Introduction

Mutualism between eukaryotes and viruses is rare, since most viruses have parasitic associations with their hosts [1,2,3]. A group of double-stranded DNA (dsDNA) viruses called polydnaviruses (PDVs) have symbiotic associations with thousands of parasitoid wasps (order Hymenoptera), which parasitize immunocompetent lepidopteran larvae to enable successful reproduction [4]. PDVs have co-evolved with wasps and present a unique opportunity to investigate genome rearrangements associated with these unique mutual symbiotic relationships [2, 5]. PDVs are broadly classified into two distinctly evolved genera: Bracovirus and Ichnovirus [6, 7]. Bracoviruses are common within a monophyletic group of wasps known as the Microgastroid complex [8]. It is thought that bracoviruses evolved from integration of a nudivirus into the genome of a Microgastroid complex ancestor approximately 100 million years ago (mya) [9]. Mutualism between wasps and bracoviruses developed over time, and functional association is estimated to date back to around 73.7 ± 10 mya [10]. Bracoviruses exist in two forms: a linear provirus integrated into the host genome that mediates vertical transmission as Mendelian traits, and as circular dsDNA virions [11, 12]. Viral replication, particle production and packaging into virions occur exclusively in a specialized part of the wasp ovaries (the calyx) [12], and precede injection alongside one or more wasp eggs into the parasitized caterpillar host during wasp oviposition [13]. Virions are replication deficient and their dsDNA is only expressed by the caterpillar host’s cellular replication machinery [6, 14]. The virion particles encode proteins that compromise the caterpillar host immune defense, thus preventing recognition, encapsulation and destruction of the parasitoid eggs and larvae [9, 15]. However, lack of genes that independently encode viral structural proteins has elicited a debate on whether bracoviruses are of viral origin or a ‘genetic secretion’ of the wasps [12]. An example is the bracoviral virion DNA in the wasp Cotesia congregata that consists of cellular genes of wasp origin, several viral genes and transposable elements [15]. Phylogenetic analysis of its functional bracoviral genes has highlighted sugar transporters of wasp origin [11]. Transfer of these wasp genes into the provirus was facilitated by transposable elements, and subsequently followed by co-evolution with the host’s genome, to become more specialized [2, 5, 9].

Main text

The recently sequenced genome of the tsetse fly, Glossina morsitans morsitans (order Diptera), has revealed numerous homologous bracoviral genes (n = 305), widely spread across the genome; in addition to a large DNA hytrosavirus, the Glossina pallidipes salivary gland hypertrophy virus (GpSGHV) [16]. Although GpSGHV has been associated with reduced fecundity, life span, and causes salivary gland pathology in Glossina, its value as a potential entry point as a tsetse fly control strategy is has to-date not been explored. Perhaps more interesting is the finding of bracoviral sequences that bear close similarity (Basic Local Alignment Search Tool (BLAST), E values of < 1e−50) to those identified in the parasitic wasps (order Hymenoptera) Glyptapanteles flavicoxis and Cotesia congregata, where they occur as PDVs [2, 16]. Although the role of PDVs is well characterized in parasitic wasps, their organization, composition and functions in the tsetse fly genome is not known; indeed, their presence is new information. Molecular dating estimates that the orders Diptera (includes the tsetse fly and the house fly) and Hymenoptera (includes wasps) diverged ~ 350 mya [8], which is prior to the estimated date of first integration of bracoviruses into the ancestral wasp genome [9]. This raises the possibility that these genes may be remnants of PDVs acquired before this separation, and tsetse flies lost bracoviral mutualism after they adapted to larviparity (development of a single larva in its uterus as opposed to laying multiple eggs). An alternate hypothesis is that an undetermined braconid wasp may have parasitized the tsetse fly ancestor [16].

Tsetse flies are important vectors that transmit African trypanosomiasis to humans (sleeping sickness) and cattle (nagana). Approximately 70 million people and 50 million cattle are at risk of disease in tsetse-fly infested areas [17]. There are limited strategies for trypanosomiasis management primarily resulting from undesirable side effects of trypanocidal drug treatments; and there are emerging reports of multi-drug resistance [18,19,20,21]. According to the Pan African Tsetse and Trypanosomiasis Eradication Campaign (PATTEC), eradicating tsetse populations is the most viable approach of controlling trypanosomiasis in sub-Saharan Africa [22]. Identification of Glossina genes regulating vectorial capacity is thus a priority, as their manipulation would provide important clues for the development of effective vector control strategies, which will greatly facilitate trypanosomiasis control [18]. The tsetse fly, unlike other members of the order Diptera, does not lay eggs, but bears a fully developed larva (obligate adenotrophic viviparity) [23]. This makes it challenging to study PDVs in Glossina since during tsetse fly reproduction they are not replicated, excised from the host insect genome and packaged into viral particles that are mixed with semen like in parasitoid wasps. Wasp PDVs can easily be studied by first specifically extracting viral particles from the host [6]. Moreover, most bracoviruses consist of genes of host cell origin with protein domains conserved across metazoans, which further complicates analysis [24].

In this study, we aimed to identify polydnaviruses (PDVs) present in five recently sequenced tsetse fly genomes (G. austeni, G. brevipalpis, G. fuscipes, G. m. morsitans, and G. pallidipes) and the housefly (Musca domestica) [16, 25], by using references described in three parasitoid wasps (Cotesia sesamiae Mombasa bracovirus, Cotesia congregata, and Glyptapanteles flavicoxis) [26].

Methods

Identification of bracoviral orthologs

The proteomes of G. austeni, G. brevipalpis, G. fuscipes, G. m. morsitans, G. pallidipes, and M. domestica were retrieved from VectorBase (www.vectorbase.org) [27]. Bracoviral orthologs (n = 305) present in the G. m. morsitans proteome, previously described by the International Glossina Genome Initiative [16], were retrieved using a Perl script and used as sequence data references, in both nucleic and amino acid format (VectorBase assembly: GmorY1). The retrieved proteomes alongside previously described reference sequences were assigned to homologous clusters using OrthoMCL with default settings (BLASTP E-value cut-off 1e−5 and inflation index 2.5) [28]; using the G. m. morsitans proteome bracoviral orthologs (n = 305). Mapped orthologs were subsequently processed using BMX, as described in detail elsewhere [29,30,31].

Sequence alignment and phylogeny reconstruction

Multiple sequence alignments were performed using MUSCLE [32]. Maximum likelihood (ML) phylogenetic analysis of the multiple aligned sequences with bootstrap values of 100 replicates was performed using PHYML version 3.520 [33]. Phylogenetic reconstruction using species-specific concatenated bracoviral orthologs was preceded by aligning sequences within individual ortholog cluster files to ensure joined sequences were of the same length.

Results

We identified 53 bracoviral ortholog clusters, and a total of 2020 orthologs present across five Glossina species and M. domestica genomes (Fig. 1). The distribution varied across species: G. austeni (n = 333), G. brevipalpis (n = 303), G. fuscipes (n = 334), G. morsitans (n = 304), G. pallidipes (n = 332), and M. domestica (n = 414). Most of the identified orthologs were homologous to those present in Glyptapanteles indiensis (n = 603) and Glyptapanteles flavicoxis (n = 1109). The protein kinase cluster had the most abundant number of orthologs (n = 286) (Fig. 1a), which were all homologous to those present in either Glyptapanteles indiensis or Glyptapanteles flavicoxis (Fig. 1b). We then established evolutionary relationships of bracoviral orthologs identified in the five Glossina species and M. domestica genomes by phylogenetic reconstruction using species-specific concatenated bracoviral orthologs. This revealed congruence with previously reconstructed insect species phylogenies [34]; tsetse fly species in the same group clustered closely together (Fig. 2). Interestingly, the concatenated bracoviral orthologs from G. m. morsitans appeared to be more distant to those from G. brevipalpis (same genus), compared to the distance to M. domestica (same order).

Fig. 1
figure 1

A total of 2020 bracoviral orthologs categorized into 53 clusters were identified in the five Glossina species and M. domestica genomes. a The distribution of orthologs identified across Dipteran insect species: G. austeni (n = 333), G. brevipalpis (n = 303), G. fuscipes (n = 334), G. morsitans (n = 304), G. pallidipes (n = 332), and M. domestica (n = 414). b The distribution of orthologs by wasp species with homologous genes for each of the 53 clusters

Full size image
Fig. 2
figure 2

Phylogenetic reconstruction using species-specific concatenated bracoviral orthologs. There is congruence with previously reconstructed insect species phylogenies

Full size image

Discussion

Bracovirus represent a unique symbiotic relationship between eukaryotes and endogenous viruses. Endogenous bracoviral sequences identified in the genomes of parasitoid wasps, some moth and butterfly lineages, and Glossina morsitans morsitans are plausibly due to rearrangements of ancient integrations. [12, 16, 35, 36].

We identified orthologs of polydnaviruses (PDVs) in recently sequenced G. austeni, G. brevipalpis, G. fuscipes, G. pallidipes, and M. domestica genomes. Although bracoviruses in wasps are co-opted to ensure their successful reproduction, their role in Diptera that do not share this mode of reproduction was unclear. Our findings suggest that PDVs are descended from a single ancestor after initial host integration before the evolutionary radiation of different insect orders, and their presence in the reference G. m. morsitans is not a single random genetic introgression event. It is unclear when exactly parasitoid wasp and Dipteran PDVs separated and diversified in their different hosts. Our findings support previous suggestions that bracoviruses are descended from a common ancestor in the Paleozoic Era, and raise the possibility of integration of PDVs before the separation of Hymenoptera, Coleoptera, Lepidoptera and Diptera [8]. We also show that PDVs vary in size and display phylogenetic diversity, which suggests intra-genomic PDVs rearrangements while co-evolving with the specific host’s genome to adapt to different environments. PDVs progressively decay after integration, with minimal effects on the host’s fitness, as they evolve blurring genetic detection [36, 37]. Fine-scale analyses of genetic variation underscore retention of adaptive alleles and loss of non-adaptive genes mediated through selection pressure at bracoviral genes [38]. Accumulation of mutations, recombination, and/or deletions leads to dissolution of bracoviral genes in the host genome, and genes that acquire function for the host are under positive selection [36]. PDVs that co-evolve with the recipient insect genome to provide new physiological function must adapt to the eukaryotic expression machinery [12, 35, 36]. Phylogeny reconstruction using concatenated within-species PDVs showed congruence with previously reconstructed insect species phylogenies [34], suggesting that adaptive bracoviral evolution within the order Diptera is associated with the host insect’s environment. It is intriguing to note that G. m. morsitans is more distant to G. brevipalpis than the housefly, which is of a different genus. G. brevipalpis is closest to the root, implying that its PDVs are the least diverse.

Understanding the genetic composition and organization of bracoviruses has led to new vector control strategies using transgenic approaches [12]. For example, the polydnavirus Oryctes rhinoceros nudivirus (OrNV) has been used as a biological control agent in palm tree farming against the rhinocerous beetle [12]. Successful delivery of bracoviral genes by wasps into lepidopteran larvae has also inspired notable agricultural applications. Currently, teratocyte secretory protein (TSP14) producing transgenic plants effectively reduce Manduca sexta growth and development, thus protecting the plants from insect damage [39]. Detrimental agricultural effects have also been observed, for example acquisition of horizontally transferred genes by lepidopterans, braconid wasps, and mites that detoxify inhibitory alkaloids and cyanide have increased pest fitness allowing to overcome plant defences produced upon attack [36, 40]. Our findings suggest that the presence of bracoviruses is neither a result of a pathogenic virus contamination of the reference Glossina morsitans genome, nor a single case of being parasitized by bracoviruses of wasp origin. This newfound knowledge provides better understanding of tsetse biology, and highlights possible novel intervention target points.

Limitations

The limitation in this study was the absence of laboratory experimental validation using PCR of identified bracoviral sequences to those previously established in wasps. We anticipate that this would refine the number of orthologs to a smaller set of bracoviral homologs found in wasps. We were able to partly circumvent this challenge by using very stringent BLAST p-values (BLASTP E-value cut-off 1e−5 and inflation index 2.5).