Wolbachia Prophage DNA Adenine Methyltransferase Genes in Different Drosophila-Wolbachia Associations

Wolbachia is an obligatory intracellular bacterium which often manipulates the reproduction of its insect and isopod hosts. In contrast, Wolbachia is an essential symbiont in filarial nematodes. Lately, Wolbachia has been implicated in genomic imprinting of host DNA through cytosine methylation. The importance of DNA methylation in cell fate and biology calls for in depth studing of putative methylation-related genes. We present a molecular and phylogenetic analysis of a putative DNA adenine methyltransferase encoded by a prophage in the Wolbachia genome. Two slightly different copies of the gene, met1 and met2, exhibit a different distribution over various Wolbachia strains. The met2 gene is present in the majority of strains, in wAu, however, it contains a frameshift caused by a 2 bp deletion. Phylogenetic analysis of the met2 DNA sequences suggests a long association of the gene with the Wolbachia host strains. In addition, our analysis provides evidence for previously unnoticed multiple infections, the detection of which is critical for the molecular elucidation of modification and/or rescue mechanism of cytoplasmic incompatibility.


Introduction
Wolbachia pipientis is an obligate intracellular symbiont belonging to the a-proteobacteria. It is thought to be present in an estimated 66% of all insect species, including disease vectors of animals and plants [1]. It has also been found in terrestrial isopods, spiders, mites, springtails and nematodes. Some Wolbachia strains can modify host reproduction and distort sex ratio by inducing parthenogenesis, feminisation, male killing or cytoplasmic incompatibility (CI) (reviewed in [2]).
Wolbachia phage particles were first observed by Wright et al. [3]. Masui et al. [4] described prophage WO, a genetic element in Wolbachia strain wTai, containing about 26 open reading frames (ORFs) in 25 kb. The phage genome includes ORFs coding for capsid proteins, baseplate assembly proteins, integrase genes, several ankyrin-like proteins, as well as a potential methyltransferase. Further studies resulted in the isolation and characterization of the bacteriophage WO from the Wolbachia strain wCauB [5][6][7]. The genome of the purified phage is linear double-stranded DNA of about 43 kb, containing 47 ORFs (wCauB2) and 45 kb and 46 ORFs (wCauB3) [7]. Sequence analysis indicated that this phage genome includes ORFs coding for a DNA packaging protein, capsid proteins, baseplate assembly proteins, tail structural proteins and for several putative toxin-like secretory proteins [6,7].
The Drosophila melanogaster wMel genome was released in 2004 [8]. Since then, three other Wolbachia genomes have become available [9][10][11] and a number of other sequencing projects are currently in progress. Two divergent prophage WO families, WO-A and WO-B have been identified. Family WO-B can be divided into three clades [4,12]. Distribution surveys indicate that WO-B homologs occur in at least 89% of the two main lineages of Wolbachia that infect arthropods [12,13]. Five WO-B-like prophage regions are present in the wPip genome, with some genes identical or highly similar between prophage copies, while other genes are unique. It seems likely that extensive recombination, duplication and insertion events have occurred between copies [10,14]. In the highly recombining wRi genome, 4 prophage segments have been detected [11], while no prophage elements could be identified in the mutualistic wBm strain of the filarial nematode Brugia malayi [9]. Insertion sequences (IS) are frequently found in WO genomes and are considered to be a major factor driving phage recombination [8,10,11].
Since prophage regions have been found only in Wolbachia strains having a parasitic relationship with their hosts, it has been hypothesized that they contribute to the CI phenotype [4]. However, a study in different Culex pipiens strains detected no correlation between prophage orf7 gene type and CI [15]. Another survey showed that phage associated protein Gp15 is similar to a bacterial virulence factor. This gene was partially correlated with CI expression, suggesting that it could be linked to a CI gene [16]. However, sequence analyses found no phylogenetic clustering of phage genotypes congruent with the four major Wolbachia-induced sexual alterations [13]. In Nasonia vitripennis, 12% of Wolbachia cells were found to show lytic phage development. The density of the bacteriophage correlated inversely with the density of Wolbachia bacteria [17]. While density is one of the most critical determinants of penetrance of Wolbachia-induced phenotypes [18][19][20], only virion-free Wolbachia were observed to contact the host spermatids [17]. These observations led to the Phage Density Model hypothesis, which suggests that lytic phages negatively control Wolbachia densities and expression of symbiont functions (reviewed in [21]).
Both WO-A and WO-B prophages in wMel carry a gene that encodes a putative DNA adenine methyltransferase. Methyltransferase genes are carried by many bacteriophages [22][23][24], and modified bases are common in phage genomes [25,26]. It is thus quite possible that a WO prophage methyltransferase has a phagespecific function or acts on the bacterial genome. Virulence and lysogeny of many pathogenic bacteria such as Escherichia, Salmonella, Yersinia, Vibrio, Haemophilus, Pasteurella, Aeromonas, Actinobacillus, Klebsiella, Brucella and Rickettsia are subject to control by adenine methylation [23,[27][28][29]. In some strains, methylation is essential for viability of the pathogen [30,31]. Adenine methylation is also involved in cell-cycle regulation of bacteria [32]. In all these cases, methylation is performed by solitary or ''orphan'' methyltransferases, i.e. those that are not part of a restrictionmodification (R-M) system. R-M systems protect bacteria against foreign DNA or act as selfish elements [33]. Negri et al. [34] provided the first evidence that a feminizing strain of Wolbachia interferes with host genetic imprinting through cytosine methylation of leafhopper DNA.
In the context of the fact that Wolbachia modifies the genome of its host insects, the presence of a DNA methyltransferase gene in its genome becomes even more intriguing. We therefore used molecular and phylogenetic approaches to detect and characterize the Wolbachia-phage embedded gene which encodes a putative DNA methyltransferase in a number of different Wolbachia strains of known CI properties. The ability of these strains to induce (mod status) and rescue CI (resc status) has previously been characterized [18,[35][36][37][38]. The potential involvement of this methyltransferaselike protein in host-Wolbachia symbiosis and in the induction of reproductive alterations is discussed.

Results
Sequence and distribution of the phage methyltransferase ORFs in different Wolbachia strains Different Wolbachia strains of known CI phenotype (Table 1) were screened for the presence of the WO methyltransferase-like genes in order to test for a possible correlation with the CI phenotype. Degenerate and specific primers were designed for each of the two wMel methyltransferase genes ( Table 2) and used to amplify DNA from a number of different Wolbachia-infected Drosophila strains. Wolbachia strains wMel, wMelCS, wAu and wHa contain both methyltransferase ORFs. In contrast, strains wRi, wNo, wYak, wTei and wSan only have one of the two ORFs (met2), while wMa and wMau do not contain any of the two methyltransferase ORFs. This was confirmed by Southern blot analysis using the two wMel methyltransferase ORFs as probes (Fig. 1). The wHa met ORFs were not further characterized.
The met1 ORFs are more closely related to each other than to the respective met2 ORFs. ORF met2_wAu contains a 2 bp GC deletion after codon 92, leading to a truncated ORF lacking the Cteminal methyltransferase domain.
Further PCR analysis using met_102F/met_269R primers indicated the presence of three PCR products in wTei: While the first was of the expected size for a met2 product, the other two were about 500 and 900 bp larger ( Fig. 2A). Sequencing of the largest product revealed the presence of ISWpi1, a Wolbachiaspecific insertion sequence belonging to the IS5 family [39], in met2 gene. The intermediate sized PCR product could not be cloned in several attempts, probably because of the presence of unstable repeat sequences. Surprisingly, when PCR with the same primers was performed using DNA from D. simulans STCP flies transinfected with wTei [38], only two PCR products were detected, with a striking difference in the intensity ratio ( Fig. 2A). Cloning and sequencing of the methyltransferase ORFs from the closely related wSan, wTei and wYak strains also detected two different met2 ORFs; one closely related to the met2 sequences of supergroup A Wolbachia strains (met2_wTeiA for wTei), while the other is closely related to B supergroup strains (met2_wTeiB for wTei). Specific reverse primers were designed to anneal only to the A-grouplike met2 (met_1024R; Fig. 2B) or the B-group-like met2 (TeiB_1024R; Fig. 2C). PCR reactions with these primers revealed that strains wSan, wTei and wYak bear a copy of the B-group-like met2; these genes were successfully transferred by microinjections into D. simulans STCP lines by Zabalou et al. [38] (Fig. 2C). The ISWpi1-disrupted met2 ORF of wTei was also transferred together with wTei into STCP, but not the intact copy of met2_wTeiA (Fig. 2B).

Phylogenetic Analysis
All methods used to reconstruct phylogenies yielded similar results. The three methods, distance, parsimony and maximumlikelihood (ML), make different evolutionary assumptions, thus their congruence provides strong support for the deduced phylogeny. We show only the tree derived by ML estimation (Fig. 3). The phylogenetic clustering of the met2 ORFs is similar to the currently accepted clustering of the respective Wolbachia strains: all met2 gene sequences coming from supergroup A strains cluster together (wMel, wMelCS, wRi, wAu, wTei, wYak and wSan strains). The met2 gene sequences from supergroup B strains also form a cluster and are distantly related to the supergroup A gene sequences. This suggests a long association of the methyltransferase genes, and consequently of the phages harbouring them, with the respective Wolbachia strains.

RT-PCR analysis
Transcriptional analysis of met2 genes was performed by RT-PCR on cDNA samples prepared from young adult male and female flies. Met2 transcripts were detected in all samples tested (Fig. 4). The met2 copy of wTei, which is disrupted by ISWpi1, was found to be transcriptionally silent (data not shown).

Discussion
The release of the first Wolbachia genome (wMel strain) revealed that it contains two DNA methyltransferase genes met1 and met2, encoded by two prophages, WO-A and WO-B respectively [8]. This finding is intriguing in the light of the fact that Wolbachiainduced CI involves modification of the insect host chromosome [40]. The presence of phage-like particles in Wolbachia-infected hosts [3,5,17,41] suggests an active role of the phage in Wolbachia biology. Thus, it is tempting to speculate that, beyond controlling lysogeny of the phage, the methyltransferases might be involved in triggering reproductive alterations imposed by Wolbachia on its host. We therefore undertook a survey of Wolbachia strains of known CI status for the presence of the methyltransferases, determined their sequences and reconstructed their phylogeny.
The met1 gene is only present in a few of the tested strains; there is no correlation with CI. Wolbachia strains wMel, wRi, wNo and wPip induce CI in permissive hosts [35,[42][43][44], and they all contain at least one functional copy of the met2 gene. In contrast, strain wAu [36] does not induce or rescue CI and its only met2 ORF is disrupted. Wolbachia strains wMau and wMa do not contain the met2 gene and also do not induce CI in their hosts [43,45,46]. On the other hand, wYak, wTei and wSan have recently been shown to fully rescue the wRi modification, while they are unable to induce CI in their native hosts [37]. wTei is, however, able to induce 100% CI after transfer into the permissive host D. simulans [38]. In this context, it is interesting to note that our data suggest a double or multiple infection of the original host of wTei, of which not all Wolbachia strain(s) were transferred upon transinfection of D. simulans (Fig. 2). Cloning and sequencing of PCR products, using different methyltransferase primer sets, supports the presence of a hidden double infection of D. teissieri. This could explain the phenotypic shift in CI properties that is observed between the natural host and the engineered strains.
When the rescue properties of all A group strains are examined, resc 2 strains lack a functional A-group-like met2 gene, which is always present in resc + strains (Table 3); a possible correlation between met2 and CI rescue should therefore be considered. B group strains (wNo, wMa, wMau) are all resc + , nevertheless they do not possess an A-group-like met2 gene. The genomes of these strains have not been sequenced and the presence or absence of prophage copies has not yet been documented. While wMa and wMau do not contain any met gene, wNo has a B-group-like met2 ORF; this could reflect a different mechanism regulating CI in B group Wolbachia strains.
Transgenic expression of wMel met2 (WD0594) in D. melanogaster was recently reported using the UAS/GAL4 system [47]. This study revealed no modification of phenotype in flies expressing met2 ubiquitously and, similarly, when expressed specifically in the ovaries, no rescue phenotype was apparent in CI crosses. Although these data suggest that constitutive expression of the met2 gene does not alone drive the CI phenotype, it is still unclear what type of regulation met2 or any of the phage-related genes are subject to and how this affects the mechanism of CI.
Southern blot analysis indicates the presence of a met-like gene also in the Wolbachia strain wUni, which is known to induce parthenogenesis in the parasitic wasp Muscidifurax uniraptor (data not shown). The distribution of the met gene in parthenogenesisinducing Wolbachia strains remains to be investigated. Interestingly, the mutualistic Wolbachia strain, which is present in the filarial nematode Brugia malayi, neither induces reproductive alterations nor carries a copy of the DNA methyltransferase genes.
Additionally, and important for any interpretation of the role of met2, we demonstrated expression of the gene in all Wolbachia strains with RT-PCR (Fig. 4). The Wolbachia phage DNA methyltransferase may be involved in the methylation of phage, bacterial, insect host genes or a combination of them. Although Drosophila had for a long time been considered to be free of DNA methylation, both the presence of methyltransferase genes in its genome [48,49], and of 5-methylcytosine residues in the early stages of embryonic development [50,51] have been demonstrated. Interestingly, a Dam-like methyltransferase has been implicated in male sterility in plants [52].
Base modification in bacterial genomes is performed by two classes of DNA methyltransferases: (i) those associated with restriction-modification systems, and (ii) solitary methyltransferases that do not have a restriction enzyme counterpart. Examples of the latter are the N6-adenine methyltransferases Dam and CcrM [53,54]. In a-Proteobacteria, CcrM methylation regulates the cell cycle in Caulobacter crescentus, Rhizobium meliloti and Agrobacterium tumefaciens and plays a role in Brucella abortus infection (reviewed in [55]). Overexpression of CcrM in these bacteria results in the accumulation of multiple chromosomes, indicative of overinitia-  tion of DNA replication [56,57]. Wolbachia prophage methyltransferase could regulate several aspects of the symbiont's cell cycle by imposing a specific epigenetic signal.
In silico analysis of Wolbachia prophage methyltransferase has predicted an N-terminal ParB-like nuclease domain (data not shown) similar to the ParB of the parCBA operon in E. coli, which is important for plasmid stability and resolving dimeric or multimeric plasmids. ParB nucleases have also been reported in several other plasmid genomes. ParB nucleases are Ca++ dependent endonucleases with 59 -39 exonuclease activity [58,59].
The methyltransferase genes, met1 and met2, are closely related (Fig. 5). The phylogenetic clustering of the methyltransferase genes, in particular of the met2 gene, is similar to the currently accepted clustering of the arthropod Wolbachia strains, both on the level of the major division of the Wolbachia strains into two supergroups, A and B, as well as on the lower level of clades and strains. Specifically, the met2-based tree is similar to the respective wsp-based tree (data not shown). This suggests a long association of the methyltransferases, and consequently of the phages carrying them, with the harbouring Wolbachia chromosomes (Fig. 3). However, translocation of met2 gene from the phage genome to Wolbachia chromosome cannot be excluded for any of the strains studied; such an event could explain why met2 phylogeny correlates with Wolbachia phylogeny.
Wolbachia exhibits a fascinating array of host manipulations. The elucidation of the molecular basis of the host-symbiont interaction will allow insight in the regulation of fundamental cell biological processes. Future studies will address any potential direct or indirect effect of the methyltransferase(s) in the establishment of symbiosis and/or the induction of reproductive manipulations.

Insect lines
Insect lines and Wolbachia strains used in the present study are listed in Table 1. Flies were routinely grown at 25uC on standard cornmeal medium in uncrowded vials. The Drosophila simulans STCP lines were produced by Zabalou et al. [38] who transferred Wolbachia wYak, wTei, wSan strains into the same host background using embryonic cytoplasmic injections.

PCR analysis
DNA was extracted from adult flies using the NucleoSpin Tissue kit (Macherey-Nagel) or the CTAB protocol, as previously described [60]. The DNA was used as template for PCR reactions and Southern blot analysis. About 50 to 100 ng total DNA from adult flies were used as template in PCR reactions of Wolbachia targets. The presence of Wolbachia was initially determined using Wolbachia-specific 16S rDNA primers [61]. The primers used to amplify methyltransferase gene sequences are listed in Table 2. Standard PCR analysis was performed using GoTaqH Flexi DNA polymerase (Promega). To generate DNA templates for sequencing, PCR reactions were done using the Elongase Amplification System (Invitrogen, Glasgow, UK). The PCR products were Atailed, cloned into the pGEM-T Easy vector (Promega, Wisconsin, USA) and transformed into competent E. coli XL1-Blue MRF cells

Southern blot analysis
Genomic DNA was prepared as reported previously [62] and digested with HaeIII. Agarose gel electrophoresis of DNA and blotting to nylon membranes were carried out using standard procedures [63]. DNA probes were prepared by random hexanucleotide priming [64]. Hybridization of 32 P-labelled probes to blotted DNA was performed using standard procedures [63].

RT-PCR analysis
For each of the tested Wolbachia-Drosophila associations, total RNA from young male and female adult flies was extracted using TRIzol (Invitrogen) and treated with RNase-free DNase (Invitrogen). First-strand cDNA was synthesized from 5 mg of total RNA using reverse transcriptase (SuperScript III; Invitrogen) and random primers (Promega) and the reactions were treated with RNase H. RNA integrity was assessed using the universal Wolbachia wsp 81F/691R primers [65]. All RNA samples were tested for genomic DNA contamination by performing PCR using wsp 81F/691R primers in DNaseI-treated RNA samples which were not reverse transcribed. Transcription of met2 was detected using GoTaqH Flexi DNA polymerase (Promega).

Alignment and Model Selection
Wolbachia adenine methyltransferase nucleotide and amino acid sequences were aligned using the ClustalW Multiple Alignment algorithm implemented in Geneious v.5.3.3. [66]. The appropriate evolutionary model JTT+ C was selected by the Akaike Information Criterion (AIC) using ProtTest v.2.4 [67]. Models of substitution for nucleotide alignments were selected using AIC in jModeltest v.0.1.1. [68]. The appropriate evolutionary model was TPM1uf+I+C.

Phylogenetic analysis
The evolutionary history was inferred by maximum likelihood criterion using PAUP* v.4.0b10 for the nucleotide alignment and PHYML for the protein alignment [69,70]. Phylogenetic trees were generated using ML bootstrap analysis in PAUP (100 pseudoreplicates of heuristic search with 10 random sequence). The maximum likelihood method conducted using PHYML was performed using 100 bootstrap replicates, a fixed proportion of invariable sites, an estimated gamma distribution parameter and optimized topology, branch lengths and rate parameters. ML trees generated are midpoint rooted using Archaeopteryx v. 0.957b [71].

Methyltransferase nomenclature
The methyltransferase gene sequences were named based on the following system. Each methyltransferase gene name is composed of ''met'' (in italics) denoting methyltransferase, followed by the numbers 1 or 2, indicating their origin from phage WO-A or WO-B, respectively (phage nomenclature according to [8]) and concluded by the name of the Wolbachia strain harbouring the methyltransferase gene (eg wRi, wMel, etc). Using this nomenclature, the two methyltransferase genes present in the wMel strain are named met1_wMel and met2_wMel.