Nematode samples from infected humans and animals were obtained with the approval of the ethics committees and regulatory authorities of all institutions and countries involved in this study.

Specimens with blood containing L. loa microfilariae, adult L. loa worms and skin biopsies extirpated by ophthalmologists and other physicians from human patients had been sent to the Bernhard Nocht Institute for diagnosis. Several specimens of a spleen extirpated from a 24-year-old German, who had been travelling in Nigeria and Cameroon, were kindly supplied by Prof. GD Burchard 24. Further adult worms and a sample of spleen were collected in Cameroon from an experimentally infected two-year-old drill (Mandrillus leucophaeus) born in captivity.

Third stage larvae were collected from Chrysops silacea fed on microfilaremic human volunteers from Cameroon, who had expressed informed consent. Infective larvae from these C. silacea were used to infect the drill and seven months later adult L. loa worms and the spleen were recovered. Samples of the spleen were embedded in Paris and sent to Hamburg. Fragments of additional adult L. loa worms from previous experimental infections of monkeys were available in Paris. Six of these worms belonged to the human strain and three to a monkey strain.

Onchocercomas with microfilariae and adult O. volvulus and specimens of other filarial worms embedded in paraffin were available from several previous studies 1325. To test the reactivity of the antibodies against Wolbachia proteins from various hosts, we also examined females of the sand flea Tunga penetrans removed from patients in Ghana 26 and the mosquito Culex pipiens from a laboratory colony maintained at the Istituto di Parassitologia, Università die Roma La Sapienza.

Electron micrographs that had been taken during previous studies on L. loa 171827 and on O. volvulus were re-examined (28 and unpublished studies). The L. loa worms had been collected in Lambarene (Gabon) during herniotomies or they had been sent to the Bernhard Nocht Institute for diagnosis by physicians in northern Germany. The O. volvulus worms had been removed from patients in Liberia. The filariae had been fixed in 2% buffered glutaraldehyde and 1% osmium tetroxide, embedded in araldite, epon or Spurr's ERL medium and processed for electron microscopy as usual. Two years ago, L. loa microfilariae from the blood of a Cameroonian patient were processed for immunogold electron microscopy 1926. For light microscopic controls, semi-thin sections had been stained with azure II and methylene-blue. Larger pieces of onchocercomas were also embedded in methacrylate and stained by methylene-blue.

The worms and the biopsies from organs had been fixed in 80% ethanol or 4% buffered formaldehyde. They were embedded in paraffin and stained conventionally with haematoxylin & eosin. Biopsies from organs also were stained with Giemsa or Pappenheim stain. Sections of all adult L. loa worms selected for this study (Table 1) and selected onchocercoma sections were stained with silver using the Warthin-Starry method conducted under strictly controlled staining conditions 29. For immunohistology, the alkaline phosphatase anti-alkaline phosphatase (APAAP) method was applied according to the manufacturer's recommendations (Dako Diagnostika, Hamburg, Germany). Antisera against filarial, wolbachial or Yersinia enterocolitica proteins or against human leukocytes were used as primary antibodies. We applied as secondary antibodies anti-rabbit mouse immunoglobulins (clone MR12/53, Dako Diagnostika, Hamburg, Germany) for rabbit sera against filarial and bacterial proteins and anti-mouse antibodies for the monoclonal immunoglobulins against the proteins of human immune cells. Fast Red TR salt (Sigma, Deisenhofen, Germany) was used as chromogen and haematoxylin (Merck, Darmstadt, Germany) functioned as the counterstain.

As primary antibodies for all selected L. loa worms (Table 1), O. volvulus and other filarial worms we used rabbit antisera against the following recombinant proteins: Wolbachia surface protein of Dirofilaria immitis Wolbachia (Wol-Di-WSP, dilution 1:1 000 – 1:4 000, 3031), aspartate aminotransferase of Wolbachia from O. volvulus (Wol-Ov-AAT, dilution 1:60 – 1:200, 32), and Y. enterocolitica heat shock protein-60 (Y-HSP60, dilution 1:1 000, 33), which had been supplied by Prof. IB Autenrieth, Tübingen. The specificity of the antisera against Wol-Di-WSP and Y-HSP60 to label Wolbachia had been shown by immuno electron microscopy 26. Wol-Di-WSP labelled only WSP, Y-HSP60 also labelled filarial HSP60 in mitochondria and other tissues, and Wol-OV-AAT labelled both wolbachial and the filarial AAT especially in sperms. In addition, for selected sections we applied antibodies against human HSP60 (clone LK2, Sigma, dilution 1:5), HSP60 from O. volvulus Wolbachia (Wol-Ov-HSP, dilution 1:500, 19) supplied by PD Dr. KD Erttmann, a catalase from O. volvulus Wolbachia (Wol-Ov-CAT, dilution 1:30 – 1:50, 14) supplied by PD Dr. K. Henkle-Dührsen, Düsseldorf, and monoclonal antibodies against human neutrophil granulocyte elastase (Dako Diagnostika, dilution 1:150, 34), and neutrophil defensins (Dianova, Hamburg, Germany, dilution 1:1 000 – 1:4 000, 34). Furthermore, antibodies against proteins of eosinophil granulocytes, macrophages 34, mast cells and other filarial and wolbachial proteins were applied. As positive control a Wolbachia-positive onchocercoma section was included for each stained set of L. loa sections.

DNA was prepared from ethanol preserved or paraffin embedded adult L. loa worms using the Dneasy Tissue Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. Microfilariae were isolated from two patients from Cameroon with confirmed loiasis by blood filtration using a 5 μm polycarbonate filter. Filters with approxiamtely 500 microfilariae each were washed using 500 μl TE buffer (0.01 M Tris, 0.1 mM EDTA), lysed in 250 μl DSP buffer (0.02 M Tris, 0.05 M KCL, 2.5 mM MgCl₂, 0.5% tween 20, 150 μg/ml proteinase k) and 1 μl template was used in a 50 μl PCR assay. Three different sets of primers were used to amplify Wolbachia DNA: one Wolbachia-specific primer pair targeting the 16S rDNA, one primer pair targeting the gene encoding the FTSZ cell cycle protein and one primer pair targeting the gene encoding the WSP as described in detail previously 26. These primers were known to amplify not only DNA of Wolbachia from filariae but also from T. penetrans. As positive control, DNA of untreated O. volvulus worms was used. In addition, to test the quality of the DNA template, the 5S rDNA of the filarial host was amplified by PCR as previously described 35.

Onchocerca volvulus represents a typical filaria that contains obligatory symbiotic Wolbachia. We have therefore examined the characteristics of the obligatory symbiotic Wolbachia in more than 800 adult O. volvulus worms and in several other filaria species. It was found that each intact living worm harboured Wolbachia in large numbers (Figure 1, Figure 2, Figure 3, Figure 4, Figure 5). This concerned worms from patients from Uganda, Cameroon, Benin, Togo, Ghana, Burkina Faso, Mali, Liberia, Mexico, Guatemala, Brazil, and Yemen. We observed numerous endobacteria in the hypodermis of female worms in all portions from the nerve ring to the posterior end.

In many oocytes, embryos and microfilariae the endobacteria were not detected by immunohistology. However, when the endobacteria had expressed the respective proteins, for which the antisera applied were specific, all live (intact) oocytes or morulae presented endobacteria (Figure 5A, 5C, 5E). Using electron microscopy we counted up to 14 bacteria in one oocyte section. We assume that all live embryos developing later to microfilariae harbour at least ten bacteria (and probably more since an ultra-thin section of 0.1 μm covers only a thin layer of the oocyte). As far as it can be concluded from the limited numbers of worms examined, we assume that this occurrence of numerous endobacteria found in O. volvulus applies also to O. ochengi (Figure 4C; Figure 5C), O. dukei (Figure 3E), O. gibsoni (Figure 5E), O. fasciata (Figure 4A), O. jakutensis (Figure 2E36), Litomosoides sigmodontis 12, Wuchereria bancrofti, Dirofilaria immitis, and Dirofilaria repens.

Based on these findings, we define a filaria species harbouring obligatory symbiotic Wolbachia as one with numerous endobacteria in each adult worm and several bacteria in each oocyte and embryo that will develop to a mature microfilaria. The studies described in the following paragraphs aimed to search for Wolbachia in the L. loa worms in numbers, as they were observed in the above-mentioned filaria species containing obligatory symbiotic Wolbachia.

Screening several dozen electron micrographs from the previous studies on L. loa and O. volvulus 17182728 we often found endobacteria in the oocytes of O. volvulus (Figure 1A). In contrast, no endobacteria were observed in the oocytes of the ovary (Figure 1B), the uterus (Figure 1C) or in the early morulae (Figure 1D) of L. loa. Immunogold electron microscopy using the anti-Y-HSP60 serum showed well-labelled mitochondria but no endobacteria in the cells of L. loa microfilariae.

In semi-thin sections stained with azure II and methylene-blue we detected granular structures in the hypodermis (Figure 2A,2C), oocytes and embryos of dozens of O. volvulus worms but never any in L. loa worms (Figure 2B,2D). Using silver staining of paraffin sections, we found endobacteria-like granules in consecutive sections of O. volvulus and O. jakutensis (Figure 2E) precisely where Wolbachia were seen after labelling with specific antisera against Wolbachia antigens. In contrast, none of the L. loa worms selected for this study (Table 1) displayed such silver-stained granules (Figure 2F). Furthermore, no endobacteria-like granules stained by haematoxylin or Giemsa stain were seen in the hypodermis of L. loa.

The reactivity of Wolbachia with our antisera was examined in various hosts. The antiserum against Wol-Di-WSP reacted strongly with Wolbachia belonging to wolbachial clade C: D. immitis, D. repens, O. volvulus, O. gutturosa, O. dukei, O. gibsoni, O. fasciata, O. armillata, O. ochengi, O. jakutensis, O. tarsicola; to clade D: Brugia malayi, Brugia pahangi, L. sigmodontis, W. bancrofti, and three other filarial species and with the Wolbachia from the insects Cx. pipiens (clade B) and T. penetrans. The antiserum against Y-HSP60 reacted well with all the Wolbachia of the above-mentioned filariae and T. penetrans (except that it was not tested with those of B. pahangi and Cx. pipiens). We conclude that these two antisera react with all Wolbachia. The other antisera were examined with some or most but not all of the above mentioned Wolbachia.

Having shown the suitability of these anti-wolbachial sera, L. loa worms were compared with O. volvulus and other Wolbachia-positive filariae. Using these antisera against Wolbachia we easily detected endobacteria in many filariae in the hypodermis of male (Figure 3E) and female worms (Figure 3A,3C; Figure 4A,4C; Figure 5E). We found the Wolbachia well labelled by the antisera in the oocytes of the ovary (Figure 5A) and in portions of the uterus (Figure 5C), in zygotes, in morulae (Figure 5E) or in other embryos and microfilariae (Figure 6A), when the respective wolbachial proteins had been expressed. However, to interpret these photos and other previously published photos 12143236373839 it has to be taken into account that it would have been easy to present as many photos from other sections of the same worms, in which no endobacteria occurred or where the endobacteria had not expressed the respective proteins. Such non-reactive endobacteria were often well stained by haematoxylin, Giemsa or toluidine-blue.

Before we selected the L. loa worms for our study, we examined them with several antisera against filarial proteins. Antisera against the ankyrin-related protein 40 and glutathione S-transferase of O. volvulus did not react with any of the L. loa worms. Y-HSP60 cross-reacting with filarial proteins 3639, reacted weakly with L. loa proteins. However, using immuno electron microscopy, mitochondria of microfilariae were distinctly labelled by this antibody. Good cross-reactivity was achieved by the antiserum against Wol-Ov-AAT, which labelled the Wolbachia-free sperms of L. loa and with two other antisera. All L. loa worms that did not react well with Wol-Ov-AAT were excluded as unsuitable for immunohistology. Those that were well labelled (Figure 3B; Figure 4B; Figure 5D) were selected for the study. As shown in Table 1, seven complete female L. loa worms producing embryos and microfilariae, six complete male worms, and some fragments of 14 other adult worms were examined. Each complete worm had been cut into small pieces before embedding in one block that was cut until nothing was left. All sections were analysed by either anti-Wol-Di-WSP or anti-Y-HSP60, except for a few sections stained with anti-Wol-Ov-AAT or by the silver method. This procedure yielded about 20 slides from each L. loa worm, each slide containing many worm sections. No Wolbachia was detected in any of these sections (Figure 3B, Figure 3D,3F; Figure 4B,4D,4E,4F; Figure 5B,5D,5F; Figure 6B,6D). Furthermore, no Wolbachia were observed in L. loa when antisera against human HSP60, Wol-Ov-HSP60, Wol-Ov-CAT and another antiserum were applied.

Wolbachia could easily be observed with all antisera in microfilariae in onchocercomas (Figure 6C), skin, and lymph nodes. In contrast, no Wolbachia were found in several hundred L. loa microfilariae from the spleen of a patient and of an experimentally infected monkey. Sections of L. loa microfilariae from a sample of human blood were also non-reactive with the antisera.

Previously it has been shown that the accumulation of neutrophil granulocytes around adult O. volvulus worms is dependent on Wolbachia 36. Such accumulation could also be seen around the other endobacteria-positive filariae: O. jakutensis 36, O. dukei (Figure 3E), O. ochengi (Figure 4C), O. gibsoni, and D. repens in subcutaneous nodules of human patients. In contrast, no neutrophil granulocytes were found around the two female L. loa worms in skin biopsies from patients (Figure 4E). Since live adult L. loa are mobile these observations were not sufficient for a conclusion. Therefore, degenerated microfilariae in the spleen were examined. We had previously shown the activity of neutrophils against O. volvulus microfilariae in onchocercomas 41 and in the skin after treatment with diethylcarbamazine 34. Such activity of neutrophils against microfilariae could clearly be shown after staining with antisera specific for proteins of neutrophils (Figure 6E,6G). However, neutrophils present in human and monkey spleen tissue, where degenerating L. loa microfilariae could be observed, were not attached to the parasites (Figure 6F,6H). The degenerated or disintegrating microfilariae were attacked only by eosinophil granulocytes, macrophages and small giant cells, as shown previously by Duke 42. We conclude from these findings that the microfilariae of L. loa do not contain sufficient numbers of Wolbachia, if any, to attract neutrophil granulocytes.

The 5S rDNA spacer of L. loa was amplified from all samples included in the study, indicating the absence of significant PCR inhibitor in the samples. Using the Wolbachia 16S rDNA primers, the ftsZ primers and the wsp primers, PCR products of about 530 bp, 510 bp and 490 bp, respectively, were obtained in all O. volvulus samples, which functioned as controls. In contrast, no bands were visible on the ethidium bromide stained agarose gel when the DNA from the three males and six female L. loa worms and two batches of L. loa microfilariae samples were tested.