Bats (order Chiroptera) represent the second largest order of mammals after rodents (order Rodentia). Classically, bats are divided into two suborders: megabats (Megachiroptera) and microbats (Microchiroptera). Megabats, also referred to as fruit bats, are assigned to a single family, Pteropidae, whereas microbats are taxonomically more diverse1. Both megabats and microbats host numerous, taxonomically diverse viruses. Examples of megabat-borne viruses that are highly virulent for humans are Marburg virus and Nipah virus. Severe acute respiratory syndrome coronavirus is an infamous example of a microbat-transmitted human pathogen1. Consequently, characterization of bats, their viromes, and cross-species transmission of bat-borne viruses have become research priorities.
Much less effort has thus far been invested in understanding the role of bat ectoparasites in maintaining viruses in bat populations or potentially transmitting them to humans or mammals of other species. The high degree of specialization and diversity of certain bat ectoparasites suggests that they could, in fact, be reservoirs for certain viruses, maintaining them in their bat hosts. Alternatively, bats could be refractory to infection with ectoparasite viruses, but nevertheless these viruses could be infectious or even pathogenic for other mammals, including humans, and be transmitted through incidental bat ectoparasite bites.
Bat flies are eyeless, wingless, hematophagous dipteran insects (Brachycera: Muscomorpha: Hippoboscoidea) that are obligate bat ectoparasites with off-host breeding life stages. They are assigned to two families, the monophyletic Nycteribiidae and the probably paraphyletic Streblidae, and they infest bats throughout the Old and New Worlds2, 3. Bat flies of each family have evolved exquisite morphological and behavioral adaptations to life on bats, reflecting a long history of co-evolution2. Bat flies host a diverse community of bacteria, including bartonellae, some of which are zoonotic4, 5. Bat flies also vector hemosporidian parasites (Plasmodiidae: Polychromophilus melaniferus) that cause “bat malaria”6. On the other hand, evidence for a role of bat flies as reservoirs or vectors of viruses is scant. Only two viruses have been unambiguously identified in bat flies: the putative orthoreovirus Mahlapitsi virus and the putative orthobunyavirus Wolkberg virus, which were both found in the nycteribiid Eucampsipoda africana on pteropodid Egyptian rousettes (Rousettus aegyptiacus)7, 8. In addition, rhabdovirus RNA-like sequences were detected in nycteribiids and bats in Spain, but the sequences were too short (108 nt) to unambiguously substantiate virus infection9.
Here, we report the discovery and coding-complete genome sequence of a novel rhabdovirus, Kanyawara virus (KYAV), in a previously unknown nycteribiid bat fly collected from an unclassified megabat in western Uganda. Phylogenetic and genomic analyses of KYAV and its relatives offer new insights into the evolutionary and ecological associations of rhabdoviruses with both bats and arthropods.
Bat flies were found on six of nine pteropodid bats trapped at the edge of Kibale National Park, western Uganda, in 2010. Next-generation sequencing (NGS) of bat flies yielded 0.11 × 106 to 1.59 × 106 reads per sample. After quality trimming, rhabdovirus-like sequences were detected in five bat flies, each from a different bat. These sequences mapped with low similarity to conserved regions of rhabdovirus genomes (order Mononegavirales, family Rhabdoviridae). De novo assembly yielded a contiguous sequence of 10,843 nt in one bat fly sample (MPK004), with five open reading frames matching the canonical rhabdovirus genome organization (Fig. 1)10. Subsequent analysis of bat fly reads mapped 448 to 206,726 individual reads to this sequence, yielding coding-complete genomes in three other bat fly samples. Rhabdovirus coding genome sequences from bat flies of individual bats were 99.9% and 99.8% similar at the nucleotide and deduced amino acid levels, respectively. Viral read frequencies in the five positive bat fly samples ranged from 8,611 to 262,258 per million, with coverage ranging from 5-fold to 3,632-fold.
Sequencing of sera from the bats on which the bat flies were found yielded 1.17 × 106 to 2.97 × 106 reads per sample, but no reads mapped to the detected rhabdovirus genome. Application of this method at this sequencing depth is approximately as sensitive as real-time quantitative PCR11; therefore, bat sera could confidently be classified as negative for the virus. For further confirmation, however, we also tested all bat sera by PCR, and results were congruent with NGS results (i.e., all bat sera tested negative for the new rhabdovirus). Oral and urogenital swab samples from all bats also tested negative for the new rhabdovirus by PCR.
Phylogenetic analysis (Fig. 2A; Supplementary Table S1) indicates the rhabdovirus to be a new member of the recently established genus Ledantevirus
12, 13. We named this virus Kanyawara virus (KYAV) after the village closest to the roost from which the bats were sampled. Sequence similarity between KYAV and other ledanteviruses based on concatenated, codon-based alignments of the canonical N, P, G, M, and L genes ranged from 62.4% (Mount Elgon bat virus) to 47.6% (Yngjiā tick virus 2) at the nucleotide level and from 59.3% (Mount Elgon bat virus) to 38.7% (Kern Canyon virus) at the deduced amino acid level, respectively. KYAV fulfills four of the five criteria of the International Committee on Taxonomy of Viruses (ICTV) Rhabdoviridae Study Group for classification in the genus Ledantevirus: A) the deduced amino acid sequence of the KYAV RNA-dependent RNA polymerase (L) diverges >7% from that of other ledanteviruses (KYAV:Mount Elgon bat virus = 35.2%); B) the deduced amino acid sequence of the KYAV glycoprotein (G) diverges 15% from that of other ledanteviruses (KYAV:Mount Elgon bat virus = 49.0%); C) KYAV has the same genome organization as other ledanteviruses (Fig. 1); and E) KYAV occupies a different ecological niche than other ledanteviruses. Criterium D (“can be distinguished in serological tests”) could not be evaluated due to the absence of a replicating KYAV isolate, but the high divergence of the sequence of KYAV G, the only ledantevirion surface protein, strongly suggests that KYAV is also serologically distinct14.
An analysis of the CpG content of the KYAV genome and related rhabdoviruses revealed significant variation (analysis of variance [ANOVA] F = 11.443; 6 degrees of freedom; P <0.0001), with low relative CpG depletion in sigmaviruses, vesiculoviruses, and the Sandjimba virus group accounting for this trend (Holm T-statistic values ranging from 3.77 to 6.86; P values all <0.01; Supplementary Table S2). Figure 3 shows average CpG depletion by virus group and gene. CpG depletion was least pronounced for the insect-only sigmaviruses15, 16, but more pronounced in the mammal-specific lyssaviruses17, 18. These CpG variation patterns were generally consistent across the five canonical rhabdovirus genes N, P, G, M, and L within each virus group (Fig. 3). Within the genus Ledantevirus, KYAV and Oita virus have the lowest CpG depletion values (KYAV: 0.69; Oita virus: 0.72); these values were comparable to values for the insect-specific sigmaviruses (Supplementary Table S2). Variation in CpG frequency also differed significantly among rhabdovirus groups (Levine’s W statistic = 3.29; 6 degrees of freedom; P = 0.008). The coefficient of variation in CpG depletion was lowest for sigmaviruses and lyssaviruses and notably higher for the other virus groups (Fig. 3).
Phylogenetic analysis of mitochondrial DNA sequences from the collected bat flies revealed them to be members of the nycteribiid subfamily Cyclopodiinae, representing a putative new species of the genus Dipseliopoda. These sequences are approximately as divergent from bat flies of the most closely related cyclopodiine bat flies (D. biannulatus) as are the cyclopdiine bat flies of the species Eucampsiopoda inermis and E. sundaica (Fig. 2B).
Phylogenetic analyses of the sampled bats revealed them to be members of a putative new species, clustering as an outgroup to Angolan soft-furred bats (Myonycteris angolensis) and approximately as divergent from those bats as are bats of other species pairs within the genus Myonycteris (Fig. 2C).
Viruses of the family Rhabdoviridae infect vertebrates, invertebrates, and plants around the world10, 19. Their broad host range and wide geographic distribution reflect a deep evolutionary history of lineage-specific adaptation to particular host assemblages and ecologies of transmission10, 19–21. Bats are disproportionately represented among mammalian hosts of rhabdoviruses18, 19. For example, many viruses of the rhabdoviral genus Lyssavirus, including rabies virus, cause bat-borne zoonoses18, 19, and bats are the dominant vertebrate hosts for at least two of the three subgroups of the genus Ledantevirus
10, 12. The reasons for this association are not clear but may reflect the unique diversity, biology, or social systems of bats1, 22, 23.
Viruses of the family Rhabdoviridae also have deep evolutionary relationships with arthropods, as do numerous viruses of other families within the order Mononegavirales
24, 25. These relationships are evident today in the strong ecological associations that many rhabdoviruses maintain with arthropods. Viruses of the genus Sigmavirus, for example, are transmitted only vertically among insects20, 26, whereas viruses of the genera Ephemerovirus, Tibrovirus, and Vesiculovirus may infect mammals but typically are vectored by biting midges, mosquitoes, sandflies, or ticks10, 19. Similarly, plant rhabdoviruses (genera Cytorhabdovirus and Nucleorhabdovirus) are vectored by aphids, leafhoppers, or plant hoppers, and even fish rhabdoviruses transmitted directly through water may have associations with arthropods10, 19, 27. Rhabdovirus genome fragment integration into genomes of arthropods belonging to widely divergent lineages also supports a long history of rhabdovirus-arthropod coevolution19, 24, 28.
Despite this family-wide dual adaptation to arthropods and bats, vector-borne transmission of bat-associated rhabdoviruses has proven difficult to confirm. For example, Binger et al. searched for the vector of Kumasi rhabdovirus by trapping 1,240 female mosquitoes of six genera close to a large transient breeding colony of African straw-colored fruit bats (Eidolon helvum) in Ghana. No infected mosquitoes were identified29.
KYAV is a new putative member of the rhabdoviral genus Ledantevirus, sorting within subgroup B, which contains bat-associated viruses (Fig. 2A)12. The discovery of KYAV in nycteribiid bat flies suggests that KYAV could be a vector-borne virus, with bat flies as vectors. However, we did not find KYAV in the blood or on mucosal surfaces of the bats from which the bat flies were collected. This negative finding may indicate limited or transient viremia in bats, as is characteristic of, for instance, rabies virus30–32; however, other rhabdoviruses have been recovered from mucosal surfaces of bats9, 33. Alternatively, KYAV may be an insect-specific virus that does not infect bats. The detection of KYAV in 5 out of 6 (83%) bat flies sampled is consistent with this notion because infection rates of arthropod vectors with vector-borne viruses tend to be much lower than this rate, typically below 10%34.
The relative CpG dinucleotide frequency in viral genomes varies widely among taxa35 and within virus groups36. CpG depletion has been used as an index of viral host adaptation16, 37, 38, although a recent study by Di Giallonardo et al. found the measure to be useful only for comparisons of higher taxonomic ranks such as Arthropoda compared to Vertebrata36. CpG frequencies in KYAV and related rhabdovirus genomes (Fig. 3) therefore likely reflect a combination of virological factors and host adaptation, with emphasis on the former36. In this light, our analyses show that CpG depletion was lowest among the insect-specific sigmaviruses. Genomes of lyssaviruses (including rabies virus), which are transmitted directly between mammals in the absence of arthropod vectors, had higher levels of CpG depletion17, 18. Genomes of ephemeroviruses, hapaviruses, and vesiculoviruses, which infect vertebrates but are vectored by arthropods, had levels of CpG depletion comparable to those of the mammal-specific lyssaviruses, if not somewhat higher. KYAV and Oita virus genomes stand out among ledantevirus genomes by having very high CpG frequencies, similar to insect-specific sigmaviruses. We again caution that dinucleotide composition appears to be shaped more by virus taxon than by host species36; however, this metric remains useful for comparing similar viruses that infect very different hosts (e.g., mammals versus arthropods)16, 39, 40.
The nycteribiid bat flies in which we found KYAV are representatives of a putative new cyclopodiine species within the genus Dipseliopoda (Fig. 2B)2. This assessment is currently based solely on the phylogeny created here from mitochondrial cytochrome oxidase subunit II and cytochrome C DNA sequence data. Formal classification of these bat flies will have to await morphological characterization and additional genetic analyses.
Likewise, at the time of sampling, we thought based on morphologic characteristics that the collected bats were Angolan soft-furred bats (Myonycteris angolensis, also known as Lissonycteris angolensis
41). However, our analysis of mitochondrial DNA sequences placed these bats as outgroup to Angolan soft-furred bats. They appear to be novel members of the Epomophorinae in the genus Myonycteris divergent enough to merit consideration as members of a separate species. This assessment is preliminary as it presently relies only on a mitochondrial cytochrome oxidase subunit I and cytochrome C phylogeny (Fig. 2C). Morphological characterization and additional genetic analyses will be required to confirm this taxonomy. Nevertheless, the discovery of a putative new pteropodid bat is surprising given that Kibale National Park is one of the most extensively researched forested areas in Africa42, 43.
Overall, our data demonstrate that our understanding of the diversity of megabats, their ectoparasites, and their viruses is still fragmented. Viruses of bats are diverse in part because bats themselves are taxonomically diverse23, 44. Therefore, identifying unknown taxa of megabats would be important for understanding the true diversity of their ectoparasites and associated viruses. Unfortunately, limited biological material and a remote field location in the present case precluded other desirable analyses, such as serologic assessment of bats or other mammals. Future studies using enzyme-linked immunosorbent assay (ELISA) or western blots targeting the major antigenic proteins of KYAV (likely N and/or G) might help elucidate the ecology of this virus in bats and animals of other species. Such studies may also resolve whether the absence of circulating KYAV in the tested bat sera reflects transient viremia, as we speculate, or lack of infection.
Bat flies occasionally bite people2, 45. Therefore, enigmatic cases of human infection with bat-associated rhabdoviruses may have resulted from incidental bites by bat flies or other bat-associated arthropods. For example, in 1969, Le Dantec virus infected a British dockworker who was bitten by an unidentified insect while unloading peanuts from a ship that had come from Nigeria29. Novel, divergent rhabdoviruses have also been found in apparently healthy people in Africa, suggesting unknown pathways of zoonotic transmission46. Our identification of an unknown rhabdovirus on unknown bat flies of unknown bats suggests further research on the diversity of these insects and their role as disease vectors might prove fruitful.
All data generated during the current study are available in GenBank (accession numbers KY385385-385392) or are included in this published article and its Supplementary Information files.