For the genome sequencing of the naturally infected Wolbachia strain of G. m. morsitans (wGmm), approximately 250 ovaries were dissected from adult females from the Gmm colony maintained in the Yale University insectary. DNA was prepared using Qiagen DNeasy kit (Qiagen, Inc., Valencia, CA). The complete genome sequence was determined using whole-genome shotgun pyrosequencing using the Roche 454 GS sequencer FLX Titanium system (454 Life Sciences, Branford, CT, USA).

In order to improve the wGmm draft genome, Illumina read libraries from the tsetse genome assembly were used. These were obtained from: (a) a pool of five tsetse flies. and (b) the first larval progeny of tetracycline-treated female. Two sets of Illumina reads were used: a PCR-free small fragment (∼300 bp) library and Hi-Seq mate-pair libraries with an insert of approximately 1.6 kb.

The tsetse ovary DNA used for wGmm sequencing contained a mixture of host genetic material, as well as cytoplasmic (cyt) and chromosomal (chr) Wolbachia DNA. A customized informatics pipeline was developed to computationally distinguish between sequence reads. An initial assembly was performed using MIRA [50]. First, all host sequences were removed by mapping the 454 reads to the Wolbachia reference genomes (wMel, wRi, wPip and wBm). The filtered sequence reads contained chromosomal and cytoplasmic reads. The chromosomal reads were further removed using MIRA by mapping the filtered sequences to the chromosomal Wolbachia contigs (99% cut-off). The same procedure was followed for the Illumina data. The resulting 454 and Illumina reads were de novo assembled using MIRA. This initial assembly was subsequently improved using approaches described in the PAGIT protocol [51]. In brief, the contigs were aligned to the wMel genome using ABACAS [52], creating one large scaffold that consisted of the contigs successfully mapped to the wMel genome and a set of contigs that did not map. An attempt was made to close the gaps in the large scaffold using IMAGE [53] with the PCR-free small fragment library. After gap closing, the large scaffold was reduced once more to a set of contigs by breaking it around any of the unclosed gaps. This is because there are usually many genome rearrangements between different Wolbachia strains, and we would therefore expect a number of rearrangements to exist between the wMel and wGmm genomes. Breaking the scaffold makes allowance for these gaps. Finally, scaffolding was then performed on this reduced set of contigs using SCARPA [54] with the Hi-Seq mate-pair libraries. The statistics for the assembly at each stage of the process are given in Table S1.

The genome was annotated with XBASE and RAST [55], [56], followed by manual curation. Putative protein-encoding genes were identified using GLIMMER [57] and tRNA by tRNAscan-SE [58]. Predicted proteins were examined to detect frame-shifts or premature stop codons to identify pseudogenes using ARTEMIS [59]. Those for which the frame-shift or premature stops were of high quality by examining re-mapped reads in these regions were annotated as “authentic” mutations. This Whole Genome Shotgun project has been deposited at DDBJ/EMBL/GenBank under the accession AWUH00000000. The version described in this paper is version AWUH01000000.

The Sanger and 454 reads used in the tsetse genome assembly were obtained from flies treated with tetracycline as described previously [41]; therefore, these reads did not contain cytWol sequences. As mentioned above, Wolbachia specific sequences were filtered out from WGS reads of each sequencing technology with MIRA [50] using the complete genomes of wMel (AE017196), wRi (CP001391), and wBm (AE017321) as reference sequences. We obtained 5,306 (Sanger), and 10,978 (454) Wolbachia-specific sequences respectively. All the filtered putative Wolbachia-specific sequences were further examined using blast and a custom made Wolbachia database.

ChrWol-specific sequences were assembled with MIRA and AMOS [50], [60] using as a reference sequence the wGmm draft genome. The statistics for the two chrWol assemblies are as follows: N50 2970, mean contig length 1261.97, longest contig 15053, total length 527,504 bp for insertion A, while for insertion B N50 2791, mean contig length 1092.82, and total length 484,123. Genes were identified with Glimmer [61], followed by a round of manual curation using Blastn [62] and MegaBlast [62] against the non-redundant and custom made Wolbachia databases. The predicted CDSs were translated and used to search the NCBI non-redundant database, KEGG, and COG databases. The tRNAScan-SE tool [58] was used to identify tRNA genes.

Phylogenetic analyses were performed using Maximum-Likelihood (ML) and Neighbor-Joining (NJ) estimation for a concatenated set of six phage genes from wGmm, wRi, wMel and wPip. The genes used for the phylogeny included HK97 family phage major capsid protein (wGmm_0882, WD_0458, WRi_002750, WP0102), phage integrase family site-specific recombinase (wGmm_0004, WD_1148, WRi_009900, WP0980), phage SPO1 DNA polymerase-related protein (wGmm_0674, WD_0164, WRi_000900, WP0922), prophage LambdaW5 baseplate assembly protein W (wGmm_0971, WD_0640, WRi_005480, WP0303), prophage LambdaW1, baseplate assembly protein J (wGmm_0970, WD_0639, WRi_010130, WP0302) and a prophage LambdaW1, site-specific recombinase resolvase protein (wGmm_0960, WD_0634, WRi_005400, WP0342).

In addition, a concatenated set of ten genes (DNA-directed RNA polymerase, DNA polymerase III (alpha subunit), DNA gyrase B, translation elongation factor G, aspartyl-tRNA synthetase, CTP synthase, glutamyl-tRNA(Gln) amidotransferase B, GTP-binding protein, cell division protein FtsZ, fructose-bisphosphate aldolase) from the identified Gmm chromosomal insertions, wGmm, wRi, wMel, wPip, and wBm were used.

All sequences were aligned using MUSCLE [63] and ClustalW [64] as implemented in Geneious 5.4 [65], and adjusted manually. ML and NJ trees were constructed using MEGA 5.0 [66] with gamma distributed rates with 1000 bootstrap replications and the method of Tamura-Nei as genetic distance model [67].

To determine the number of chromosomal insertions, genomic DNA from tetracycline-treated Gmm females and normal Gmm individuals were restricted with HindIII endonuclease, electrophoresed on 1% agarose gel in 1× TBE buffer, and transferred to a positively charged nylon membrane according to Southern protocol [68]. The membrane was hybridized at 55°C with 350 ng of a 569 bp probe corresponding to part of the wsp gene labeled with the Gene Images Alkphos Direct labeling system (GE Healthcare, Little Chalfont, UK) using the random primer method following manufacturer protocols. Signal detection was performed using CDP-star followed by exposure to autoradiographic film (X-OMAT AR, Kodak). The absence of cytWol from the tetracycline-treated Gmm DNA was confirmed by a PCR assay, which resulted in only a single 16S rRNA amplification product originating from the chromosomal insertions [45].

Mitotic chromosome spreads were obtained from freshly deposited larvae from the Slovakia Academy of Sciences Institute of Zoology tsetse laboratory Gmm strain. Briefly, larval nerve ganglia were incubated on a slide in 100 µl 1% sodium citrate for 10 min at room temperature, and sodium citrate was replaced with methanol-acetic acid (3∶1 solution) for 4 min. The tissue was disrupted by pipetting in 100 µl 60% acetic acid for fixation and dropped onto clean slides heated on a hot plate at 70°C until acetic acid evaporation. After dehydration in 80% ethanol, slides were stored at −20°C for at least 2 weeks.

For in situ hybridization experiments, multiple probes specific for Wolbachia 16S rRNA, fbpA and wsp genes were amplified from the Slovakian strain DNA [45], [69]. To generate the labeled probes, 1 µg of DNA resuspended in 16 µl ddH₂O was denatured by boiling for 10 min. 4 µl of labeling mix (Biotin High Prime kit; Roche, Basel, Switzerland) were added and the reaction was incubated overnight at 37°C. After the reaction was stopped, ddH₂O (5 µl), 20×SSC buffer (25 µl) and formamide (50 µl) were added and 25 µl of denatured probe was placed on each pre-treated slide. The hybridization was performed at 37°C overnight in a humid chamber and detection of hybridization signals was performed using the Vectastain ABC elite kit (Vector Laboratories, Burlingame, CA, USA) and Alexa Fluor 594 Tyramide (Invitrogen). Chromosomes were DAPI stained and the slides were mounted using the VECTASHIELD mounting medium (Vector Laboratories). Chromosomes were screened under an epifluorescence Zeiss Axioplan microscope and images were captured using an Olympus DP70 digital camera. For the localization of signals on mitotic chromosomes the karyotype description of Willhoeft [69], [70] was adopted.

Natural samples of Gmm used to examine HGT fragments originated from four populations collected in Zambia, Zimbabwe and Tanzania (Table 1). DNA was isolated from adult flies stored in EtOH using the Qiagen DNeasy kit (Qiagen, Valencia, CA) following the manufacturers' instructions and stored at −20°C. The aposymbiotic (Wolbachia-free) Gmm line [41] was used as a control. For detection of Wolbachia, a PCR assay that amplified a 438 bp 16S rRNA fragment was used with the specific primer set wspecF and wspecR [71]. For input DNA control, a 377 bp fragment of the mitochondrial 12S rRNA gene was amplified with the primer set 12SCFR and 12SCRR [72]. The PCR amplification protocol was 10 min at 95°C, 35 cycles of 30 sec at 95°C, 30 sec at 54°C and 1 min at 72°C, and 10 min at 72°C.

The identification of the Wolbachia strain infections was based on MLST (gatB, coxA, hcpA, fbpA and ftsZ) and wsp-based genotyping approaches [45], [69]. PCR reactions were performed using the following program: 5 min of denaturation at 95°C, followed by 35 cycles of 30 sec at 95°C, 30 sec at the appropriate temperature for each primer pair (52°C for ftsZ, 54°C for gatB, 55°C for coxA, 56°C for hcpA, 58°C for fbpA and wsp) and 1 min at 72°C. All reactions were followed by a final extension step of 10 min at 72°C. Both strands of the products were sequenced using the respective primers. In addition, PCR products of 16S rRNA, wsp and MLST genes from the Gmm populations analyzed were cloned in pGEM-T Easy Vector System, and PCR products from several clones generated by the primers T7 and SP6 were sequenced in both directions using the BigDye Terminator v3.1 Cycle Sequencing Kit (PE Applied Biosystems) and were analysed using an ABI PRISM 310 Genetic Analyzer (PE Applied Biosystems). All Wolbachia gene sequences were manually edited with SeqManII by DNAStar and aligned using MUSCLE [63], as implemented in Geneious 5.4 [65], and adjusted manually.

To determine if genes from the chromosomal insertions were potentially expressed in locations other than the gonotrophic tissues, we utilized mapping of Illumina datasets from other studies, that included transcriptome reads from somatic tissues [73]–[75]. Reads were mapped to the chromosomal insertions using CLC Genomics Workbench (CLC Bio, Cambridge, MA) allowing no mismatches per reads, a maximum of 10 hits per read and 80% of the gene must match at 95%. Predicted open reading frames (ORF) from the insertions were extracted and the following criteria were utilized to determine the possibility of expression: 1) at least 25 reads were recovered from the ORF and 2) those represented had coverage of over 85% of the ORF. This filtering approach excluded genes with a high number of mapped reads that were only present in small limited sections of the ORFs. These sections with high read numbers mapping but low coverage could be where sequence similarity between Gmm, Wigglesworthia or Sodalis is high enough to yield mapping to the chromosomal insertions.

The draft genome of cytoplasmic wGmm contains 201 contigs of 1,019,687 bp, comprised of 800 putative functional coding sequences (CDS) and 16 pseudogenes (Figure 1 and Table 2). The GC content of wGmm is 35.2%, in the range observed for the other sequenced Wolbachia genomes (Table 2). Although, the wGmm genome is not complete, based on comparison of the identified contigs, it is most similar to the two Wolbachia strains associated with Drosophila melanogaster and D. simulans, wMel and wRi, respectively (Table S2). It is more distantly related to the genomes of the Wolbachia strains associated with Culex pipiens and Brugia malayi, wPip and wBm, respectively (Table S2). The majority of the regions and genes missing from the wGmm genome relative to the wMel and wRi genomes encode phage, ankyrin and hypothetical proteins (Tables S3 and S4).

Both PCR-based evidence from Wolbachia infected tsetse flies, and analysis of the Gmm annotated genome data indicated the presence of Wolbachia gene fragments inserted in the host genome. We mined the final assembly of the Gmm host genome and were able to identify 261 contigs that carried chrWol DNA sequences. Based on nucleotide diversity, close examination of the 261 contigs indicated that these represented at least three different events, which we refer to as insertions A, B and C. Manual editing and implementation of the AMOS snps script enabled the separation of the contigs into different insertions, with insertions A and B being the largest in size. Figure 2 shows the mapping of these two insertions on the wGmm reference genome. The observed pattern suggests that at least two large Wolbachia genome segments of 527,507 and 484,123 bps have been integrated into the Gmm chromosomes indicating that at least 51.7% and 47.5.% of the draft Wolbachia genome were transferred to the host nuclear genome. Sequence analysis of insertion A predicted 197 putative functional coding sequences, 148 pseudogenes, and 15 tRNAs. Remnants of 163 pseudogenes were discovered that are greater than 100 bp in size and that have either partially been integrated into the host genome, or only represent part of the pseudogene. For insertion B, sequencing analysis revealed the presence of 159 putative functional coding sequences, 148 pseudogenes and 13 tRNAs. In insertion B, 157 remnants of pseudogenes were also identified. Thus, on average more than 60% of the genes transferred to the tsetse nuclear genome have been pseudogenized. The average length of the putative functional coding sequences is slightly smaller than wMel, wRi and the cytoplasmic wGmm at 690 bp for insertion A and 677 bp for insertion B (Table S4). The GC% content for insertion A and B is 35.1%. Comparison between the chromosomal insertions A and B and the wGmm draft genome using Blastn and lastz indicated that: (a) the two insertions are very similar to each other (Figure S4) and (b) at least four genes, three hypothetical proteins and hemK are present in the chromosomal insertions but not in the cytoplasmic Wolbachia genome. The sequence identity between chromosomal and cytoplasmic genes and phylogenetic analysis based on ten concatenated genes clearly suggests that the chromosomal insertions A and B are closely related to the cytoplasmic wGmm genome (Table 5 and Figure S3). In more detail, comparison of the sequence identity in eleven chromosomal genes indicates that the majority of them exhibit a high sequence identity with the wGmm sequences (Table 5). The third Wolbachia HGT segment, insertion C, is only 2,089 bp in size and sequence analysis predicted the presence of only six pseudogenes.

A number of different types of mutations were identified in insertions A and B present in the host nuclear genome, and these shed light on the pseudogenization process. Our analysis suggests that more than 80% of the mutations that accumulated in the putative functional coding sequences represent single nucleotide polymorphisms (SNPs) (Figure 3). The majority of the genes that have been pseudogenized accumulated mutations that consist of nucleotide polymorphisms with deletions (NPD) and NPs. In both insertions, genes that have been pseudogenized contain mutations that combine NPs and deletions (NPDs) are more than those pseudogenized by NPs (Figure 3). In addition, we identified two additional types of mutations, NPs with insertions and NPs with deletions and insertions, associated with both chromosomal Wolbachia insertions but to a much lesser degree. A list of partial and full genes corresponding to the chrWol insertions is available in Tables S6, S7 and S8.

Based on our results, there were very few ORFs that met our criteria for expression from chromosomal insertions. In general, there were multiple ORFs that had high number of mapped reads (>100), but in nearly all cases the coverage of the mapping was below 30% indicating that these may represent reads from another symbiont or tsetse transcripts. Results were similar for the three transcriptomes analyzed from heads, salivary glands and the bacteriome. However, three putative ORFs satisfied our criteria: serB, ccmB and a degenerate transposase located at both insertions (102636-102894 for insertion A and 97255-97523 for insertion B). These analyses suggest that most of the genes present in the chromosomal insertions are likely not expressed, but the few specific genes we identified may have low levels of expression. Further studies will be necessary to validate their expression.

Hybridization of the wsp probe to Gmm female DNA restricted with the HindIII enzyme produced five bands of about 1200, 1600, 2150, 2600 and 2700 bp (Figure 4, lanes 1 and 3). DNA from tetracycline-treated females (cytWol-free) had a similar profile, except that the 2700 bp band, corresponding to the expected cytWol wsp fragment, was absent (lane 2). Untreated male DNA displayed an additional band of 1500 bp, indicating the presence of insertions on the Y chromosome (Lane 4). This banding pattern suggests the presence of at least five independent wsp chromosomal insertions, including one on the Y chromosome, supporting the in silico analyses.

To determine the location of Wolbachia insertions on Gmm chromosomes, we performed FISH analyses on mitotic spreads using wsp, 16S rRNA and fbpA specific probes. The Gmm mitotic complement, comprising the supernumerary dispensable chromosomes (B chr) [78] is depicted in Figure 5, where the AT-rich heterochromatic nature of Y and B chromosomes is indicated by the strong DAPI-staining. The two autosomes, L1 and L2, as well as the X chromosome, appear to contain heterochromatic regions on both sides of the centromere. FISH results indicate that the Wolbachia genes 16S, fbpA, and wsp consistently display a biased location on the distal part of the X, Y and B chromosomal arm. Although tyramide labeling generates strong and site-specific signals, it is difficult to detect the presence of multiple insertions on one chromosome if these events are localized in close proximity. The 16S rRNA signal detected on the short arm of the X chromosome appears to be particularly strong and diffused, and may thus represent more than one insertion event in that region.

Our previous characterization of the laboratory Gmm strain by Wolbachia-specific 16S rRNA-based PCR screening, the wsp-based and the MLST typing system revealed several HGT events [45]. Our results presented above indicate that these transfer events are in fact more extensive than previously considered. We next investigated the presence of HGT events in natural populations of Gmm originating from Zambia, Tanzania and Zimbabwe. We detected the pseudogenized fragment of the 16S rRNA gene carrying a deletion of 142 bp (Figure 6), similar to that we described in Gmm colony DNA prepared from the tetracycline-treated (cytWol-free) samples [41], [45]. We observed a similar phenomenon for fbpA, where a pseudogenized gene fragment could be amplified containing two deletions of 47 and 9 bp from the same four natural populations, as well as from the cytWol-free Gmm laboratory strain DNA sample. Finally, the HGT event of the Wolbachia wsp gene, which has been pseudogenized through a deletion of 7 bp, was also detected in two natural samples (Figure 6). Unlike the laboratory line of Gmm, in which all individuals analyzed carried the cytWol strain (100% infected), the prevalence of Wolbachia varied in the different populations and was not fixed (Table 1).

wMel genome (AE017196), wRi genome (CP001391), wPip genome (AM999887), wBm genome (AE017321), DNA-directed RNA polymerase (WD_0024/GI 42409679, WRi_000230/GI:225591874, WP0554/GI:190357240, Wbm0647/GI:58419220), DNA polymerase III (alpha subunit) (WD_0780/GI:42410358, WRi_006220/GI:225592374, WP0658/GI:190357336, Wbm0499/GI:58419072), DNA gyrase B (WD_0112/GI:42409755, WRi_001420/GI:225591969, WP1103/GI:190357759, Wbm0764/GI:58419337), translation elongation factor G (WD_0016/GI:42409671, WRi_000140/GI:225591866, WP0562/GI:190357248, Wbm0344/GI:58418918), aspartyl-tRNA synthetase (WD_0413/GI:42410026, WRi_003280/GI:225592127, WP0387/GI:190357090, Wbm0012/GI:58418589), CTP synthase (WD_0468/GI:42410077, WRi_002850/GI:225592086, WP1235/GI:190357884, Wbm0169/GI:58418745), glutamyl-tRNA(Gln) amidotransferase B (WD_0146/GI:42409786, WRi_003090/GI:225592108, WP0087/GI:190356829, Wbm0445/GI:58419018), GTP-binding protein (WD_1098/GI:42410645, WRi_012740/GI:225592933, WP0891/GI:190357560, Wbm0032/GI:58418609), cell division protein FtsZ (WD_0723/GI:42410305, WRi_007520/GI:225592482, WP0577/GI:190357261, Wbm0602/GI:58419175),fructose-bisphosphate aldolase (WD_1238 GI:42410776, WRi_012130/GI:225592879, WP1081/GI:190357738, Wbm0097/GI:58418674), HK97 family phage major capsid protein (WD_0458/GI:42410067, WRi_002750/GI:225592077, WP0102/GI:190356840), phage integrase family site-specific recombinase (WD_1148/GI:42410690, WRi_009900/GI:225592678, WP0980/GI:190357644), phage SPO1 DNA polymerase-related protein (WD_0164/GI:42409803, WRi_000900/GI:225591926, WP0922/GI:190357589), prophage LambdaW5 baseplate assembly protein W (WD_0640/GI:42410229, WRi_005480/GI:225592306, WP0303/GI:190357018), prophage LambdaW1, baseplate assembly protein J (WD_0639/GI:42410228, WRi_010130/GI:225592699, WP0302/GI:190357017) and a prophage LambdaW1 site-specific recombinase resolvase protein (WD_0634/GI:42410223, WRi_005400/GI:225592300, WP0342/GI:190357056)