The widespread nature of intact GAG-POL-ENV elements across metazoan genomes.

(A) Open reading frame arrangement of intact errantivirus copies. GAG encodes the structural capsid-like protein, POL encodes the enzymatic protein containing protease (PRO), reverse transcriptase (RT), RNase H and Integrase (IN) domains, and ENV encodes the envelope-like fusogen. tRNA primer binding sequence is used for reverse transcription. (B) Summary of Errantiviruses found in Metazoan phyla with type of ENV protein indicated. (C) Copy number of intact errantivirus copies in metazoans. Column represents a genome assembly, each tile representative of annotated errantivirus, colour of tile represents copy number.

Phylogeny of POL extended RT/connection region from intact metazoan errantiviruses.

Maximum likelihood tree constructed on representative POL extended RT/connection regions of intact errantiviruses (see methods) and rooted to the midpoint. The “ancient errantiviruses” and “insect errantiviruses” phylogenetic groups are highlighted in green and white respectively. Positions of known dipteran errantiviruses, Gypsy_ZAM (1), Gypsy_Gypsy5_Dgri (2), Gypsy_Gypsy (3), Gypsy_Idefix (4), and Gypsy_DM176 (5) are indicated. Host classifications of errantiviruses are annotated in colour, non-arthropod Protostomia (yellow), Deuterostomia (red), Diploblasta (orange), non-insect Arthropoda (green), Insecta (blue). One cnidarian element within the group of insect errantiviruses is marked by an asterisk.

env classifications of intact metazoan errantiviruses.

(A) Shown is a POL extended RT/connection maximum likelihood tree of the representative errantiviruses (see methods) with host (*) and type of ENV protein (**) indicated. Colours represent non-insect animals (*) (yellow), insects (*) (blue), glycoprotein F (F) (**) (orange), and glycoprotein B (HSV/gB) (**) (purple). Clades containing hymenopteran elements in “Insect errantiviruses” that carry either F or HSV/gB env are annotated as 1 and 2. (B) Close up of clades 1 and 2 in “Insect errantiviruses”. F-type and HSV/gB-type errantiviruses that are both found in Lasioglossum lativentre are marked by asterisks.

Structure and phylogeny of glycoprotein F-type ENV found in intact errantiviruses.

(A) Domain arrangement of glycoprotein F ectodomain in Errantiviruses. (B) A Sankey plot of POL extended RT/connection and glycoprotein F-type ENV ectodomain maximum likelihood trees for representative errantiviruses. In the Sankey plot, each ribbon links the clade assignment of the same element in the Pol extended RT/connection tree and the F-type ENV ectodomain tree. The plot is intended as a qualitative summary of global congruence between the two trees. Ribbon crossing can result from the relative ordering or rotation of clades in the two tree drawings and should not by itself be interpreted as recombination. Instead, broad one-to-one mapping between Pol-defined and ENV-defined clades indicates concordance, whereas extensive splitting or many-to-many connections indicates discordance consistent with recombination, env exchange or differential phylogenetic resolution. Bootstrap values for nodes are given. The “ancient errantivirus” group is highlighted in green. (C) Monomeric and trimeric structures of the “pre-fusion configuration” and the “post fusion configuration”. Schematic cartoons of F-type ENV ectodomain domain 4 are shown in (C’) and (C’’) Respectively. Structures are taken from Diptera_Sciomyzidae_Cmar_errantivirus_19 and Branchipoda_Diplostraca_Dcar_errantivirus_3 for “pre fusion” and “post fusion” structures in (C).

Structure of glycoprotein HSV/gB-type ENV found in intact errantiviruses.

(A) Structural arrangements of Ectodomains, Transmembrane domains and signal peptides found in HSV/gB-type errantiviruses. (B) Shown are the arrangement of domains of the HSV/gB-type ENV ectodomain, and Alphafold 2-predicted 3D monomer and trimer structures of the ectodomain. The structure is taken from Hemiptera_Pemphigidae_Elan_errantivirus_1. (C) Fusion loops and conserved cysteine bridges on structure of the errantivirus HSV/gB-type ENV ectodomain. Shown are variable cysteine bridges in the errantivirus HSV/gB-type ENV ectodomains, in domain I and II (Indicated with orange bar. Proximity of cysteine bridge to fusion loops of domain I are indicated for structures within C1, C2 and C3. Structures taken from Cnidaria_Hydrozoa_Chem_errantivirus_5, Bryozoa_Gymnolaemata_Cpal_errantivirus_2, and Tunicata_Ascidiacea_Clep_errantivirus_1, for C1, C2 and C3 configurations, respectively. (D) Shown is the summary of the cysteine bridge configurations found in HSV/gB-type ENV of different metazoan orders. Positions for cysteine bridge configurations within Nematoda and Platyhelminthes shown on Simple schematic. Exact positions of all cysteine bridges can be found in Supplementary Figure S8.

RNase H domains found in the bridge region of intact errantiviruses.

(A) Alphafold 2-predicted structures of domains found in the bridge regions of errantiviruses, with topology maps of α helices (grey) and β sheets (red). Structures are taken from Annelida_Polychaeta_Apac_errantivirus_7, Bryozoa_Gymnolaemata_Cpal_errantivirus_3, Coleoptera_Coccinellidae_Npum_errantivirus_1, Annelida_Polychaeta_Slim_errantivirus_9, Diptera_Cecidomyiidae_Smos_errantivirus_1 and Hymenoptera_Apoidea_Obic_errantivirus_2 for RNase H/Tether, ‘mini’ RNase H, ‘pseudo’ RNase H domains, “Mini” domain 2, “Pseudo” domain 2 and “Pseudo” domain 3, respectively. (B) Shown is the POL extended RT/connection maximum likelihood tree of representative errantiviruses, with RNase H domain architecture in bridge region of pol highlighted. “Insect errantiviruses” with RNase H + pseudo–RNase H (blue), “insect errantiviruses” with bridge region outliers (dark blue), “ancient errantiviruses” with RNase H + mini domain (green), “ancient errantiviruses” with bridge region outliers (yellow). (C) The structure of the bridge region of yeast Schizosaccharomyces pombe Ty3 element tf1 predicted by Alphafold 2.

Trees of known Ty3/gypsy elements and newly identified Errantiviruses.

Midpoint-rooted maximum-likelihood tree of the POL extended RT/connection region from representative env-containing errantiviruses and selected non-env-containing Ty3/gypsy retrotransposons. Representative non-env-containing Ty3/gypsy elements from animals, fungi and plants are included to place the newly identified errantiviruses within the broader diversity of Ty3/gypsy retrotransposons. Host Order and env type given for each clade, as well as solitary elements on the tree are shown. Errantiviruses containing chromodomains are marked in yellow. The “ancient errantivirus” group is coloured in green and “insect errantiviruses” in black. Known Ty3/gypsy elements includes Gypsy Group 1 OSIRIS (Gypsy-21_DAn), Gypsy Group 1 OSVALDO (Dypsy-3_DEl), Gypsy Group 2 BICA (Gypsy-6_DFi), Gypsy Group 2 BLASTOPIA (Gypsy-32_DWil), Gypsy Group 2 MDG3 (INVADER3, MDG3_DM), Gypsy Group 2 MICROPIA (Gypsy-7_DSe), Gypsy Group 2 SACCO (Gypsy-10_DAn), Gypsy Group 3 412/MDG1 (BLOOD, Gypsy-28_Dan), Gypsy Group 3 CHIMPO (Gypsy-7_DAn), Gypsy Group 3 17.6 (DM176, IDEFIX, ZAM), Gypsy Group 3 GYPSY (Gypsy-5_Dgri, Gypsy-Gypsy), Mag (Mag_As, MAG_Bmor), Cer (Cer1, Cer4), Tor (Tor1, Tor2, Tor4b), Athila/Tat (ATHILA0p1), and Chromoviridae (ATHILA0p1, CRM_AAM94350.1, REINA, Saccharomyces_Ty3, SCL213_Galadriel, Skipper*, Spombe_Tf1_AAA35339.1, SUSHI).

Identification of errantiviruses in metazoan genomes.

(A) Computational approach to identify and annotate env-containing retroelements in metazoan genomes. (B) A pruned maximum likelihood tree of POL extended RT/connection regions from errantiviruses with protein baits for consecutive tBLASTn searches annotated in colour, showing a comprehensive coverage of the baits in the tree. (C) POL extended RT/connection maximum-likelihood trees showing representative elements selected for AlphaFold2 structural analysis. Elements highlighted in red were used for structural prediction of the RNase H-containing bridge region, defined here as the canonical RNase H domain together with the C-terminal region between RNase H and Integrase. Representative elements used for AlphaFold2 analysis of F-type and HSV/gB-type ENV ectodomains are also indicated.

Complete errantivirus tree and annotations of tRNA primer binding sites of insect errantiviruses.

(A) Complete Maximum likelihood tree constructed on POL extended RT/connection regions of intact errantiviruses, with discrete phylogenetic groups within “insect errantiviruses” indicated (I1 – I6). “ancient errantivirus” group highlighted in green, “insect errantivirus group highlighted in white. Host phylogeny (*) annotated for insects (blue), non-insect arthropods (green), non-arthropod Protostomia (yellow), Deuterostomia (Red) and Diploblasta (Orange). ENV type (**) annotated for glycoprotein F (F) (orange), and glycoprotein B (HSV/gB) (purple). (B) A Sankey plot of POL extended RT/connection and IN maximum likelihood trees for errantiviruses, showing an overall phylogenetic congruence between them. (C) Shown are annotations of tRNA primer binding sites of “insect errantiviruses” within phylogenetic groups of POL extended RT/connection regions, demonstrating that the types of tRNAs used reflects the separation of POL extended RT/connection regions. Number of elements are indicated on Y axis. Sequence information of the tRNA PBS can be found in Table S1.

Insect errantivirus clades.

A maximum likelihood tree of POL extended RT/connection for the “insect errantivirus” group in isolation, with host phylogeny and env type, and discrete phylogenetic groups (I1-I6) indicated. Host orders (column 1) are annotated for Diptera (red), Lepidoptera (orange), Coleoptera (yellow), Hymenoptera (green), Hemiptera (cyan), Polyneoptera (pink) and Palaeoptera (purple). ENV types (column 2) are annotated for Glycoprotein F (F) (orange), and glycoprotein B (HSV) (purple). This figure provides the detailed tree underlying the insect portion of the representative POL phylogeny shown in the main figures and supports the conclusion that many insect errantivirus clades retain host-order or lower-level host-taxonomic structure.

Ancient errantivirus clades.

(A) Maximum likelihood tree of POL extended RT/connection for the “ancient errantivirus” group in isolation, with host phylogeny and env type, and discrete phylogenetic groups (A1-A11) are indicated. Host phylogeny (*) is annotated for insects (blue), non-insect arthropods (green), non-arthropod Protostomia (yellow), Deuterostomia (Red) and Diploblasta (Orange). ENV type (**) annotated for glycoprotein F (F) (orange), and glycoprotein B (HSV) (purple). (B) Detailed view of phylogenetic groupings in “ancient errantiviruses” with host (*) and env (**) annotated. Elements with the non-canonical GAG-ENV-POL arrangement are indicated (#). Bryozoa elements with GAG-ENV-POL arrangement occurring in distinct positions of the tree are highlighted as an example of host-associated elements that do not form a single monophyletic group. Each element is named for the host species and simplified to Phylum/Class/Order, Superfamily, Genus and Species. For example, elements found in Diptera Drosophilidae Drosophila melanogaster are named as Diptera Drosophilidae Dmel. (C) tRNA annotations of “ancient errantiviruses” within phylogenetic groups of POL extended RT/connection. Insect orders in the groups A1, A2 and A7 are marked by hash. Number of elements indicated on Y axis. This figure provides the detailed phylogenetic context for the host-taxonomic clustering, ENV-type distribution and tRNA PBS patterns summarised in the main text.

Host-taxonomy concordance analyses of selected POL extended RT/connection subclades.

Annelida-associated (A) and Lepidoptera-associated (B) errantivirus POL extended RT/connection subclades corresponding to tree position A6 (Fig. S4B) and I1(Fig. S3), respectively. Same tree scale for A and B. Highly supported internal subclades were identified using bootstrap support ≥95 and 3–10 descendant tips. Host-taxonomic concordance was assessed by comparing the observed distribution of host labels on the fixed POL topology with 10,000 random permutations preserving the observed label counts. In the Annelida clade, host-species labels were permuted and host-family labels were derived from the permuted species labels. Polynoidae, Sigalionidae and Nereididae are host families within Polychaeta. In the Lepidoptera clade, host-family labels were permuted and superfamily assignments were derived from the permuted family labels. Label “related to macro moths” refers to families represented in the tree from Drepanoidea, Geometroidea, Lasiocampoidea, Bombycoidea and Noctuoidea.Perfectly and mostly (≥80%) concordant subclades are highlighted in red and pink, respectively. These targeted analyses test whether host labels are non-randomly clustered on local POL trees, but do not estimate the number of horizontal transfer, duplication or loss events.

Conserved predicted Furin cleavage sites and hydrophobic fusion peptides of glycoprotein F.

Shown is a multiple sequence alignment of the middle part of the ectodomains of representative glycoprotein F. Arginine (R) residues in the putative Furin cleavage site and hydrophobic residues (Leu, Ile, Val, Phe, and Trp) in the fusion peptides are highlighted in red and green, respectively. Furin cleavage sites were predicted using ProP-1.0 (https://services.healthtech.dtu.dk/services/ProP-1.0/). Conserved α helix structures are shown upstream and downstream of the Furin cleavage site. This figure provides the alignment-level evidence supporting conservation of key F-type ENV motifs discussed in the main text.

Expanded phylogeny of F-type ENV ectodomains with viral and retroelement-associated homologues.

Representative viral and non-Ty3/gypsy retroelement F-like proteins are included for comparison, including baculovirus F proteins, paramyxovirus and pneumovirus F proteins, and the BEL/Pao-associated Drosophila Roo F-like protein. The added viral sequences form family-level clades within the broader F-type ENV tree, while errantivirus F-type ENV proteins span a comparable scale of diversity.

Conservation of cysteine bridges in errantivirus HSV/gB ectodomains.

Shown are alignments of domains I and II from errantivirus HSV/gB-type ENV ectodomains, together with representative viral and BEL/Pao-associated class III fusogens. Cysteine residues are highlighted in red and hydrophobic residues within the predicted fusion loop peptides are highlighted in green. Sequences from different metazoan orders, as well as representative viral or retroelement-associated groups, are aligned separately. Conserved cysteine bridges within and across the alignments are numbered, and ENV proteins with the GAG-ENV-POL arrangement are marked. Sequences used for structural prediction by Alphafold are indicated by asterisks, and those whose experimentally validated structures are available are marked by #. The structure of *Hymenoptera_Formicoidea_Tbic_errantivirus_1* is shown to highlight the position of the cysteine bridge in domain I of an insect errantivirus HSV/gB-type ENV ectodomain and the absence of fusion loop II. The single HSV/gB-type ENV identified in Annelida errantiviruses did not predict a trimeric structure and was therefore not included in the structural classification. For each viral family or errantivirus ENV group where there are at least three sequences, a summary of pairwise amino-acid identity is shown in the format “median x%; range y–z%; n = number of pairwise comparisons”. Raw data of pairwise amino-acid comparisons are also made available in Supplementary Table S2. This figure provides the alignment-level evidence that cysteine-bridge architectures are conserved within viral families and animal-associated errantivirus groups despite extensive primary-sequence divergence.

Presence of chromodomain in “ancient” errantiviruses

Maximum likelihood tree of POL extended RT/connection from representative errantiviruses along with known Ty3/gypsy elements. ENV type (*) annotated for Glycoprotein F (F) (orange), and glycoprotein B (HSV/gB) (purple). Chromodomain (**) annotated in purple.