Flying Syringes for Emerging Enzootic Virus Screening: Proof of Concept for the Development of Noninvasive Xenosurveillance Tools Based on Tsetse Flies

Pathogen transfers between wild and domestic animals and between animals and humans are increasing. Their dramatic consequences for public and veterinary health as well as for conservation call for innovative and user-friendly methods for pathogen surveillance in wildlife. Xenosurveillance, a method based on the use of invertebrates (e


Introduction
Emerging and re-emerging human infectious diseases have increased in recent years mostly due to the growth of human activities that promote contacts with novel sources of pathogens and favor their spread worldwide [1]. Sixty percent of diseases emerging in humans are zoonoses, and wildlife plays a key role by providing a zoonotic pool from which previously unknown pathogens may emerge [2][3][4][5]. Many emerging threats to human health are viruses, such as human immunodeficiency virus (HIV) [6], avian influenza viruses [7], Ebola virus [8], Chikungunya virus [9,10], Yellow fever virus [11], and more recently Zika virus [10], severe acute respiratory syndrome-related coronavirus SARS-CoV-1 and SARS-CoV-2 viruses [12] and Middle East respiratory syndrome-related coronavirus [13]. Most of these viruses circulate primarily within enzootic cycles that involve wild animals. However, occasionally, they spill over from their primary wild hosts to alternative new vertebrates, including wild and domestic animals that become sources of human infections. The recent SARS-CoV-2 crisis highlighted the urgent need to address the issue of viral emergences according to the One Health concept in order to improve the surveillance of enzootic viruses in their natural hosts before their spillover to the anthropic compartment and the occurrence of devastating epizootic or epidemic waves [14].
Besides human health, emerging viruses are also an important veterinary concern with major effects on food production stability and wild species survival. When affecting livestock, emerging viruses can cause dramatic economic losses, and threaten food security and trade. For example, the African swine fever virus (ASFV), which recently spread in China and neighboring Asian countries, caused the death or culling of more than one million pigs and seriously threatened the swine industry in Asia [15]. In addition to their spreading to livestock, some emerging viruses can propagate to wildlife with important conservation issues, especially for threatened wild species. For instance, gorillas have been infected by Ebola virus in Central Africa [16], wild ruminants by the peste des petits ruminants virus (PPRV) in Asia and Middle East [17], and lions by the canine distemper virus that decimated their populations in the Serengeti National Park in Tanzania [18].
The risk of emerging zoonotic infectious diseases is particularly elevated in tropical regions that are currently undergoing land-use changes and where the wild mammal diversity is high [19]. Surveillance and virus discovery efforts need to be directed toward mammals in these biodiversity hotspot areas. Combined with comprehensive field studies, more specific knowledge will help to refine and adapt surveillance strategies to better monitor diseases at the wildlife/human/livestock interface.
Therefore, novel and easy-to-implement surveillance strategies to detect many different pathogens that circulate among wild mammals are needed. Xenosurveillance or invertebrate DNA (iDNA), in which invertebrates are used to sample wild mammal tissues for pathogen screening, is one of these approaches. Xenosurveillance of viruses was first described by Grubaugh et al. [20] who assessed whether human viruses could be detected using mosquito blood meals. They found that HIV was detectable in the gut of mosquitoes up to 24 hr after the meal. However, their mosquito-based technique would be not suitable in wild environments where bloodengorged mosquitoes are difficult to collect [21]. Moreover, as their host preference is often restricted to a limited number of vertebrate species, different mosquito species would be needed to cover all vertebrate wildlife. Alternatively, previous studies in Gabon [22] and Tanzania [23] have shown the interest of tsetse flies to obtain the blood of the vertebrates they bite (mostly mammals, but also birds and reptiles) for the screening of bloodborne pathogens, particularly in Sub-Saharan Africa where Glossinidae are present. As, tsetse flies take large blood meals (ranging from 20 μL up to 100 μL for the largest species) for their activity and reproduction [24], they are useful for blood sampling without affecting wildlife. Nevertheless, it is not known whether blood meals from tsetse flies can be used to detect viruses (DNA and RNA viruses) and whether the time between blood ingestion and sample handling could affect the virus detection probability. On the other hand, these points have been already experimentally tested in other hematophagous invertebrates, for instance mosquitoes [25].
Therefore, in the present study, using an experimental approach, we assessed the relevance of using hematophagous flies as "flying syringes" for the detection/surveillance of RNA and DNA viruses of medical or veterinary interest that circulate among wild animals. Briefly, we allowed tsetse flies to feed on controlled mixtures of blood and viruses and then used molecular approaches to determine whether the virus genomes could be detected in the fly abdomens (i.e., in the blood meal) at different digestion stages. We also assessed whether two different blood meal preservation methods affected the probability of virus detection.

Tsetse Strain.
A tsetse fly strain belonging to Glossina fuscipes fuscipes (colonized in the insectarium for more than 294 generations) was used for experimental infections. The strain was provided by the International Atomic Energy Agency, Insect Pest Control Laboratory, Seibersdorf, Austria as batches of pupae that were then reared to adult stages in the INTERTRYP insectarium (Baillarguet, France) in controlled environmental conditions (25°C, 80% room hygrometry, and 12 hr light: 12 hr dark cycle). Upon reception, pupae were placed in petri dishes disposed in big cages. When they emerged, tsetse flies were sorted by sex and amputated of a single wing to avoid any risk of escape. Teneral adults (2-5 days postemergence) were used for experimental infectious blood feeding.

Virus Selection.
To assess for how long viral nucleic acids could be detected in tsetse fly guts after infectious blood meals, G. f. fuscipes males and females were given a blood meal that contained one of the following six emergent viruses, including five arboviruses (Table 1). They were (i) viruses of veterinary interest: ASFV (Asfaviridae/Asfivirus), PPRV (Paramyxoviridae/Morbillivirus), and Bluetongue virus (BTV; Reoviridae/Orbivirus) and (ii) zoonotic viruses of major concern for human health: Chikungunya virus (CHIKV; Alphaviridae/Alphavirus) and Dengue virus and Zika virus (DENV and ZIKV; Flaviviridae/ Flavivirus). Their choice was guided by several considerations: (1) availability of usable viral strains, specific agreement, and infrastructure allowing their handling; (2) diversity of genome types (ssRNA/dsRNA/dsDNA); (3) viruses corresponding to several families/genera; and (4) viruses representing different modes of natural transmission. Moreover, all selected viruses are current human or veterinary health concerns and represent viruses that could be theoretically surveyed in the wild using flying syringes because they all induce viremia and can be detected in the peripheral blood of infected animals. This includes also PPRV [36,40] although it induces a much lower viremia than arboviruses. The Transboundary and Emerging Diseases 3 characteristics of viruses and strains are summarized in Table 1 and Table S1.

Blood feeding of Tsetse Flies.
To determine the limits of pathogen detection following blood meal digestion, tsetse flies were fed with a blood/virus mixture. All blood feedings were done in BSL3 laboratories (Vectopôle of IRD, Montpellier, France for DENV, CHIKV and ZIKV; CIRAD, Baillarguet for ASFV, PPRV and BTV). For each virus, two batches containing between 40 and 90 teneral flies (one batch per sex) were fed the blood-virus mixture (2 mL of washed rabbit erythrocytes and 1 mL of viral suspension). The blood meal was supplemented with 0.5 M ATP as phagostimulant. The final viral titers in the blood meals were adapted for each tested virus by taking into account the usually reported titers in their natural hosts (Table 1). Blood/pathogen meals were provided using a Hemotek membrane feeding system. Tsetse flies were allowed to feed for 10 min through a piece of pork intestine that covered the Hemotek feeder base maintained at 37°C. After the blood meal, tsetse flies were anesthetized on ice for 5-10 min and then engorged specimens were retrieved and transferred to cardboard containers where they were maintained with 10% sucrose in controlled conditions in a climatic chamber For RNA viruses, the detection and quantification were performed by one-step reverse transcription-quantitative polymerase chain reaction (RT-qPCR). The RT-qPCR assays were modified (i.e., by using a one-step RT-qPCR kit instead of two steps) from several previously published protocols to generate 68-209 bp amplicons located in the conserved region (see Table S2 for details). Assays were performed using the SuperScript III Platinum One-Step Quantitative Kit (Invitrogen): 1x reaction mix, 10 pmol PCR primers and probes, and 10 µL RNA in a final volume of 30 µL. Amplification was performed on a LightCycler 96 real-time thermocycler (Roche Diagnostics, Meylan, France) (see Table S3 for details). For ASFV (DNA virus), qPCR was performed using the SensiFast Lo-ROX Genotyping Kit (Bioline) with 1x reaction mix, 8 pmol primers, 2 pmol probe, and 2 µL DNA in a final volume of 20 µL. Negative controls (i.e., by replacing RNA or DNA with water instead) were used in each run. In addition we tested, a set of flies (five per time point) fed with blood without virus to search for nonspecific amplification at the different time points postfeeding.

Comparing FTA Cards and RNAlater for Nucleic Acid
Storage. To preserve the nucleic acids present in the blood meals of tsetse flies collected in the field, the RNAlater solution (Sigma-Aldrich) and Whatman FTA cards (WB120305, Sigma-Aldrich) were compared. Twenty-four abdomens of tsetse flies engorged with rabbit blood and DENV were isolated at 0, 6, 24, 48, 72, and 144 hr postingestion (four flies per time point), individually ground in 250 μL 1X DPBS, and centrifuged at 15,000 rpm for 1 min. For each abdomen, an equal quantity of supernatant (65 μL) was transferred to 100 μL RNAlater and on a 25 mm diameter FTA card sample area. RNAlater samples were centrifuged at 15,000 rpm for 1 min, the supernatant removed, and replaced by 100 μL 1X DPBS. RNA was extracted from 50 μL of this solution and from half of the FTA card sample area using the NucleoSpin RNA Kit, followed by DENV qPCR analysis as described above. The efficiency of the two preservation methods was compared by calculating the positivity rates and cycle threshold (Ct) values.

Statistical
Analyses. The linear model for each virus was adjusted to compare the Ct values between T0 and each time point after virus ingestion. The sensibility of virus detection was estimated using the brglm function with binomial distribution to estimate the error. Time points were considered as factors because they were represented by different batches of tsetse flies. The Student's t-test was used to determine the storage method effect on Ct. All statistical analyses were performed with the R software (version 1.2.13) [41].

Kinetics of Virus Detection in Blood Meals.
Overall, we tried to experimentally feed 1,613 tsetse flies with a bloodvirus mixture. In total, 852 flies were fully engorged with the infectious mixture, and the other 759 tsetse flies did not eat and were not considered for this analysis. The engorgement rate (Table 2) ranged from 43.7% to 68.2% in function of the blood-virus mixture. The mortality rates at the different postingestion time points were comparable for the different blood-virus mixtures, except for the PPRV mixture that led to higher mortality especially at 48 hr (Table 2).
When analyzing tsetse flies fed with blood without virus, we detected a single nonspecific amplification (1/40) for the BTV RT-qPCR system with an associated Ct value of 42.65. We, therefore, decided not to consider positive samples with Ct values higher than 40 for this system, and we excluded from the analysis three samples that showed BTV amplifications (Ct values of 40.99, 41.11, and 42.6) after 144 hr postfeeding. The virus positivity by RT-qPCR was higher than 50% until 72 hr postingestion for all viruses, and remained high at 144 hr postingestion, except for ASFV for which the positivity rate decreased to 5% (Table 2 and Figure 1). For all tested viruses, the viral genome remained detectable in the abdomen up to 6 days (144 hr postengorgement), although the number of negative results increased significantly with

Discussion
Our objective was to evaluate and validate the use of tsetse flies as "flying syringes" for the detection and monitoring of viruses that circulate in wildlife. Specifically, we wanted to define the temporal window of virus detection after their ingestion during a blood meal.

Virus Detection Capacity.
Our blood-feeding experiments showed that viruses can be detected with a rather high sensitivity in blood meals of G. f. fuscipes. Nevertheless, the positivity rates decreased over time, from 67% to 100% just after the blood meal ingestion to 5%-88% 6 days later. This decrease in positivity rate was particularly high between day 3 and day 6 postingestion, while the rate remained roughly the same between day 0 and day 3. For all viruses (except ASFV for which the positivity rate was close to 0 after 6 days), the positivity rate remained rather high after 6 days of blood digestion (42%-88%), suggesting that tsetse flies can be effectively used to detect viruses even if the blood meal has been largely digested. For tsetse fly, the time required for a complete digestion of the blood meal content varies between 48 and 120 hr (e.g., host DNA remains detectable in the tsetse fly gut up to 7 days after blood ingestion, ravel, personal observation), while the rate of digestion has been shown to be modulated by several factors, including ambient temperature, tsetse fly activity, and experimental versus field conditions [42,43]. Blood meal analysis of mosquitoes experimentally infected with different pathogens [44] already showed that the Ct values increase with time, probably due to the degradation of viral particles (i.e., viral genome) or of nucleic acids (DNA and RNA) during the digestion process.
In mosquitoes, pathogens could be detected 24 hr after blood digestion, whereas in tsetse flies this was possible at least up to day 6 due to a longer digestion process. This indicate that using tsetse flies as "flying syringes" increases the probability of virus detection and reinforces their usefulness for virus surveillance. These observations have practical implications for field studies. Indeed, viruses could be detected in almost all engorged tsetse flies and up to 6 days after blood feeding. This means that all collected flies are suitable for virus screening and not only those with a fresh blood meal, as previously done [22]. All abdomens can be stored (e.g., using FTA cards or RNAlater, as tested here) to increase the probability of virus detection. This should largely reduce the field efforts to sort flies and to prepare and store specimens (the blood meal does not need to be isolated after dissection).
We found that the positivity rate varied among the six tested viruses. ASFV (the only DNA virus included) displayed the lowest positivity rate, with a mean detection probability of only 53.5%. Similarly, PPRV, which was the only nonarbovirus virus in our panel [45], was the second less well-detected virus (64.5%). On the other hand, the mean positivity rates for remaining arboviruses were higher with mean detection probabilities of 81.7%, 87.2%, 90.3%, and 95.3%, respectively, for ZIKV (Flavivirus), BTV (Orbivirus), DENV (Flavivirus), and CHIKV (Alphavirus). Differences in starting titers or in the molecular method sensitivity could explain the observed detection rate differences.

Choice of Blood Meal Storage Methods in the Field.
Although we did not test the effect of storage temperature and time, our results suggest that the two tested storage methods (FTA cards and RNAlater) are comparable. Nevertheless, FTA cards are more expensive than the RNAlater solution (1.20 Euros vs. 0.07 Euros/sample). Moreover, the RNAlater solution can even be prepared in the laboratory, for a lower cost. In addition, the use of FTA cards is impractical because it is necessary to keep part of the sample area and to extract the other part, and this require sterile cutting. From a high throughput perspective (i.e., high number of tsetse flies collected and analyzed), the use of one FTA card sample area for one fly abdomen could be restricting, whereas 10-100 specimens can be pooled together in the RNAlater solution. Other traditional transport media also could be used to store blood meals, such as viral transport medium (VTM), at even lower cost. Regardless of the storage methods, further work using field conditions will be useful to determine the best way to maintain the sample integrity under varying environmental parameters (including storage time, temperature, and humidity).

Conclusions
Our study demonstrated that xenosurveillance methods, such as tsetse flies as flying syringes, could be used to monitor enzootic viruses that circulate in wildlife. As tsetse flies are easy to trap (little or no specific expertise required), this method will allow the simultaneous and long-term monitoring in different Sub-Saharan Africa areas. The interest of this method is increased by several studies showing that blood meals could also be used to detect circulating antibodies against specific viruses, thus broadening the information obtained from blood meals [23,47,48]. Overall, tsetse flybased xenosurveillance could be very useful to monitor the circulation of enzootic viruses in the wild and at wild/human interfaces in future One Heath programs in sub-Saharan Africa.

Data Availability
The databases related to positivity rates/Ct values and scripts used for statistical analyses with R to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
The authors declare that they have no conflicts of interest.

Authors' Contributions
CP, FP, VR, FB, BM, and PYBE conceived the study. CP, FP, DJ, and AV designed the methodology. AV and SR reared tsetse flies. DJ, LV, VM, OK, AP, PF, and DM performed the virus production/titration. DJ and AV carried out experimental tsetse infectious feeding and molecular analyses.
AV and AP did the statistical analyses. AV, DJ, FP, and CP analyzed the results and wrote the manuscript. All authors edited, commented, and accepted the final version of the manuscript. Adeline Valente and Davy Jiolle are the co-first author. Franck Prugnolle and Christophe Paupy comanaged the work.