The Complete Genome of Chelonus insularis Reveals Dynamic Arrangement of Genome Components in Parasitoid Wasps That Produce Bracoviruses

ABSTRACT

Bracoviruses (BVs) are endogenized nudiviruses in parasitoid wasps of the microgastroid complex (family Braconidae). Microgastroid wasps have coopted nudivirus genes to produce replication-defective virions that females use to transfer virulence genes to parasitized hosts. The microgastroid complex further consists of six subfamilies and ∼50,000 species but current understanding of BV gene inventories and organization primarily derives from analysis of two wasp species in the subfamily Microgastrinae (Microplitis demolitor and Cotesia congregata) that produce M. demolitor BV (MdBV) and C. congregata BV (CcBV). Notably, several genomic features of MdBV and CcBV remain conserved since divergence of M. demolitor and C. congregata ∼53 million years ago (MYA). However, it is unknown whether these conserved traits more broadly reflect BV evolution, because no complete genomes exist for any microgastroid wasps outside the Microgastrinae. In this regard, the subfamily Cheloninae is of greatest interest because it diverged earliest from the Microgastrinae (∼85 MYA) after endogenization of the nudivirus ancestor. Here, we present the complete genome of Chelonus insularis, which is an egg-larval parasitoid in the Cheloninae that produces C. insularis BV (CinsBV). We report that the inventory of nudivirus genes in C. insularis is conserved but are dissimilarly organized compared to M. demolitor and C. congregata. Reciprocally, CinsBV proviral segments share organizational features with MdBV and CcBV but virulence gene inventories exhibit almost no overlap. Altogether, our results point to the functional importance of a conserved inventory of nudivirus genes and a dynamic set of virulence genes for the successful parasitism of hosts. Our results also suggest organizational features previously identified in MdBV and CcBV are likely not essential for BV virion formation.

IMPORTANCE Bracoviruses are a remarkable example of virus endogenization, because large sets of genes from a nudivirus ancestor continue to produce virions that thousands of wasp species rely upon to parasitize hosts. Understanding how these genes interact and have been coopted by wasps for novel functions is of broad interest in the study of virus evolution. This work characterizes bracovirus genome components in the parasitoid wasp Chelonus insularis, which together with existing wasp genomes captures a large portion of the diversity among wasp species that produce bracoviruses. Results provide new information about how bracovirus genome components are organized in different wasps while also providing additional insights on key features required for function.

INTRODUCTION

Integration of all or portions of a viral genome into the germ line of a eukaryotic host is referred to as endogenization (1, 2). Most endogenous virus elements (EVEs) rapidly decay but some have become fixed in the populations of hosts (1, 2). Some EVEs have also been coopted (domesticated) by hosts, in which their products provide beneficial functions (3). The most complex known example of endogenous virus domestication are the bracoviruses (BVs) found in parasitoid wasps of the family Braconidae (Hymenoptera) (4–6). BVs evolved from a virus in the family Nudiviridae (genus Betanudivirus) that integrated into the germ line of a braconid ancestor ∼100 million years ago (MYA) (7–9). Speciation events have since resulted in ∼50,000 BV-associated wasps in six subfamilies (Microgastrinae, Cardiochilinae, Miracinae, Mendeseliinae, Khoikhoiinae, Cheloninae) that form a monophyletic assemblage called the microgastroid complex (10). Most microgastroid wasps further appear to be specialists whose hosts are Lepidoptera (moths and butterflies) (11). BVs are much more complex than other EVEs because microgastroid wasps retain many genes from the nudivirus ancestor that still interact to produce virions, but virion function has also been repurposed for activities that enable females to parasitize their lepidopteran hosts.

The Nudiviridae is sister to the Baculoviridae with all known species in both families infecting insects or other arthropods (12, 13). Nudiviruses and baculoviruses also similarly have large, circular, double-stranded (ds) DNA genomes (80–230 kb), and share a partially overlapping set of core genes that primarily have functions in producing enveloped virions (12, 13). BV genome components are integrated into the germ line of all microgastroid wasps and also produce enveloped virions containing circular dsDNAs by replicating in ovary calyx cells of females (14). Calyx cell lysis releases virions into the lumen of the reproductive tract where they are stored with mature eggs, and females inject both into hosts. Virions rapidly infect host cells followed by the expression of BV genes that alter immune defenses and growth in ways that enable wasp offspring to develop (4, 15–17). However, the virions wasps inject into hosts cannot replicate, which results in BV genome components only being transmitted vertically (4).

Insights into BV genome evolution primarily derive from sequencing two wasp species in the subfamily Microgastrinae (Microplitis demolitor and Cotesia congregata) that produce M. demolitor bracovirus (MdBV) and C. congregata bracovirus (CcBV), respectively. BV genome components in both wasp species are extensively rearranged relative to nudiviruses. Nudivirus genes with known or predicted functions in producing virions are expressed in calyx cells but none are located in the DNA domains that are amplified, circularized, and packaged into virions (6, 18, 19). Almost half of these genes are also located in an ∼100 kb region of the wasp genome called the nudivirus cluster (6, 18, 19). Gene content and order of the nudivirus cluster is almost identical in M. demolitor and C. congregata, which has been suggested to be both functionally important and a remnant of the nudivirus that integrated into the common ancestor of the microgastroid complex (6, 18, 19). However, other nudivirus genes with functions in virion formation are widely and disparately dispersed (18, 19). The DNA domains that are amplified and packaged into virions are called proviral segments, which contain the genes that are expressed in parasitized hosts. A majority of these proviral segments are tandemly arrayed in a large ‘macrolocus’ (6, 18, 19). Inferred orthology relationships further suggest the organization of proviral segments was already established in the common ancestor of M. demolitor and C. congregata which diverged ∼53 MYA (7, 10, 18). The inventory of genes on proviral segments partially overlaps between M. demolitor, C. congregata and other species in the Microgastrinae (4, 16, 20). A majority of these genes also contain introns and share homology with genes from wasps, other insects, or other eukaryotes that suggest they have been coopted from diverse sources for functions in parasitizing hosts (4, 16, 20).

Altogether, the preceding results indicate the nudivirus ancestor of BVs was domesticated through genome rearrangements and regulatory alterations to produce replication-defective virions that wasps use to transfer virulence genes to hosts. Evolutionary constraints potentially maintain architectural features like the nudivirus cluster and macrolocus, but it is also possible gene content and organization differ in wasps outside the Microgastrinae. Among the subfamilies in the Microgastroid complex, the Cheloninae is of most interest because it diverged earliest from the Microgastrinae (∼85 MYA) after endogenization of the nudivirus ancestor (7, 9). Unlike microgastrines that parasitize larval stage hosts, chelonines parasitize hosts during their egg stage, which could also select for differing traits. Some BV genome components were previously identified from Chelonus inanitus bracovirus (CiBV) from the wasp Chelonus inanitus (8, 21–23). However, no chelonine wasp genome has been sequenced, which prevents assessment of how chelonine and microgastrine BV genome components compare to one another. In this study, we sequenced the genome of Chelonus insularis (Braconidae: Cheloninae) that produces C. insularis bracovirus (CinsBV) (24). We report that the inventory of nudivirus genes in C. insularis significantly overlaps with M. demolitor and C. congregata but they are dissimilarly organized. Proviral segments in C. insularis share organizational features with M. demolitor and C. congregata but gene inventories exhibit almost no overlap. Altogether, our results reveal a more dynamic arrangement of nudivirus genes and proviral segments in the genomes of BV-carrying wasps than was previously known.

RESULTS

De novo sequencing generated a draft genome for C. insularis with high levels of completeness and contiguity.

Long-read PacBio sequencing together with short-read Illumina sequencing were used to generate a reference genome for C. insularis (Table 1). After removing likely bacterial contaminants with blobtools, the assembly yielded 455 scaffolds with an N50 of 1,162,728 bp, and an overall size of 135 Mb (average cumulative coverage = 233×; Table 1). This assembly was similar in quality to the M. demolitor genome assembly Mdem2 (25), but was more fragmented compared to the chromosome-level assembly available for C. congregata (18). The genome size was slightly larger than the estimate (122.5 Mb) from kmer analysis with KAT. An assessment of genome completeness using BUSCO indicated that 98.8% of the BUSCO “Insecta” protein-coding gene set was identified. The GC content of the genome assembly was 30.5%, which is similar to other parasitoid genomes (26). Genome annotation with the NCBI Eukaryotic Annotation Pipeline yielded 11,442 genes or pseudogenes, including 10,548 containing protein-coding regions, and 19,220 annotated mRNA transcripts (Table 2). Most predicted transcripts (N = 18,106) were fully supported by the RNA-Seq data sets we generated from different C. insularis life stages and parasitized S. frugiperda larvae that wasp larvae had been removed from but were CinsBV infected (Table 3). A total of 894 noncoding genes, 113 tRNAs, 848 lncRNAs and other genome components were also identified (Table 2).

TABLE 1.

Raw reads generated for the genome assembly

SRA accession	Library strategy	Sequence data (gbp)	Coverage
SRR11678241	Pacbio	18	136×
SRR11678242	Illumina	108.1	97×

TABLE 2.

Gene annotation summary statistics

Feature	Count	Mean length (bp)	Median length (bp)	Min length (bp)	Max length (bp)
Genes	11,442	7,970	2,844	68	763,150
Protein-coding	10,548
Non-coding	894
All transcripts	20,679	2,954	2,225	68	56,124
mRNA	19,220	3,057	2,306	277	56,124
misc_RNA	400	2,944	2,331	244	15,328
tRNA	113	74	73	71	84
lncRNA	848	1,246	923	157	9,042
CDSs	19,220	2,170	1,521	105	54,753
Exons	77,264	399	218	2	15,371
Introns	63,902	1,406	120	30	176,056

TABLE 3.

RNA-Seq reads from C. insularis samples used for annotation

Sample type	SRA accession	Raw reads sequenced
C. insularis
Adult males (N = 4)	SRR11845185	16.7 million
Adult females (N = 4)	SRR11845186	15.5 million
Pupae (red eye stage, N = 3)	SRR11845184	11.9 million
Larvae (third instars, N = 3)	SRR11845190	14.3 million
Ovaries from adult females (N = 20)	SRR11845189	16.2 million
Parasitized S. frugiperda first instars (N = 10)
Mapping to proviral segments	SRR11967921	1,863
Non-proviral segment reads	SRR11967920	13.5 million
Parasitized S. frugiperda fourth instars (N = 3)
Mapping to proviral segments	SRR11967919	11,938
Non-proviral segment reads	SRR11967918	12.3 million

The C. insularis genome contains 43 homologs of known nudivirus genes.

To characterize CinsBV genome components, we focused first on identifying nudivirus genes in the C. insularis genome. Our homology-based search identified 43 nudivirus genes on 21 scaffolds (Table 4). Twenty-four were dispersed in the genome while 19 were in small clusters of 2–5 genes on 7 scaffolds (Fig. 1; Table 4). The clustered nudivirus genes on five of these scaffolds (Cluster 1–3, 6, 7) were closely aligned and flanked by nonviral wasp genes, while pairs of nudivirus genes <75 kb apart were on two scaffolds (Clusters 4, 5) with almost no wasp genes (Fig. 1). All sequenced nudiviruses share 21 core genes with baculoviruses while an additional 11 genes have been proposed to be nudivirus-specific, which yields a total nudivirus core gene set of 32 genes (13, 27). The Nudiviridae is also currently subdivided into two recognized genera (Alphanudivirus, Betanudivirus) (13) with certain genes potentially being genus specific. For the nudivirus core gene set, 21 were present in C. insularis but expansion of 11k into a 3-member gene family resulted in a total of 23 homologs. Four of the nudivirus core genes (p47, lef-4, lef-8, lef-9) encode the subunits of a predicted DNA dependent RNA polymerase (RNApol), one (lef-5) encodes a predicted transcriptional initiation factor, while three (helicase, vlf-1, and int-1) have predicted replication functions (Table 4). Thirty-one nudivirus genes were classified as virion envelope (odv-e66-1 through −6, pif-0 through −6, pif-8, p33, HzNVorf64-like) or nucleocapsid components (int-1, vlf-1, p33, 38k, vp39, 27b-like, HzNVorf9-1 and −2-like, HzNVorf93-like, HzNVorf118-like, HzNVorf106-like, HzNVorf128-like, HzNVorf140-like-1 and −2, K425_459-like) on the basis of a recent proteomic analysis of MdBV virions (Table 4) (28). None of the nudivirus genes in C. insularis had introns except HzNVorf128-like (see below). The most prominent absence from the nudivirus core gene set was a baculovirus/nudivirus-like DNA polymerase gene. We note that odv-e66 is duplicated in some nudiviruses but in C. insularis formed a six-member family although two (odv-e66-2 and −5) were truncated on their 3′ ends and thus likely pseudogenized (Table 4). We primarily used our unreplicated RNA-seq data sets for annotation purposes, but recognized they could provide qualitative information on expression patterns. Since BVs only produce virions in ovarian calyx cells, we expected the nudivirus genes to be primarily if not exclusively expressed in the adult female and ovary RNAseq samples we generated. Consistent with this expectation, FPKM values for nearly all of the nudivirus-like genes were higher in female wasps and ovaries than other wasp life stages while none were detected in host stages (Table S1). The only exceptions were odv-e66-2 and −5 that were largely not detected in any sample which further supported that both were pseudogenes (Table S1).

TABLE 4.

Nudivirus genes identified in the Chelonus insularis genome compared to baculoviruses, exogenous nudiviruses and three other BVs (CiBV, MdBV, CcBV)

Genes universally conserved in nudivirus and baculovirus genomes are indicated in bold type. Boxes shaded in gray are part of the core gene set, are present in all members of the Alpha- or Betanudivirus, or a given BV. Current ICTV nomenclature is used to categorize nudiviruses into two genera (Alphanudivirus and Betanudivirus) with Tipula oleracea nudivirus being most closely related to other members of the Betanudivirus genus (85). Boxes shaded in blue indicate gene products that were detected as virion components in baculoviruses (86), Tipula oleracea nudivirus (>2 unique peptides detected) (85), or a given BV (only partial data exists for CiBV and CcBV) (8, 23, 28). An exception is 11k, which was detected as a structural component of Tipula oleracea nudivirus but has not been identified in BV virions to date. An asterisk (*) indicates the gene is single copy. The ampersand symbol (&) indicates the gene has duplicated into a multimember family.

FIG 1.

Locations of nudivirus gene clusters on C. insularis genome scaffolds. Boundaries of nudivirus genes or genes encoding putative structural components of CinsBV virions (BVpp12-like and BVpp13b-like) are shown as red boxes with names listed above. Nonviral wasp gene boundaries are shown as white boxes.

The C. insularis genome contains 20 proviral segments in 7 loci.

We next identified the boundaries of proviral segments representing excision sites in the C. insularis genome by Illumina sequencing the DNAs in CinsBV virions, mapping the reads to the wasp genome, and identifying regions with marked differences in sequence read depth. Results identified 20 proviral segments (CinsV1-CinsV21) in 7 loci (Fig. 2A; Table 5) that were named to be consistent with naming of the partial inventory of segments identified in CiBV virions from C. inanitus (see below for information about segment homology between CinsBV and CiBV). Locus 1 contained 8 tandemly arrayed segments, loci 2–4 contained 2–4 segments, and loci 5–7 each consisted of a single segment (Fig. 2A). CinsV3 (locus 2) was at the end of scaffold NW_023276388.1 while CinsV8 (locus 7) was at the end of scaffold NW_023276780.1, which resulted in both being incomplete (Fig. 2A). The relative abundance of each viral segment in virions was estimated from the average sequence read depth of sites between segment boundaries. The abundance of the incomplete viral segments CinsV3 and CinsV8 differed enough to make it unlikely they derive from a common proviral segment that was broken between two scaffolds (Table 5). Instead, our read depth data much more strongly support that each of these proviral segments are unique but incompletely assembled. Read depth data further suggested CinsV6.1 and CinsV14.1 contained nested segments, named CinsV6.2 and CinsV14.2, that were generated by a lesser-used alternative excision site extending further into the genome on one side of each segment (Table 5). Sequence homology searches indicated that CinsV10 and CinsV12 in locus 1 shared partial homology despite being separated by several other proviral segments. No appreciable intersegmental homology was identified among the other segments (Fig. S1).

FIG 2.

A. Locations of proviral segments in the C. insularis genome. Proviral segments reside in seven loci containing 1–8 segments at each locus. Proviral segment excision boundaries are shown with blue transparent boxes with names beneath, while proviral gene boundaries are shown with blue opaque boxes with names indicated above. Nonviral wasp gene boundaries are shown as white boxes. Two segments have alternative excision sites as indicated by dashed boxes (CinsV6.1 and CinsV6.2, and CinsV14.1 and CinsV14.2). B. Wasp Integration Motifs (WIMs) for all CinsBV proviral segments. 200 nucleotides (nt) surrounding the 5′ and 3′ WIM sites of each segment are aligned, respectively. Similarity for each site is colored in shades of green. The tetramer AGCT and other conserved motifs are highlighted with black and gray boxes, respectively. Conservation of each site in the motifs is indicated in bits.

TABLE 5.

Virulence genes identified in proviral segments of CinsBV

Proviral segments in M. demolitor and C. congregata are initially amplified as replication units that can span one or more proviral segments in a given locus (29, 30). The replication units are amplified in one of two forms (head-tail or head-head/tail-tail), which are each associated with specific sequence motifs (18, 30). The circular dsDNA segments that are packaged into BV virions are then generated by excision and recombination, which occurs in association with conserved sequences named wasp integration motifs (WIMs) or direct repeat junctions (DRJs) containing the tetramer AGCT that identifies the site of recombination to produce circularized DNAs containing one WIM (29–33). There are additional, lesser-conserved motifs in the 10 bp immediately flanking the 5′ ends of WIMs and within proviral segments approximately 80 bp upstream of 3′ WIM sequences (TGAAT) (18, 19). Amplification is also non-equimolar, which results in some segments that are individually packaged into virions being more abundant than others (17, 19, 34). For C. insularis, read mapping data indicated proviral segments were differentially amplified with CinsV1, 2, 3, 16.8 and 10 being the most abundant (50–104× relative coverage) in CinsBV virions and the two nested segments (CinsV6.2 and 14.2) being the least (1–6× relative coverage) (Table 5). CinsV10 was much more abundant than the other segments (13–32×) in locus 1 suggesting it belonged to a different replication unit (Fig. 2A; Table 5). Abundance varied across all three segments (24–63×) in locus 2, but were less variable in loci 3 and 4 (Fig. 2A; Table 5). Loci 5–8 that consist of single segments also varied in abundance (Fig. 2A; Table 5). None of the specific sequence motifs associated with amplification units in M. demolitor or C. congregata were identified in C. insularis. However, all of the complete proviral segments in C. insularis were flanked by WIM sequences (Fig. 2B). We also identified two other conserved motifs in positions conserved with other bracovirus-producing wasps: one that was in the flanking region upstream of the 5′ AGCT sequence and another in the proviral domain upstream of the 3′ WIM sequence (GAAT) (Fig. 2B). Comparing proviral segments to the predicted circularized segments in virions indicated most were processed at several sites in a ‘window’ surrounding the conserved AGCT motif that ranged from 1–5 nucleotides (nt) at the 5′ end and 1–8 nt at the 3′ end. Within each window, most excision sites were used with relatively equal frequency (Fig. S2). Several MdBV and CcBV segments integrate into the genome of infected host cells with a conserved inverted repeat named the host integration motif (HIM) identifying the site of integration (17, 35). In contrast, only one CinsBV segment (CinsV10) contained a HIM domain (Fig. S3).

Gene coding densities on CinsBV segments are very low.

Proviral segments in C. insularis ranged from 6.6–46.1 kb which summed to an overall size of 341 kb, but coding densities were very low with only 35 predicted protein-coding genes and 7 noncoding RNAs (Table 5). CinsV1 encoded five genes; CinsV11, CinsV14.1, and CinsV21 each encoded four, while the remaining proviral segments encoded three or fewer, including CinsV7 and CinsV8 with none (Fig. 3; Table 5). Half of the protein-coding genes further belonged to three expanded gene families named the ank, CiV19.5g2-like, and CiV14g2-like genes (Table 5). Unlike the nudivirus genes, 64% of the genes on proviral segments contained one or more introns. Most genes were named by the segment they resided on because they shared no significant homology with other genes outside Chelonus inanitis bracovirus (CiBV) (Fig. 3, Table 5) (see below). The exception was the four-member ank gene family on 2 segments (CinsV1, CinsV3) that was so named because of the presence of a similar ankyrin repeat domain (Fig. 3; Table 5).

FIG 3.

Locations of virulence genes in CinsBV proviral segments. Exons, introns, and transcripts are colored in dark blue, white, and gray, respectively. Distinct loci containing proviral segments are indicated by dashed boxes, while the edges of proviral segments are indicated by bars with ticks indicating proviral segment positions in genome scaffolds. Names of proviral segments are located under scale bars with sizes of segments in parentheses in kilobases (kb).

We expected the genes on proviral segments would primarily be expressed in parasitized hosts. Our RNAseq data sets indicated this was the case for 11 CinsBV genes, including CinsV16.8_orph1 and ank-CinsV1-3 that had normalized expression values that were >500 in 4th instar parasitized hosts, and ank-CinsV1-3, CinsV16.8_orph1, CinsV2_orph1, CinsV6_orph1, and CinsV16.8_orph2 that had high expression values in parasitized hosts compared to the samples we prepared from different wasp stages (Table S2). However, some genes on proviral segments were detected in both parasitized hosts and one or more wasp stages, others were preferentially detected in adult female wasps and/or ovaries, and a few were not detected in any of our RNAseq samples (Table S2).

Chelonine BV genome components partially overlap.

As previously noted, comparative data from other chelonines are largely restricted to C. inanitus where ovary expressed sequence tags (ESTs) and proteomic analysis of virions identified some nudivirus genes while eight of the DNA segments in CiBV virions were cloned and sequenced (8, 21, 23, 36). A very early study also generated N-terminal sequence data (VGILDTVLSNTIQPH) for a 41 kDa virion protein from Chelonus sp. near curvimaculatus (37). C. insularis contained orthologs to all of the nudivirus genes identified in C. inanitus (Table 4) while the N-terminal sequence data for the 41 kDa protein from Chelonus sp. near curvimaculatus identified it as a vp39 homolog. A second point of interest was that previously generated proteomic data for CiBV virions from C. inanitus identified several proteins and corresponding cDNAs that at the time had no similarity to known nudivirus genes (23). One of these genes, 27b, was subsequently identified in a nudivirus (27) but not the others. Using these CiBV data, we identified orthologs to 23 genes in C. insularis, which included 27b that we classified as a nudivirus gene (Table 4, Table 6). For the remaining, 10 were single copy genes and 12 formed 4 small families (Table 6). Seven of the single copy genes and all of the genes in families 1, 3 and 4 were intronless, while the other single copy genes plus all genes in family 2 contained multiple introns (Table 6). These genes are present in the C. insularis genome in six clusters of two genes, three of which appear to be generated by local duplications (Table 6). Our RNA-seq data sets further indicated each of these genes were preferentially expressed in female wasps and ovaries (Table S3).

TABLE 6.

Genes encoding putative structural components of CinsBV virions

Boxes shaded in gray indicate homologous genes are present in C. congregata that produces CcBV or M. demolitor that produces MdBV. Boxes shaded in blue indicate the gene product is also detected in MdBV virions.

Also as noted above, we named CinsBV proviral segments in a manner consistent with how CiBV segments had been earlier named. This was because our analysis indicated five CinsBV and CiBV segments shared significant similarity, which suggested they were homologous (Fig. S4). We thus named the homologous CinsBV segments CinsV12, CinsV14.2, CinsV14.5, CinsV16.8 and CinsV21 to correspond to CiBV segments CiBV12, CiBV14.2, CiBV14.5, CiBV16.8 and CiBV21, while the other, nonhomologous, CinsBV proviral segments were named numerically starting with CinsV1 and extending through CinsV18 (Fig. 3; Table 5). Five single copy genes on the CinsBV segments that were homologous to CiBV segments were orthologs of single copy CiBV genes (CiV15.8g1-like, CiV15.8g2-like, CiV16.8g6-like, CiV19.5g1-like, CiV21g1-like) of which three also resided in similar locations on homologous CiBV and CinsBV segments (CiV14.2-like-3, CiV16.8g6-like, and CiV21g1-like). In contrast, the other two single copy homologous genes were on different CinsBV and CiBV segments (Fig. 3; Table 5). We further determined that genes forming 3- (CiV14g2-like-1 through −3) and 9-member (CiV19.5g2-1 through −9-like) families on multiple CinsBV segments shared homology with two genes identified on CiBV segments (Fig. 3; Table 5). Semiquantitative data previously indicated CiBV segments CiV12 and CiV16.8 were approximately five times more abundant than CiV14, CiV14.5, and CiV21 (36). Our read mapping data likewise classified CinsV16.8 as a higher abundance segment (relative coverage 53×), but suggested CinsV12 (13×) and the other CinsBV segments that were homologous to CiBV segments were lower abundance (5–18× relative coverage) (Table 5).

Chelonine and microgastrine wasps encode overlapping inventories of nudivirus but not proviral segment genes.

A complete genome for C. insularis also allowed us to compare BV gene content and organizational features to M. demolitor and C. congregata. The inventory of nudivirus genes substantially overlapped with homologs of all of the 43 nudivirus genes in C. insularis being present in M. demolitor and C. congregata, respectively (Table 4). DhNVorf067-like in C. insularis, which shared 25.6% identity with its homolog in Dikerogammarus haemobaphes nudivirus (38) was absent in the nudivirus gene inventories of M. demolitor and C. congregata; however, two uncharacterized genes in M. demolitor (LOC106693848) and C. congregata (CAJNRD030001114.1, position 2079873 to 2081417), both with 47% identity with DhNVorf067-like, were identified using blastp or tblastN. Eight candidate nudivirus genes in M. demolitor and C. congregata (HzNVorf94-like, K425_438, K425_456, K425_461, int-2, fen-1, PmNVorf87-like, and ToNVorf54-like) were absent in C. insularis (Table 4). Other differences included odv-e66, which as noted above was modestly expanded into a 6-member family in C. insularis but is greatly expanded into 21 and 36 member families in M. demolitor and C. congregata, respectively (6, 18, 19). Three genes orthologous with C. insularis, odv-e66-1, −3, and −6, were identified in partial data from C. inanitus, odv-e66-3 (CBA62617.1), odv-e66-1 (CBA62593.1), and odv-e66 (CAR40197.1), respectively. Unlike C. insularis, the int genes have expanded into a multimember family in M. demolitor and C. congregata (18, 19). A maximum likelihood tree placed int-1 from C. insularis, C. inanitus, M. demolitor and C. congregata in one clade that was separate from int-2, which was consistent with duplication of these genes in the microgastrines occurring after divergence from chelonines (Fig. S5). In contrast, this analysis suggested HzNVorf140-like had duplicated prior to divergence of the Cheloninae and Microgastrinae (Fig. S5). Like C. insularis, all of the nudivirus genes in M. demolitor and C. congregata were intronless except HzNVorf128-like. However, the introns in HzNVorf128-like were not in the same place in C. insularis (in the 5′ untranslated region and the coding sequence) and M. demolitor (in the 5′ untranslated region) which suggested they evolved independently. For the C. insularis genes that were first identified as CiBV virion components but are unknown from nudiviruses, M. demolitor and C. congregata also encoded 17a, 35a, 30b and 58b, while M. demolitor but not C. congregata encoded a BVpp69-like gene (Table 6).

The number of proviral segments (N = 20) and genes (N = 42) on proviral segments in C. insularis were both much lower than in M. demolitor (26 proviral segments, 95 genes) or C. congregata (38 proviral segments, 222 genes) (6, 18, 19). Most genes on CinsBV segments also shared no homology with MdBV or CcBV. The one possible exception was the presence of ankyrin domain-containing genes that also exist on proviral segments in all studied microgastrine BVs and ichnoviruses (IVs) from parasitoids in the family Ichneumonidae (16, 39). Genes with ankyrin domains also exist in the genomes of M. demolitor and other microgastroids, other wasp species, as well as other insects. A phylogenetic analysis of wasp-associated ank genes identified a clade containing the CinsBV ank genes plus several other wasp ank genes with a high support value that was outside the clade containing ank genes on microgastrine BV and IV proviral segments (Fig. S6). This finding suggested independent origins for the chelonine and microgastrine BV ank genes although uncertainty also existed because most clades were supported by low bootstrap values due to high sequence divergence.

Organizationally, chelonine and microgastrine proviral segments were similar in regard to some being tandemly arrayed and all being flanked by WIM domains (18, 19, 21, 36, 40, 41). However, the process of proviral segment amplification may differ in microgastrine and chelonine species, given the differential abundance of segments derived from the same loci and the lack of replication unit associated motifs in C. insularis. However, as earlier noted recombination occurs variably within a 1–8 bp window surrounding the AGCT motif of CinsBV proviral segments which was also found for several CiBV proviral segments (21, 36), whereas it is restricted to the 4 bp AGCT motif in microgastrine proviral segments (17–19, 29, 30). Thus, the conserved AGCT motif may be important for WIM recognition by presumptive integrases in chelonine wasps but sequence specificity may be less essential for recombination. The greatly reduced prevalence of HIM domains in CinsBV proviral segments relative to MdBV and CcBV further suggested fewer chelonine DNA segments integrate into the genome of infected host cells.

C. insularis and M. demolitor exhibit little synteny.

Comparing the C. insularis and M. demolitor genomes identified only 493 syntenous blocks containing 5 or more genes with average block size being only 10 genes. None of the blocks in C. insularis corresponded to the nudivirus cluster in M. demolitor but 3 small clusters of nudivirus genes in C. insularis shared syntenic features. Cluster 4 contained 27b, vp39, and HzNVorf118-like with a wasp gene located between vp39 and HzNVorf118-like, which was similar to 27b, vp39, pif-3, and HzNVorf118-like in the same order in the M. demolitor nudivirus cluster (Fig. 4A). Int-1 and 38k in cluster 3 and K425_459-like, HzNVorf106-like, and HzNVorf9-1-like in cluster 5 also exhibited syntenic order with the M. demolitor nudivirus cluster (Fig. 4A), which overall suggested a nudivirus gene cluster existed in the common ancestor of microgastroid wasps that has persisted in microgastrines like M. demolitor and C. congregata but has largely dispersed in chelonines like C. insularis. Four other small clusters of nudivirus genes were also identified in C. insularis that exhibited syntenic features with M. demolitor. In C. insularis, p47, lef-8, and HzNVorf128-like clustered together while in M. demolitor lef-8 and HzNVorf128-like reside on the same scaffold with intervening wasp genes gained and p47 moving to another location in the genome (Fig. 4B). We found that pif-1 and pif-2 were located together in both genomes albeit in different locations as evidenced by different neighboring wasp genes (Fig. 4C). Paralogs of odv-e66 in both species independently underwent localized duplications next to the same wasp gene (nucleolar preribosomal associated protein), while pif-8 in both species was located next to a conserved-position wasp gene (zwei Ig domain protein zig-8-like) (Fig. 4D and E).

FIG 4.

Regions of nudivirus gene microsynteny between C. insularis and M. demolitor. Boxes identify genes as defined in Fig. 1. Shading between scaffolds indicate orthologous genes.

Microsynteny of genes surrounding proviral segments has been used to determine homologous segment locations in genomes as well as segment gain and loss events between species in the Microgastrinae (18, 19). However, we identified no syntenous blocks of greater than three genes between any of the proviral segments in C. insularis and M. demolitor. While most nudivirus-like genes reside distantly from proviral segments, the small number near proviral segments in M. demolitor and C. congregata have been suggested as evidence that certain nudivirus genes and proviral segments reflect an ancestral association (18, 19). For example, pif-0 resides within 50 kb of locus 2 that contains five tandemly arrayed proviral segments in M. demolitor, while lef-5 is distant from any proviral segment but resides next to vlf-1 (19). Yet in C. insularis lef-5 is distant from vlf-1 but resides within 40 kb of locus 1 that contains 8 tandemly arrayed proviral segments (Fig. 2), while pif-0 is located on scaffold with no proviral segments (Table 4). Rather than reflecting an ancestral association, these results may simply be a product of different genome rearrangements that have occurred in C. insularis and M. demolitor due to movement of proviral segments or erosion of synteny surrounding proviral segments over time.

DISCUSSION

BVs retain many genes from their Betanudivirus ancestor, but function as transducing agents that female wasps use to transfer genes to hosts for the benefit of their offspring. Dispersal of nudivirus genes to locations outside proviral segments prevents BVs from being able to replicate outside wasps. However, the nudivirus cluster in M. demolitor and C. congregata has been suggested to be important for virion formation, while the clustering of proviral segments could be important for amplification, processing, or the evolution of virulence gene variants in molecular arms races with hosts (18). Our analysis of the C. insularis genome indicates several nudivirus genes and organizational features of proviral segments are likely conserved across all wasps in the microgastroid complex, while other features like the nudivirus cluster are not.

Twenty-one core genes are conserved in all baculovirus and nudivirus genomes: dnapol and helicase (DNA replication); lef-4, lef-5, lef-8, lef-9, p47 (transcription of viral genes); 38k, p33, p6.9, vlf-1, vp39 (DNA packaging, virion production and assembly); pif-0 through pif-6, pif-8 (infectivity); and ac81 (unknown function) (12, 13). Nudivirus gene inventories in C. insularis, M. demolitor and C. congregata indicate all except dnapol and ac81 are present (p6.9 is difficult to detect but may be present in all three species [18]) (Table 4). The absence of a baculovirus/nudivirus dnapol in all three of these species underscores this gene was lost while an unknown wasp DNA polymerase was likely recruited early in BV evolution to replicate proviral segments. Eleven additional genes present in all sequenced nudiviruses have been suggested as nudivirus-specific core genes: helicase-2, integrase (int), fen-1, three thymidine kinases (tk1-3), 11k, HzNVorf143-like, OrNVorf18-like, PmNVorf62-like, and GbNVorf19-like. Among these genes, only the int, OrNVorf18-like, and 11k genes remain recognizable in C. insularis, M. demolitor and C. congregata (Table 4). Like other large DNA viruses, baculoviruses and nudiviruses encode a number of other genes that are present in some but not all lineages or that are unique to particular species (12, 13). C. insularis, M. demolitor and C. congregata all encode odv-e66 genes that are present in many but not all baculoviruses and nudiviruses as well as several genes that are present in beta- but not alphanudiviruses (Table 4). Altogether, the genes shared by C. insularis, M. demolitor and C. congregata may comprise a BV conserved gene set, which consists of most of the core genes shared by baculoviruses and nudiviruses, three of the proposed nudivirus-specific core genes (int, OrNVorf18-like, and 11k), odv-e66, and certain genes (HzNVorf9, HzNVorf64, HzNVorf93, HzNVorf106, HzNVorf118, and HzNVorf128) known only from betanudiviruses.

It is possible other genes from the nudivirus ancestor are also shared among all microgastroid wasps but cannot be conclusively categorized as such because of their absence from currently sequenced nudivirus genomes. Due to their presence in the M. demolitor and C. congregata nudivirus cluster, four intronless genes are counted as nudivirus genes (K425_438, K425_456, K425_459, K425_461) (18, 19), but results from this study indicate only K425_459 is present in C. insularis. Additionally, 15 intronless genes were identified as encoding structural components of CiBV virions that are also present in C. insularis (Table 6), with only 17a, 35a, and 30b also present in M. demolitor and C. congregata. C. insularis further lacks nudivirus genes like HzNVorf94-like, int-2, fen-1, PmNVorf87-like, and ToNVorf54-like that are present in M. demolitor and C. congregata, suggesting some genes from the nudivirus ancestor have persisted in certain lineages of microgastroid wasps but not others, which in turn could result in differing replication or structural properties of BV virions among lineages.

Our results overall support that the single large nudivirus cluster present in Microplitis and Cotesia spp. derives from a syntenic domain from the ancestral virus that has largely been lost in C. insularis. However, this finding also indicates the large nudivirus cluster present in M. demolitor and C. congregata is likely not essential for high-level production of virions in calyx cells, which does not appear to differ among Chelonus, Microplitis and Cotesia spp. (14, 42, 43). The dispersal of BV genomes also does not functionally impede high level replication of proviral segments or virion formation, which in the case of MdBV exceeds replication rates of baculoviruses (44). We thus conclude that three other previously noted features (19) conserved between chelonine and microgastrine wasps are likely more important for virion formation. The first is the baculovirus/nudivirus-like RNA polymerase from the p47, lef-4, lef-8, and lef-9 genes. Similar to baculoviruses, the viral RNA polymerase in M. demolitor specifically transcribes nudivirus structural genes through promoter recognition which likely enables it to transcribe these genes regardless of their location in different microgastroid species (45). The second is the still unknown DNA polymerase(s) that amplifies proviral segments by also recognizing conserved motifs, while the third is the nudivirus-like integrase/recombinase genes (vlf-1, int) that experimental data from M. demolitor implicate in recognizing the WIM domains that likely flank all proviral segments in microgastroid wasps to produce the circularized segments in virions (45).

The similarities in proviral flanking sequences between C. insularis, M. demolitor and C. congregata support shared ancestry across the microgastroid complex although the origin of these motifs remains unclear as no sequenced nudiviruses encode similar domains. However, it is also possible integrase/recombinase gene function has potentially diverged between chelonine and microgastrine wasps given the greater variability in the sites of circularization for segments produced by C. inanitus and C. insularis versus M. demolitor and C. congregata. In contrast, duplication of genes into families appears to be a key means of generating novelty on proviral segments in all microgastroid wasps given the prevalence of multimember gene families across all BVs that have been examined to date (4, 5). Previous studies of M. demolitor, C. congregata and other species in the Microgastrinae indicate some genes and gene families on proviral segments are recent acquisitions from wasps (46–48), while others show evidence of acquisition from other insects or are very ancient and thus could have originated from the nudivirus ancestor (20, 44, 48). Most striking from sequencing C. insularis is that CinsBV proviral segments and associated genes share some homology with CiBV but gene inventories greatly differ from MdBV and CcBV, indicating gene gain and loss occurs frequently. Some of the differences in viral segment gene content and activity between microgastrine and chelonine species could be a consequence of the greater time of divergence compared to within-microgastrine comparisons. However, it seems likely that the inventory of genes on proviral segments is also strongly shaped by the specific interactions that have evolved between different lineages of microgastroid wasps and the host stages and/or species they parasitize. In the case of C. insularis, differences in proviral segment gene inventories may reflect differing host usage strategies given that chelonine braconids are egg-larval parasitoids that cause hosts to precociously initiate metamorphosis, while most microgastrine braconids are larval parasitoids that usually prevent hosts from pupating (4, 49, 50).

Like all insects, the lepidopteran hosts of chelonine braconids grow as larvae by molting which is followed after the last instar by a metamorphic molt to the pupal stage (51). Larval-larval versus larval-pupal molts are also primarily regulated by the titers of two key hormones: 20-hydroxyecdysone (20E) which stimulates insects to molt when its titer rises, and juvenile hormone (JH) that at high titer when 20E is released results in molting to another instar (larva) but at low titer results in molting to a pupa (52, 53). Chelonine braconids are referred to as solitary, egg-larval parasitoids because females oviposit a single egg into the egg stage of their hosts which is followed by the host hatching and developing as a larva. In turn, the wasp egg inside the host hatches into a larva that develops by feeding on hemolymph in the host larva (54). Hosts parasitized by chelonine braconids undergo ‘precocious’ metamorphosis, which refers to the host larva initiating behavioral and developmental processes associated with molting to the pupal stage one instar earlier than normally occurs (54). Precocious metamorphosis of the host is further associated with the mature wasp larva emerging from the host’s body to pupate while the host itself is unable to complete a pupal molt. Prior studies of C. inanitus establish that precocious metamorphosis correlates with host JH titers declining to undetectable levels one instar earlier than occurs in nonparasitized hosts. Concurrently, 20E titers do not increase normally, which is associated with the host larva being unable to complete a pupal molt (55, 56). Results from C. curvimaculatus and C. inanitus implicate both the wasp larva and BVs in regulating precocious metamorphosis but the specific genes involved are unknown (57–59).

Results from this study do not identify the gene products that cause precocious metamorphosis but they do suggest that similar to CiBV (22, 60–62) certain CinsBV genes are preferentially expressed in early or late instar parasitized hosts while others are expressed at similar levels. We hypothesize that CinsBV genes expressed similarly between host stages could have functions in processes like protection against host immune defenses that wasp progeny would be expected to require over the course of development. In contrast, genes that are preferentially expressed in early versus late instar hosts are potential candidates for having functions in regulating the endocrine processes that control host growth, including precocious metamorphosis. Lastly, we were surprised that some genes on C. insularis proviral segments were only detected in wasps or both wasps and hosts because few genes on MdBV and CcBV proviral segments are expressed in any wasp stage (62–64). We currently are uncertain about the function of these genes but speculate that they could be recently introduced into proviral segments from other locations in the wasp genome and as a result may still be responsive to regulatory factors and have functions that result in continued expression in wasps. It is also possible differences in methodology underestimate expression levels of some genes on proviral segments in species like M. demolitor and C. congregata, which would indicate functions in wasp biology may not be restricted to BVs associated with wasps in the subfamily Cheloninae.

MATERIALS AND METHODS

Insects.

Chelonus insularis was collected at the University of Florida Everglades Research and Education Center in Belle Glade, Florida and then maintained at the University of Georgia on its host, Spodoptera frugiperda. S. frugiperda larvae were fed corn leaves while adult moths were provided 10% sucrose in plastic containers covered with paper towels which is where females laid eggs. C. insularis was reared by allowing adult females to oviposit into S. frugiperda eggs. After hatching, parasitized host larvae were provided corn leaves until wasp offspring emerged to pupate. Adult wasps were then kept in cages and fed a 10% sucrose solution in water. All cultures were maintained at 23°C, 40–50% humidity, with a 12 h light: 12 h dark photoperiod.

Genome Sequencing, Assembly and Validation.

The Blood and Cell Culture DNA minikit (Qiagen) was used to isolate high molecular weight (HMW) DNA from adult male wasps. PacBio sequencing was performed using the PacBio Sequel System at the Georgia Genomics and Bioinformatics Core (GGBC). Genomic DNA for Illumina sequencing was extracted from one male wasp by homogenizing in lysis buffer (500 μl of 1xPBS, 2% sarkosyl and 0.5 mg/ml proteinase K) at 62°C for 1 h. After phenol:chloroform extraction, DNA was precipitated with 0.3 M sodium acetate (pH 5.2), 25 μg glycogen, and 100% isopropanol. The DNA pellet was then resuspended in 20 μl Nuclease-Free water. The sample library was prepared and sequenced (2 × 75 bp reads) on a NextSeq 500 system at GGBC.

Pacbio raw reads were assembled using Flye V2.7.1 with a genome size setting of 200 Mb (65). The assembly was polished using Pilon V1.22 with Illumina paired-end reads (66). First, reads were adapter trimmed and quality filtered with Trimmomatic V0.36 (program settings: ILLUMINACLIP:2:20:10:1 LEADING:20 TRAILING:20 SLIDINGWINDOW:4:20 MINLEN:36) (67). Processed reads were then mapped to the assembly using BWA V0.7.17 with the MEM option (68). Finally, Pilon was performed to improve assembly quality with default settings. Blobtools V1.1.1 was used to remove potential contaminants (69). The completeness of the genome assembly was assessed by BUSCO V4.0.5. with the “Insecta” data set (70) and assembly statistics were generated with QUAST V5.0.2 (71). We further performed a kmer analysis using KAT V2.4.1 to assess the completeness and heterozygosity of the assembly (72).

RNA purification and sequencing.

RNA was extracted from different wasp stages (larvae, pupae, adults), ovaries from adult females, and parasitized first and fourth instar hosts from which wasp larvae were removed by dissection (Table 3). Samples were first extracted using the RNeasy minikit with an on-column DNase step (Qiagen), followed by re-extraction with acid phenol:chloroform and ethanol precipitation in the presence of 0.5 M NaCl. Samples were then treated with the Ambion TURBO DNA-free kit (Invitrogen). Standard strand-specific Illumina-compatible libraries were then constructed from more than 1 μg of total RNA starting material for each sample, and sequenced (2 × 150 bp reads) by GGBC using the NextSeq system.

DNA extraction and sequencing from virions.

Ovaries were dissected from 50 adult female wasps and homogenized in 200 μl of TURBO 1× DNase buffer. The tissue was gently spun down at 200 × g and the supernatant containing CinsBV virions was passed through a 0.45 μm filter. DNase was added to the supernatant and incubated at 37°C for 1 h to remove exogenous DNA outside virions. After adding RNase A (Qiagen) to 0.8 mg/ml for 2 min, the DNase was inactivated by adding EDTA to 10 mM. DNA was then extracted from virions using proteinase K, sarkosyl and phenol-chloroform extraction method (44) followed by sequencing on a NextSeq system as described above.

Genome annotation.

Annotation of the C. insularis genome assembly was performed using the NCBI Eukaryotic Genome Annotation Pipeline (https://www.ncbi.nlm.nih.gov/genome/annotation_euk/process/). RNASeq data for C. insularis were compared to NCBI RefSeq protein sets for Diachasma alloeum, Microplitis demolitor, Fopius arisanus, Nasonia vitripennis, Apis mellifera, Chelonus inanitus, 39,059 other Insecta RefSeq proteins, and 106,743 GenBank Insecta proteins for gene prediction. Statistics and the evidence used for annotation are available at https://www.ncbi.nlm.nih.gov/genome/annotation_euk/Chelonus_insularis/100/.

Nudivirus gene annotation.

Nudivirus homologs in the C. insularis genome were identified using two strategies. First, ORFs from the C. insularis genome assembly and annotations generated by NCBI were searched against a database of all nudivirus-like proteins from C. inanitus, M. demolitor, and C. congregata using BLASTP (E value < 0.01) (8, 18, 19). Second, the set of genome ORFs or annotations was searched against a custom protein database containing all viral protein sequences from the NCBI nr database (downloaded July 2020). All of the identified ORFs or gene annotations were then manually converted into annotated gene models with the C. insularis jBrowse/Apollo instance on the i5k workspace (https://i5k.nal.usda.gov/available-genome-browsers). Manual annotations were merged with NCBI annotations to generate a C. insularis Official Gene Set OGSv1.0 (DOI 10.15482/USDA.ADC/1523023). Bigwig coverage blots generated from the transcriptome data sets generated were then used to delineate nudivirus gene transcription boundaries.

Proviral segment analysis.

Paired-end Illumina sequence reads from DNA that was isolated from CinsBV virions were mapped to the C. insularis genome using BWA V0.7.17 with the MEM option. A BAM file was imported to Tablet to view the mapping result (73). Regions of scaffolds with high read coverage and clear coverage boundaries indicating the presence of segments were then selected. Relative coverage was calculated for each proviral segment by normalizing coverage values to set the segment with the lowest coverage to 1×. Segments previously cloned and sequenced from CiBV virions (21) were used to search the proviral segments identified in C. insularis using BLASTN (E value < 1e-20). Coordinates of each alignment were plotted as line segments on grids representing pairwise comparisons between segments from C. inanitus and C. insularis with ggplot2 in R. The genes on segments were predicted by NCBI and supplemented with manual annotations from tblastn results of CiBV proteins against the C. insularis genome (E value < 0.01) and the FGENESH gene-finder (74), which were then searched against the nr database and manually annotated with Apollo. Wasp Integration Motifs (WIMs) were identified using read mapping data as described above. A conserved tetramer AGCT in WIMs was used as segment start and end coordinates (31, 32). A 100 bp flanking sequence at both ends of each segment was extracted and aligned using MAFFT (75) to identify segment direction and conserved motifs. Host integration motifs (HIMs) from M. demolitor and C. congregata were used as queries for blastn searches (E value < 0.01), and were also used to construct a Hidden Markov Model (HMM) to search against the C. insularis proviral segments using hmmsearch (18, 19). Candidate C. insularis HIMs were then further aligned with the M. demolitor and C. congregata HIMs using MAFFT. Sequence logos of the conserved motifs were generated using the WebLogo server (76).

Expression analysis of CinsBV genes.

Raw reads from samples listed in Table 3 were adapter-trimmed and quality filtered with Trimmomatic v0.36 using the above program settings. Expressed genes in different wasp or host stages were determined by read mapping to the assembled genome with HISAT2 (77). Transcript assembly and abundance estimation were performed using StringTie and read counts per gene were obtained using Ballgown (78).

Phylogenetic analysis of CinsBV genes.

Integrase/recombinase family members, int-1 (integrase-1), vlf-1 (very late expression factor-1) and HzNVorf140-like, were identified in C. insularis using select sequences from other BVs, nudiviruses and baculoviruses that were used previously (44) as queries in a blastp search (E value <0.01). Proteins containing ankyrin domains from C. insularis and M. demolitor were identified using hmmsearch with the PFAM Hidden Markov Models Ank, Ank_2, Ank_3, Ank_4, and Ank_5. Ankyrin domain-containing sequences from other BVs and ichnoviruses (IVs) were retrieved using accession numbers reported in earlier studies (79, 80). Amino acid sequences of each gene were aligned with MAFFT using the l-INS-i model (75). Poorly aligned positions were excluded by trimAI V1.2 (81). Substitution models for each gene were determined with jModelTest2 (82). Maximum Likelihood trees were inferred with RAxML-HPC2 via the CIPRES Science Gateway portal (83, 84).

Data availability.

RNA-seq and Pacbio reads from Chelonus insularis samples are available in the Sequence Read Archive under accession numbers SRR11845184 to SRR11845190, SRR11967918 to SRR11967921, and SRR11678241 and SRR11678242.