Susceptibility of anurans, lizards, and fish to infection with Dracunculus species larvae and implications for their roles as paratenic hosts

Dracunculus spp. are parasitic nematodes that infect numerous species of mammals and reptiles. The life cycles of Dracunculus species are complex, and unknowns remain regarding the role of paratenic and transport hosts in transmission to definitive hosts. We had two primary objectives: to assess the susceptibility of several species of anurans, lizards, and fish as paratenic hosts for Dracunculus species, and to determine the long-term persistence of Dracunculus infections in African clawed frogs (Xenopus laevis). Animals were orally exposed to copepods infected with infectious third-stage larvae (L3s) of either Dracunculus insignis or D. medinensis. Dracunculus L3s were recovered from four anuran species, two lizard species, and one fish species, demonstrating that Dracunculus can infect tissues of a diversity of species. In long-term persistence trials, D. medinensis L3s were recovered from African clawed frogs tissues up to 58 days post-infection, and D. insignis L3s were recovered up to 244 days post-infection. Our findings regarding the susceptibility of novel species of frogs, lizards, and fish to infection with Dracunculus nematodes, and long-term persistence of L3s in paratenic hosts, address pressing knowledge gaps regarding Dracunculus infection in paratenic hosts and may guide future research regarding the transmission of Dracunculus to definitive mammalian hosts.

Dracunculus nematodes are endoparasites which infect, as definitive hosts, a high diversity of mammal and reptile species on multiple continents 1 . The most well-known species in this genus is Dracunculus medinensis, or the human Guinea worm, for which humans are considered the primary definitive host 2 . This parasite has been targeted for eradication by a global Guinea Worm Eradication Program (GWEP), which has succeeded in decreasing human case numbers by over 99.9% 3 . In recent decades, an increasing number of D. medinensis infections have been reported from domestic dogs (Canis lupus familiaris), particularly in Chad, Africa, which poses a serious challenge for the GWEP 2,4 . There are also multiple species of Dracunculus that infect mammals in the Americas, including Dracunculus insignis which infects a wide range of mammals (e.g., raccoons [ 1 . In definitive hosts, the large, gravid female Dracunculus typically migrates to the distal extremities of the infected host where a blister is formed at the eventual site of emergence 5 . The anterior end of the worm emerges from the blister and, if the worm is submerged in water, hundreds of thousands of free-swimming, first-stage larvae (L1s) will be released 5 . Cyclopoid copepods (the intermediate host) must ingest these larvae for them to develop into third-stage larvae (L3s), which are infectious to subsequent paratenic or definitive hosts, within the copepod 5 . In the life cycle of Dracunculus spp., the term transport host is used separately from paratenic host and refers to hosts in which infected copepods are ingested, but larvae never leave the gastrointestinal tract. However, ingestion of these transport hosts while L3s are present in the gastrointestinal tract may result in infection (distinction from paratenic hosts is discussed in references) 6 www.nature.com/scientificreports/ classically occurs by drinking water contaminated with infected copepods 8 . Emergence of the female worm typically occurs one year after infection 5 .
While the life cycle of D. medinensis and its transmission to humans is relatively well studied and understood, there remain gaps in scientific knowledge regarding the primary mode of natural infection for other mammalian hosts. It is hypothesized that direct ingestion of copepods may not explain patterns of D. medinensis transmission seen in dogs in Sub-Saharan Africa, because, when drinking, many animals lap (e.g., dogs and opossums) or suck (e.g., many snake species) from the top of the water column [9][10][11][12] . Copepods are often found lower in the water column, especially when infected with Dracunculus 13,14 . For these reasons, it has been suggested that Dracunculus transmission to wildlife and domestic animal hosts (as well as, potentially, humans) may also occur through alternative infection routes (e.g., the use of a paratenic or transport host infected with L3s) 6,7 .
Tissues of wild-caught fish have been inspected for the presence of Dracunculus larvae, but no larvae have been recovered 17,19 . Fish have previously been experimentally inoculated with or fed Dracunculus L3s and Dracunculus-infected copepods; however, L3s have been recovered from only a small number of fish and the number of larvae recovered from each animal was low, suggesting that they are not likely common paratenic hosts 6 . However, some fish species have been shown to serve as short-term transport hosts for Dracunculus species, successfully transmitting D. insignis and D. medinensis to ferrets 7 .
Historically, subcutaneous worms detected in Nile monitor lizards (Varanus niloticus) have been reported to be adult Dracunculus, although few morphologic features were noted and none have been genetically confirmed to be Dracunculus 20 . More recently, suspect subcutaneous nematodes found in Nile monitor lizards in Chad were determined not to be Dracunculus species, but similar to Ochoterenella species 17 . Although these investigations have predominately focused on adult subcutaneous worms, the possibility remains that Nile monitors may be potential paratenic hosts, susceptible to infection with Dracunculus sp. larvae that do not mature into adult worms.
The objectives of this study were to investigate the susceptibility of several anuran, lizard, and fish species to infection with Dracunculus spp. L3s and to determine the long-term persistence of Dracunculus larvae in anurans. We hypothesized that anurans would become most readily infected and that lizards and fish would develop few, if any, infections. We also hypothesized that larvae would persist at least several weeks, likely longer, within infected paratenic hosts. The insight gained from this work will help to better understand the role that these animals may play as paratenic hosts for Dracunculus transmission to domestic animals and wildlife hosts. (45%) anurans that were exposed by group batch with 20 copepods offered per individual. All recovered larvae (n = 1-14) were from muscle tissue ( Table 1). No larvae were recovered from anurans that were exposed by group batch methods with only 10 copepods offered per individual (Table 1).

Infection of potential paratenic hosts with
Dracunculus medinensis larvae were recovered from three out of seven (43%) of the anurans exposed to 20 infected copepods (Table 1). Larvae from D. medinensis-infected anurans were all recovered at six days postinoculation (DPI) ( Table 1). All recovered D. medinensis larvae were from muscle tissue, except one, which was recovered from the viscera of an American toad.

Lizards.
A single D. insignis larva was recovered each from two of five (40%) Nile monitor lizards (Table 1). A single D. insignis larva was recovered from gastrointestinal or visceral tissue of one monitor lizard at DPI 13 and a single D. insignis larva was recovered from the muscle from the other monitor lizard on DPI 14. Four D. insignis and/or D. medinensis larvae were recovered from the single exposed green anole (Anolis carolinensis) at DPI 6 ( Table 1). These four Dracunculus sp. larvae were recovered from the abdomen (n = 2), tail or legs (n = 1), and viscera (n = 1) of the anole.

Ability of Dracunculus L3s from a paratenic host to infect another paratenic host. No larvae
were recovered from the African clawed frogs that were exposed to D. medinensis larvae recovered from previous paratenic hosts. www.nature.com/scientificreports/ months post-infection. A single D. medinensis larva was recovered from an adult African clawed frog at approximately two months post-infection, but not beyond. All larvae remained L3s (Table 2).

Discussion
This study demonstrated that several anuran genera (Xenopus, Lithobates [Rana], Hyla, and Anaxyrus [Bufo]), as well as Nile monitor lizards, green anoles, and featherfin catfish, are susceptible to infection with D. insignis and/or D. medinensis L3s. We also found that D. insignis and D. medinensis larvae can persist in anuran tissues for at least eight and two months, respectively, although the number of L3s recovered from each infected animal was generally low. Regardless, these data show that these animals could serve as paratenic hosts if they ingest infected copepods in nature and are subsequently ingested by an appropriate definitive host. We exposed animals using two different methods (group batch or by mouth [PO]), but aimed to primarily batch expose animals as that better mimics natural exposure. A few anuran species (i.e., American toads, Cope's gray treefrogs, and adult African clawed frogs) were exposed to D. medinensis-infected copepods PO, as they had metamorphosed into adults before D. medinensis larvae became available for use and would be unlikely to ingest all copepods autonomously. Our primary goal in this study was to determine susceptibility to Dracunculus infection. www.nature.com/scientificreports/ Six anurans that were exposed as tadpoles underwent metamorphosis to froglets before being necropsied. Dracunculus L3s were recovered from two of these animals, supporting previous findings that D. insignis larvae can persist in anuran tissues through metamorphosis 14 . The persistence of larvae in the tissues through metamorphosis may facilitate Dracunculus transmission from aquatic to terrestrial food chains. This could be an important factor in transmission, as the majority of definitive hosts of Dracunculus nematodes are terrestrial. This study found that, in addition to X. laevis and Lithobates spp. (which have previously been infected with Dracunculus spp. larvae), Anaxyrus sp. and Hyla sp. can also become infected with Dracunculus L3s 14,21 . The infection of Anaxyrus sp. and Hyla sp. is particularly interesting, as members of these genera transition to a terrestrial or arboreal existence as adults, compared to Xenopus sp. and Lithobates spp. which remain completely or predominantly aquatic, even as adults. This transition to a terrestrial habitat could carry infectious larvae further from water sources, making them available to definitive hosts more widely across the landscape. However, the role of these animals in Dracunculus transmission would still depend on many other factors, including the natural history of these amphibian species, diets of definitive hosts, and how long Dracunculus L3s persist in paratenic hosts, as terrestrial anurans would be unlikely to acquire new infections after metamorphosis.
During a previous experimental study, D. insignis L3s persisted in amphibian paratenic hosts for up to 37 DPI, at which time the animals were necropsied 16 . In this long-term infection trial, we found that D. insignis larvae persisted for at least 244 days (approximately eight months), while D. medinensis larvae persisted for at least 58 days (approximately two months). These results demonstrate that infection of a paratenic host can extend the time that L3s may persist in the environment well beyond the lifespan of a copepod 21 . As we had a limited supply of D. medinensis L3s, we were unable to conduct sufficient trials to determine whether D. insignis may persist longer in paratenic hosts than D. medinensis. If this difference was found to exist, it could contribute to the higher proportion of wild-caught adult frogs found to be infected with D. insignis than with D. medinensis during field surveys 18,19 . Further testing with an increased sample size would be required to determine whether the persistence of larvae actually differs between Dracunculus species or paratenic host species.
No Dracunculus larvae were recovered from the two adult African clawed frogs that were fed D. medinensis L3s that had been recovered from other paratenic hosts. It is likely that our very small sample size (two animals) and the prolonged period before necropsy (4 months) explain these negative results. In our persistence trials, there was attrition over time so these animals should have been examined earlier after exposure. Future efforts to investigate transmission of Dracunculus between different paratenic hosts should use larger sample sizes and shorter infection periods. It would also be interesting to know if predatory animals, such as Nile monitor lizards, which can experimentally become infected with Dracunculus sp. larvae could become infected by ingesting other paratenic hosts.
Fish were investigated for their potential role in Dracunculus transmission, as many fish species consume copepods as part of a natural diet 25,26 . Despite this, Dracunculus larvae have not been recovered during multiple studies screening wild-caught fish 17,19 . Dracunculus insignis L3s have rarely been recovered from previous experimental trials with fish 16 . When larvae were recovered from fish, larval recovery rates were very low (0.6-2.0% recovery; 1-2 larvae per fish) and only 3/43 (7.0%) of the fish harbored Dracunculus larvae upon necropsy 16 . In a separate trial, fish experimentally functioned as short-term transport hosts of D. medinensis and D. insignis to infect domestic ferrets 7 . Our findings from this trial were surprising, as we recovered up to 6 D. medinensis L3s from the tissues of three out of four (75%) exposed featherfin catfish. This fish species is common in the Chari River Basin area in Chad, Africa where high numbers of D. medinensis infections are reported in domestic dogs living in fishing villages, and is consumed by both people and dogs 17 . Dogs in these villages often eat discarded small fish or fish viscera 4 . Although our sample size was small, our current findings are evidence that some fish species may be more capable of serving as paratenic hosts for Dracunculus than those that have been previously tested. This finding further supports the continuation of the screening of wild fish muscle tissues for Dracunculus larvae.
Lizards were included in this study because large, subcutaneous nematodes (believed to be Dracunculus sp.) were historically reported from Nile monitor lizards and these lizards are consumed by people 20,22 . However, a lack of contemporary reports and recent work in Chad, Africa, determining that large, subcutaneous nematodes recovered from wild Nile monitor lizards were not Dracunculus sp. but actually most similar to Ochoterenella Table 2. Results of experimental infection and long-term persistence trials of African clawed frogs (Xenopus laevis) exposed by mouth (PO) to 50 copepods infected with third-stage Dracunculus insignis or Dracunculus medinensis larvae. All larvae recovered remained L3s. www.nature.com/scientificreports/ sp., suggest that monitor lizards in this region are not definitive hosts for D. medinensis 17 . This current study confirms that Nile monitor and green anole lizards could become infected with Dracunculus larvae. As the diet of Nile monitor lizards can include amphibians and fish, were those prey to contain Dracunculus larvae, it is possible that monitors could serve as paratenic hosts, either by ingestion of larvae in fish intestines or in tissues of amphibians or fish, although these modes of transmission to paratenic hosts have not been confirmed 23,24 . It is unlikely that green anoles would become naturally infected with Dracunculus spp. due to their diet and primarily arboreal habitat; however, their infection demonstrates that multiple, distantly related lizard species are susceptible to experimental infection. Although anoles were exposed to both D. insignis and D. medinensis larvae, it is most likely that the recovered larvae were D. insignis, as only two D. medinensis-infected copepods were administered (in addition to 23 D. insignis-infected copepods). Species identity of these larvae could not be confirmed, however, as Dracunculus larvae can only be identified to species using molecular diagnostic techniques, which would destroy the sample, and these larvae were used in an experimental infection trial after recovery. Exposure of a ferret PO to the four larvae recovered from this anole (as part of a separate study) did not yield an infection, which is unsurprising given the low dose of larvae used. A previous study has shown that as few as 10 Dracunculus larvae may lead to infection of a ferret when administered interperitoneally (IP) (which was a more effective infection route than PO inoculation), therefore, four larvae administered PO would be unlikely to yield infection of a ferret 27,28 .
In all trials, infection occurred only in those animals that were inoculated with or exposed to at least 20 copepods per individual, suggesting an impact of parasite dose-dependent infection probability for Dracunculus infection in paratenic hosts. As copepod infection rate during this study was estimated to be ≥ 25%, it is likely that animals ingesting 20 copepods would consume at least 5 Dracunculus sp. larvae. Previous studies demonstrated that 10 larvae (administered IP) were sufficient to infect a ferret, but that percent recovery was higher with IP infection than PO 27,28 . It is likely that a similar minimum infectious dose also exists for paratenic hosts and may differ by paratenic host species and mode of infection. Parasite dose-dependent infection probability of Dracunculus spp. merits further investigation, as understanding this relationship could help researchers to more effectively study transmission in the laboratory by performing experimental infection trials with greater reliability.
Despite the variable sample sizes and exposure routes in this study, we demonstrated that a wide range of animals (anurans, fish, and lizards) were susceptible to infection with D. insignis and/or D. medinensis L3s. Importantly, one exposed fish species (Synodontis eupterus) was susceptible, opening up further concerns that certain fish species could serve as transport and paratenic hosts of Dracunculus species. Nile monitor lizards and anoles were successfully infected with L3s, demonstrating the first experimental infection of lizards with Dracunculus larvae. Dracunculus larvae remained L3s in the tissues of tested anurans for up to 244 days, extending the known persistence time of infectious larvae. Although no larvae were recovered from frogs that were fed L3s recovered from other paratenic hosts, continued investigation into the possibility of paratenic host to paratenic host transmission would be particularly interesting in determining if some predatory frogs (tadpoles or adults), fish, or lizards may concentrate higher numbers of L3s over time through predation of other infected paratenic hosts. Despite this study not determining how infectious larvae recovered from each of these paratenic hosts would be to another host, our findings contribute to a better understanding of the ability of these paratenic hosts to harbor Dracunculus L3s. This information is valuable to understanding how transmission to animal definitive hosts may be occurring, in addition to informing GWEP management decisions aiming to decrease transmission of D. medinensis to humans and animals.

Methods
Copepods and Dracunculus larvae. Dracunculus insignis larvae used in this study were obtained from raccoons from Georgia, USA in April and May of 2016 19 . Dracunculus medinensis larvae were obtained from infected dogs in Guinea worm endemic zones along the Chari River in Chad, Africa in May through July of 2016. Lab-raised copepods (from wild-caught stock from Athens, Georgia, USA) of the genus Macrocyclops were used for D. insignis trials and wild-caught African copepods most similar to the genus Mesocyclops were used for D. medinensis trials. Copepod identification was confirmed by sequence analysis of the partial cytochrome oxidase subunit 1 (COI) gene (Genbank accession numbers: MW522586 and MW522587) 29 . Copepods were infected by adding Dracunculus L1s to their water in a 1:3 ratio and allowing copepods to feed for 72 h. After two weeks, copepods were checked via microscopy to ensure Dracunculus larvae had molted to L3s. Copepod infection rate and Dracunculus development was assessed as previously described 7 . Batches of copepods with ≥ 25% infection rate (most often one larva per infected copepod) were used in trials.
Animals used in this study. Aquatic or semi-aquatic animals that may be prey for definitive hosts of Dracunculus spp. were chosen for their suspected involvement in transmission. African species were chosen primarily for their potential to transmit D. medinensis and North American species were chosen for their potential to transmit D. insignis and other North American Dracunculus species. Five species of anurans were used. African clawed frogs were captive-bred (Xenopus Express, Brooksville, Florida, USA). All other anuran species were collected from the wild in Georgia, USA (American bullfrog [Lithobates (Rana) catesbeianus], American toad, Cope's gray treefrog, and southern leopard frog). Tadpoles were also identified to species 30 . Anurans were assigned to age groups based on Gosner stage: tadpoles were categorized as Gosner stages 26-42, froglets were categorized as Gosner stages 43-45, and adults were Gosner stage 46 and older. The lizards used were juvenile, captive-bred Nile monitor lizards (Backwater Reptiles, Rocklin, California, USA) and adult, wild-caught green anoles from Georgia, USA. Two species of captive-bred, commercially sourced fish, bichir and featherfin catfish, were included in the study. www.nature.com/scientificreports/ Infection methods. Animals were exposed by one of two methods: group batch or PO (Tables 1, 2). Animals that were exposed via group batch were added to a 500 ml beaker (n = 1-8) containing infected copepods and allowed to feed up to 72 h, until all copepods were ingested. For those species that did not readily consume copepods or were too large to expose in a small beaker, infection PO was performed by concentrating infected copepods in a small volume of water and performing oral gavage with a pipette. The pipette was rinsed to ensure all copepods were ingested. If copepods were recovered during the rinse, oral gavage was repeated until all copepods were ingested.
Infection of potential paratenic hosts with Dracunculus-infected copepods. The number of animals exposed to Dracunculus larvae depended on the availability of infectious larvae and infected copepods. Fifty-nine individual anurans of four species were exposed to D. insignis-infected copepods via group batch (Table 1). Thirty-seven anurans were exposed in batches with 10 D. insignis-infected copepods per individual; 22 anurans were exposed in batches with 20 D. insignis-infected copepods per individual (Table 1). Seven anurans, all adults, were exposed PO to 20 D. medinensis-infected copepods (Table 1). Five individual Nile monitor lizards were exposed PO to D. insignis-infected copepods ( Table 1). The single green anole was exposed PO with both D. insignis-and D. medinensis-infected copepods ( Table 1). Eight individual fish (four bichir and four featherfin catfish) were exposed to D. medinensis larvae PO or by group batch methods with infected copepods (Table 1). Animals were euthanized after a varying number of days post exposure (6-140) and examined for the presence of Dracunculus larvae ( Table 1).

Ability of Dracunculus
L3s from a paratenic host to infect another paratenic host. Some D.
medinensis L3s recovered from experimentally infected animals (Hyla chrysoscelis and Synodontis eupterus ( Table 1)) were administered PO to two adult African clawed frogs (11 and 15 larvae were given to each animal, respectively). These two African clawed frogs were euthanized four months post-inoculation and examined for the presence of Dracunculus larvae.
Recovery of Dracunculus larvae from paratenic hosts. Animals used in this study were humanely euthanized using a buffered MS-222 bath (anurans and fish) or isoflurane inhalation (lizards) followed by pithing and cervical dislocation 31 . Gastrointestinal tract and muscle tissues were placed in separate Petri dishes, macerated, and allowed to sit at room temperature in water. Petri dishes were observed under a dissecting microscope for movement of larvae immediately after necropsy and tissue preparation and then again at four, eight, and 24 h. For larger animals (i.e., fish, adult African clawed frogs, and lizards), tissues were divided into multiple dishes.
Ethical approval and informed consent. All animal procedures in this study were reviewed and approved by the University of Georgia Institutional Animal Care and Use Committee (A2018 01-010). Additionally, all methods performed were in accordance with ARRIVE guidelines and within the aforementioned approved animal use protocol.