Long-Term Evolutionary Stability of Bacterial Endosymbiosis in Curculionoidea: Additional Evidence of Symbiont Replacement in the Dryophthoridae Family

Abstract

Bacterial intracellular symbiosis (endosymbiosis) is well documented in the insect world where it is believed to play a crucial role in adaptation and evolution. However, although Coleopteran insects are of huge ecological and economical interest, endosymbiont molecular analysis is limited to the Dryophthoridae family. Here, we have analyzed the intracellular symbiotic bacteria in 2 Hylobius species belonging to the Molytinae subfamily (Curculionoidea superfamily) that exhibit different features from the Dryophthoridae insects in terms of their ecology and geographical spanning. Fluorescence in situ hybridization has shown that both Hylobius species harbor rod-shaped pleiomorphic symbiotic bacteria in the oocyte and in the bacteria-bearing organ (the bacteriome), with a shape and location similar to those of the Dryophthoridae bacteriome. Phylogenetic analysis of the 16S ribosomal DNA gene sequences, using the heterogeneous model of DNA evolution, has placed the Hylobius spp. endosymbionts (H-group) at the basal position of the ancestral R-clade of Dryophthoridae endosymbionts named Candidatus Nardonella but relatively distant from the S-clade of Sitophilus spp. endosymbionts. Endosymbionts from the H-group and the R-clade evolved more quickly compared with free-living enteric bacteria and endosymbionts from the S- and D-clades of Dryophthoridae. They are AT biased (58.3% A + T), and they exhibit AT-rich insertions at the same position as previously described in the Candidatus Nardonella 16S rDNA sequence. Moreover, the host phylogenetic tree based on the mitochondrial COI gene was shown to be highly congruent with the H-group and the R-clade, the divergence of which was estimated to be around 125 MYA. These new molecular data show that endosymbiosis is old in Curculionids, going back at least to the common ancestor of Molytinae and Dryophthoridae, and is evolutionary stable, except in 2 Dryophthoridae clades, providing additional and independent supplementary evidence for endosymbiont replacement in these taxa.

Introduction

Associations between insects and bacteria are common in nature and include a wide range of interactions from parasitism to mutualism. In insects subsisting on nutritionally deficient habitats, symbiotic associations may occur and involve intracellular bacteria (endosymbionts) living within specialized host cells (the bacteriocytes) that sometimes form an organ, the bacteriome (Buchner 1965). Endosymbionts are thought to provide limiting components to insects (Douglas 1998), thus improving their fitness and their invasive power (Heddi et al. 1999). They could also be involved in insect temperature resistance (Montllor et al. 2002), protection against parasites (Oliver et al. 2003), or broadening of food plant range (Tsuchida et al. 2004). Endosymbionts often transmit vertically to insect offspring for hundreds of generations, resulting in cospeciation between the bacteria and the insect (Baumann et al. 1995). As a result, this route of symbiont transmission leads to a peculiar symbiont DNA evolution. The most striking features are the tendency to genome size reduction (Shigenobu et al. 2000; Gil et al. 2002; Nakabachi et al. 2006; Pérez-Brocal et al. 2006), a global enrichment in AT nucleic bases (Moran 1996; Heddi et al. 1998) and fast nucleotide substitution rates (Moran 1996).

Molecular phylogenetic studies have provided a broad insight into the evolutionary history of many insect group endosymbioses, including aphids (Moran et al. 1993), carpenter ants (Schröder et al. 1996), tsetse flies (Chen XA et al. 1999), psyllids (Thao et al. 2000), weevils (Lefèvre et al. 2004), mealybugs (Downie and Gullan 2005), and leafhoppers (Takiya et al. 2006). Cospeciation appears as the dominant pattern of evolution between the primary endosymbiont and the host lineages, and the stability of the association is so strong that, in insects harboring multiple endosymbionts, multiple cocladogenesis has been shown to occur in some cases (Takiya et al. 2006). However, recent molecular data have indicated that associations with a given endosymbiont are not necessarily persistent and might have often been lost and replaced by different microbes during insect evolution. Such symbiont replacements have been suspected in insect groups wherein multiple bacterial types are coexisting within a single host lineage (Moran and Baumann 1994; Gomez-Valero et al. 2004). Such a replacement event could be governed by a switch of the host habitat and may result from the competition between primary endosymbionts with a reduced exhausted genome and secondary endosymbionts that express a beneficial role for host fitness (Chen DQ et al. 2000; Sabater et al. 2001; Montllor et al. 2002; Koga et al. 2003; Gomez-Valero et al. 2004; Lefèvre et al. 2004; Tsuchida et al. 2004).

The Curculionoidea superfamily is the largest group of the most species-rich order of living organisms, the Coleoptera. There are more than 60,000 species of weevils worldwide that may be found in a variety of habitats, ranging from seashores to mountain tops and from deserts to rainforests, where they are particularly abundant and diversified (Zimmerman 1993). These insects have succeeded in thriving on a large range of plant structures, such as roots, stems, branches, flower heads, fruits, and seeds of both angiosperms and gymnosperms. Some are pests of major concern to agriculture and forestry, and others are beneficially used in the biological control of noxious weeds. However, in spite of their ecological and geographical importance, very little is known about intracellular symbiosis’ evolution and its role in weevil adaptation. So far, only the Dryophthoridae weevils—that are mainly distributed in tropical regions and also represent important pest species in temperate regions—have been studied with modern microbiological and molecular techniques (Lefèvre et al. 2004). In this family, 3 clades of bacterial symbionts that differ in their evolutionary rates and their nucleotide composition have been defined. The D-clade (53.4% Guanine/Cytosine content) includes endosymbionts of Diocalandra frumenti and Trigonotarsus rugosus; the S-clade (53.8% GC content) comprises the endosymbionts of Sitophilus rugicollis, Sitophilus granarius, Sitophilus zeamais, and Sitophilus oryzae; and the R-clade (40.5% GC content) includes the endosymbionts of Yuccaborus frontalis, Rhynchophorus palmarum, Cosmopolites sordidus, Sphenophorus abbreviata, Scyphophorus yuccae, Metamasius hemipterus, and Metamasius callizona. All these Dryophthoridae belong to the Rhynchophorinae subfamily, except Y. frontalis that belongs to the Orthognathinae. The R-clade, which includes the endosymbionts of the most ancestral Dryophthorid insect analyzed (i.e., Y. frontalis), is the oldest clade that was established 100 MYA (Lefèvre et al. 2004). The S- and D-clade endosymbionts have become associated more recently, probably by symbiont displacement.

To expand our knowledge on endosymbiosis occurrence and evolution in the Curculionoidea and test the hypothesis of evolutionary succession of endosymbionts in the Dryophthoridae family, we have investigated endosymbiosis in 2 species of the Molytinae subfamily, Hylobius abietis and Hylobius transversovittatus. Molytinae members are widely distributed in the northern and southern hemispheres. Furthermore, most Hylobius weevils, including H. abietis, are restricted to conifer trees and breed in the roots of dying or recently dead conifers. Hylobius transversovittatus insects are among the few exceptions that breed in the lignified roots of the perennial plant Lythrum salicaria, and they are used as a biological control agent of the invasive host plant in North America (McAvoy et al. 2002).

We show that both Hylobius species harbor pleiomorphic endosymbiotic bacteria in a very conserved bacteriome structure, similar to that of the Dryophthoridae. Host and endosymbiont phylogenetic tree construction revealed a strong concordance between the endosymbionts and their hosts from Molytinae and Dryophthoridae and indicated that Molytinae and their endosymbionts branched with the Dryophthoridae insects and endosymbionts (R-clade), respectively. We conclude that Curculionoidea endosymbiotic associations are older than previously estimated for the Dryophthoridae insects and that symbiont displacement may have occurred at least twice in the S- and D-clades of the Dryophthoridae.

Materials and Methods

Insect Collection and Bacterial DNA Extraction

Insects were collected in the field and kept alive on host plant organs until dissection. Hylobius abietis and H. transversovittatus were collected in France, on Pinus pinaster (Ait.) in a clear-cut forest site (44°01′58″N and 1°14′41″W) and on L. salicaria (L.) in a meadow next to a wetland (45°05′45″N and 5°47′31″E), respectively.

Ovaries and larval bacteriomes were dissected in buffer A (25 mM KCl, 10 mM MgCl2, 250 mM sucrose, and 35 mM Tris–HCl, pH 7.5) and homogenized in 250 μl of buffer STE (Sodium Chloride, Tris, EDTA: 100 mM NaCl, Tris–HCl, pH 8, and 1 mM Na2-ethylenediaminetetraacetic acid [EDTA]), and Tris–Hcl, pH 8). DNA was extracted by adding 3 μl of proteinase K (18 mg/ml) and then incubating for 2 h at 55 °C. RNA contamination was removed with the addition of 3 μl of RNaseA (10 mg/ml) for 1 h at 37 °C. DNA was then purified with phenol–chloroform extraction and precipitated with 0.7 v/v of isoamelic alcohol. For insect DNA extraction purposes, imago flight muscles were dissected in buffer A and DNA was extracted using a DNeasy Tissue Kit (Qiagen, Hilden, Germany), following the instructions of the manufacturer.

Bacterial 16S rDNA Polymerase Chain Reaction Amplification and Sequencing

Because we had no prior indications regarding bacterial phylogenetic origin, bacterial 16S rDNA amplification was performed using eubacterial universal primers: 008 forward 5′-AGAGTTTGATCMTGGCTCAG-3′ and 1,389 reverse 5′-GACGGGCGGTGTGTACAA-3′ (nucleotides numbering from Escherichia coli, GenBank accession number J01859). Reactions were carried out in a volume of 50 μl consisting of 2.6 units of Taq DNA polymerase (Roche, Basel, Switzerland), 1.75 mM MgCl₂, 0.35 mM deoxynucleoside triphosphate (dNTP), 0.3 μM primers, and 10 ng of DNA template. The polymerase chain reaction (PCR) parameters were as previously described (Lefèvre et al. 2004). Sequencing was performed on PCR products by Genome Express (Grenoble, France), according to the Applied Biosystems protocol, on an automatic sequencer (ABI Prism 3100).

Fluorescence In Situ Hybridization Procedure

Larval bacteriomes, apexes of the ovaries, and mature oocytes were dissected in buffer A and squashed on poly-L-lysine-coated microscope slides. Slides were heated for 5 min on a hot plate at 60 °C and then treated for 5 min with acetic acid (70%), at 45 °C, in a moisture chamber. They were rinsed for 5 min with phosphate buffer saline (PBS, pH 7.2), dehydrated in different alcohol solutions, and digested with 100 μg/μl pepsin in 0.01 N HCl for 10 min at 37 °C. Larval bacteriomes were fixed during 1 week in Finefix solution (Milestone, Bergamo, Italy) and embedded in paraffin. Five micrometer thick sections were mounted on poly-L-lysine-coated microscope slides. After methylcyclohexane dewaxing and rehydration, sections were digested with 100 μg/μl pepsin in 0.01 N HCl for 10 min at 37 °C.

The specific oligonucleotide probe used was designed by sequence alignment of Hylobius endosymbionts 16S rDNA (5′-ACCCCCCTCTATGAGAC-3′) and 5′ end labeled with rhodamine. Hybridization was performed at 45 °C in a dark moisture chamber in a 0.9 M NaCl, 20 mM Tris–HCl, 5 mM EDTA, 0.1% sodium dodecyl sulfate, and 10× Denhardt solution. After a 30-min preincubation period, 25 ng/μl of probe were added and incubation continued for 3 h. Slides were washed twice in the same buffer at 45 °C for 15 min, rinsed with PBS, and mounted in Gel Mount (Biomeda, Foster City, CA) medium containing 4′,6-diamidino-2-phenylindole (DAPI). Tissue imaging was carried out on Zeiss microscope (Axiovert). To confirm the specific detection of the endosymbionts, several control experiments were undertaken including Rnase digestion negative control, signal-decrease control with excess of unlabeled probe, no-probe control, and the use of a Wolbachia 16rDNA probe (Heddi et al. 1999).

Phylogenetic Analyses and Topology Test for Bacteria

Endosymbiont DNA is characterized by evolutionary rates significantly higher and a genomic GC content that is generally lower, when compared with closely related free-living bacteria (Heddi et al. 1998). The among-site rate variation and base composition heterogeneity challenge the standard reconstruction of phylogenetic trees, particularly when organisms are highly AT biased and have evolved at relatively high rates. Hence, the long AT-rich branches of insect endosymbionts have always been difficult to place in the phylogenetic tree with free-living relatives (Lefèvre et al. 2004; Herbeck et al. 2005). Therefore, phylogenetic trees were constructed using the heterogeneous model of DNA sequence evolution developed by Galtier and Gouy (1998; Galtier et al. 1999) implemented in the nhml software (ftp://pbil.univ-lyon1.fr/pub/mol_phylogeny/nhml). This model implements both variations of substitution rates among sites and unequal base composition across lineages and provides, as output, the estimated equilibrium GC content of each sequence (i.e., the GC content, it would have if it were to evolve indefinitely with the evolutionary bias of its terminal node) and the estimated GC contents in ancestral sequences.

The 16S rDNA sequences alignment, provided as Supplementary Material online by Lefèvre et al. (2004), was used as a reference. Sequences of Hylobius endosymbionts were added, and a new alignment was performed using ClustalW, implemented under BioEdit (Hall 1999), followed by manual refinement. Regions of ambiguous alignment were removed from all analyses, resulting in a data set of 1,388 bp in total.

Alternative topologies were compared using the exploration method developed by Lefèvre et al. (2004) for a similar data set. This method involved taking the a priori fixed topology of weevil endosymbionts and free-living bacteria published in Lefèvre et al. (2004) and adding Hylobius endosymbionts, either one by one or together, on each branch of the reference topology. For each tested endosymbiont position, the tree log likelihood was estimated. A bootstrap was performed on the 16S rDNA data set (500 replicates) in order to test for the final phylogenetic tree. Alternative topologies were compared using t-tests on their log-likelihood scores (500 values for each topology). In addition, the internal structure of the final tree was tested with local “pruning regrafting” procedures implemented in the NHML software (Non-homogenous maximum likelihood for phylogenetics: ftp://pbil.univ-lyon1.fr/pub/mol_phylogeny/nhml). Only the subtree formed by the endosymbionts of the R-clade and the branches of the Hylobius endosymbionts were allowed to be rearranged. Finally, to assess confidence interval (CI) on the most likely tree, we performed an approximately unbiased (AU) test using CONSEL 1.19 (Shimodaira and Hasegawa 2001). Sitewise log likelihoods under this model were computed using nhPHYML (Boussau and Gouy 2006) and were used as input for the tests implemented in CONSEL with default scaling and replicate values. The topology obtained by Lefèvre et al. (2004) using the heterogeneous model of Galtier and Gouy (1998) was again used as a backbone on which all possible positions of Hylobius endosymbionts were added, yielding 63 topologies. Additionally, to evaluate the likelihood of weevil endosymbiont monophyly, a subset of the permutation of the 4 clades was tested given that S,[R,D] was found to be the most likely topology when Dryophthoridae endosymbionts were tested alone (Lefèvre C, personal communication), that is, S,[D,[R,H]] versus H,[S,[R,D]] versus S,[H,[R,D]] versus S,[R,[D,H] allowed to move on the tree backbone formed exclusively with free bacteria. Then, the most likely topologies under hypothesis of monophyly (designated by the AU test) were compared with topologies constructed under a hypothesis of polyphyly, that is, D with Haemophilus/Pasteurella clade plus [R,S] branched on the enteroclusters and with the H-group allowed to move on every node versus H,[R,D] with the S-clade allowed to move on every node.

Relative Rate Tests

The substitution rates of the 16S rDNA gene were compared between endosymbionts of the Hylobius clade and the Dryophthoridae clades and with several representatives of free-living bacteria and other insect endosymbionts, using the RRTree software (Robinson and Huchon 2000) and Aeromonas haywardensis (AF015258) as an outgroup. The null hypothesis was that the rate of substitution of the tested clade is the same as that of the reference group.

Insect COI PCR Amplification and Sequencing

Two specific primers were designed for Hylobius COI mtDNA amplification: 76 forward 5′-CGGAATAGTTGGTACATC-3′ and 1,278 reverse 5′-AATCCTAAGAAATGTTGAG-3’ (nucleotides numbering from Drosophila yakuba GenBank accession number X03240). Reactions were carried out in 25 μl containing 1 unit of Taq DNA polymerase (Ampligold, Roche), 2 mM MgCl₂, 0.2 mM of each dNTP, 1 μM primers, and 2 μl of DNA template. The PCR amplification reaction included an initial 10-min denaturing step at 95 °C followed by 40 cycles at 95 °C for 30 s, 48 °C for 30 s, 72 °C for 1 min, and a final elongation step at 72 °C for 7 min. COI mtDNA sequencing was performed on an ABI 3100 capillary automatic sequencer using the big dye terminator sequencing Kit V3.1 from Qiagen.

Nucleotide sequences were deposited in the GenBank database, and the other endosymbiont and weevil sequences used were obtained from the same database (GenBank accession numbers given in figs. 2 and 3, respectively).

Phylogenetic Analysis of Host–Insect Relationships

Two outgroup taxa were chosen within Curculionoidea. Lepidapion cretaceum was randomly selected among the COI sequences of Apionidae available in Genbank (accession number AJ717656), and Tanysphyrus lemnae was chosen within Erirhinidae (DQ768215). In addition, another Molytinae species was added, namely Pissodes strobi (AY131096), a North American species living on conifers, in which no endosymbiotic bacteria have so far been observed.

A total of 16 sequences were aligned using ClustalW and all regions with ambiguous alignment were excluded, resulting in a total alignment of 1,246 bp. Parsimony analysis revealed a high rate of homoplasy (CI = 0.43), which is not surprising given the available gene used (i.e., COI mtDNA) and the presumably high level of divergence between the studied taxa. Therefore, nucleotide sequences were translated into amino acid sequences using the appropriate reading frame. This considerably reduced homoplasy (CI = 0.68), allowing for a more reliable phylogenetic reconstruction. A phylogenetic analysis was performed on this data set using the substitution model MtRev for mitochondrial proteins implemented in PHYML (Guindon and Gascuel 2003).

Statistical Test of Cospeciation

A statistical test of cospeciation between the weevils and their symbiotic bacteria was performed using the program ParaFit (Legendre et al. 2002). ParaFit statistically assesses the fit between host and symbiont phylogenetic distance matrices mediated by the matrix of host–symbiont links. Matrices of patristic distances were built from both endosymbiont and host–insects phylogenies. Each matrix in turn was then transformed into a matrix of principal coordinates that was used as input for ParaFit. Probability for the test of cospeciation was computed after 999 permutations.

Results

Endosymbiosis in Hylobius Species

Symbiotic bacteria were histologically examined in the oocytes, the apexes of ovarioles, and the larval bacteriome of both H. transversovittatus and H. abietis. These species were shown to harbor intracellular bacteria that permanently infect the oocyte and the apex of ovarioles, as described in Dryophthoridae species (Nardon et al. 2002). The Hylobius larval bacteriome is also of a similar shape and location as the bacteriome of Dryophthoridae insects (Scheinert 1933; Buchner 1965). It surrounds the foregut/midgut junction as a single, ring-shaped structure consisting of numerous lobelets of different sizes but without any communication with the gut (fig. 1B and C). A fluorescence in situ hybridization (FISH) experiment was conducted, with a specific probe, in order to confirm that the 16S rDNA sequences, generated by PCR from H. transversovittatus and H. abietis bacteriomes, were derived from the endosymbionts we had observed. Endosymbiotic bacteria appear to be pleiomorphic in all the samples examined (fig. 1). They are rod shaped and their size varies from 1.5 to 50 μm, depending on the insect stage and tissue. Although in the oocyte, bacteria are short rodlets that measure from 1.5 to 5 μm, they increase in size in the bacteriomes and become filamentous, from 6 to 30 μm in the ovariole bacteriome and up to 50 μm in the larval bacteriome.

Phylogenetic Molecular Analyses of the 16S rDNA Sequence from Hylobius Endosymbionts

Phylogenetic Tree Construction

Prior to the phylogenetic analysis using the heterogeneous model of evolution, distance, maximum likelihood and maximum parsimony analyses were performed. In all these analyses, high bootstrap values supported the basal position of Hylobius endosymbiont sequences to the R-clade of the Dryophthoridae (see Supplementary Material online).

The Hylobius endosymbiont sequences were highly AT biased and exhibited AT-rich insertions in the same positions, as previously described in the bacterial R-clade of Dryophthoridae by Lefèvre et al. (2004) and in Buchnera endosymbionts (Lambert and Moran 1998). Three AT-rich insertions were found to be common to the R- and Hylobius clades and absent in all other endosymbionts analyzed (including the 16S sequences of primary endosymbionts from 10 aphids, 3 whiteflies, and 7 carpenter ant species and weevils ES from S- and D-clades): positions 150–200, 215–245, 885–895, and 1085–1095 (see Supplementary Material online). These 3 AT-rich regions are synapomorphies that support the monophyly of ES from R- and H-clades and exclude the S- and D-clades and therefore strongly supporting the hypothesis of symbiont replacement in the Sitophilus species: R-, S-, and D-clades clearly are not monophyletic. The GC contents were 41.7% and 42.1% for the Hylobius clade and for the Dryophthoridae R-clade, respectively.

When tested alone, each of the Hylobius bacterial sequences branched together with the R-clade. When tested simultaneously, the highest likelihood tree grouped the Hylobius endosymbionts together with the R-clade subtree (position 51 on the reference tree, see fig. 2). Among the 205 trees evaluated by the shake procedure, the highest likelihood topology was the same as the one obtained by the topology exploration method. Statistical comparisons of the topologies confirmed this topology as the most significant (node 51; log-likelihood score: −7727.23; t-test, P < 1.10⁻⁴). This was consistent with the best tree found by the shake procedure when branches of the R-clade and the 2 Hylobius endosymbionts were allowed to be rearranged. The final topology is given in figure 3. The name of “H-group” was assigned to the group defined by the 2 Hylobius endosymbiont sequences.

Monophyly versus Polyphyly of the Weevil Endosymbionts

When sequences of the H-group were allowed to move freely on the tree backbone, the tree with the highest likelihood among the 63 possible topologies corresponded to a position of the Hylobius sequences basal to the whole R-clade. The topology with H sequences grouped with Yuccaborus had a lower likelihood but was not significantly different, whereas all other topologies were significantly less likely (threshold: P > 0.05: positions 51 and 53, see fig. 3 and table 1A).

Under the hypothesis of monophyly, the topology with the highest likelihood corresponded to the tree where S,[D,[R,H]] branched with the Haemophilus/Pasteurella clade (table 1B, threshold: P > 0.05).

The global comparison of polyphyletic versus monophyletic topologies for the weevil endosymbiont clades showed unambiguously that the tree, where H sequences are grouped with R sequences, is significantly the likely than all other topologies (table 1C, threshold: P > 0.05).

Relative Rates of Evolution of Hylobius Endosymbionts

The H-group showed rates of evolution significantly different from all other bacterial clades, except the R-clade (table 2). Therefore, as it has been deduced for sequences of the R-clade of Dryophthoridae, the Hylobius endosymbionts evolved more rapidly than both free-living enteric bacteria and the endosymbiotic bacteria from the S- and D-clades. Thus, we define an “Rh-group” comprising R-clade and H-group sequences, to be used further in the text. Given an estimated rate of evolution of 0.128 substitutions per site per 100 Myr for the R-clade (Lefèvre et al. 2004), the H-group would have diverged approximately 115 MYA.

Host–Insect Phylogenetic Analysis

The amino acid alignment showed 73 parsimony informative characters out of 415. The tree obtained by the maximum likelihood analysis is given in figure 3. Molytinae weevils were grouped together with Dryophthoridae with a high bootstrap support. This topology was also strongly supported by distance analysis (data not shown).

The molecular clock hypothesis was not verified on the whole weevil phylogeny, that is, COI substitution rates were not constant among the different weevil lineages (likelihood ratio test P = 3.5 × 10⁻⁵). However, relative rate tests performed with RRTree, using the 2 outgroups as reference sequences, showed no significant difference in sequence evolution between Dryophthoridae and Molytinae (P = 0.76). Given that the oldest known Molytinae fossil is ∼55-Myr old (Britton 1960; Zherikhin 2000), the divergence between Dryophthoridae and Molytinae must have taken place earlier. Accepting an age of 100 Myr for the Dryophthoridae group (O'Meara 2001; Lefèvre et al. 2004) gives an estimate for the divergence between the Dryophthoridae and the Molytinae between 125 and 165 MYA, that is, between the mid-Jurassic and the end of the late Cretaceous period.

Cospeciation between the Hosts and the Endosymbionts

In a first step, the global test of cospeciation performed with ParaFit on trees comprising all weevils and their endosymbionts was highly significant (13 insect–bacteria couples, P = 0.009). However, a detailed examination of the results showed that the test was significant only for the host–symbiont couples of the S-clade and the H-group. In a second step, when performing a test on the R-clade and the H-group only, the global test was highly significant again (9 insect–bacteria couples, P = 0.001), but this time all the pairwise tests of cospeciation were significant, that is, for the R-clade and the H-group taxa.

Discussion

Endosymbiosis Occurrence in the Curculionoidea Superfamily

It is well recognized and documented that insects are the most diverse group of macroscopic organisms. Throughout the world, some 900,000 different kinds of living insects are known, which approximates to 80% of the world's species. According to Erwin (1983), the number of living species of insects has been estimated to be 30 million. Out of this huge diversity, only a few (less than one thousand) have been studied with regards to intracellular symbiosis, which greatly underestimates the extent of symbiosis within this large animal group. Among approximately 30 orders of insects, Coleoptera is the largest order in the entire animal kingdom. There are about 350,000 named species of beetles in the world and many more unnamed species (Chadwick 1998). So far, less than 20 species, all belonging to the Dryophthoridae family, have been investigated molecularly and were shown to harbor intracellular bacteria. Here, we have extended the modern microbiological characterization of bacterial endosymbionts in the Curculionoidea superfamily beyond the Dryophthoridae group. We show that those Molytinae species also harbor bacteria related to Candidatus Nardonella described in Lefèvre et al. (2004). Interestingly, endosymbiont examination in both Hylobius species has revealed that the bacteriome structure is histologically similar to the structure observed in the Dryophthoridae, which supports a deep-seated and long-term relationship between the Curculionoidea and their endosymbionts. Furthermore, Hylobius exhibits similar pleiomorphism as in the Dryophthoridae, with a relatively small shape in the oocytes that becomes bigger and filamentous in both the larval and ovariole bacteriome. Scheinert (1933) hypothesized that endosymbiotic bacteria might undergo a process of “degeneration” during their passage from the larval stage of the host to the adult sexual cells. The FISH study showed that the bacterial strain observed in bacteriomes and in oocytes is the same despite the significant morphological modifications. Along with the previous observation in the Dryophthoridae, this finding suggests that bacterial pleiomorphism is of a host–symbiont interaction origin and that beetles might use a conserved molecular mechanism to control endosymbiont cell division and growth, presumably through the inhibition of bacterial cell septation during the insect development.

Endosymbiosis Evolution within the Curculionoidea Superfamily

The host–insect and endosymbiont trees were strongly congruent for the Rh-group as certified by the cocladogenesis tests, supporting a scenario of cospeciation between host–insects and their endosymbionts from an ancestor common to the Hylobius species and the Dryophthoridae and a single bacterial infection. Positive signal of global cospeciation between the complete set of bacteria and their host–weevils was obtained with a separate treatment of the S-clade and the Rh-group. This indicated that a cospeciation scenario is supported both within the S-clade and the Rh-group but that another evolutionary event had to occur to explain the global pattern observed. This scenario leads to the following evolutionary arguments on the weevil endosymbiosis.

First, the age of endosymbiosis in weevils could be at least 25 Myr older than previously estimated from the Dryophthoridae group alone (∼100 Myr, Lefèvre et al. 2004). Given an age of divergence of 125 Myr for weevil endosymbionts, the estimation of the 16S rDNA evolutionary rate is 0.125 substitutions per site per 100 Myr (s/s/100 Myr), very close to the rate estimated for the R-clade alone (0.128 s/s/100 Myr, Lefèvre et al. 2004) and higher than those calculated for Buchnera (0.058 s/s/100 Myr), as well as for the free-living enterobacteria (0.007–0.018 s/s/100 Myr, Clark et al. 1999), but lower than those estimated for the carpenter ant endosymbiont Blochmannia (0.24 s/s/100 Myr, Degnan et al. 2004).

Second, symbiont inheritance across such a large time span indicates strong benefits for the insect host. The biological significance of endosymbiosis has not been addressed in Hylobius weevils so far. However, given the low digestibility and the weak nutritional value of the deadwood diet, which is supposed to be the ancestral habitat of weevils (Marvaldi et al. 2002), early association with bacteria able to provide complementary nutrients might have represented a key innovation in this group of insects. In addition to chemical and nutritional characteristics, the nature and the availability in space and time of the Molytinae diet resources might act as a constraint on the biology and behavior of the insect. Recent studies have supported a high dispersal ability in H. abietis (Nilssen 1984; Conord et al. 2006), similar to many other Curculionoidea insects. This particularity might have evolved because the adult is feeding on young conifer seedlings, while the larva develops in the root cambial tissues of freshly dead conifers, which could be considered as a scarce and ephemeral resource. Furthermore, most suitable host plant tissues are exhausted when the adult emerges so that the new generation must migrate and find new breeding sites. This life cycle requires that insects fly from their feeding sites to their egg laying and developmental sites, which could be considerably distant and, hence, energy consuming. Regarding this aspect, it was noteworthy that Sitophilus endosymbionts were shown to highly increase insect mitochondrial energetic metabolism (Heddi et al. 1993, 1999) and to improve the adult flight ability (Grenier and Nardon 1994). Hence, it is likely that endosymbiosis might also impact on the flight performance of other weevils and enhance their dispersal ability and invasive power.

Finally, this work has provided additional support for the occurrence of bacterial replacement in Dryophthoridae. In addition to the molecular evidence discussed by Lefèvre et al. (2004), the branching of the Hylobius endosymbionts within the R-clade and relatively distant from the S-clade has confirmed the polyphyly of the Dryophthoridae endosymbionts and therefore sustains symbiont displacement in Sitophilus weevils. However, the factors favoring symbiont replacement, either internal (weevil physiology) or external (ecological influences), remain speculative. Whether a host plant shift was favored by symbiont replacement or, on the other hand, whether habitat shift has increased the probability that a new symbiont acquired horizontally (or a secondary endosymbiont) outcompetes the ancestral endosymbiont is still an open question. Seed-eating species of the Molytinae have been described, which could provide a means of testing for an eventual link between endosymbiont replacement and the change in the nature of the weevil's diet (Lyal 1996; Vanin and Gaiger 2005). This could be addressed by exploring endosymbiosis in monophyletic groups where drastic host shifts have likely occurred.

In conclusion, this work has opened new evolutionary and symbiotic perspectives in the Curculionoidea, the largest insect superfamily in terms of species number. Moreover, endosymbiosis in the Curculionoidea may have been established before the divergence between the Dryophthoridae and the Molytinae, at least 125 MYA. This strongly argues in favor of a long-term stability of symbiosis in the Curculionoidea. Finally, symbiosis stability does not mean symbiont persistence and continuity as endosymbiosis in the Curculionoidea may have evolved several times through symbiont competition and displacement.

Updating the taxonomic status of “Candidatus Nardonella endosymbionts”: Proposition of “Candidatus Nardonella dryophthoridicola” as the taxonomic name of the R-clade endosymbionts and of “Candidatus Nardonella hylobii” as the taxonomic name of the Rh-clade endosymbionts, according to the rules published in Murray and Stackebrandt (1995).

In their paper, Lefèvre et al. (2004) proposed the generic name Candidatus Nardonella for the symbiotic bacteria of the R-clade. Given the similarities for several traits between the R- and Rh-clade endosymbionts (e.g., similarities in shape and position of the larval bacteriome, in shape and size pleiomorphism of the bacteria, and nucleic synapomorphies on the 16S rDNA sequences described in the present paper), we suggest that the Rh-clade endosymbionts will be included within the genus Nardonella. To distinguish between the R- and Rh-clades, we further suggest different species names: Candidatus Nardonella dryophthoridicola name will replace “Candidatus Nardonella” name previously attributed to the R-clade endosymbiont in Lefèvre et al. (2004). The Candidatus Nardonella dryophthoridicola describes hence the Dryophthoridae R-clade endosymbionts strictly.

Candidatus Nardonella hylobii name will be used to design the Hylobius Rh-clade endosymbionts. The description of Candidatus Nardonella hylobii is as follows: phylogenetic position, γ3 subclasses of Proteobacteria; cultivation, not cultivated on cell-free media; morphology, rod shape, pleiomorphic size from 1.5 to 50 μm in length, 1–2 μm in diameter; basis of assignment, 16S rDNA sequences; association and host, intracellular symbionts of Molytinae weevils (described from oocytes, ovariole, and larval bacteriomes in H. abietis and H. transversovittatus); mesophilic; authors, Conord et al. (this study). The generic name honors Professor Paul Nardon (Lefèvre et al. 2004), and the species name is based on the genus name of the 2 weevil host species.

The authors would like to thank 2 anonymous reviewers for their help in the improvement of the paper. They also thank R. Douzet for his help with Lythrum salicaria, F. Nicolè and M. Quivrin for their help with the weevil collections, B. Boussau for advices on phylogenetic analyses, and V. James and D. Wainhouse for the manuscript reading and the English correction. This work was partly supported by the French Ministère de la Recherche.

References

Author notes