1. New filarial tissues infected with Wolbachia

It is clear that tissues other than the female germline and the hypodermal lateral chords may be infected with Wolbachia. One of these infected tissues is the somatic gonad (epithelial layer), once briefly reported previously [52]. This Wolbachia localization was evidenced in Mansonella (Cutifilaria) perforata and Cercopithifilaria japonica (Table 2; Figures 5C,E). Interestingly, they are both members of the supergroup F of Wolbachia.

The real novelty is the Wolbachia tissue localization in the intestinal wall; it was only observed in M. (Cu.) perforata, but in all sectioned and whole mount samples (Figures 3C–F, 5D).

These divergent localizations suggest a more complex and diversified relationship between the bacteria and filariae. They also raise the question of how and when Wolbachia bacteria reach the appropriate filarial host tissues. It is likely that it is an early event, since it was shown in Brugia malayi an asymmetric distribution of bacteria in the egg followed by a preferential segregation in defined blastomeres [21].

2. Recent capture of Wolbachia type F suggested by Cercopithifilaria japonica, parasite of the Japanese bear

The species screened in this study generally confirm the previously identified types of Wolbachia in the Onchocercidae (Figure 6; see for instance [9]). The newly screened Loxodontofilaria is placed among the species of Onchocerca [53]. In addition, endosymbionts from this filaria belong to Wolbachia supergroup C (Figure 6). There is a major phylogenetic congruence discrepancy between Wolbachia and their hosts and it occurs in the genus Cercopithifilaria and the supergroup F of Wolbachia. One African and seven Japanese species of Cercopithifilaria have no Wolbachia, while one species in Japan is Wolbachia positive. The filarial hosts belong to a well-supported genus, Cercopithifilaria as evidenced by adult morphology [45], [54], larval morphology [55], 12S rDNA gene sequences [53], [56], the transmission by hard ticks and the skin-dwelling microfilariae [45]. A parsimonious interpretation of the Wolbachia screening is that a single acquisition event took place in C. japonica. This hypothesis is supported by the co-cladogenetic analyses (Figure 7).

The supergroup F is intriguing as it is presently the only Wolbachia type infecting both insects and onchocercid nematodes [7]. The Wolbachia supergroup F contains the species of Mansonella studied so far: M. (M.) ozzardi and M. (Esslingeria) perstans [13], [15], and in this study, M. (Tetrapetalonema) atelensis amazonae, M. (Cutifilaria) perforata. Cercopithifilaria japonica in supergroup F suggests a transversal transmission event, likely recent due to limited occurrence among the species of this genus. C. japonica is a parasite of the Japanese bear in which it coexists with a species of Mansonella of the subgenus Mansonella [57]. This Mansonella species, M. (M.) akitensis [58], has not been screened for Wolbachia but it likely harbours the type F Wolbachia. The bacterial host switching might have occurred between the two filarial parasites of the bear, perhaps via an oral infection route. Indeed filariae, despite their apparent small mouth, can ingest particles from their environment, such as red blood cells [59] and larger bodies, such as microfilariae released in the coelomic cavities of the filarial host [60].

3. Absence of Wolbachia

The filarial species in which Wolbachia were not detected appeared more numerous than it was thought, based upon previous observations. In [34], the percentage of negative species was 10.5% (2 negative among 19), in [5] with a larger sample, it was 37%. In this study, it is twice more elevated, 63%. It has to be emphasized that the negative results were not due to DNA degradations or bad extractions because all of the Wolbachia PCR negative samples gave positive amplifications using filarial nematode specific primers. However, we must take into account the fact that, in a few species, Wolbachia were not detected in all the specimens. This can partly be explained in species which do not harbour Wolbachia in the lateral chords, or at very low density, as Onchocerca dewittei japonica (Figures 5A,B) and Loxodontofilaria caprini (Figures 2H,I). In M. (Cu.) perforata, the bacteria are in the intestine wall of the female worms (Figures 3C,D, 5D) and any part of female body would be Wolbachia positive, but this is not the case (Tables 2, S1 and S2). Thus, in M. (Cu.) perforata, presence/absence of Wolbachia may occur, as suggested by [15] for another Mansonella species, M. (Esslingeria) perstans from humans. It is interesting to note that in both of these cases, the Wolbachia supergroup is F. Further, research performed on deeply studied filariae, such as Brugia malayi, have also shown that the amount of Wolbachia carried by a worm may vary greatly over time and be stage-dependent [21]. This dynamic probably has little impact when considering developing larval stages, because these are transient and the chance of recovering them in the wild is extremely low. More interesting is the observation that the female worms recovered in the wild are not all fully gravid in some species (Figure 2F). This was not the case of filariae from frogs, lizards, bats, birds, but was the case from some parasites of Japanese ungulates. It is worth to note that we paid great attention to the part of worm sampled to ensure that the germline was screened in all species.

4. Systematic position of Wolbachia negative filariae

It is clear that there is a need to increase the number of screened specimens for more solid results. However, in several cases, the global results are impressive and the distribution of the Wolbachia negative species does not appear random. As a matter of fact, at present, the species parasitic in frogs and lizards are negative for Wolbachia if they are Waltonellinae (three species of Ochoterenella), Oswaldofilariinae (one species of Piratuba), or Dirofilariinae (two species of Foleyella). It was also shown here that the first screened species parasitic in birds, an Aproctella in the Splendidofilariinae, was Wolbachia negative (7 females screened with PCR, from two passeriform species, totalising 7 specimens hosts). Wolbachia were not detected also in several species of Onchocercinae from mammals. The first is Litomosa chiropterorum, from an African bat (8 specimens screened from 8 Miniopterus schreibersi), an unexpected observation since the single Litomosa species screened previously, Li. westi from a North American rodent Geomyoidea, was Wolbachia positive [5].

The second species is Monanema martini (8 females screened from 8 murid specimens). The absence of the bacteria is an important feature, considering that this filaria was used, in the past, as a model of onchocerciasis, a true limit due to the important role played by the bacteria in the pathology of the disease [39], [61]–[63]. The absence of Wolbachia could explain partly the weak ocular pathology induced in the murid host [49]. Severe or mild ocular onchocerciases have been related to a strain-dependent variation of density of Wolbachia per filarial genome [64].

Other species in which Wolbachia were not detected belong to the genus Cercopithifilaria; 8 out of 9 species were Wolbachia-negative (Table 2; Figures 4; 5F).

Two hypotheses might be taken into account to explain the absence of Wolbachia. In the first case, Wolbachia could be considered to be present and then were subsequently lost [38]. This could have occurred before the mutually dependent symbiosis between symbiont and host developed. The second hypothesis is that the bacteria were not yet acquired within that filarial group.

The first hypothesis seems to possibly explain some of the observed cases, such as Litomosoides yutajensis (5 samples: 2 males previously and now 3 more females), a species without Wolbachia, in contrast with 5 congeneric species infected with Wolbachia. In the cases of Acantocheilonema viteae and Onchocerca flexuosa [38], the absence of Wolbachia appears to be a secondary loss, because some genes of the bacteria were incorporated in the filarial genomes. However, it is not clear whether this loss occurred before symbiosis was established or not. This is of interest because if loss was after symbiotic establishment, perhaps some of the genes incorporated into the host genome were those required by the host now provided by Wolbachia, but the extent of this event is still to be understood.

The hypothesis that the Wolbachia negative species might never have been infected, might be considered for the filarial groups that are supposed to be “ancient” such as the Oswaldofilariinae [44], [45], [65]. Estimation of dates of divergence has been proposed for some groups of nematodes based on molecular phylogenetic analyses [66]. In filariae, data from morphology, biology, geographic distribution, host range and palaeontology led to the proposal that the Oswaldofilariinae emerged during the late Jurassic, at the beginning of the Gondwanian break up, 140 mya [44], [45]. This is before the hypothesized ancestral acquisition of a Wolbachia by an onchocercid [3]; it follows that the absence of Wolbachia in Oswaldofilariinae could be primitive.

Foleyella is another parasite of saurians, which appears to have no Wolbachia; it is presently placed in the Dirofilariinae, which includes the Wolbachia positive Dirofilaria, but this systematic position needs to be revised because the characters of the infective larvae are very distinct [55]. A solid phylogeny of Onchocercidae linking traditional and molecular data is needed and warrants further investigations.

The subfamily Setariinae, parasitic in ungulates and in which Wolbachia were not detected, as first shown by [67], also deserves a comment. Based on larval morphology, it has been hypothesized that it evolved separately from the other onchocercids and derived from a group of spirurid Habronematinae [65]. Until now no spirurids have been found infected with Wolbachia [5], [34].

Did the bacterial infection never occur, or was the useful part of Wolbachia genome incorporated in the host genome and subsequently the bacteria eliminated, to reply to some adverse constraint? Further genomic analyses will solve this question concerning the absence of Wolbachia in ancestral filariae.

At present, the features observed on tissue or specimen distribution, and a very probable recent lateral transfer, suggest complex evolutionary dynamics of interactions between the symbionts, their host filariae and the nematode hosts. In the sampling studied, which will be further enlarged, it may become possible to use some defined species to decipher these questions.

Specimens and species

Specimens of filariae were recovered during dissections of the vertebrate hosts captured in the wild from different geographic areas [50], [54], [68]–[72].

All experiments, procedures and ethical issues were conformed to the competent national ethical bodies: Venezuelian animals were captured according to Licencia con Fines Cientificos N° 2192 dated June 18, 2007 and Contrato Marco Acceso Recursos Genéticos N° 33/2007 both granted by Ministerio del Ambiente de la Republica Bolivariana de Venezuela. Japanese serows, sika deer, bears, and wild boars were killed by hunters who have an individual permit to kill wild animals in accordance with the conservation and control policies of the Ministry of the Environment of Japan. Italian samples were collected by veterinarians and physicians and no permits were necessary. African rodents and agama do not belong to protected species and were obtained from local hunters. African mountain reedbuck, plain zebra, gemsbok and porcupine were also obtained from local hunters. Bats from South African bats have been collected for previous studies [48] in which no permits were necessary. Brazilian birds were donated by the Brazilian Institute of the Environment and Renewable Resources (IBAMA), Region of Juiz de Fora, Minas Gerais for the laboratory of Taxonomia and Ecology of Helminths of the Department of the Zoology.

Many of the filariae from large mammals were extracted from the subconnective tissue, dermis, or tendons of limbs. Host animals kept at 4°C were sent to the laboratory after they were killed. Afterwards the parts of the animals were dissected to collect living filarioids for Wolbachia study. From frogs, lizards, birds, bats and rodents filariae were generally recovered immediately after host death. Samples used for positive and negative PCR controls were laboratory strains.

Species identification of the specimens was done with morphological studies performed by several of us (OB, SU, RG, SL, MD). Some species are not yet named, but all are under study, and morphological analysis and sequencing of coxI gene in this study and in [53] showed that they represent distinct molecular entities. Co-infection of a host specimen by several congeneric or non-congeneric filarial species was rather frequent in large mammals, but in a few cases, the same filarial species was recovered from two host species. The supraspecific levels of taxonomy followed the systematic works of [46], [73],[74] and more recent studies for some taxa: [65], [71], [75] for the subgeneric divisions of Mansonella Faust, 1929; [54], [56] for the genus Cercopithifilaria Eberhard, 1980 (created as a subgenus); and [54] for the genus Loxodontofilaria Berghe & Gillain, 1939.

Abbreviations used in the text are: A. for Aproctella; C. for Cercopithifilaria; Cu. for Cutifilaria; D. for Dirofilaria; Di. for Dipetalonema; F. for Foleyella; L. for Litomosoides; Li. for Litomosa; Lo. for Loxondontofilaria; M. for Mansonella; Mo. for Monanema; O. for Onchocerca; Oc. for Ochoterenella; P. for Piratuba; S. for Setaria; T. for Tetrapetalonema. The material screened in this study is detailed in Figure S1, Tables S1 and S2.

Filarial worms were fixed and kept in absolute alcohol at 4°C for PCR analyses, in 4% PFA (paraformaldehyde) for overnight at 4°C for immunohistochemical staining and whole mount fluorescent analysis. In many cases, filarial specimens were cut into anterior (a), median (m) and posterior (p) parts, which were fixed for the different analysis approaches. Since Wolbachia are transmitted by female filariae, almost all of the studies were based on female worms; male specimens were rarely examined, and we analysed this sex alone in a single case, Di. gracile (Table S3).

Molecular screening for Wolbachia on filarial nematodes

PCR screening for Wolbachia was conducted according to [5], [12], using general Wolbachia primers for 16S rDNA (99f and 994r [76]), originally designed to work on Wolbachia from the supergroups A and B, and primers for 16S rDNA (16SWolbF and 16SWolbR3), originally designed to work on Wolbachia from the supergroups A–D [12], but whose target sites are also conserved in Wolbachia from supergroups E and F [8].

PCRs were performed in a 20 µl final volume under the following conditions: 1x buffer (containing 1.5 mM MgCl₂, Eppendorf™), 0.2 mM of each dNTP, 1 µM of each primer, and 0.5 U of Taq DNA Polymerase (Eppendorf™). The thermal profile used was: 94°C 45 sec, 52°C 45 sec, and 72°C 90 sec for 40 cycles.

When the PCRs were negative under the above PCR conditions, a nested-PCR approach was implemented in order to improve the sensitivity of the PCR screening [5], [77]. The first PCR was performed using the general eubacterial primer 27F [78] combined with 16SWolbR3; PCR volumes and conditions were as above. Five µl were visualised on a 1.5% w/v agarose gel and one µl of the first PCR was diluted 1/10 and 1/100 in water, and then both used as templates in a second PCR, performed using primers W-EF and W-ER [70]. W-ER and W-EF recognize sites that are conserved in supergroups E–F and that are internal to the primers used in the first PCR. PCR conditions for this amplification were as described in [79].

Of samples remaining negative after the two PCRs approaches described above, PCRs with primers 16SWolbF and 16SWolbR3 were performed varying the following parameters: MgCl₂ concentrations at 2.5, 4 and 6 mM and annealing temperatures of 52°C +/−5°C.

DNA preparations from filarial species harbouring Wolbachia (D. immitis and Brugia pahangi) and from a Wolbachia-infected strain of mosquitoes (Culex pipiens) [5] were included in the screening as positive controls. DNA preparations from a filarial species not harbouring Wolbachia (A. viteae) [5] were included as negative controls.

Of the samples positive for PCR screening homologous to Wolbachia 16S rDNA, dnaA, ftsZ and groEL were also amplified using the primers described in [5], [80] under the following conditions: 1x Eppendorf buffer including 1.5 mM MgCl2, 0.2 µM of each dNTP, 1 µM each of forward and reverse primers, and 0.5 units MasterTaq (Eppendorf). The thermal profiles we used were: (1) dnaA, 94°C 45 sec, 52°C 45 sec, and 72°C 90 sec, for 40 cycles; (2) groEL, 94°C 45 sec, 60°C 45 sec, 72°C 80 sec, for 5 cycles, and 94°C 45 sec, 55°C 45 sec, and 72°C 80 sec, for 34 cycles; (3) ftsZ, 94°C 30 sec, 60°C 45 sec, 72°C 90 sec, for 5 cycles, and 94°C 30 sec, 57°C 45 sec, and 72°C 90 sec, for 34 cycles.

Amplifications were performed in 20–50 µl volumes.

In all cases, in order to ascertain the DNA conditions before Wolbachia screenings and to confirm morphological identification, coxI amplification was performed as described in [12].

Sequencing conditions

coxI sequencing was performed as described in [12]. dnaA, ftsZ and groEL sequencing were performed as described in [5], [80]. From the Wolbachia PCR positive samples, almost the full length of the 16S rDNA gene of Wolbachia was sequenced using primers 27F and 16SWolbR3. The amplifications obtained (about 1400 bp) were gel-purified (using the QIAquick® PCR Purification Kit, Qiagen) and directly sequenced using ABI technology. The sequences obtained have been deposited in the EMBL Data Library. The dnaA, groEL and ftsZ sequences were not obtained from all the taxa included in this study mainly caused by the scarcity of certain specimens, and amplification/sequencing problems for some of the species examined. Where DNA of the host was amplified, the PCR product was purified as above and directly sequenced using ABI technology.

A list of the sequences including accession numbers is available in Table S2.

Immunohistochemical staining of worm sections

Immunohistochemical staining was performed according to [24], [48]. A rabbit polyclonal antiserum raised against the WSP of the Wolbachia from B. pahangi (designed by [81]) has been used to stain samples of 13 filarial species (see Tables S2 and S3). After paraffin inclusion, 4 µm sections were obtained and placed on Silane coated glass slides (3-aminpropyltriethoxysilane) and then kept at 63°C overnight, to avoid sections detaching from slides. To verify that the 63°C overnight did not alter the specific epitopes, Wolbachia positive L. sigmodontis laboratory strain were incubated at both 63°C and room temperature overnight and then stained: no significant differences in the staining were observed. Negative controls were carried out by omitting the primary antibody.

Fixed female worms were divided in three main parts, posterior (p), median (c) and anterior (a) in order to observe the different regions of the genital tract. Transverse sections were made at different levels of each part, and few of them were stained with hemalun-eosin for anatomical identification. Lateral cuticular internal crests were identified to orient the worm section; hypodermal chords extend above and on the side of the crests. The filarial species used for histology are opisthodelphic and the initial part of the ovaries is in the posterior part of the worm; the distal region of the ovary is composed of a cytoplasmic axis, the rachis [82], [83], and an outer cytoplasmic layer with the nuclei of oogonia and oocytes (in the text we will refer to these states as oocytes, and germline to describe the whole production of the gonad). An epithelial layer and an outer muscle layer surround the gonad, both referred to as somatic gonad in the text [25]. Uteri occupy almost the whole body and are found in median and anterior part of worms. The different uterine contents are ovulae, spermatozoa, divided eggs and microfilariae. Eggs were identified as aborted by hemalun-eosin staining when divided eggs were eosinophilic and nuclei not discernible. The laboratory strain of Litomosoides sigmodontis, which has been shown in several studies to harbour Wolbachia [84]–[86] was used as a positive reference for Wolbachia immunostaining.

Whole mount fluorescent analysis

Worms were cut with a razor blade to expose the different tissues to RNAse A (15 mg/mL, Sigma) in rotating tubes overnight at 4C. They were rinsed in PBS, and incubated with a fluorochrome-conjugated Phalloidin (atto−488 Phalloidin, Fluka, at 10 nanomolar) overnight to stain F-actin, followed by a Propidium Iodide (Molecular Probes, 1.0 mg/mL solution) incubation for DNA staining for 20 minutes in PBS (1∶50) and a 5 minutes wash. Tissues were mounted in Vectashield (Vector Laboratories) [25]. The species analyzed were C. crassa, C. japonica, Lo. caprini, M. (Cu.) perforata and O. d. japonica (Table S1).

Phylogenetic reconstruction

The bacterial 16S rDNA, dnaA, groEL and ftsZ sequences and filarial 12S rDNA and coxI sequences generated were aligned with the sequences available in the databases (for the ribosomal genes according to their secondary structures) using ClustalX2 [87].

The alignments were analysed using Maximum Likelihood (ML) and Bayesian Inference of phylogeny (BI) methods. The appropriate model of sequence evolution for ML and BI was estimated via likelihood ratio test using Modeltest 3.7 [88]: the model selected for the filarial concatenated dataset was HKY+G, while the model for the Wolbachia concatenated dataset was GTR+I+G. Phylogenetic analyses were performed using PAUP* 4.0 b10 [89] and MrBayes [90]. In addition, a phylogenetic tree was inferred with a GTR+Γ₄ nucleotide substitution model in a Bayesian framework using MrBayes version 3.0 [90]. Two independent runs were performed, each using 1 million steps with four chains sampling every 100 steps. The first 10% of the trees were removed and posterior probabilities were calculated from these post-burnin trees.

Test for recombination

Four Wolbachia alignments (relative to the genes 16S rDNA, ftsZ, groEL and dnaA) were screened for presence of recombination events by using a set of nonparametric detection programs: RDP, GENECONV, Bootscan, MaxChi, Chimaera, SisScan and LARD [91]–[97]. The first six programs search for putative recombination breakpoints in a set of aligned DNA sequences and are implemented in RDP3 software package, whilst LARD checks signals detected by other methods [92]. Sequences were auto-masked for optimal recombination detection. General recombination settings were as follows: sequences were considered linear and the highest acceptable P-Value was set to 0.01 and overlapping signals were disentangled.

Method specific options were as follow: MaxChi and Chimaera were run with a variable window size; Bootscan and SisScan were forced for exploratory screening. Successive phases of refining analysis and manual tests were performed where needed.

Cocladogenesis analysis

To test for congruence, phylogenies for the hosts (based on a concatenated dataset of coxI and 12S rDNA) and for the bacterium (based on a concatenated dataset of 16S rDNA, ftsZ, dnaA and groEL) were compared with two methods. The first tested whether there was a greater than random correspondence between reconstructed nodes for host and symbiont. This was performed in Component (R. Page, University of Glasgow, UK) using the “compare tree with” function, with 1,000 randomized trees and the four available tree-comparison metrics (partition; triplets; quartets, nearest neighbour interchange). The second test examined the null hypothesis that the endosymbionts have undergone cocladogenesis with their hosts. ML trees for each dataset were first estimated using the successive approximation method [89]. Then, the scores for each of these ML trees based on the host dataset were compared using the [98] test. This was repeated using the parasite dataset. A single uninfected filarial (T. callipaeda) and a non-filarial Wolbachia (from C. quinquefasciatus) were used as outgroups.