The insufficiency of circulating miRNA and DNA as diagnostic tools or as biomarkers of treatment efficacy for Onchocerca volvulus.

Skin snip evaluation for onchocerciasis has insufficient sensitivity when skin microfilarial (mf) densities are low, such as following ivermectin treatment. Mf density is suitable for assessing microfilaricidal efficacy but only serves as an indirect indicator of macrofilaricidal activity. We assessed circulating nucleic acids from Onchocerca volvulus as an alternative to skin snips. We screened a plasma sample set of infected individuals followed up at four, 12 and 21 months after microfilaricidal (ivermectin, n = four), macrofilaricidal (doxycycline, n = nine), or combination treatment (n = five). Two parasite-derived miRNAs, cel-miR-71-5p and bma-lin-4, and O-150 repeat DNA were assessed. Highly abundant DNA repeat families identified in the O. volvulus genome were also evaluated. miRNAs were detected in two of 72 plasma samples (2.8%) and two of 47 samples (4.3%) with microfilaridermia using RT-qPCR. O-150 DNA was detected in eight (44.4%) baseline samples by qPCR and the number of positives declined post-treatment. One doxycycline-treated individual remained O-150 positive. However, only 11 (23.4%) samples with microfilaridermia were qPCR-positive. Analysis by qPCR showed novel DNA repeat families were comparatively less abundant than the O-150 repeat. Circulating parasite-derived nucleic acids are therefore insufficient as diagnostic tools or as biomarkers of treatment efficacy for O. volvulus.


Methods
Study design and participants. Plasma samples were collected as part of a randomised community-based trial registered with the current controlled trials registry (trial registry no. ISRCTN48118452) 42 . The trial experimental protocol was designed in accordance with the general ethical principles outlined in the Declaration of Helsinki. The trial was approved by ethics committees of the Tropical Medicine Research Station, Kumba and the Research Ethics Committee of The Liverpool School of Tropical Medicine. Written informed consent was obtained from all participants, with the exception of those who were illiterate, where a literate witness signed on behalf of the participant and the participant added a thumbprint.
The double-blind, randomised field trial was undertaken in six satellite villages (Bifang, Ebendi, Eka, Ngalla, Dinku and Olurunti) in the North West Province of Cameroon, starting on 1 July 2003 and finishing on 31 March 2005 42 . All enrolled individuals (adults of both sexes aged 15-60) had O. volvulus microfilaridermia >10 mf/snip and were assigned to one of three drug regimens: (1) DOXY: Doxycycline for six weeks plus dummy pill at month four.
(3) IVM: Dummy pill for six weeks and ivermectin at month four.
Plasma and skin snips were collected from participants prior to first treatment and at sequential time points (four, 12 and 21 months) after initiating the study. Two skin snip samples of approximately 1 mg were taken from the rear of the leg using a Walser skin punch. Skin snips were placed in 200 µl saline containing 2 mM EDTA and incubated overnight at room temperature. O. volvulus microfilaridermia and L. loa/Mansonella perstans microfilaraemia were assessed by microscopy at each time point.
For the purposes of this study we omitted participants with filarial co-infections, so the final sample-set included: DOXY (n = 9), DOXY + IVM (n = 5) and IVM (n = 4). The median age was 30, 40 and 42.5 years for miRNA primers and primer validation. miRNAs were reverse transcribed using the Universal cDNA Synthesis Kit II (Exiqon, Denmark). Locked Nucleic Acid-enriched miRNA-specific qPCR primers were obtained from Exiqon (Denmark).
Parasite miRNAs lin-4 and miR-71 are here referred to as bma-lin-4 (accession no. MI0013320) and cel-miR-71-5p (accession no. MIMAT000003), respectively, following the miRBase 43 (release 21) naming convention. Assay efficiencies were determined using five log 10 serial dilutions of O. ochengi cDNA, performing each dilution in triplicate. Experiments were repeated three times to calculate inter-and intra-assay coefficient of variation (CV). To determine the assay limit of detection (LOD), cel-miR-71-5p and bma-lin-4 PCR products were purified using the QIAquick PCR Purification Kit (Qiagen, UK). miRNA amplicon stocks were quantified using Qubit 3.0 Fluorometer (ThermoFisher, UK) and copy numbers were determined with Science Primer 44 . A 1:10 dilution series spanning 10 5 to 10 0 copies was prepared for both cel-miR-71-5p and bma-lin-4, with nine reactions per dilution. The 95% LOD was determined by probit regression analysis (SPSS, Version 23, IBM Corp). Assay specificity was evaluated by melt curve analysis, and by using European control plasma and 'no template control' (NTC) reactions.
miRNA RT-qPCR. The miRCURY LNA Universal RT microRNA PCR (Exiqon, Denmark) was used for miRNA RT-qPCR. For cDNA synthesis, reactions were prepared to 10 µl final volumes following the manufacturer's instructions. RNA input volume was optimised for clinical samples. Total RNA from O. ochengi was added at 10 ng per reverse transcription (RT) reaction. A 'no reverse transcriptase' (-RT control), substituting enzyme mix with nuclease-free H 2 O, was prepared for each sample. RT was conducted using the TC-4000 Thermal Cycler (Bibby Scientific Ltd, UK) with thermocycling parameters: 42 °C for 60 min, 95 °C for 5 min, and a cool down to 4 °C. qPCR was performed in 10 µl reaction volumes, consisting of: 2x ExiLENT SYBR Green Master Mix (Exiqon, Denmark), primer mix (1 µl) and cDNA template (4 µl). cDNA from the worms was diluted to the desired concentration and dilution volumes were optimised for clinical samples. The CFX384 C1000 Thermal Cycler (Bio-Rad, UK) was used for qPCR, with thermocycling parameters: 95 °C hold for 10 min, 40 cycles of 95 °C for 10 sec, and 60 °C for 1 min with a ramp rate of 1.6 °C/sec. Fluorescence was monitored during the 60 °C step, using the FAM channel. Melt curve analysis was performed between 60 and 95 °C at a ramp rate of 0.5 °C/sec. For each experiment, samples were tested in duplicate reactions and a positive control, NTC and -RT control (for each sample) was included.
Samples were considered positive if amplification occurred in fewer than 40 cycles and in both reactions. A single peak at the correct melting temperature (T m) was required for each product. The T m was determined from standard curves prepared from O. ochengi cDNA. Samples positive in one reaction were retested in triplicate and considered positive if amplification occurred in two or more reactions. www.nature.com/scientificreports www.nature.com/scientificreports/ Novel DNA targets: repeat region identification. A species-specific repeat library of O. volvulus was generated using RepeatModeler (Version 1.0.11) 45 with default parameter settings. This repeat library was used, alongside a general repeat library from RepBase (RepBase RepeatMasker edition 20170127) 46 , as a filter against the O. volvulus genome 28 using RepeatMasker (Version 4.0.8) 47 . This generated a list of repeat families and their predicted location, number of occurrences and total genome coverage.
The five most abundant repeat families and five abundant repeat families that shared one or more contigs with the O-150 repeat region were selected for further investigation. Specificity was confirmed via BLAST searches of the Onchocerca genus, L. sigmondontis and L. loa genome repositories and the M. perstans nucleotide repository.
Primers and probes for DNA-based experiments. Assay sequences are provided in Supplementary   Table S2. A TaqMan qPCR assay 19 (IDT, Belgium) was used to detect the O-150 sequence (accession no. J04659). Details of assay design are reported elsewhere 19 . Assay linearity was assessed from standard curves prepared from five 1:2 serial dilutions of O. volvulus DNA, with three replicates per dilution. Experiments were repeated three times to determine intra-and inter-assay CV. The assay LOD is reported elsewhere 19 . Assay specificity was confirmed using control plasma and NTCs.
Up to three primer sets were designed for each novel repeat family using Primer-BLAST 48  Cycling parameters consisted of: a hold at 95 °C for 15 min, 40 cycles of 94 °C for 15 sec and 57 °C for 30 sec, and extension at 72 °C for 30 sec. Fluorescence was monitored during the 72 °C step, using the FAM channel. Melt curve analysis was performed between 60 °C to 97 °C at a ramp rate of 0.2 °C/sec. Statistical analysis. Probit regression analysis was performed in SPSS (Version 23, IBM Corp). As sample sizes were unequal and small, non-parametric statistical analyses were performed in GraphPad Prism 5 49 . Levels of plasma reference targets were compared between treatment groups at each time point using Kruskal-Wallis one-way ANOVA. The consistency of endogenous control levels within treatment groups over time was assessed by Friedman's test, followed by Dunn's post hoc test (with 95% confidence intervals). Significance level was set to alpha = 0.05.

Results
Parasite miRNA RT-qPCR. The parasite miRNAs bma-lin-4 and cel-miR-71-5p were detected in RNA extracted from both adult male O. ochengi and L4 L. loa worms. The miRNA assays tested negative in European control plasma and in NTC and -RT control reactions. The efficiency of the cel-miR-71-5p qPCR assay was 99.9%, with an R 2 value of 0.979. Melt analysis detected a single peak at a T m of 69.5 °C. The bma-lin-4 qPCR assay efficiency was 97.3%, and the R 2 was 0.993. Melt analysis detected a single peak at a T m of 71.5 °C. Further details of assay validation, including the inter-and intra-assay CV, the average qPCR efficiency across experiments, the linear dynamic range and the LOD, are provided with Supplementary Fig. S1.
Detection of parasite miRNAs in clinical plasma samples. RNA and cDNA inputs for RT and qPCR, respectively, were determined empirically by evaluating several RNA input volumes and cDNA dilutions ( Supplementary Fig. S3). Plasma from 18 microfilaridermic individuals was screened for the presence of Cel-miR-71-5p was detected in one DOXY patient (patient a.5) with 2 mf/snip at month four; the average Cq and standard deviation (± SD) from three replicates was 38.02 ± 0.339. Bma-lin-4 was detected in one DOXY patient (patient a.7) with 1 mf/snip at month 12; the average Cq from three replicates (± SD) was 36.75 ± 0.212. All control reactions were negative.
The endogenous control miRNA hsa-miR-16-5p was consistently detected (Supplementary Fig. S4) and there was no significant difference in expression levels between groups at any time point. The IVM and DOXY + IVM groups also showed no significant change in expression in individuals over time. However, a significant difference was detected within the DOXY group over time (Friedman's test: P = 0.0191). Dunn's post hoc test indicated the difference was between months four and 12 (P < 0.05). Hsa-miR-16-5p, and to a lesser extent the spike-in UniSp5, had a higher average Cq value in samples at month four relative to month 12. This is likely due to a small degree of inhibition. Overall, the synthetic low abundance miRNA spike-in UniSp5 was consistently and comparably detected across all samples ( Supplementary Fig. S4).  (Table 1). In each treatment group, the proportion of O-150 positive patients was higher at baseline and at month four. Four months into the trial, the number of positive patients had declined from four to one (75% decrease) with DOXY, and three to one (66.7% decrease) with DOXY + IVM. The number of O-150 positive individuals initially increased in the IVM group from one to two (50% increase), but IVM was not provided until month four of the trial. Two individuals, one treated with DOXY + IVM and one with IVM, were positive at both baseline and month four, but negative at later time points. One individual treated with DOXY, who was also positive for miRNA bma-lin-4, remained O-150 positive over the trial duration.

Onchocerca volvulus
Of   The endogenous control GAPDH was detected in all samples, and relative levels did not differ significantly between or within treatment groups over time. The viral control spike-in PhHV-1 confirmed DNA was extracted uniformly across all samples. This data is shown in Supplementary Fig. S6.  Table S3). While the O-150 repeat region was predicted to occur 46 times, several repeat families were predicted to occur over 1,000 times.

Identification and validation of novel
We evaluated five highly abundant families (families A, B, C, D, E) and an additional five families (families F, G, H, I, J) located on contigs shared with the O-150 repeat (Supplementary Table S4). Up to three primer sets were designed for each of the ten repeat families, with primers targeting different regions where possible. The novel repeats were evaluated alongside the O-150 repeat to compare the relative abundance in O. volvulus DNA ( Table 2). The average Cq value (± SD) was 20.60 (± 0.03) for O-150, and the 10 repeat families ranged from 24.95 (± 0.08) to 34.00 (± 0.14). We planned to optimise five primer sets, but primer efficiencies ranged between 97.1-110.9%, and so optimisation was unlikely to considerably improve assay performance. Contrary to our bioinformatic predications, the novel targets were comparatively less abundant than O-150 in the O. volvulus genome.

Detection of novel DNA repeat families in clinical plasma samples. To determine whether the new O.
volvulus targets were detectable in plasma, primers for repeat families A (primer set A-III) and G (primer set G-II) were evaluated with a random selection of nine of the 18 baseline plasma samples (participants: a.2, a.6, a.7, a.9, b.1,  b.2, b.4, c.3, c.4). These two primer sets were selected as they were the best performing amongst the 25 primer sets.
Only one individual (a.7) was considered positive (Cq ± SD) for circulating O-150 (32.97 ± 0.73) and repeat family G (36.04 ± 0.28). Due to sample limitations, samples with one amplified reaction could not be retested in triplicate (data provided in Supplementary Table S5)

Discussion
This study reports the first use of RT-qPCR and qPCR to detect O. volvulus miRNAs and DNA, respectively, in plasma from individuals in an onchocerciasis-endemic community before and after macrofilaricidal or microfilaricidal treatment. We confirmed the parasite miRNAs cel-miR-71- sigmodontis (adults localised in the pleural cavity with mf in blood) 37 , but they were not identified in the serum of mice infected with Heligmosomoides polygyrus (localised in the gut lumen) 37 .
A recent study investigated circulating cel-miR-71-5p and bma-lin-4, as well as 15 other putative O. volvulus miR-NAs, using RT-qPCR 40 . This study also highlighted that levels of extracellular parasite-derived miRNAs are very low, if at all detectable. As miRNAs from blood-localised parasites appear to be more abundant in host circulation relative to parasites in other tissues 34,37 , circulating miRNAs could have originated from another parasite. We attempted to reduce the risk of L. loa /M. perstans co-infections by eliminating samples from patients infected with other mf species, although we acknowledge there could be occult infections. The study by Lagatie et al. 40 also highlighted inadequate specificity as a concern due to the detection of worm-derived miRNAs in uninfected samples, as well discrepancy in miRNA qPCR melt curves. By comparison, our study validated worm miRNA assays with relevant biological reference species, and the melt curve for the assays was consistent when tested with O. ochengi, L. loa and patient plasma cDNA.
Use of an O. volvulus-specific DNA assay improved the diagnosis of onchocerciasis cases, identifying 44.4% of individuals who were skin snip positive at baseline. The number of individuals O-150 positive then declined in all groups after first treatment. While more samples from infected participants were correctly identified by testing O-150 DNA relative to parasite miRNAs, the majority of microfilaridermic individuals had plasma samples that were negative by qPCR. Furthermore, the Cq values recorded for most of the O-150 positive samples were very high (>37). We re-tested samples and used robust controls in experiments, including a positive control, a NTC, European control plasma and plasma endogenous control, to ensure that we did not accept false positives and plasma nucleic acids were not degraded or affected by plasma-derived inhibitors. Therefore, although TaqMan and qPCR enable detection of targets with high sensitivity and specificity, the very low levels of circulating O. volvulus DNA were not sufficient to detect infection.
We attempted to identify novel and more abundant DNA targets to exploit as circulating diagnostic markers or biomarkers of treatment efficacy. The bioinformatic approach used showed similar results, in terms of overall coverage, to those identified in the original genome sequencing paper 28 , in particular the predicted small frequency of O-150 repeats in the genome. Therefore, a novel repeat region could be more suitable as a diagnostic marker. We were unable to identify a superior repeat family after qPCR analysis of O. volvulus DNA and clinical plasma DNA, but we can confirm that the true copy number of O-150 repeats is significantly higher than what is bioinformatically predicted. These predicted values did not include the possibility of collapsed contig repeats, an issue previously noted 28 , and this could affect the predicted repeat frequency of O-150 and novel repeat families. Due to these issues, there is a possibility that a different repeat family with an extremely high 'true' occurrence rate (like O-150) may exist, but the information available to this study was insufficient to identify it.
We have demonstrated that O. volvulus nucleic acids are variably detectable at low to undetectable concentrations in host plasma. Circulating nucleic acids from O. volvulus are therefore insufficient as diagnostic markers for onchocerciasis infection or as biomarkers of treatment efficacy. Among the nucleic acids evaluated the O-150 DNA sequence showed the greatest frequency, although well below the sensitivity of parasitological diagnosis. While we did not identify a more abundant novel DNA target from O. volvulus, we acknowledge the possibility that a different highly abundant DNA repeat family may still exist.

Data availability
The datasets supporting the conclusions of this article are included within the article and the Supplementary Information.