Ethics statement

All research performed in this study conforms to relevant ethical regulations. All animal experiments using jirds were approved by the ethical committees of the University of Liverpool and Liverpool School of Tropical Medicine (LSTM) and conducted under Home Office Animals (Scientific Procedures) Act 1986 (UK) requirements (license numbers P86866FD9 and PP6173839). Animals had free access to food and water throughout the duration of the studies, were checked daily for welfare and were weighed weekly.

Human serum was obtained from patients enrolled in a double-blind placebo-controlled randomized clinical trial conducted in Cameroon and was approved by the ethics committees of the Tropical Medicine Research Station, Kumba, Cameroon; the Research Ethics Committee of the LSTM, Liverpool, UK; and National Health Service (NHS) National Research Ethics Service (09/H1001/81, Northwest 4 REC). Written informed consent was obtained from all participants, with the exception of those who were illiterate, for which a literate witness signed on behalf of the participant and the participant added a thumbprint (trials registry, number ISRCTN48118452)41. Human sera from Uganda, Cameroon, Togo, Nigeria and Ecuador were obtained from the National Institutes of Health (NIH), National Institute of Allergy and Infectious Diseases (NIAID) FR3 (http://www.filariasiscenter.org/), which provides blanket approval for the research community to use as part of a NIAID-funded sample repository. These samples were curated by the EMCF in 198543 and the CDC onchocerciasis serum bank. Ethical approval for use of UK uninfected control sera was obtained from the NHS Research Ethics Committee (16/NW/0170) and the Central Liverpool Research Ethics Committee (protocol number UoL001207). All donors had given approval and consent to be included as part of additional experiments (such as those described in this paper) upon sample collection.

Ethical clearance for the collection of O. volvulus parasite material from Cameroon was obtained from S. M. Ghogomu of the Ethics Review and Consultancy Committee, Cameroon Bioethics Initiative, P.O. Box 31489, Biyem-Assi, Yaoundé, Cameroon, reference number CBI/ 443/ ERCC/CAMBIN. Ethical clearance for the collection of O. volvulus parasite material from Ghana was obtained via a full board review from the Kintampo Health Research Centre Institutional Ethics Committee, Ghana, study identification number KHRCIEC/2018-18. Written informed consent was obtained from all participants, apart from those who were illiterate, for which a thumbprint was used instead.

Statistics and reproducibility

Randomization was not relevant for this study, as no particular experimental variable or covariate was being tested. We were interested only in the presence or absence of viruses in datasets or parasite material that is currently available to us. Blinding was also not relevant for this study, as this work was focused on analysing the presence or absence of viruses in material in which no previous knowledge is available. However, as the investigators had no previous knowledge on the parasite’s infection status with viruses (and their jird host’s exposure level to the viruses), data collection and analysis were randomized and performed blind to the conditions of the experiments, except for experiments using serum from jirds and humans whose infection status with their respective parasite was known. No statistical method was used to predetermine sample sizes for experiments performed in this study, as no information on viral infection status is available for sample size calculations. Instead, the investigators used the maximum number of samples possible from previously collected material40,41,43. No animals or data points were excluded from the analysis.

Parasite maintenance

The life cycle of B. malayi was maintained in mosquitoes and male jirds as previously described40. Briefly, microfilariae (mf) were collected via peritoneal catheterization from jirds infected for more than 12 weeks. The mf were purified using PD10 columns in RPMI 1640 (1% penicillin–streptomycin and 1% amphotericin B) and mixed with heparinized human blood to 15–20,000 mf per ml; the blood with mf was fed to female Aedes aegypti mosquitoes from the FR3, using an artificial membrane feeder (Hemotek). After 14 days, infective L3 were collected from infected mosquitoes by crushing and Baermann’s filtration. Jirds were inoculated with 100 L3 larvae via the intraperitoneal route, and adult B. malayi were collected from the peritoneal cavity at necropsy, either 6 months or 18 months after infection. Infections with O. ochengi were performed by surgical implantation of adult male parasites into jirds, and sera were collected 15 days after infection40.

All jirds used to maintain the life cycle at LSTM were purchased from Janvier Laboratories and reared to ethical standards as described previously. Jirds used were all males and aged 3 months when infected with B. malayi. Parasites that were maintained in-house can be termed the ‘LSTM strain’, but are descendants of the TRS (now FR3) strain, and were originally obtained from TRS Laboratories. The exact age of the parasites used for experimental processing cannot be monitored. Infected jirds were processed for parasites and/or serum between 18 and 22 months of age. Uninfected, naive jirds were processed between 2 and 3 months of age for serum. The parasites used in this paper were from this ‘LSTM’ strain, unless specifically mentioned (for example, FR3 strain).

Virus genome assembly and annotation

Transcriptome datasets were downloaded from the European Nucleotide Archive (ENA) through the ENA Browser Tools application59 (v1.5.3) via the BioProject accession number, or a list of accessions identified as RNA sequencing from the project details. A full list of this is available in Supplementary Table 1. All datasets chosen for analysis were from experiments originally designed to investigate the parasite transcriptome specifically, either for genome assembly and annotation, or investigation of nematode biology. These datasets used either Illumina sequencing or 454 Titanium pyrosequencing technologies, and assembly. These were trimmed for adapters and quality using bbduk60 (v38.79) for a minimum PHRED quality score of 20 and minimum trimmed read length of 50 nucleotides. The reference libraries ‘adapters.fa’, ‘nextera.fa’ and ‘truseq.fa’ from the distributed bbduk package were combined and used for adapter trimming. The resultant reads were then assembled using the program rnaviralSPAdes61,62 (v.3.15.3) with default settings, without filtering reads against the host genome. This was done as an initial assembly to conduct a full screen of any virus-like sequences that may be present in the transcriptome or integrated into the host genome. Once assembly was completed, sequences were then analysed using the program Virsorter2 (ref. 63; v.2.2.3) using a minimum assembled length of 1,000 base pairs and to screen only against the RNA group of viruses. As additional validation, we also used the program geNomad64 (v1.6.1) to screen for viruses. All sequences identified by Virsorter2 and/or geNomad were first validated for quality and completeness using the program CheckV21 (v1.0.1). This program looks for potential host contamination in the assembled virus sequences and assigns a predicted ‘completeness’ score to the sequences based on a curated hidden Markov model gene database. Results from CheckV are available in Supplementary Table 2. Validated sequences were subsequently taken for further analysis via BlastX and BlastN against non-redundant protein and nucleotide databases in NCBI, respectively, to identify the closest relatives or presence of potential genome integrations with the nematode host. This was done by limiting the searches to either ‘Viruses (taxID 10239)’ or ‘Nematoda (taxID 6231)’, respectively. Sequences were annotated using PROKKA65 (v.1.11) with the standard genetic code to identify reading frames and protein sequences, which were further analysed using InterProScan66 to identify functional domains, such as the RdRP. Visualization of virus genomes was done using Artemis67 (v16.0.17), and coding regions were plotted using ggplot2 (ref. 68; v3.4.2) and the gggenes package69 (v0.5.0).

A second reassembly of the reads was also performed after removal of host parasitic nematode transcripts. Genomes for 38 out of the 41 different parasite species were identified from WormBase70 (Supplementary Table 1). The same quality-trimmed sequencing reads used in the first pass-through were aligned to their respective parasite species genome (where available) using BWA-mem71, run as part of the NVIDIA Clara Parabricks toolkit72 (v4.0.1-1) using default parameters. Reads that did not align to the host genome were then removed using Samtools view73, with the flags ‘-13’ for paired-end reads and ‘-4’ for single-end read datasets. These remaining reads were then reassembled and analysed using the method described previously.

Based on a combination of results from Blast and InterProScan, all identified viruses were assigned to a putative order, which was used to build sequence alignments for phylogenetics. This analysis was performed using a pre-existing alignment used for virus phylogenetics23 downloaded from Figshare supplementary data. The identified RdRP sequences from this study were then added to these alignments based on their putative virus orders. Additional RdRP sequences from viruses of flatworms that were more than 50% complete (less than 50% of identified amino acids were ‘X’) were also added for completeness, with all sequences realigned using MAFFT74,75 (v.7.505) using the E-INS-I strategy. Uninformative parts of the sequence were then removed using Trimal76 (v1.2rev59), with the gappyout strategy specified. The gappyout strategy allows the removal of ‘gap-rich’ regions (areas of the alignment where there are gaps across many or all sequences used in the alignment) while preserving ‘gap-poor’ regions (areas with few gaps across the sequences). The final alignment was then used for phylogenetic analysis using IQTree77 (v.1.6.1) at 1,000 non-parametric bootstraps. ModelFinder78 was also used to identify the best evolutionary models (Supplementary Fig. 2 caption). Phylogenetic trees were then visualized and annotated using the Interactive Tree of Life79. The full phylogenetic tree can be found in Supplementary Fig. 2, with alignment information and Newick formatted trees found on Figshare (Data Availability statement). In addition to these larger trees built for the different virus families, more ‘focused’ trees were constructed for BMRV1 and OVRV1 specifically. In these cases, branches of interest were identified from the larger overall trees, and the original amino acid sequences downloaded and reprocessed following the above method for the generation of clade-specific phylogenetic trees. These ‘focused’ trees were visualized in Figtree v1.4.4 (ref. 80) and annotated using iTOL79.

In addition to the phylogenetic trees for the different viruses, non-parametric bootstrap trees for a group of totiviruses, as well as their Trichinella hosts, were constructed to investigate their co-evolutionary relationship with their parasite host. To construct the phylogenetic tree of the Trichinellidae parasite hosts, single-copy orthologue genes were identified from the parasite proteome (downloaded from WormBase ParaSite) using the program OrthoFinder81 (v2.5.1). Protein sequences from these genes were aligned individually, before the aligned sequences were concatenated together. This was then used to build a phylogenetic tree at 1,000 non-parametric bootstraps, using the evolutionary model JTT + F + I + G4 (as defined by ModelFinder78). A phylogenetic tree of the totiviruses was created by aligning the RdRP amino acid sequence, using 1,000 non-parametric bootstraps and the evolutionary model JTT + F + G4. These two trees were then manually compared to construct a tanglegram to evaluate potential virus–parasite co-evolution (Supplementary Fig. 1).

Frequency of virus prevalence within datasets was performed by first concatenating the virus-like sequences to the end of the nematode host genome (where available, nematode genome accessions used are available in Supplementary Table 1), thus acting as a false chromosome. This concatenated genome was used as a ‘reference genome’, allowing competitive alignment and ensuring that sequencing reads from the nematode host did not align to the virus. Alignment was performed using the same quality-trimmed sequencing dataset used as input for assembly (Supplementary Table 1), and BWA-mem71, run as part of the NVIDIA Clara Parabricks toolkit72 (v4.0.1-1). This is a computational workflow designed to take advantage of graphical processing units to increase the speed of alignment and variant calling exercises compared with standard workflows82. Resultant bam files were then processed using the program MosDepth83 (v0.3.2) to calculate average read depths in 50-nucleotide ‘bins’ across the virus ‘chromosome’ specifically (options ‘–chrom’ and ‘–by’). The tabular output ‘regions.bed’ file was then used in R, and the mean read depth was calculated across the virus genome. A virus was considered present in a dataset based on an arbitrary mean read depth of 0.5 (Supplementary Table 1) to account for chance matches, as well as the possibility of low viral titres in any one sample.

Additional visualization of read alignments for BMRV1 and OVRV1 identified in B. malayi and O. volvulus was further performed by graphing the outputs from MosDepth83 (v0.3.2) into ggplot2 (ref. 68; v3.4.2).

Visualization of read alignments to BMRV1 and OVRV1 identified in B. malayi and O. volvulus, respectively, was performed with (PRJNA772674 (ref. 84) for BMRV1, PRJEB2965 (ref. 32) for OVRV1). The virus-like sequences were first concatenated to the end of the nematode host genomes to act as a false chromosome (nematode genome accessions used available in Supplementary Table 1).

O. volvulus sample collection, RNA extraction and sequencing

Nodulectomies were performed on identified nodules of persons with an O. volvulus infection from both Ghana and Cameroon. All nodulectomies were performed under aseptic conditions and local anaesthetics by trained medical personnel. Once the nodules were recovered, the human tissue was digested using 0.5 mg ml−1 collagenase in RPMI for 9 h at 37 °C, shaking at 90 rpm. All worms were either pooled by gender or sequenced separately (Supplementary Table 3). The worms collected from Cameroon were washed three times and then placed in RNALater to be stored at −80 °C. Worms collected in Ghana were handled in a similar manner but placed in alcohol to be stored at −20 °C to be later transferred to −80 °C in saline for long-term storage and shipment to Belgium.

All samples were processed according to the NetoVIR protocol85 to enrich them for viral-like particles. The original protocol was slightly modified by adjusting the initial sample preparation for nematode homogenization with the Precellys homogenizer in three intervals of 30 s at 4,500 rpm with the addition of Precellys 2.8 mm zirconium oxide beads (Bertin Technologies). Paired-end sequencing of samples was performed on a NovaSeq 6,000 system for 300 cycles (2 × 150 bp) at the Nucleomics Core sequencing facility (VIB).

Confirmation of OVRV1 in adult O. volvulus nematodes

We performed additional sequencing of O. volvulus parasites collected from Ghana and Cameroon (available under PRJEB67302). This sequencing yielded between 24.40 and 53.28 million sequences per sample (Supplementary Table 3), which were trimmed for quality and adapters with trimmomatic86 (v0.39). Good-quality reads were subsequently mapped to the OVRV1 genome specifically with Bowtie2 (ref. 87) (v2.4.2) on default settings. Consensus sequences were extracted from the resulting BAM files using samtools73 (v1.15), bcftools88 (v1.8) and bedtools89 (v2.29.2). Finally, the consensus sequences were manually curated in Geneious Prime (v2023.1.2). Visualization of aligned reads was performed as previously described.

Nucleic acid extraction from B. malayi and Onchocerca nematodes

For production of RNA, mf, L3, and adult male and female B. malayi were collected via the life cycle maintained at LSTM, with additional adults obtained from the FR3. O. volvulus and O. ochengi adults stored in liquid nitrogen were also used from previous studies40,41. B. malayi material generated at LSTM was placed directly in TriZol for storage at −20 °C, while adults from FR3 were placed in RNALater before shipping to LSTM. The mf were divided into 10 aliquots of 30,000 (used for RT-qPCR) or 4 lots of 50,000 (used for PCR) and placed in Bertin Instrument tubes (model VK05-2ml) suspended in a total volume of 1 ml of TriZol. Individual adults and batches of L3 (50 per batch) were transferred from their storage tubes into 1 ml of TriZol in Bertin Instrument tubes (model SK38-2ml). Archived O. volvulus and O. ochengi parasite material were removed from liquid nitrogen, allowed to defrost and then transferred into 1 ml of TriZol in Bertin Instrument tubes (model SK38-2ml). Batches of material were homogenized at 6,000 rpm (Minilys, Bertin Instruments) for 3 × 60 s, cooling on ice for 60 s in between. The material was then left overnight at −80 °C, and the next day subjected to heat-shock at 75 °C for 1 min to lyse any remaining material before being placed on ice. Adult samples were further homogenized for one additional round. The resultant extracts were then removed from the tubes and placed into 2 ml microcentrifuge tubes, which were then spun to pellet any cellular debris. The supernatant was then transferred into a fresh 2 ml microcentrifuge tube, to which an equal volume (~700 µl) of 100% ethanol was added. From here, the Zymo Direct-zol RNA Miniprep kit (catalogue number R2062) was used following the manufacturer’s protocol, but excluding on-column DNAse treatment, with RNA eluted in the final step using 25 µl of supplied DNAse- and RNAse-free water.

We generated additional gDNA for B. malayi, O. volvulus and O. ochengi. gDNA material for B. malayi was obtained from 10,000 mf using Qiagen’s QIAamp DNA kit (catalogue number 51306), following the manufacturer’s instructions for isolation of genomic DNA from tissues. gDNA material for Onchocerca parasites was also obtained from previously mentioned archived samples40,41, using the same DNA extraction kit and method.

Concentration of RNA was then quantified using a Nanodrop spectrophotometer (IMPLEN NanoPhotometer NP80) with two measurements of 1 µl volumes. A third measurement was taken if there was a discrepancy in detected concentration of more than 2 ng µl−1. For the RNA extracts, the remaining volume of nucleic acids was then DNAse-treated in-solution using Lucigen’s Baseline-ZERO DNase (catalogue number DB0715K) following the manufacturer’s protocol, scaled-up to the correct volume. Up to 9.5 µl of the resulting RNA was used for the synthesis of cDNA, using Jena Bioscience’s SCRIPT cDNA synthesis kit (catalogue number PCR-511L), using random primers and following the manufacturer’s instructions.

Small RNA sequencing and analysis

RNA was extracted from pools of BMRV1-positive B. malayi female (three pools of three) and male (three pools of eight) adults, and small RNA sequencing was undertaken at Beijing Genomics Institute using the DNBSEQ unique molecular identifier (UMI) small RNA sequencing technology platform. Briefly, small RNA of read lengths 18–30 nt were first selected via polyacrylamide gel electrophoresis (PAGE), and a 5′-adenylated and 3′-blocked adaptor was ligated to the 3′ end. A UMI-labelled primer was then added, followed by the ligation of an adaptor to the 5′ end. First-strand synthesis was performed with the UMI-labelled primer, followed by second-strand synthesis. Fragment selection was carried out to isolate fragments of the desired size. Finally, the double-stranded PCR products were circularized by the splint oligo sequence, forming single-strand circular DNA sequenced in a single-end 50 bp format. The six libraries were sequenced, with final clean tags of all libraries between 25.18 and 25.37 million reads. Basecalled fastq files were quality and adapter trimmed using SOAPnuke90 with parameters ‘n 0.001 -l 13 -q 0.1–highA 1–minReadLen 15–maxReadLen 44–ada_trim’. Trimmed reads were then mapped to the BMRV1 genome using Bowtie2 (v2.4.5)87, with local mapping flag (–local). The trimming and mapping statistics summary is available in Supplementary Table 4. The histogram of mapped read lengths and first base pair bias was generated using bash scripts available from github.com/rhparry/viral_sRNA_tools/, specifically 3_bam_sRNA_histogram.sh, which uses samtools (v1.16.1). Output BAM files were filtered to include only the 23 nt reads, and coverage for each position of the BMRV1 genome was extracted using bedtools genome coverage tool (v.2.27.1)89 and visualized using GraphPad Prism (v10.0.2).

PCR for visualization of virus presence or absence

Nucleic acid material generated as previously described (both RNA and gDNA) were used for confirming the presence of RNA virus genomes. Four primer sets for BMRV1 were designed using the NCBI PrimerBlast program, screening against B. malayi material from FR3 and additional microfilariae material generated at LSTM. The four sets were chosen based on their location in the virus genome, with two sets overlapping the gene junctions between the polyprotein and capsid protein of BMRV1, and the remaining two overlapping the polyprotein’s methyltransferase and RdRP domains. An additional four primer sets for OVRV1 were designed using the same tool, screening against O. volvulus. The four sets targeted different regions of the virus genome, with two overlapping with the L-protein of OVRV1, and the remaining two overlapping with the virus’ N-protein and phosphoprotein genes. The analysis was done using cDNA as well as gDNA templates generated as described earlier. PCR was performed using Jena Bioscience’s Taq Core Kit (catalogue number PCR-214S) following the manufacturer’s protocol, with all samples using 10 µl input cDNA and gDNA into 20 µl volume assays. Cycling conditions were as follows: 2 min at 95 °C before being subjected to 35 cycles of 20 s at 95 °C, 20 s at 60 °C and 40 s at 72 °C. The amplicons were run on a 2% Tris–borate EDTA agarose gel with 1× GelRed (Biotium) and visualized using a UV transilluminator (Syngene). Primer sequences are available in Supplementary Table 5.

qPCR for analysis of virus genome copies per life cycle stage

Primer sequences for qPCR analysis of virus genome copy number are described in Supplementary Table 5, with the primer efficiency of amplification validated to 100% (±10%) using five serial dilutions of cDNA material, replicated on at least two separate experiments. All qPCRs were done using a total volume of 20 μl, with 2 μl of cDNA from each sample, forward and reverse primer concentrations of 0.5 μM and 10 μl of QuantiTect SYBR Green PCR kit master mix (Qiagen). All reactions were run in duplicate and first heated for 2 min at 50 °C, followed by 15 min at 95 °C, before being subjected to 40 cycles of 30 s at 94 °C, 30 s at 60 °C and 30 s at 72 °C. A melt-curve analysis was then performed as the final step, from temperatures 55 °C to 95 °C.

A stock of template cDNA for the primers was then generated by performing PCR on extracted cDNA, using Jena Bioscience’s Taq Core Kit (catalogue number PCR-214S). This was then run on a 2% Tris–borate EDTA agarose gel with 1× GelRed (Biotium) and visualized using a blue light transilluminator. The resultant amplification bands were then cut out, nucleic acids purified using Zymo’s Gel DNA Recovery Kit (catalogue number D4007) and the concentration of eluted cDNA quantified by Nanodrop spectrophotometery (IMPLEN NanoPhotometer NP80). Based on the weight of recovered cDNA and known amplicon size from primer design, copy numbers were then deduced using the following formula:

$$}\;}\;}=\frac}\;}\;}\;}\times 6.0221)\times ^}}\;}\;}\times 660)\times ^}$$

This template was then diluted to a concentration of 107 copies per µl and used as starting material for serial dilutions to create a standard curve for virus genome quantification. All samples were run using the Quantstudio 5 system, following the qPCR protocol as described earlier. The standard curve was used for exact quantification of virus genome copy numbers in each individual sample. This was then first normalized based on the input amount of cDNA used for the reaction, determined by the concentration of extracted RNA. A second normalization step was then performed, by dividing this copy number concentration by the number of nematodes used for the RNA extraction process (that is, 30,000 for mf, 50 for L3, individuals for adult males and females).

FISH

Probes for FISH were designed using the OligoMiner online application91, with modification to the following settings: minimum length of 20 bp, minimum GC content of 40%, all instances of salt (Na+) concentrations at 750 mM, linear discriminant analysis temperature model at 37 °C and secondary structure prediction at 37 °C. Probes were designed to target BMRV1 in this manner and had an ATTO-647 dye attached to the 5′ end (IDT). Probes targeting the Wolbachia endosymbiont were used as a technical positive control, with sequences taken from previously published literature92 and an ATTO-488 dye attached to the 5′ end. An additional probe targeting the Flock House Virus93 using an ATTO-647 dye (IDT) was also included to act as a non-binding negative control. All probe sequences used as part of this work are listed in Supplementary Table 5.

Adults or mf of B. malayi nematodes were either shipped from FR3 (6-month-old adults) or collected from jirds maintained at LSTM (mf or >12-month-old adults), and stored at −80 °C until fixation and preparation. Nematodes were first fixed in 70% ethanol and 1× PBS and left overnight at 4 °C, with a second fixation step in 4% paraformaldehyde and 1× PBS for 15 min, before two washes in 1× PBS. They were then permeabilized using 10 µg ml−1 of pepsin solution for 10 min at 37 °C. For mf, each change of supernatant and buffer was preceded by centrifugation at 3,000 × g for 10 min to pellet the mf. Additionally, permeabilization of mf was performed using 5 µg ml−1 of pepsin for 30 min. All samples were then incubated at 37 °C overnight with 100 µl hybridization buffer (50% formamide, 5× saline sodium citrate (SSC), 200 g l−1 dextran sulfate, 250 mg l−1 poly(A), 250 mg l−1 salmon sperm DNA, 250 mg l−1 tRNA, 0.1 M dithiothreitol (DTT), 0.5× Denhardt’s solution) and 2 µl of each target probe (combination of Wolbachia, BMRV1 or none depending on the experiment). After overnight hybridization, the supernatant was removed and the nematodes were washed in 100 µl of hybridization buffer without any probes for 15 min at 37 °C, followed by two washes of 1× SSC buffer with 10 mM DTT, another two washes with 0.1× SSC buffer and 10 mM DTT, and a final wash with 1× PBS. For mf, only one wash with each buffer was performed instead of two. All wash steps were carried out at room temperature using a microcentrifuge tube rotator for 5 min each, removing the supernatant after each wash. After the washes, the nematodes were mounted on slides using Vectashield Mounting Medium with DAPI (Vector Labs). The nematodes were then visualized with a confocal laser-scanning microscope (Zeiss LSM 880, software Zeiss ZenBLUE Edition (v3.7.97.02000)) and images captured with ×40 or ×60 oil objectives.

Expression of recombinant virus proteins

The full-length BMRV1 capsid protein cDNA was obtained from B. malayi adult female cDNA (SAW96MLW-BmAF) in the λ-Uni-ZAP XR vector library94 (courtesy of S. Williams, Smith College). This was amplified from the bacteriophage DNA by PCR using forward and reverse primers corresponding to the ends of the open reading frame sequence. All primer sequences used are detailed in Supplementary Table 5.

The C-terminus of the OVRV1 glycoprotein was obtained via PCR of an O. volvulus L3 cDNA library constructed in λ-Uni-ZAP XR94 (Stratagene; Smith College) using forward and reverse primers as described in Supplementary Table 5. This resulted in 624 bp nucleotides that were expressed as a protein of 24 kDa with a iso-electric point (PI) of 6.58.

The amplified PCR products were subcloned into the TA vector, pCR2.1-TOPO (Invitrogen). The insert from pCR2.1-TOPO was re-amplified and subcloned into the expression vector, pJC40 (ref. 95). The constructs were then transformed into E. coli DE3 cells, induced with isopropyl-1-thio-d-glucopyranosides (IPTG) and expressed as a fusion with an N-terminal poly-histidine tag. Results were checked by nucleotide sequencing of constructs (Source Bioscience) and protein sequencing (Functional Genomics and Proteomics Laboratories, University of Birmingham) to confirm that the final products were correct. Recombinant proteins were then purified by affinity chromatography on a nickel column (Probond resin, Invitrogen) according to the manufacturer’s instructions.

Production of polyclonal antibodies

Antibodies to recombinant BMRV1 capsid proteins and to C-terminal OVRV1 glycoprotein were produced in rabbits by immunization with recombinant antigen as described above. In brief, the products were emulsified in Freund’s complete adjuvant (for the first inoculation) and Freund’s incomplete adjuvant (subsequent inoculations). The process was carried out by Alta Bioscience, with 200 µg of protein antigen administered by subcutaneous injection on three occasions, at an interval of 3 weeks. Rabbits were bled 8 days after boosting. No experimental work was performed using rabbits.

Gel electrophoresis and immunoblotting

Proteins were extracted from parasites by boiling for 10 min in electrophoresis sample buffer (3% (w/v) SDS; 62 mM Tris–HCl, pH 6.8; 15% (v/v) glycerol) containing 5% 2-mercaptoethanol. The lysates were homogenized and vortexed before boiling. Insoluble material was removed by centrifugation for 5 min at 16,000 × g. The parasite extracts from different life cycle stages were then fractionated on either 12% NuPAGE Bis–Tris protein gels using morpholinoethanesulphonic acid (MES) buffer or NuPAGE 3% to 8% protein gel using Tris–acetate buffer (Invitrogen). The protein markers used included phosphorylase b (94 kDa), bovine serum albumin (68 kDa), ovalbumin (43 kDa), carbonic anhydrase (30 kDa), soybean trypsin inhibitor (20 kDa) and α-lactalbumin (14 kDa). Separated proteins were electrophoretically transferred to nitrocellulose, and the membranes were blocked by incubation in 5% skim milk in TST buffer (0.01 M Tris, pH 8.5; 0.15 M sodium chloride; and 0.1% Tween) for 1 h at room temperature. Blots were incubated with rabbit antisera to BMRV1 capsid protein or OVRV1 glycoprotein at 1:2,000 dilution in 5% milk overnight at 4 °C. Control blots were probed with preimmune rabbit sera. The detection was performed using IRDye anti-rabbit specific antibodies 1:15,000 in 5% milk for 1 h at room temperature (IRDye 680RD goat anti-rabbit IgG (Li-COR Biosciences)). The results were visualized in 700 nm fluorescence channels by Odyssey FC (LI-COR Biosciences).

To detect jird immune responses against virus surface proteins, we used freshly collected jird blood from the existing life cycle or naive, uninfected jirds (for the detection of BMRV1 capsid protein), or archived jird serum40 (for OVRV1 glycoproteins). For BMRV1, naive and B. malayi-infected adult gerbils were euthanized by CO2 inhalation. Blood was collected from the animal heart by cardiac puncture and centrifuged at 13,000 g for 10 min at 4 °C to separate the blood and collect serum (repeated if necessary). Recombinant virus proteins were purified as described earlier and transferred via gel electrophoresis and immunoblotting. After transfer onto a nitrocellulose membrane, each protein lane was cut, labelled and blocked, and then placed into a 5 ml tube for reaction with individual gerbil serum overnight at 4 °C with agitation. The membrane was then washed five times with TST buffer, then incubated with rabbit anti-Mongolian gerbil IgG (H + L), (catalogue number BS-0403R, Bioss) for 1 h at room temperature. The membrane was washed 5 times with TST buffer, then incubated with IRDye anti-rabbit-specific antibodies 1:10,000 in 5% milk for 1 h at room temperature (IRDye 680RD goat anti-rabbit IgG (Li-COR Biosciences, 926-68071). After incubation and washing in TST, the results were visualized in 700 nm fluorescence channels by an Odyssey FC imager (LI-COR Biosciences).

ELISA for OVRV1 glycoprotein using human sera

Human sera were obtained from the FR3, originally derived from the EMCF43 and the CDC43. Samples originated from endemic regions within Cameroon (n = 200), Togo (n = 67), Nigeria (n = 54), Uganda (n = 88) and Ecuador (n = 54). Sampled individuals ranged in age from 3 to 95; the median age per country was 51, 35, 39, 33 and 33, respectively, wi