.

The currently valid taxonomy of Rickettsiella bacteria (Garrity et al. 2005) is primarily based on the identity of a specific type strain's original host. Recognized taxa are the nomenclatural type species Rickettsiella popilliae (Dutky & Gooden), Rickettsiella grylli (Vago & Martoja), Rickettsiella chironomi (Weiser) and Rickettsiella stethorae (Hall & Badgley). Rickettsiella isolates have been further defined by pathotypes which have often been placed in synonymy with one of the recognized species. For instance, the pathotypes ‘Rickettsiella melolonthae’, i.e. the causative agent of the ‘Lorsch disease’ of white grubs of European cockchafer species (Coleoptera: Scarabaeidae) (Wille and Martignoni 1952; Krieg 1955), and ‘Rickettsiella tipulae’, a pathogen of the crane fly, Tipula paludosa (Diptera: Tipulidae) (Müller-Kögler 1958; Huger and Krieg 1967), are considered ‘subjective synonyms’ of the species R. popilliae.

Due to their early perception as ‘rickettsiae of insects’, Rickettsiella bacteria had originally been assigned to the taxonomic order Rickettsiales (Weiss et al. 1984) that currently belongs to the class Alphaproteobacteria, in contrast to an alternative classification in the order Chlamydiales that had been considered as well (Yousfi et al. 1979; Federici 1980). However, based on 16S rRNA sequencing results from a strain of R. grylli, which revealed the highest sequence identity to the orthologous gene from Coxiella burnetii (Roux et al. 1997), the genus Rickettsiella has been reassigned to the taxonomic family Coxiellaceae in the order Legionellales of the Gammaproteobacteria (Garrity et al. 2005).

On a genomic basis, this reorganization has been largely confirmed for R. grylli (Leclerque 2008a) and receives additional support from the determination of 16S rRNA-encoding sequences from further Rickettsiella pathotypes from a wide range of arthropods as, for instance, ticks (Kurtti et al. 2002; Vilcins et al. 2009; Carpi et al. 2011), collembola (Czarnetzki and Tebbe 2004), crustaceans (Cordaux et al. 2007), aphids (Tsuchida et al. 2010) as well as coleopteran (Leclerque and Kleespies 2008a; Leclerque et al. 2011a; Kleespies et al. 2011) and dipteran insects (Leclerque and Kleespies 2008b). However, further arthropod-associated bacteria originally described as Rickettsiella pathotypes (Drobne et al. 1999; Radek 2000) were, on the basis of 16S rRNA sequence data, removed from this taxon and re-classified instead in the candidate genus ‘Candidatus Rhabdochlamydia’ of the order Chlamydiales (Kostanjsek et al. 2004; Corsaro et al. 2007). The significance of these findings for the monophyly of the genus Rickettsiella has been critically discussed (Cordaux et al. 2007; Leclerque 2008b).

In view of the inapplicability of the classical species concept to asexually reproducing micro-organisms, comparison of orthologous gene sequences has been widely used to infer phylogenetic relationships among bacteria, and 16S ribosomal RNA-encoding sequences have become the standard molecular chronometer in prokaryote phylogenetics (Woese 1987). However, a complementary approach based on phylogenetic information derived from the analysis of additional protein-encoding marker gene sequences, termed multilocus sequence analysis (MLSA), is often useful or required to improve resolution at lower taxonomic ranks, e.g. where species or subspecies delineation is intended, or to evaluate and eliminate phylogenetic inconsistencies due to, for instance, lateral gene transfer (Kämpfer and Glaeser 2012). Recently, a comparative genomics approach has identified MLSA schemes for numerous bacterial genera and species, but without considering members of the taxonomic order Legionellales (Larsen et al. 2012).

Within the taxonomic family Coxiellaceae, several protein-encoding genes have been investigated as possible markers for phylogenetic studies beyond the 16S rRNA gene level (Sekeyová et al. 1999; Leclerque and Kleespies 2008a,c; Mediannikov et al. 2010), but often with rather limited success. For the genus Rickettsiella, a systematic whole genome based evaluation of possible phylogenetic markers that operate reasonably well at the infra-generic level has revealed a set of MLSA markers comprising the gidA gene encoding glucose inhibited cell division protein A, the rpsA gene encoding the 30S ribosomal protein 1, the sucB gene encoding dihydrolipoamide succinyltransferase and the ftsY gene encoding the bacterial homolog of the eukaryotic signal recognition particle receptor subunit alpha involved in protein translocation (Leclerque et al. 2011b). Previously, ftsY had been identified as the most appropriate single gene marker for the estimation of the G + C content in prokaryotic genomes (Fournier et al. 2006). This four marker MLSA scheme has been employed to clarify the infra-generic taxonomic characterization of a R. grylli strain associated with ixodid ticks (Leclerque and Kleespies 2012).

Infection by Rickettsiella bacteria is difficult to diagnose as infected insects show few external signs of disease, but loose vigour over time and sometimes appear white to blue in advanced infections (Jurat-Fuentes and Jackson 2011). Rickettsial infections have been found among larvae of the Scarabaeidae in different parts of the world (Jackson and Glare 1992). Both the New Zealand grass grub, Costelytra zealandica (White), and larvae of several species of the manuka beetle, Pyronota spp. Bois, (both: Coleoptera: Scarabaeidae), live in the soil and feed on plant roots causing extensive, economically important damage in pastures. The occurrence of cyclic outbreaks with grub populations reaching as much as 1000 individuals per m2 makes their control particularly challenging. While carrying out surveys for the presence of disease in populations of C. zealandica and Pyronota setosa, examination of moribund and whitened insects followed by microscopic diagnosis of fat body tissue indicated intracellular infection with microbes resembling Rickettsiella bacteria.

A recent analysis of 16S rRNA encoding sequences isolated from the P. setosa pathogen has confirmed its identity within this taxonomic genus where it is referred to as pathotype ‘Rickettsiella pyronotae’ (Kleespies et al. 2011), whereas the C. zealandica pathogen has not been studied previously at the genetic or molecular taxonomic level.

In the study presented here, 16S rRNA gene based phylogenetic reconstruction has been combined to MLSA for the comparative genetic characterization and infra-generic taxonomic classification of ‘R. pyronotae’ and the new Rickettsiella-like pathogen of C. zealandica.

Materials and methods

Infections with Rickettsiella bacteria of P. setosa and C. zealandica were initially detected by phase contrast microscopy of squash preparations of fat body tissues. For additional ultrastructural studies, larval fat body tissues of P. setosa were fixed in 2·0% osmium tetroxide in Veronal buffer (pH 7·2) for 17 h for electron microscopy. Tissues were embedded in a 7 : 3 mixture of butyl- and n-methylmethacrylate after dehydration in ascending ethanol series. Thin sections were examined in a Zeiss EM 902 electron microscope (Zeiss, Oberkochen, Germany) after double-staining with uranyl acetate and lead citrate. Micrographs were taken with a CCD camera TrS sharp:eye (Troendle, Moorenwies, Germany).

Genomic DNA was extracted from infected insects according to the protocol reported by Kleespies et al. (2011). PCR amplifications were performed with Phusion High-Fidelity DNA polymerase (Finnzymes) using the oligonucleotide primers and annealing temperatures described in Table 1 in a reaction comprising an initial denaturation step of 2 min at 94°C preceding 30 reaction cycles of a 15 s denaturation step at 94°C, a 30 s annealing step at the primer pair specific temperature and a 2 min elongation step at 72°C, followed by a final elongation step of 5 min at 72°C. For each marker gene, PCR products from three independent amplification reactions were purified by passage over a Qiaquick column (Qiagen) and sequenced on both strands. Raw sequence data were analysed, combined into a single consensus sequence and wherever applicable translated into peptide sequences using the dna strider 1.3 software tool.

Table 1. Marker genes with PCR primers and annealing temperatures (Ta) used for amplification
Gene acronymFunction of gene productPCR primersTa (°C)Length of amplified sequences (bp)
  1. 16S rRNA gene primers are from Weisburg et al. (1991), multilocus sequence analysis marker primers from Leclerque et al. (2011b).

rrs 16S ribosomal RNAfwd: 5′-TGAAGAGTTTGATCCTGGCTCAG521304
rev: 5′-CCTACGGCTACCTTGTTACGACTT
ftsY Signal recognition particle-receptor subunit alphafwd: 5′-AGYTTNCKNCCNCKCCAYTGNCCNCC50828
rev: 5′-ACCATYTCRAARTCYTCYTTCATNGC
gidA Glucose inhibited cell division protein Afwd: 5′-GAATACAATGGCGTACATTGAATGC50786
rev: 5′-AAGACGGAAAAAGATCGCGTGGATC
rpsA 30S ribosomal protein S1fwd: 5′-AAAGTAAAAGGCGGTTTTACNGTNGA48879
rev: 5′-GAAATACGTTCTCGTTCNGGRTCDAT
sucB Dihydrolipoamide succinyl-transferase component E2fwd: 5′-TAGAAGTACCGGCAYCCGCYGACGG50885
rev: 5′-ACATCATCGGTCGAATAACCACTTG

Orthologous sequences from the genomes of selected Alpha- and Gammaproteobacteria as well as Chlamydiae (Fig. 1) were identified in the GenBank database using the BlastN or tBlastN software tools (Altschul et al. 1997; Zhang et al. 2000) for the ribosomal RNA- and protein-encoding marker genes, respectively. For the reconstruction of a 16S rRNA gene phylogeny BlastN hits were ranked according to their maximal sequence identity, i.e. independent of the respective percentage sequence coverage, with a value of <90% being applied as a cut-off criterion. For groups of sequences stemming from the same source, the single sequence entry displaying the highest maximum identity was retained for phylogenetic analysis.

Figure 1.

Bacterial Maximum Likelihood phylogeny generated from 16S ribosomal RNA encoding sequences. Terminal branches are labelled by genus, species, pathotype, strain and/or original host designations, GenBank accession numbers and – where appropriate – pairwise nucleotide sequence identity percentages as calculated from a p-distance matrix with respect to the ‘Rickettsiella costelytrae’ sequence. Numbers on internal branches indicate bootstrap support values; branches that do not receive >90% bootstrap support are represented by dashed lines. The phylogram has been rooted using Blattabacterium as the outgroup. The size bar corresponds to 10% sequence divergence. To enhance resolution, the upper clade of the phylogram has been extended into a cladogram. Partial sequences AM490937AM490939 show only 68% sequence coverage with the remaining 16S sequences. The overall topology of the Maximum Likelihood tree as well as the 100% bootstrap support value for the root of the presumed Rickettsiella clade are reproduced in the corresponding ME and Neighbor Joining phylogenies (data not shown).

Sequence alignments were performed by means of the clustal w function (Thompson et al. 1994) of the mega 4 program (Tamura et al. 2007) using an IUB DNA or a Gonnet protein weight matrix, respectively. Protein-encoding markers were aligned at the deduced amino acid sequence level, with the corresponding nucleotide sequence alignments being generated from these alignments under conservation of codon boundaries. The consistency of 16S rRNA gene comparisons was assessed by alternative alignment using the respective tool of the T-Coffee software package (Notredame et al. 2000; Di Tommaso et al. 2011). The Tree-Puzzle 5.2 software (Schmidt et al. 2002) was used to estimate data set specific parameters as nucleotide and amino acid frequencies, the percentage of invariable sites, transition/transversion ratios and the α parameter for the Γ-distribution based correction of rate heterogeneity among sites.

For phylogenetic reconstruction from nucleotide sequence alignments, the most appropriate models of DNA sequence evolution were chosen according to the rationale outlined by Posada and Crandall (1998). Organism phylogenies were reconstructed with the Maximum Likelihood (ML) method as implemented in the phyml software tool (Guindon and Gascuel 2003) using the HKY model of nucleotide substitution (Hasegawa et al. 1985). Protein-encoding nucleotide data were filtered by systematic suppression of third codon positions. For ribosomal RNA encoding markers, additional Neighbor Joining (NJ) and Minimum Evolution (ME) phylogenies were reconstructed in mega 4 from unfiltered nucleotide sequence data under, respectively, the MCL (Tamura et al. 2004) and the K2P (Kimura 1980) models of nucleotide substitution. For protein-encoding markers, NJ and ME phylogenies were generated from hypervariability-filtered nucleotide data using a modified Nei-Gojobori model (Nei and Gojobori 1986) with Jukes-Cantor corrected frequencies of nonsynonymous substitution. Moreover, organism phylogenies were reconstructed from amino acid sequence alignments using the JTT (Jones et al. 1992) model of substitution with the ML, NJ and ME methods. In all cases, a Γ-distribution based model of rate heterogeneity (Yang 1993) allowing for eight rate categories was assumed. Tree topology confidence limits were explored in nonparametric bootstrap analyses over 1000 pseudo-replicates. Consensus tree topologies were generated by means of the consense module of the phylip 3.6 software package (Felsenstein 2004). Pairwise sequence identity percentages were assessed from p-distance matrices calculated in mega 4 from unfiltered nucleotide alignments under pairwise deletion of alignment gaps and missing data.

Results

Ultrastructural and histopathological features were typical of Rickettsiella diseases. Fat bodies showed massive accumulations of bacteria and associated crystals as shown in Fig. 2 for P. setosa. Massive accumulations of bacteria and associated crystals were observed in hypertrophied fat body cells, and myriads of pathogens were floating in the haemolymph. Typically, tiny bacteria dancing in rapid Brownian movement were found in squash preparations of infected fat body cells with phase contrast microscopy (not shown). Further characteristic features of Rickettsiella infection like intracellular vesicles and vacuolar structures filled with bacterial cells were observed as well. In electron microscope investigations, areas of relatively dense stroma had been found which finally were replaced by masses of bacteria (Fig. 2). Round to oval shaped ‘giant bodies’ were particularly conspicuous, most of them carrying either one large compact crystal or a cluster of smaller crystals (Fig. 2).

Figure 2.

Electron micrograph of ultrathin section of fat body of Pyronota setosa heavily infected with Rickettsiella bacteria (R). Accumulations of Rickettsiella-like bacteria finally replace areas of relatively dense stroma material (S). Many giant bodies (G) transforming into associated crystals (C) can also be observed.

The near complete 16S rRNA encoding gene sequence and internal partial sequences of the ftsY, gidA, rpsA and sucB genes were amplified from the C. zealandica associated bacterium together with the gidA, rpsA and sucB sequences from ‘R. pyronotae’. In all cases, triplicate independent amplification reactions gave rise to unambiguous consensus sequences. The 16S rRNA and FtsY encoding genes from ‘R. pyronotae’ had been determined and described previously (Kleespies et al. 2011).

The results of phylogenetic reconstruction based on the alignment of each of these genetic markers with orthologous sequences from a range of selected bacteria are presented in Figs 1 and 3. In the 16S rRNA gene phylogeny (Fig. 1), both the grass grub and the manuka beetle pathogen were located in a clade unambiguously distinct from those representing other groups of arthropod-associated prokaryotes such as (i) gamma-proteobacterial Buchnera or Wigglesworthia, (ii) alpha-proteobacterial Rickettsia or Wolbachia and (iii) chlamydial arthropod pathogens currently described as ‘Candidatus Rhabdochlamydia’. Instead, both bacteria were firmly placed among gamma-proteobacterial Rickettsiella pathotypes, with the presumed Rickettsiella clade receiving maximal bootstrap support. This result was fully corroborated by the four single MLSA marker phylogenies as well as by phylogenetic reconstruction based on concatenated marker sequences (Fig. 3).

Figure 3.

Bacterial phylogenies generated from ftsY, gidA, rpsA and sucB gene sequences as well as a concatenation of these. Terminal branches are labelled by genus, species, pathotype, strain and/or original host designations, GenBank accession numbers and pairwise nucleotide sequence identity percentages as calculated from a p-distance matrix with respect to the ‘Rickettsiella costelytrae’ sequence. Trees were rooted using Escherichia coli as the outgroup. The ftsY sequence GU289825 from ‘Diplorickettsia massiliensis’ shows only 83% sequence coverage with the remaining ftsY sequences. Left-hand trees: Maximum Likelihood phylograms based on hypervariability filtered nucleotide data. Numbers on internal branches indicate bootstrap support values; branches that do not receive >90% bootstrap support are represented by dashed lines. The size bar corresponds to 5% sequence divergence. Right-hand trees: Cladograms representing the extended majority rule consensus tree topology generated from the six phylogenies reconstructed for each marker gene by applying the Maximum Likelihood, ME or Neighbor Joining method to hypervariability filtered nucleotide or deduced amino acid sequence alignments. Internal branches collapsing under the strict consensus criterion are represented by dashed lines, and their frequency of occurrence across the aggregated set of phylogenies is indicated as a percentage value aside the respective branch; the asterisk (*) denotes 100% frequency of occurrence.

Estimation of the genomic G + C content following the ftsY gene based approach of Fournier et al. (2006) gave calculated G + C contents of 37·8 and 37·9% for ‘Rickettsiella costelytrae’ and ‘R. pyronotae’, respectively, that were very similar to the corresponding values for ‘R. melolonthae’ (37·3%), ‘R. agriotidis’ (37·6%) and ‘R. tipulae’ (37·2%) and lower than the 39·2% G+C content calculated from the ftsY gene of R. grylli; however, the real G + C content obtained from the R. grylli genome sequence has been found to be 37%. All these values are well within the range of genomic G + C contents (36–41%) predicted for Rickettsiella bacteria on the basis of DNA hybridization studies (Frutos et al. 1994). In contrast, the ftsY gene sequences of Coxiella burnetii and Legionella pneumophila gave rise to a considerably higher calculated G + C content of 42·5% for both organisms.

In the ribosomal RNA phylogeny, both ‘R. costelytrae’ and ‘R. pyronotae’ clustered with a group of Rickettsiella strains comprising, among others, the pathotypes ‘R. melolonthae’ and ‘R. tipulae’, i.e. presumed synonyms of the nomenclatural type species, ‘R. popilliae’ (Table 2). This clustering was mainly supported by differential pairwise sequence identity values, whereas there was no convincing bootstrap support (Fig. 1). However, the closest phylogenetic relationship of ‘R. costelytrae’ and ‘R. pyronotae’ to these synonyms of R. popilliae was unambiguous according to both criteria – sequence identity and bootstrap support – when applied to the MLSA marker phylogenies (Fig. 3).

Table 2. Rickettsiella popilliae – synonymized pathotypes referred to in this study
Pathotype designationOriginal hostGeographic originReference
Rickettsiella costelytraeCostelytra zealandica (Coleoptera: Scarabaeidae)Rakaia Gorge, Canterbury, New Zealand 43°23′S 171°34′EThis study
Rickettsiella pyronotaePyronota setosa (Coleoptera: Scarabaeidae)Cape Foulwind, Buller, New Zealand 41°47′S 171°30′EThis study, Kleespies et al. (2011)
Rickettsiella melolonthaeMelolontha sp. (Coleoptera: Scarabaeidae)Lorsch, GermanyLeclerque and Kleespies (2008a)
Rickettsiella tipulaeTipula paludosa (Diptera: Tipulidae)Burscheid, GermanyLeclerque and Kleespies (2008b)

Within the clade combining R. popilliae synonymized pathotypes, i.e. at a supposedly infra-specific level, both entomopathogens from New Zealand formed a tight – presumably geographic – cluster. Compared 16S rRNA, rpsA and sucB gene sequences from both strains were pairwise identical, whereas the respective FtsY and GidA encoding sequences coincided in 99·6% of their nucleotides. In contrast, pairwise sequence comparisons between a MLSA marker gene from ‘R. costelytrae’ or ‘R. pyronotae’, on the one hand, and its ortholog from ‘R. melolonthae’ or ‘R. tipulae’, on the other hand, gave rise to much lower sequence identity values ranging between 94 and 97%, as did pairwise MLSA marker comparisons between ‘R. melolonthae’ and ‘R. tipulae’ (94·6–97·9%, data not shown), i.e. both R. popilliae synonymized pathotypes from Germany did not form a similarly tight geographic cluster.

The 16S rRNA encoding sequences were found insufficiently phylogenetically informative at this taxonomic level, uniformly giving rise to sequence identity values >99% across all these pairwise comparisons.

Interestingly, under an evolutionary perspective, these pairwise sequence similarities translate into internal topological structures of the clade comprising the four presumed R. popilliae synonyms from New Zealand and Germany that are controversial for the MLSA marker phylogenies. Geographic clustering is firmly supported by the ftsY trees, i.e. both the ML phylogeny reconstructed from hypervariablity-filtered nucleotide data and the consensus tree aggregating the data from six individual phylogenies. At the other extreme, the corresponding rpsA phylogenies placed the pathotype ‘R. tipulae’ in an outgroup position with respect to the three further R. popilliae synonyms indicating that pathotypes ‘R. melolonthae’, ‘R. costelytrae’ and ‘R. pyronotae’, i.e. Rickettsiella pathogens of coleopteran insects, might be more closely related among each other than to the dipteran insect pathogen ‘R. tipulae’. The gidA and sucB phylogenies as well as the trees reconstructed from concatenated marker sequences are intermediate in that they display a respective clade structure consistent with geographic clustering, but with the ‘R. melolonthae’–‘R. tipulae’ subclade systematically collapsing under both the 90% bootstrap support and the strict consensus criterion, respectively.

Discussion

Histopathology and ultrastructure of the bacterial disease of the manuka beetle, P. setosa, have been described in detail by Kleespies et al. (2011). The morphological and histopathological features of bacterial infections of both P. setosa and C. zealandica as revealed by microscopic investigation of diseased larvae are consistent with and indicative of infection by Rickettsiella-like bacteria.

The molecular taxonomic analysis of both the 16S rRNA gene and protein-encoding sequence data clearly differentiated the grass grub and the manuka beetle pathogen from prominent gamma- and alpha-proteobacterial as well as chlamydial arthropod-associated bacteria and placed both strains in close vicinity to members of the genus Rickettsiella. Unequivocal distinction of several of these genera, e.g. Rickettsia or ‘Candidatus Rhabdochlamydia’, from Rickettsiella has often been found a difficult task by microscopic means alone. From the ribosomal RNA and MLSA phylogenies presented, we conclude that the new bacterial strain from C. zealandica should be assigned to the taxonomic genus Rickettsiella (Gammaproteobacteria) and, for purposes of discussion, henceforward refer to the new specimen as representative of the pathotype ‘R. costelytrae’. The previous rrs and ftsY gene based analogous generic classification of the manuka beetle pathogen, ‘R. pyronotae’, was corroborated here by an analysis of gidA, rpsA and sucB marker sequences. Moreover, our comparison of calculated G + C content data across the taxonomic order Legionellales was fully consistent with the assignment of both bacteria to the genus Rickettsiella.

At the infra-generic level, the close phylogenetic relationship of both ‘R. costelytrae’ and ‘R. pyronotae’ to synonyms of the nomenclatural type species, R. popilliae, clearly suggests an analogous synonymization of the pathotype ‘R. costelytrae’ and independently corroborates the previous respective conclusion for ‘R. pyronotae’. In particular, an alternative synonymization to the species R. grylli as – due to unavailability of the original type strain – represented by the isolates from the hard tick, Ixodes woodi, and the pill bug, Armadillidium vulgare, was ruled out by the MLSA data presented. The phylogenetic relationship to the pathotype ‘Rickettsiella armadillidii’ that is of ambiguous taxonomic status as having been perceived earlier as a subjective synonym of R. grylli (Weiss et al. 1984), but recently been both placed in synonymy with R. popilliae (Garrity et al. 2005) and claimed an independent species (Cordaux et al. 2007), appeared equally distant in the light of a comparison of 16S rRNA genes. However, to date no MLSA marker sequence data are available for ‘R. armadillidii’.

From the particularly high marker sequence similarities for ‘R. costelytrae’ and ‘R. pyronotae’ we conclude that both isolates from New Zealand should most appropriately be perceived as two different strains of the same – even subspecific – taxon. Given the nature of the geographic origins, geographic clustering will be a priori expected to be a predominant feature of the phylogenetic comparison of the R. popilliae synonymized strains from New Zealand and Germany and has been found a consistent explanation of the topologies obtained for the ftsY, gidA and sucB trees presented in Fig. 3. However, a tree topology consistent with an alternative phylogenetic hypothesis, the predominance of host adaptation over geographic origin, has been obtained for the rpsA marker. Moreover, supposedly geographic clustering of ‘R. melolonthae’ and ‘R. tipulae’ appeared flawed in the light of both bootstrap statistics and the strict consensus criterion. Taking these findings together, a picture of Rickettsiella phylogenetics emerges where clustering by host-adaptation seems to balance clustering by geographic origin – at least to a certain degree. Given the strong geographic bias present in our sample, this picture implies that host adaptation of Rickettsiella bacteria should, in contrast to what is mostly supposed, be a comparatively stable trait. If confirmed independently, this might turn out to be an exciting finding not only for evolutionary bacteriologists, but also with respect to Rickettsiella-based biological arthropod control strategies. It goes, however, without saying that the currently available data are by far too scarce to motivate drawing far-reaching conclusions.

In conclusion, the present study demonstrated by the methodological combination of electron microscopy with a molecular taxonomic approach going beyond ribosomal phylogenies that the two new bacterial pathogens of coleopteran insects from New Zealand first belong to the taxonomic genus Rickettsiella (Gammaproteobacteria) and should, therefore, be referred to as pathotypes ‘R. pyronotae’ and ‘R. costelytrae’ as long as species delineation within the genus is not placed on a stable basis. Moreover, both are closely related to a group of Rickettsiella pathotypes currently placed in synonymy with the type species, R. popilliae, and a respective synonymization of ‘R. pyronotae’ and ‘R. costelytrae’ is suggested by the data presented here. Finally, both pathotypes appear genetically very closely related to each other and will, therefore, most likely be considered two strains of the same taxon even in a future, presumably more differentiated systematic taxonomy of Rickettsiella bacteria.

Acknowledgements

The studies reported here have been financially supported by the German Federal Office for Agriculture and Food (BLE), project grant NZL 3/10-11

Ancillary