To begin with, we first collected the env sequences that have been previously reported in Table 4 and Table 6 of Vargiu et al.2016. Using the env consensus sequences (Vargiu et al., 2016) as a query, we further extracted additional nucleotide sequences for each HERV group from UCSC Genome Browser through the BLAT search function.

Secondly, group specific alignments were performed, using the Mafft algorithm (Katoh et al., 2018) of the Geneious software, for all the HERV env sequences collected from the above-mentioned resources. The env nt alignments were translated into forward frames in Geneious followed by extracting the portions from the translated regions from the alignments to avoid the stop codons. Lastly, the Env protein consensus sequences were generated for each group by aligning the extracted translated portions to the reference sequence to generate a complete representative Env sequence for each HERV group. These generated Env sequences were then aligned to the exogenous retroviruses of all classes. For this, we collected data from various databases for exogenous RVs and generated an exhaustive dataset of 58 envelope protein sequences, that included different strains as well as subtypes of exogenous retroviruses. This dataset included all the RVs from all the classes (i.e. alpha, beta, gamma, delta, epsilon, intermediate-Beta like, spuma and lentiviruses).

Using the above mentioned reconstructed HERV Env sequences we screened the primate genomes using the BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi). We used reconstructed Envs as queries to perform group specific tBLASTn searches in Hominidae, Cercopithecinae and Colobinae families of the Catarrhini parvorder and Platyrrhini parvorder. All the genome assemblies were accesses from NCBI Assembly database (https://www.ncbi.nlm.nih.gov/assembly/). A 40% sequence identity and E-value of 1E-5 were used as filters in tBLASTn algorithm. The hits obtained from all the searches were aligned and refined by removing other repetitive elements present as secondary integrations. The refined significant hits obtained from the similarity search were used for further analysis.

All the hits obtained from the tBLASTn search were filtered and were downloaded in fasta format. For each group, sequences whose length was less than 50% of the entire env gene were discarded. Followed by the refinements we performed group specific sequence alignments using MAFFT algorithm of Geneious software. These alignments were further used to carry out phylogenetic analysis with the IQTREE software (http://www.iqtree.org/) for Gamma-like, beta-like and spuma-like envs. The trees were inferred using maximum likelihood (ML) methods in IQ-TREE with a bootstrap of 1000 replicates. The trees generated were further visualized and annotated by the Figtree V1.4.3. software(http://tree.bio.ed.ac.uk/software/figtree/).

After the phylogenetic analysis of gamma, beta and spuma-like ERVs, the recombination events were detected using RDP, Geneconv, Bootscan, Maxchi, Chimaera, Siscan and 3SEQ methods implemented in RDP4 software (http://web.cbio.uct.ac.za/~darren/rdp.html) (Martin & Rybicki, 2000). The default settings of the RDP4 software that were used are (i) only potential recombination events detected by four or more of the above-mentioned methods; (ii) sequences were treated as linear and (iii) window size of 30 variable nucleotide positions used for the RDP method. The breakpoint positions and the recombinant sequences indicated for the potential recombination event were further confirmed through comparative genomics. The detected recombinant sequences were further used to perform BLAT search in the primate genomes available in UCSC Genome to confirm the position and location of recombinant sequence in the host genomes.

The initial step in understanding the evolutionary dynamics of ERVs – which encompasses their origin, proliferation within the host genome, and the mechanisms of inactivation and elimination – is to precisely reconstruct their viral genes. Reconstructing an ancestral prototype of ERVs is a very crucial step and needs to address two important factors i.e., the biases present in the ERV data used and the changes that had occured after the retrovirus was integrated into the host genome. Hence, with the purpose of providing prototype Env proteins for the defferent HERVs, we reconstructed the Env sequence of 32 HERV groups, including members of Class I (gamma- and epsilon-like, 20), II (beta-like, 10), and III (spuma-like, 2) (Table 1). To build these consensuses, we followed different approaches, we first extracted a total of 33 HERV elements that have been previously reported in Table 6 of Vargiu et al.2016, as the ones including the most intact env genes, belonging to the Class I groups HERV9, HERVFC, HERVH, HERVT, HERVW, HERVE, HERV3 and the Class II groups HML2 (18 out of 33 sequences), HML5, HML6, and HML8. For the other HERV groups not included in this list, we referred to the Table 4 of Vargiu et al., 2016, which provides a summary of 39 canonical HERV clades found in the human genome (GRCh37/hg19). More 190 sequences were extracted from the additional files of Vargiu et al., 2016 for which we generated a bed file and extracted them with the flanking of 200nt at both 3’ and 5’ ends using the table browser tool in the genome browser (Supplementary file1(S1)). These 190 sequences initially extracted were divided into canonical and non-canonical forms and further aligned group-specifically, to understand if any of the sequence codes for complete Env protein. Further, to have some sequences to be used as a reference, we extracted 19 consensus sequences from Dfam (nt and aa both), and 17 reported HERV-Env proteins from UniProt for which the accession numbers are provided in Supplementary file 2 (S2). The sequences used to perform alignments for each HERV group are described in detail in Supplementary file 3 (S3), which were then aligned to references sequences taken from Dfam, NCBI and Uniprot sequences (Supplementary file2 (S2)).

For each HERV group alignment, we translated the nt sequences in 3 forward frames to check the presence of stop codons using the Geneious software. From these translated frames we selected the portions of sequences based on two conditions: the sequence should not have any stop codons and should be in line with the reference sequences that can be either Dfam sequence, Uniprot or the consensus sequences taken from the Supplementary files of Vargiu et al. 2016. The selected translated sequence portions were then extracted and realigned with reference sequences thus generating a consensus sequence for the HERV Envs. Therefore, we reconstructed a total of 32 Envs for HERVs, which includes 17 Env for Class I i.e. gammaretroviruses, 13 Env for Class II i.e betaretroviruses and 2 Env for Class III i.e. spumaretroviruses (Table 1, Supplementary file1(S4)). A schematic representation of the Env reconstruction process if shown in Figure 1.

Since the Env reconstruction was performed based on the data available for human ERVs, to confirm the accuracy of reconstruction by its relatedness to the other RVs, we generated Env protein phylogeny of reconstructed Envs with 69 Env proteins of exogenous retroviruses (Figure 2). As expected, we observed a class specific clustering of the reconstructed Envs with that of exogenous RVs, except for HERVS which was clustered with Class I RVs and not with the Class III RVs, indicating that HERVS has a Class I Env that is related to PRIMA41, PABL and HERV1_Artodact, as also previously reported (Vargiu et al., 2016).

To track the spreading of the above HERV groups across primates’ evolution and evaluate their conservation, we screened a total of 43 species of primates to identify the env genes of class I, class II and class III ERVs. Briefly, we performed a group specific tBLASTn analysis search for HERV Env consensus sequence in each primate genomes. From the tBLASTn similarity search, we obtained 1374 hits for class I i.e. gamma-like env, 873 hits for class II i.e. beta-like env and 96 for class III i.e. spuma-like env. For all the three classes, we generated nucleotide alignments and performed maximum likelihood (ML) phylogenies using IQTREE software with a bootstrap of 1000 replicates.

For the gamma-like env 1374 hits the best fit model generated for the phylogeny using the IQTREE was TVM+F+R4 (Figure 3a, Supplementary file 2). A group-specific clustering of the env genes was observed in the phylogeny with the major grouping of HERVHF supergroup, that comprises of HERVH, HERVFA, HERVFB and HERVFC members. The supergroup HERVHF is reported to be integrated into the Catarrhini parvorder around 30-45 million years ago (mya). A widespread distribution of this supergroup was observed in Homo sapiens, Pan troglodytes, Gorilla gorilla, Pongo abelii, Nomascus leucogenys, Papio Anubis, Papio cynocephalus, Chlorocebus sabaeus, Macaca mulatta, Macaca fascicularis, Macaca thibetana, Colobus angolensis, Macaca nemestrina, Rhinopithecus bieti, Rhinopithecus roxellana Trachypithecus francoisi, Piliocolobis tephrosceles, Callithrix jacchus, Aotus nancymaae, Saimiri boliviensis, Cebus imitator (Figure 3a, Supplementary file 2). Similarly, we also observed wide distribution of ERVT, ERVW, and ERVE and ERVFRD followed by some small clusters of ERVH48, ERV9 and ERVIP. Additionally, we noticed distinct occurrences and clusters of HERV15, along with HERVR, ERVV1, ERVV2 and PABL, despite their exclusion from the blast analysis. This indicates that the derived Env consensus sequences effectively represent other related members of the supergroup that were not initially included in the collection (Figure 3a).

Likewise, phylogenetic analysis was also performed for 873 beta-like env hits obtained from tBLASTn. The best fit model for the phylogeny obtained was TVM+F+I+R3, with the major grouping obtained for HML2, which was further divided into two types i.e. type I (Np9) and type II (Rec) (Cadeddu et al., 2013; Chabukswar et al., 2022; Subramanian et al., 2011). In fact, the HML2 members are divided into type I and type II due to the presence of alternative splicing variants of env thus producing Np9 (type I) and Rec (type II), differing in the presence and absence of a 292bp deletion in the env gene, respectively (Chabukswar et al., 2022; Heyne et al., 2015; Subramanian et al., 2011) (Figure 3b, Supplementary file 3). The HML2 group is present in abundance followed by HML8 and HML10. The complete HML supergroup was observed in Homo sapiens, Pan troglodytes, Pan paniscus, Gorilla gorilla, Gorilla gorilla gorilla, Hylobates moloch, Nomascus leucogenys Macaca nemestrina, Macaca fascicularis, Macaca mulatta, Macaca thibetana, Cercocebus atys, Chlorocebus sabaeus, Papio anubis, Papio hamadryas, Theropithecus gelada, Madrillus sphinx, Rhinopithecus roxellana, Rhinopithecus bieti, Presbytis melalophos, Trachypithecus cristatus, Piliocolobus tephrosceles. These species belong Hominidae, Cercopithecinae and Colobinae subfamilies within Cercopithecidae family (Figure 3b, Supplementary file 3). A very few sequences were obtained in the Callithrix jacchus that belong to Platyrrhini parvorder. Other than clusters for HML2 Np9 and Rec, a cluster of HML2 Rec was observed which included sequences that are annotated as ERVK-7, ERVK-19 and ERVK-25 as the env of these sequences were observed to be different from HML2 Rec. The blast hits obtained in HML4 (HERVK-13) in Callithrix jacchus were clustered together with the ERVK-7, ERVK-19 and ERVK-25 group and hence to confirm the annotations, using ERVK-7, ERVK-19 and ERVK-25 sequences as query we performed the BLAT search in Platyrrhini genomes available on genome browser i.e. Marmoset (Callithrix jacchus) and Squirrel monkey (Saimiri boliviensis). Previous studies reported that the emergence of the HML2 group occurred approximately 30 mya i.e. in the Catarrhini parvorder and hence are found in abundance in the primates belonging to Hominidae, Cercopithecinae and Colobinae (Subramanian et al., 2011), but while performing the BLAT search, we obtained 6 copies of HML2 annotated as HERVK in the squirrel monkey genome while no HML2 sequences were detected in Callithrix jacchus (Table 2). Interestingly, previous studies that focuses on the characterization and identification of HERVs in the primate genome also reported the presence of HML supergroup in the Catarrhini parvorder (Grandi et al., 2017, 2021; Scognamiglio et al., 2023) and by far only HML5 has been known to be the most ancient member of the HERVK supergroup (Lavie et al., 2004), as its presence has been previously reported in marmoset and squirrel monkey. In this study, we identified the presence of members of HERVK supergroup i.e. HML1, HML2, HML4, HML8, HML9 and HML10 in the squirrel monkey genome that has never been previously reported (Table 2). The sequences identified are in few cases full-length ERV copies while in most of the cases are of very short length or fragmented. These results indicates two possible hypothesis for the presence of HML1, HML2, HML4, HML8, HML9 and HML10 in squirrel money is that either the onset of these HMLs began even before 30 mya i.e. before the split between New World Monkeys (NWM) and Old World Monkeys (OWM), that occurred around 43 mya or they got fixed in to the genome as a results of cross-species transmission of retroviruses after the split.

When spuma-like envs were investigated only 96 hits were obtained. An ML phylogenetic tree with the best fit model of K3Pu+F+R3 was generated with the bootstrap of 1000 replicates (Figure 3c, Supplementary file 4). While performing tBLASTn search for both HERVL and HERVS, hits were mostly obtained for gamma-like ERVs, specifically for ERV3-1, ERVFC-1, ERVFRD and ERV-W. These results were in line with our previous studies which states that HERVL is most similar to ERV3-1 in hyrax, tenrec, armadillo, alligator and turtle (Bénit et al., 1997; Cordonnier et al., 1995; Vargiu et al., 2016). The hits were further aligned and refined by excluding short elements and removing the sequence of other repetitive elements acquired by secondary integrations. The phylogenetic analysis performed for these groups showed that a major distribution took place in Homo sapiens followed by Macaca mulatta and other primate genomes Apart from the group specific env clustering, we also observed some small clusters hinting towards the presence of recombinant ERVs, for example, the grouping of HML4 for Callithrix jacchus with HML2 in the beta-like env phylogeny.

An important factor for the diversity of the env gene are recombination events, since recombination is a ubiquitous process and may give rise to new variants for the existing viruses, possibly accounting in changes of the original tropism. Various recombination mechanisms are known to occur in the ERVs such as template switching, env acquisition, env degradation, cross-species transmission, and ectopic recombination between homologous sequences (Chabukswar et al., 2023; Pérez-Losada et al., 2015). Such events can greatly increase the diversity among the ERVs and complicate their classification. Hence, we aimed to detect the potential recombination events in our gamma, beta and spuma-like env alignments. To this purpose, we used RDP4 software, which implements various inbuilt methods to detect recombination events, which are confirmed if four or more methods give a positive signal. The software detected multiple events of recombination in the gamma and beta alignment files generated in the present study (Figure 4a). Since the spuma-like env alignment includes only two groups i.e. HERVL and HERVS, and no recombination among them was detected, the spuma-like env were aligned with gamma-like env, to check the presence of recombination among the two classes. The analysis showed recombination events in HERVL (Figure 4a), but not HERVS. To confirm the presence of these recombinant forms in the primate genomes, we performed a BLAT search in the UCSC Genome Browser using the recombinant sequences detected by the RDP software as a query. While performing the BLAT search, we considered all the primate genomes available in Genome Browser (n=15) to also trace back the initiation of these recombination events along primates’ evolution (Figure 4b).

The recombination events detected in the gamma-like env alignment are ERVH-ERVE-MER11C, ERV15-LTR12, ERV15-MER11A, and PRIMA41-LTR5/LTR18. The first event was detected in ERVH, in which ERVE was present in the SU region while MER11C that is an LTR of HML8 group was present in the TM region of the env (Figure 4a). This event was only identified in 4 genomes i.e. humans (Homo sapiens, 8 copies), bonobo (Pan paniscus, 4 copies), chimpanzee (Pan troglodytes, 6 copies), and gorilla (Gorilla gorilla, 5 copies) (Supplementary file 5), suggesting a very recent occurrence (i.e. less than 9 mya) (Figure 4b). It can be said that this recombinant form of ERVH emerged possibly due to the co-packaging of ERVH, ERVE and HML8 and thus the recombinant form ERVH-ERVE-MER11C got fixed in the genome of human, chimpanzee, gorilla, and bonobo with the very few copies. Other recombination events were observed in ERV15 and PRIMA41, in which a LTR was detected in the TM region of the env gene. The ERV15-LTR12/MER11A is detected in 12 primate genomes i.e. Homo sapiens, Pan paniscus, Pan troglodytes, Gorilla gorilla, Pongo abelii, Nomascus leucogenys, Chlorocebus sabaeus, Macaca fascicularis, Macaca mulatta, Papio hamadryas, Nasalis larvatus and Rhinopithecus roxellana, while the PRIMA41-LTR15/LTR18 that involves the LTR of ERV15 and ERV3 groups, respectively, was observed in only 6 genomes i.e. Homo sapiens, Pan paniscus, Pan troglodytes, Gorilla gorilla and Pongo abelii (Figure 4b). Of note, these two events were present as a single copy in each genome in just one chromosome, as indicated in supplementary file 5. These single events might have occurred due to the template switching mechanism during the reverse transcription process and eventually got fixed in the chromosome. In the template switching mechanism, the two copies of RVs are co-packaged into the virus particle, a template switch occurs as RT is estimated to dissociate from one template to another resulting in intermolecular i.e. homologous or nonhomologous recombination or intramolecular i.e. deletions or insertions (Delviks-Frankenberry et al., 2011). As in the case of the recombinants detected in the present study, a template switch could have occurred between the highly conserved TM region of the env of one ERV and the LTR of the other ERV due to some sequence similarity, resulting in the recombination, where an LTR is inserted into the TM region of the env gene. According to the mechanism of template switching, the LTR found within the env gene has the features of a viral RNA genome LTR (i.e. not with the complete U5-R-U3 structure that characterizes the proviral ones), as indeed it is observed in the sequences.

Similar events were detected in beta-like env alignment, where the LTRs are detected in the TM region of the HML env gene. The events detected were HML10-LTR13, HML2-LTR13 and HML2-MER11A/MER11B (Figure 4a). Single chromosomal recombination event was detected for HML10-LTR13 and HML2-LTR13, where the presence of LTR13 which is the LTR of HML4 group was detected in the HML10 and HML2 env gene in Homo sapiens, Pan paniscus, Pan troglodytes, Gorilla gorilla, Pongo abelii, Nomascus leucogenys, Chlorocebus sabaeus, Macaca fascicularis and Rhinopithecus roxellana (Figure 4b). Similarly to ERV15, these events are also present as the single copies in the primate genomes. While on the other hand, HML2-MER11A/MER11B is present in abundance in the genomes of Cercopithecinae and Colobinae subfamilies and only one species of Hominidae. In the Nomascus leucogenys genome, 10 copies were detected followed by 17 copies in Chlorocebus sabaeus, 33 copies in Macaca fascicularis, 75 in Macaca mulatta, 18 in Papio anubis, 6 in Nasalis larvatus, and 23 in Rhinopithecus roxellana. The HML8 group contains three types of LTR i.e. MER11A, MER11B and MER11C, of which the presence of MER11A in HML2 env was detected in Chlorosebus sabaeus, Macaca fascicularis, Macaca mulatta, Papio hamadryas, Nasalis larvatus and Rhinopithecus roxellana while MER11B was detected in and Nomascus leucogenys, Chlorocebus sabaeus, and Macaca fascicularis (Figure 4b) suggesting that the template switching of HML2 happened with two different variants of HML8, as HML8 is divided into three variants based on the LTR type in the previous study(Scognamiglio et al., 2023). Since this event is absent in most of the species of Hominidae except for Nomascus leucogenys, it can be said that this event occurred after the split between Colobinae subfamily and Platyrrhini, simultaneously with the abundant spread of HML2. All the recombination detected in the HML group are hypothesized to be emerged due to the template switching mechanism as in the gamma-like HERVs and hence were fixed in the primate genomes.

Lastly for the spuma-like env, we detected two single chromosomal events for ERVL i.e. ERVL-MER11C and ERVL-LTR13 (Figure 4a): while on one hand presence of MER11C in the env was observed in Chlorosebus sabaeus, Macaca fascicularis, Macaca mulatta, Papio hamadryas, Nasalis larvatus, and Rhinopithecus roxellana, LTR13 was detected in Homo sapiens, Pan paniscus, Pan troglodytes, Gorilla gorilla and Pongo abelii (Figure 4b). As mentioned for all the above recombination events, ERVL seems to have also followed the same mechanism of template switching for both the LTRs.

Finally, to infer the functional effect of the presence of LTRs in the env region of ERVs, we performed the conserved domain (CD) search in NCBI. We first used the reconstructed Env to identify the domains present in the env and then performed the same search with the recombinant forms. In Table 3, all the CD are listed and their position in the env obtained from NCBI CD for the non-recombinant forms together with its presence or absence in the recombinant forms. Of note, a consistent pattern of loss of CD is observed for the ERVs which have the LTR in the env gene. The domains are either completely absent or truncated due to the LTR, leading to the loss of function in the corresponding protein. As shown in the Table 3, the only domain identified for ERVH is TLV coat, but other domains have been identified for the recombinant form of ERVH-ERVE-MER11A suggesting an env acquisition event for the ERVH group by the co-packaging with ERVE. From the overall results, we can say that the LTRs of HML8 (MER11A, MER11B and MER11C) are the most frequently found in the env gene of different ERVs and are associated with recombination in primate genomes. Interestingly, the recombination events involving the presence of LTRs in the env gene are mostly found in the Catarrhini parvorder, suggesting that these events mostly occurred during the process of endogenization of the ERVs, leading to the intragenomic spreading of the recombinant forms. These events are usually associated with the Env loss of function, with phenomena termed “env snatching” or “env degradation”. Such events can further lead to loss of the env gene followed by the intragenomic spreading of the env-less ERVs in the genomes. Interestingly, this is a widespread phenomenon among the retroviral integration in human lineage but have not been previously reported in the other primate genomes. The exception here is the ERVH, which is associated with the “env acquisition” i.e. the gain of env function from ERVE. Thus, such recombination events may lead not only to the generation of new retroviral elements but also to eventual spreading of ERVs throughout the genomes and therefore the co-evolution of ERVs in the primate lineage.

ERVs are the remnants of ancestral exogenous RVs whose origin and diversity are still not completely understood and are a great tool to study the coevolution of host and RVs. The present study provides a phylogenomic perspective of the retroviral env gene diversity. With this study approach we provide the evidence for inter as well as intra group recombination events. Present results demonstrate the prevalence of ERVs across the primate lineage as well as the diversification of the retroviral Env through the process of recombination over the evolutionary timescale. Overall, this analysis gives further knowledge in studying the host-virus interactions on a larger scale.