Evolutionarily divergent nidovirus with an exceptionally large genome identified in Pacific oysters undergoing mass mortality

Significance

Pacific oysters are the most widely farmed shellfish worldwide, but are increasingly affected by mass-mortality events, often of unknown cause. We present Pacific Oyster Nidovirus 1 (PONV1), identified during a mortality event in farmed Pacific oysters. PONV1 has an exceptionally large genome, is the first nidovirus reported to replicate in bivalves, exhibits a genetic architecture with protein domains not found in other nidoviruses, and represents a newly proposed family, Megarnaviridae. Moreover, PONV1 and its close relatives were detected in Pacific oysters from three continents; thus, if PONV1 causes disease in Pacific oysters, it underlines potential risks from translocation of oysters and oyster-seed. These findings highlight nidovirus diversity and host range, and have potential implications for oyster health and aquaculture.

Abstract

Metatranscriptomic data from a mass-mortality event of adult Pacific Oysters, Crassostrea (Magallana) gigas, the most widely cultivated shellfish globally, revealed a nidovirus shown to replicate in a bivalve, Pacific Oyster Nidovirus 1 (PONV1). At 64,331 bp of linear bisegmented, positive-sense single-stranded RNA, PONV1 has one of the largest genomes reported for an RNA virus. Moreover, transcriptomic data reveal that many conspecific viruses of PONV1 occur in Pacific oysters from Europe and the Pacific coasts of Asia and North America. PONV1 (tentative species Megarnavirus gigas) and its conspecific viruses represent a putative family, Megarnaviridae, that is widely divergent from other families of nidoviruses, and encodes protein domains likely involved in virus–host interaction that are undescribed for other nidoviruses. Given that PONV1 occurs in diseased Pacific oysters undergoing mass mortality, and that close relatives of PONV1 are widespread globally, it emphasizes the need for caution in the translocation of Pacific oyster spat that routinely occurs.

The Pacific oyster, Crassostrea (Magallana) gigas, holds significant economic and ecological value due to widespread wild populations and extensive cultivation (1–3). Native to the northwest Pacific Ocean, wild and cultured populations of Pacific oysters have been established worldwide with large ecological effects and economic benefits (3–6). They are the most intensively cultured oysters worldwide and are farmed in their native Japan, as well as many other countries, including Australia, Canada, China, France, South Korea, the United Kingdom, and the United States (1, 3, 7). Although generally resilient and adaptable, the health of Pacific oysters is being challenged by environmental stressors driven by climate change, human encroachment, and disease, leading to mortality events and economic loss (8–12). Mortality events are recurrent in British Columbia (13–17) and elsewhere (7, 18–22), and have been attributed to viruses (21), bacteria (16, 18), and protists (14, 17), often triggered by physiological and environmental stressors such as spawning and elevated water temperatures (8–12, 19, 20); yet, in many cases the causative agents are unknown. In addition, oyster spat are shipped from hatcheries to oyster farmers worldwide, leading to potential disease transmission on farms and in wild populations (1, 6).

Nidoviruses infect a wide range of invertebrates and vertebrates, and can cause diseases in humans, livestock, wildlife, and aquatic animals (23–33), including emergent diseases such as COVID-19 caused by SARS-CoV-2 (Severe Acute Respiratory Syndrome Coronavirus 2). Indeed, the extent of nidoviruses diversity and their role in infectious diseases is only beginning to be appreciated (24, 25, 28–30). Recent studies have revealed a remarkable diversity of nidoviruses in an expanded range of invertebrate hosts, including the discovery of multisegmented nidoviruses (34–36). Here, we use transcriptomic analysis (36–39) to interrogate farmed adult Pacific oysters during a mass mortality event in British Columbia, Canada, revealing Pacific Oyster Nidovirus 1 (PONV1) in dead and dying oysters, and is the first nidovirus shown to replicate in bivalves. Mining of extant databases shows conspecific viruses to PONV1 in Pacific oysters from the Pacific coasts of Asia and North America, as well as Europe. PONV1 and its conspecific viruses, including the recently described Crassostrea gigas nidovirus (CGNV) (34), represent a newly proposed species (Megarnavirus gigas) and family (Megarnaviridae) that are widely divergent from other nidoviruses, and are characterized by unusual genome architecture and composition. Below, we detail the features and distribution of PONV1 and its close relatives, and discuss the implications for understanding nidovirus diversity and the potential implications for translocation of oyster spat.

Results

Identification of PONV1.

To understand the etiology of diseases affecting Pacific oysters, we conducted metatranscriptomic analyses on adult Pacific oysters collected from two farms (three farm sites) during a mass-mortality event, as well as from wild oysters without obvious signs of disease, during the summer of 2020 (Fig. 1A and Dataset S1 and SI Appendix, Fig. S1). The metatranscriptomic data revealed a previously undescribed nidovirus, PONV1 (Materials and Methods). The genome of PONV1 is a positive single-stranded (ss) RNA of about 64,331 bp, with two segments (41,709 and 22,622 bp for segment 1 and 2, respectively) (Fig. 1B), larger than any RNA virus genome found in the International Committee on Taxonomy of Viruses (ICTV, version 37) and NCBI Reference Sequence (NCBI RefSeq, version 216) databases, although recently nidovirus genomes of similar size have been uncovered by mining existing metatranscriptomic data (34). The sequencing depth of segment 1 and 2 was ~250-fold and ~682-fold in mantle tissue, and ~19-fold and ~59-fold in hemolymph, respectively (SI Appendix, Fig. S2), and based on CheckV (40), the genome is estimated to be complete.

Fig. 1.

Discovery of Pacific Oyster Nidovirus 1 (PONV1) in adult Pacific oysters in British Columbia (BC), Canada. (A) Map showing the locations where wild (StnA, StnB, StnC, StnD, StnE, StnF, StnG, StnH, StnJ, and StnK) and farmed (StnI, StnL, and StnM) adult Pacific oysters were sampled. Oysters experiencing mass-mortality were sampled from two farms [farm 1 (StnI) and farm 2 (StnL and StnM)] in the Baynes Sound area of Vancouver Island in the summer of 2020 (Dataset S1). The wild oysters were sampled from ten locations (StnA-H, StnJ, and StnK) during the mortality event. The map was generated from Google Maps Platform. (B) Genomic map of PONV1. The concentric circles represent from the innermost to the outermost: i) genomic sequence; ii) predicted coding sequences (CDS) on the forward strand; and iii)–ix) various protein databases used for annotation: iii) SUPERFAMILY, iv) Pfam, v) PANTHER, vi) ProSiteProfiles, vii) SMART, viii) Conserved Domain Database (CDD), and ix) structural protein databases (via HHpred search), including the Protein Data Bank (PDB), Structural Classification of Proteins-extended (SCOPe), and the UniProt-SwissProt viral subset (UniProt). The predicted protein domains include capping 2’-O-MTase (nucleoside-2′-O-ribose methyltransferase), ExoN (Nidovirus 3’-5’ exoribonuclease), helicase, ZBD (Zinc binding domain), RdRp (RNA-dependent RNA polymerase), NiRAN (Nidovirus RdRp-associated nucleotidyltransferase), SAM-Mtase (S-adenosylmethionine-dependent methyltransferase), NADAR (NAD and ADP-ribose), YbiA-like (currently grouped into NADAR), Macro [a.k.a., ADP-ribose 1’’-phosphate processing activity (A1pp)], SH3 (Src Homology 3), Late domain-like (L domain-like), LRR (Leucine-rich repeat), and PDZ-like (PSD-95/Dlg1/ZO-1) domains.

According to estimated “abundance” (the normalized transcript representation), sequences that map to the PONV1 genome (both segments) with an average nucleotide identity (ANI) greater than 98%, were present [RPKM (reads per kilobase million) value > 0] in 20 of 33 oysters with obvious signs of disease across all three farm sites. In contrast, reads mapping to PONV1 were undetectable in “healthy” wild oysters (n = 26) sampled from ten locations, at distances ranging from a few km to >100 km north and south of the farms (Fig. 1A and SI Appendix, Fig. S3 and Dataset S1). Fisher’s exact test revealed a highly significant (P = 2.13e−07) association between viruses in the species Magarnavirus gigas (proposed) and oysters with obvious signs of disease, when compared to naturalized beach oysters with no sign of disease.

Genomic and Phylogenetic Characterization of PONV1.

The genomic landscape of PONV1 comprises two segments, each flanked by untranslated regions (UTRs) at both the 3’ and 5’ ends (Fig. 1B and SI Appendix, Figs. S4 and S7). Segment 1 encodes the replicase polyprotein through two overlapping open reading frames, ORF1a and ORF1b, with ORF1b translated via a programmed ¬1 ribosomal frameshift to produce the full-length ORF1ab polyprotein; Segment 2 encodes ten additional predicted ORFs (Fig. 1B and Datasets S2–S4). The ORF1ab polyprotein (pp1ab), resulting from the fusion of ORF1a and ORF1b, spans approximately 41,369 bp and encodes 13,789 amino acids, making it one of the largest known RNA virus polyproteins to date. Within ORF1ab, a variety of domains typically associated with nidoviruses were identified, such as RNA-dependent RNA polymerase (RdRp), helicase, 3’-5’ exoribonuclease (ExoN), Zn-binding domain (ZBD), Nidovirus RdRp-associated nucleotidyl transferase (NiRAN), 3C-like proteinase (3CPro), Macro [a.k.a., ADP-ribose 1’’-phosphate processing activity (A1pp)], NAD and ADP-ribose (NADAR), YbiA-like (currently grouped into NADAR), S-adenosyl-L-methionine-dependent methyltransferase (SAM-MTase), PSD-95/Dlg1/ZO-1 (PDZ-like), and nucleoside-2′-O-ribose methyltransferase (capping 2’-O-MTase) domains, illustrating a rich diversity of functional elements (Fig. 1B and Dataset S3). Strikingly, ORF1ab encompasses previously undocumented domains in nidoviruses, such as Src homology 3 (SH3), leucine-rich repeats (LRR), and Late domain-like (L domain-like) regions, significantly expanding our knowledge of nidovirus domain architecture.

In segment 2, eight of the ten ORFs yielded hits in HHpred search against various protein databases (Materials and Methods), with probabilities ranging from 34.5% to 93.8%, to various viral protein sequences (Dataset S5). Notably, one ORF (ORF9) showed homology to a viral fusion glycoprotein (UniProt accession: Q8V3T9), suggesting a potential structural role in virion architecture or host entry (Fig. 1B and Datasets S3 and S5).

The presence of nidovirus signature domains, such as ZBD and NiRAN, affirms the classification of PONV1 within the order Nidovirales, as determined through the RdRp-Scan pipeline (41). Further, phylogenetic analysis based on the RdRp-containing polyprotein sequences places PONV1 in a sister clade with Aplysia abyssovirus 1 (AAbV) in the Nidovirales family Abyssoviridae, in a cluster of deeply branching clades of other invertebrate-infecting nidoviruses (e.g., Mononiviridae, Euroniviridae, Roniviridae, Medioniviridae, and Mesoniviridae), and distinct from vertebrate-infecting counterparts (e.g., Coronaviridae, Arteriviridae, Nanhypoviridae, Cremegaviridae, Gresnaviridae, Nanghoshaviridae, and Olifoviridae) (Fig. 2).

Fig. 2.

Phylogenetic relationships of PONV1 with other nidoviruses. The unrooted maximum-likelihood (ML) phylogenetic tree, based on the analysis of the inferred amino acid sequences of the polyprotein containing the RNA-dependent RNA polymerases (RdRp). The phylogeny illustrates the evolutionary relationship of PONV1 to other nidoviruses found in the ICTV (version 37) and NCBI Reference Sequence (NCBI RefSeq, version 216) databases. Labels on the tree’s nodes indicate the bootstrap supporting value, with values greater than 60 being shown. Each leaf is denoted by an NCBI accession number of the respective virus (Dataset S8). Viral family affiliations are indicated by the color-coding of clades and leaves. A series of concentric color rings annotate the tree leaves, from the innermost to the outermost, representing: i) virus genome size in kilobase pairs, ii) host class, iii) virus source, iv) host habitat, and v) host group, which indicates whether the virus originated in a vertebrate or an invertebrate host. The scale bar represents the evolutionary distance along the branches of the ML tree.

The genome of PONV1 was also contrasted with other nidovirus sequences in the ICTV (version 37) and NCBI RefSeq (version 216) databases. This comparison revealed protein domains in PONV1, such as SH3, LRR, and L domain-like domains that are not found in other nidoviruses, as well as two PDZ-like domains (Fig. 3 and SI Appendix, Fig. S8) that are only known to occur in the Fathead minnow nidovirus (FHMNV; NCBI accession: NC_038295) (42) (family Tobaniviridae). In contrast, similar to other invertebrate-infecting nidoviruses, except for Xinzhou nematode virus 6 in the family Tobaniviridae, PONV1 lacks coding structures for nidovirus endoribonuclease (NendoU) (Fig. 3).

Fig. 3.

Comparative genomic overview of PONV1 and other invertebrate-infecting nidoviruses. Representatives of invertebrate-infecting nidoviruses from each family in the Nidovirales were obtained from ICTV (version 37) and NCBI RefSeq (version 216) databases. The linear representation outlines the genome organization, with the scale bar indicating the genome size in kilobase pairs. Gray arrows along the genome line represent predicted coding sequences (CDS) or polyproteins. Conserved domains within these CDS or polyproteins are depicted by colored bars, illustrating the structural and functional diversity across the nidovirus genome landscape. Similar domains between two neighboring genomes were connected using the blue panel. Abbreviations for these domains are consistent with those introduced in Fig. 1, with additional annotations for domains including Nidovirus-specific endoribonuclease (NendoU), Fibronectin type II (FN2), Torsin-1A-interacting protein 1/2 (TOIP 1/2), and Ribonuclease T2 (RNase T2) domains. The panel on the left shows the unrooted ML phylogenetic tree of these viruses, estimated using the RdRp-containing polyprotein sequences.

Screening for Putative Megarnavirus gigas in Other Oysters.

To ascertain the presence of close relatives of PONV1 in oysters from distant locations, we analyzed publicly available transcriptomes from 44 studies, spanning 13 oyster species, 54 geolocations, 14 countries, and five continents (Dataset S6). Reads were mapped to the PONV1 genome (both segments) at an average nucleotide identity (ANI) > 98%. Out of 13 oyster species, close relatives of PONV1 were detected (defined as having mapped read counts > 0 for both genome segments) in Pacific oysters from China and France, as well as in Eastern oysters (Crassostrea virginica) from eight different studies and geolocations with the “abundance” (the normalized transcript representation) ranging between 1.7 and 963.6 RPKM (Fig. 4A and Dataset S6). Sequences mapping to PONV1 were not detected in 11 other oyster species including the Hongkong oyster (Crassostrea hongkongensis), Olympia oyster (Ostrea lurida), Iwagaki oyster (Crassostrea nippona), Pearl oyster (Pinctada fucata, Pinctada margaritifera, and Pinctada imbricata radiata), Flat oyster (Ostrea edulis), Sydney Rock Oyster (Saccostrea glomerate), Australian black-lip oyster (Saccostrea sp. non-mordax lineage J), Portuguese oyster (a.k.a. Fujian oyster: Crassostrea angulata), and Mangrove oyster (Crassostrea gasar).

Fig. 4.

Detection of Megarnavirus gigas viruses in global oyster populations. (A) Prevalence of Megarnavirus gigas viruses in global oyster populations. This map shows the geographical distribution of viruses in the tentative species, Megarnavirus gigas, in different oyster species worldwide (refer to Dataset S6). The color of the circles indicates the species of oyster and where the viruses were detected, while the size of the circles reflects “abundance” (the normalized transcript representation) of Megarnavirus gigas viruses, based on the count of transcriptomic reads aligned to the PONV1 genome (both segments, ANI > 98%) and adjusted by “reads per kilobase million” (RPKM). The RPKM normalized abundance allows a semiquantitative estimate of Megarnavirus gigas viruses, facilitating comparisons across species, studies, and locations. Oyster species with undetectable RPKM values are excluded from the visualization (n = 11 species, detailed in Dataset S6). Additionally, each circle is labeled with the location, country, and study for each dataset, as well as the proportion of metatranscriptomic samples in which PONV1 was detected (PONV1-mapped reads count > 0 or RPKM-normalized abundance of PONV1 > 0), highlighted in red within the square brackets. (B) Phylogenetic relationships of PONV1-like viruses (including PONV1) to other virus families in the order Nidovirales. The rooted ML phylogenetic tree, based on the analysis of the inferred amino acid sequences of the polyprotein containing the RNA-dependent RNA polymerase (RdRp), illustrates the evolutionary histories of Megarnavirus gigas viruses and other nidoviruses found in ICTV (version 37) and NCBI RefSeq (version 216) (Datasets S7 and S8). The outgroup includes members of Picornavirales, such as Hepatitis A virus (HAV, NC_001489), Human poliovirus 1 (PV1, NC_002058), and Heterosigma akashiwo RNA virus SOG263 (HaRNAV, NC_005281). Phylogenetic clades within the tree are represented as slanted triangles, categorized by the viral families to which these nidoviruses belong (Dataset S8). The width of each triangle’s base indicates the relative diversity within the clade, with wider bases representing greater diversity. Each viral family is distinctly color-coded, with the corresponding key provided in the legend of Fig. 2. Labels on the tree’s nodes indicate the bootstrap supporting value. The tree scale bar represents the relative rate of evolution along the branches of the ML tree.

Furthermore, we assembled 12 distinct RdRp-containing nidovirus-like contigs (>1 kb; Materials and Methods) from the global oyster transcriptomes (Dataset S6) that exhibit species-level nucleotide similarity [>90% ANI across >75% alignment fraction (AF)] to PONV1 segment 1 (SI Appendix, Fig. S9 and Dataset S7). These contigs correspond to segment 1 of PONV1-like virus 1 through 11 and 15, respectively. For each virus, segment 2 was also assembled from the same metatranscriptomic sample in which segment 1 was identified (SI Appendix, Fig. S10). These PONV1-like viruses (PONV1-like virus 1 through 11, and 15), ranging in size from 12.378 to 64.310 kb (Dataset S7), were assembled from transcriptomes of adult Pacific oysters that were apparently healthy (43) (n = 8, 12.378 - 19.285 kb, from France; PONV1-like virus 1 and 3 through 9), stressed by salinity changes (44) (n = 1, 15.439 kb, from China; PONV1-like virus 11), exposed to the toxigenic dinoflagellate Alexandrium minutum (43, 45) (n = 2, 13.369 to 16.704 kb, from France; PONV1-like virus 2 and 10), or of unspecified age and health (n = 1, 64.310 kb, from China; PONV1-like virus 15) (46). Notably, PONV1-like virus 15 (64.310 kb) is assembled from the same Pacific oyster sample (NCBI SRA accession SRR15293269) in which the recently described CGNV (64.336 kb) was reported (34). Given their nearly identical genome sizes (differing by only 26 bp) and high nucleotide sequence identity across both segments (99.9% and 99.7% in the large and small segments, respectively), PONV1-like virus 15 likely corresponds to CGNV.

For several transcriptomes from Eastern oysters from the United States (47) and some Pacific oysters from China (48, 49) and France (50), in which the sequencing depths were relatively low, RdRp-containing contigs >1 kb (segment 1) were not assembled even though the RPKM-normalized abundance was > 0, and short-reads mapped to the PONV1 genome (ANI > 98%) (Fig. 4A and Dataset S6).

In addition, we assembled three other contigs >1 kb that shared species-level nucleotide similarity to PONV1 segments 1 and 2, respectively, from the diseased farmed Pacific oysters where PONV1 was identified. PONV1-like virus 12 (32.452 kb), 13 (37.023 kb), and 14 (25.546 kb) sequences were assembled from oyster samples I6, I8, and I16, respectively, from Farm 1; whereas PONV1 was assembled from oyster samples from Farm 2 (e.g., L10) (Fig. 1 and SI Appendix, Fig. S9 and Datasets S1 and S7), indicating there are other close relatives of PONV1 in Pacific oysters from North America.

To investigate whether Megarnavirus gigas (proposed) is specific to oysters, we conducted an all-versus-all BLAST (BLASTn; Materials and Methods) analysis of PONV1 and PONV1-like virus 1 through 15 against RNA virus sequences found in the ICTV (version 37) and NCBI RefSeq (version 216) databases, as well as in RNA virome databases [Serratus (36), RVMT (38), Tara Ocean (51), IMG/VR4 (52), and LucaProt (39)]. The results showed that these oyster-associated nidoviruses clustered into a single viral cluster, in this context, termed “vOTU PONV1” (i.e., proposed species Megarnavirus gigas), and are distinct at the species-level nucleotide similarity (>90% ANI across >75% AF) from other viruses in the ICTV, NCBI RefSeq, or RNA-viromic databases [Serratus (36), RVMT (38), Tara Ocean (51), IMG/VR4 (52), and LucaProt (39)], which comprise RNA viruses across a wide range of hosts, environments, and habitats. Thus, PONV1 and its close relatives appear to be specific to oysters.

Given their association with oysters, we explored the phylogenetic relationship of the RdRp-containing polyprotein sequences of PONV1 and PONV1-like virus 1 through 15. This analysis clustered PONV1 with the other 15 viruses in the species, Megarnavirus gigas (proposed), in a well-supported, long, and monophylogenetic branch with its sister clade comprising AAbV in the order Nidovirales, family Abyssoviridae (Fig. 4B). In addition, viral proteomic tree (ViPTree) (53) analysis based on the whole genome-wide translated-nucleotide sequence (e.g., protein sequences) also clustered PONV1 and PONV1-like virus 1 through 15 into one clade, with the sister clade being nidoviruses in the family Medioniviridae (SI Appendix, Fig. S11). Although the ViPTree and maximum-likelihood (ML) RdRp tree architectures are different, PONV1 and PONV1-like virus 1 through 15 are positioned in a distinct evolutionary branch within the order Nidovirales (Fig. 4B and SI Appendix, Fig. S11). Genome-wide translated-nucleotide-sequence similarity analysis also corroborates PONV1 and its 15 close relatives form a separate cluster outside of extant groups of nidoviruses (Fig. 5). Together, these findings suggest a divergent evolutionary trajectory for PONV1 and viruses in the proposed species, Megarnavirus gigas.

Fig. 5.

Genome-wide translated-nucleotide sequence similarities among nidoviruses. This heatmap illustrates the genome-wide inferred amino acid sequence similarities across various nidoviruses, including PONV1, PONV-like virus 1 through 15, and those found in ICTV (version 37) and NCBI RefSeq (version 216) (Datasets S7 and S8). Similarities were assessed using an all-to-all tBLASTx comparison, which translates nucleotide sequences into amino acid sequences across all six reading frames, applied to the predicted coding sequences of nidoviruses; analysis was conducted employing ViPTreeGen (53) with default settings. The heatmap matrix values, represented by normalized tBLASTx scores (SG), quantify the level of similarity among these inferred protein sequences across the entire genome. The color gradient from black to red represents a spectrum of similarity from low to high, respectively. Dendrograms on the left and top of the heatmap, based on Bray–Curtis dissimilarity of the SG scores, cluster the viruses to reflect their genome-wide evolutionary relationships.

Discussion

Pacific oysters are the most widely grown shellfish worldwide (1), but oyster farming is increasingly threatened by various pathogens (8, 10, 14, 16–18, 21). Our investigation into the etiology of mass-mortality in Pacific oysters in British Columbia, Canada, led to the identification of PONV1 and 15 other viruses (PONV1-like virus 1 through 15) in the tentative species, Megarnavirus gigas, in Pacific oysters from several continents. PONV1 is remarkable in being the first nidovirus reported to replicate in a bivalve, having one of the largest genomes reported for an RNA virus, and being associated with mortality (Fisher’s exact test: P < 0.001) (Figs. 1 and 4). The presence of PONV1 transcripts in mantle tissue and hemolymph demonstrates its replication in Pacific oysters. The identification of PONV1 highlights the diversity and distribution of previously unknown nidoviruses and their potential involvement in shellfish disease, as well as expanding the wide spectrum of aquatic vertebrates and invertebrates that are known to be infected by nidoviruses, including gastropods, crustaceans, fish, and mammals (28–30). As well, PONV1 puts the spotlight on the potential risks of translocating oysters and other marine organisms across wide geographic areas, when many of their pathogens remain unknown. Below, we elaborate on our evidence that PONV1 and its close relatives represent a previously unrecognized family of nidoviruses, their close association with Pacific oysters, and their relationship to disease and stress.

PONV1 and Its Conspecifics Represent the Newly Proposed Virus Family, Megarnaviridae.

PONV1 and its conspecific, the recently described CGNV (34), are remarkable in possessing the largest genomes of any described RNA virus. In addition, they are markedly different from other viruses in genome sequence and architecture. Yet, several characteristics place PONV1 and its relatives in the Nidovirales. As for all nidoviruses, PONV1 and its conspecifics have positive-sense single-stranded RNA genomes encoding RdRp, helicase, NiRAN, ZBD, 3C-like proteinase, and/or ExoN, which are essential for replication and transcription (25, 54); in particular, ZBD and NiRAN signature domains are only known in nidoviruses (25) (Figs. 1 and 3 and SI Appendix, Figs. S9 and S10 and Dataset S3). Despite having signature domains of nidoviruses, ML phylogenetic analysis of RdRp-containing polyprotein sequences (Fig. 4B) shows that PONV1 and its conspecifics are on a divergent evolutionary branch with their closest neighbor being Aplysia abyssovirus 1 (AAbV) that infects the gastropod Aplysia californica (55) and is the sole representative of the deeply branched family, Abyssoviridae. The distant relationship of PONV1-like to other nidoviruses is also supported by viral proteomic tree (ViPTree) analysis (SI Appendix, Fig. S11), and comparisons of genome-wide translated-nucleotide sequence similarity (Fig. 5), which confirm that PONV1 and its conspecifics belong to the proposed family, Megarnaviridae.

The genetic architecture of PONV1-like viruses is distinct from other nidoviruses, although there are a few features in common. Features not reported from other nidovirus genomes include leucine-rich repeats (LRR) that can be involved in host–pathogen interactions, immune evasion, and cell signaling modulation (56, 57), and sequences encoding putative viral Late domain (e.g., L domain-like domain) that are involved in virion assembly and budding (58–60). As well, there is a potential SH3 domain that can be involved in protein–protein interactions mediating host cell signaling and cytoskeletal reorganization (61, 62), which are unknown in other nidoviruses (Figs. 1B and 3 and SI Appendix, Fig. S9), and PDZ-like domains that modulate host cellular pathways, including apoptosis and immune signaling (63–65), which have only been reported in a nidovirus from a snake (42).

The presence of LRR, SH3, L domain-like, and PDZ-like domains in PONV1-like viruses suggests distinct aspects of their interaction with host cellular machinery and possibly its pathogenesis mechanism. These domains can confer enhanced modulation of host-cell signaling pathways, immune-evasion strategies, and interactions with the host cellular machinery. The SH3 domain, for example, has been shown to facilitate the recruitment of host proteins to the viral replication complex (66), while LRR domains can be involved in immune recognition and/or evasion (67). Similarly, L domain-like regions can play roles in virus assembly and release, while the PDZ-like domains may influence host-cell architecture or signaling (60, 68, 69). Speculatively, PONV1 possesses PDZ-like domains that could manipulate host cellular pathways by mimicking host PDZ domains, a strategy distinct from the PDZ-binding motifs (PBMs) detected in coronaviruses (Coronaviridae) including SARS-CoV, SARS-CoV-2, and the Middle East Respiratory Syndrome Coronavirus (MERS-CoV) (70–73), as well as in other viruses, including Human adenovirus (Adenoviridae), Human papillomaviruses (HPV, Papillomaviridae), Human T-lymphotropic virus type 1 (HTLV-1, Retroviridae), Hepatitis B virus (HBV, Hepadnaviridae), Influenza A virus (Orthomyxoviridae), Tick-borne encephalitis virus (TBEV, Flaviviridae), and Rabies virus (Rhabdoviridae) (63, 74). These viruses bind to specific host PDZ-containing proteins, subverting their activity to facilitate disease progression (63, 70, 74, 75). Unlike these viruses, PONV1 does not encode typical PDZ-binding motifs (PBMs). Instead, its PDZ-like domains may function as structural modules that mediate either direct viral protein–protein interactions or perform functional mimicry of host PDZ domains, thereby altering host-cell architecture or signaling pathways, a similar mechanism detected in another PDZ-domain-containing virus such as the Vaccinia virus in the family Poxviridae (order: Chitovirales) (68, 76). Thus, together, these domains (LRR, SH3, L domain-like, and PDZ-like domains) suggest PONV1’s pathogenic mechanisms and interactions with the host are different from those of other nidoviruses.

PONV1 and its close relative, CGNV (34), expand the reported genome sizes for RNA viruses. At approximately 64 kb, they represent the largest RNA virus genomes reported to date. This exceptional size is likely enabled by the presence of an exonuclease (ExoN) domain, which enhances replication fidelity and is thought to play a critical role in allowing the expansion of RNA genomes beyond the size limits imposed by high mutation rates (34, 77, 78). Notably, PONV1 and its conspecific, CGNV, have bisegmented genomes, with structural component encoded on a separate RNA segment, a genomic organization recently reported in other nidoviruses (34–36). We identified both genome segments of PONV1 through their consistent co-occurrence across 96 oyster samples from British Columbia (Mantel test: r = 0.912, P = 0.001), as well as from oysters sampled globally (Mantel test: r = 0.806, P = 0.001). Segment 2 consistently exhibited higher read coverage than segment 1, and was always present when the segment 1 was detected, suggesting coordinated replication and a modular genome architecture that may confer regulatory flexibility, particularly in the expression of structural genes.

Another striking feature of PONV1 and its close relative CGNV, is the absence of a recognizable nidoviral endoribonuclease (NendoU), a gene commonly found in vertebrate-infecting nidoviruses and likely involved in bypassing or evading the host’s antiviral system (79) (Fig. 3 and SI Appendix, Fig. S8). This absence, along with its segmented genome, unique protein domains, and highly divergent genome including the replication machinery, suggests that PONV1 and its conspecifics are shaped by distinct evolutionary pressures and life-history strategies (Figs. 1 and 3 and 5 and SI Appendix, Figs. S8 and S11 and Dataset S3). Collectively, these features highlight the distinct evolutionary trajectory of marine bivalve-infecting nidoviruses and underscore the extraordinary plasticity of RNA virus genomes. PONV1 and its conspecific, CGNV (34), challenge established boundaries for genome size, architecture, and replication strategies within the Nidovirales, reinforcing the importance of studying viral diversity in underexplored marine invertebrate hosts.

Viruses in the Proposed Species Megarnavirus gigas are Unique to, and Widespread in Oysters.

Pacific oysters from France and China and the Pacific coast of Canada were hosts for viruses closely related to PONV1, as shown by mapping metatranscriptomic reads to PONV1 (>98% ANI) and assembling contigs that share species-level similarity with PONV1 (>90% ANI across >75% AF) (Fig. 4 and Datasets S6 and S7); thus, Megarnavirus gigas is globally widespread in Pacific oysters (Fig. 4 and SI Appendix, Figs. S9 and S11). For example, the 64.310 kb genome of PONV1-like virus 15 (41.705 and 22.605 kb for segment 1 and 2, respectively), was assembled from the transcriptome of a Pacific oyster from Shandong province, in NE China (NCBI SRA accession number SRR15293269). This virus has striking similarity to PONV1 (64.331 kb; 41.709 and 22.622 kb for segment 1 and 2, respectively) (Fig. 4 and SI Appendix, Figs. S9 and S11 and Dataset S7), including an overall genome similarity of 95.9%, closely matched gene content and layout, and a smaller ORF1a and a larger ORF1ab that are 94.2% and 95.1% identical at the amino acid level, respectively, between the viruses. Additionally, the genomes have identical organization and domain structures, implying remarkable resemblance in their genomic and functional attributes (SI Appendix, Fig. S9).

Recently, a 64.336 kb “bisegmented” nidovirus, CGNV, was reported (34) from the same Pacific oyster sample (NCBI SRA accession SRR15293269) in which we detected PONV1-like virus 15 (64.310 kb). Similar to PONV1 (64.331 kb), CGNV has a 41.715 kb large segment (coding replicative proteins) and a smaller segment of 22.621 kb (coding structural proteins) (34). CGNV likely corresponds to PONV1-like virus 15, with 99.9% and 99.7% nucleotide sequence identity in the large and small segments, respectively, and thus also belongs to the species Megarnavirus gigas.

The species Megarnavirus gigas appears to be largely confined to Pacific oysters, as their sequences were undetectable in metatranscriptomic datasets from other species of oysters, with the exception of a few reads (~4.7 RPKM) from Eastern oysters (C. virginica) collected in the northeastern United States (47) that aligned with PONV1 (>98% ANI) (Fig. 4A). However, the evidence for Megarnavirus gigas in Eastern oysters (C. virginica) is inconclusive given the low sequence coverage and inability to assemble contigs >1 kb with similarity to PONV1.

Furthermore, interrogation of sequence and transcriptomic databases revealed no species-level genome similarity (>90% ANI across >75% AF) between PONV1 and its close relatives, and other RNA viruses in the ICTV, NCBI RefSeq, and RNA viromic databases [Serratus (36), RVMT (38), Tara Ocean (51), IMG/VR4 (52), and LucaProt (39)]. Given the expansive range of hosts and environments in these datasets, the finding implies a coevolutionary relationship between Megarnavirus gigas and oysters.

The Enigma of Megarnavirus gigas and Disease in Pacific Oysters.

Our investigations into potential pathogens associated with mass-mortality events in farmed Pacific oysters led to the identification of an unusual nidovirus, PONV1. Oysters collected from three farm sites (belonging to two farms) in British Columbia, Canada, revealed replication of PONV1 and closely related viruses in 20 of 33 oysters with obvious signs of disease, but not in 26 apparently healthy “wild” oysters sampled from ten locations ranging in distance from a few kilometers to more than 100 km away from the farms. If the wild oysters are considered as controls, albeit imperfect controls given the differing growth conditions, there is a significant association of Megarnavirus gigas with disease (Fisher’s exact test: P < 0.001).

The association of Megarnavirus gigas with disease is more tenuous in Pacific oysters from Asia and Europe. Although disease was not reported in other studies, replication of Megarnavirus gigas viruses was found in Pacific oysters affected by environmental stressors (44) and exposure to the toxigenic dinoflagellate Alexandrium minutum (43, 45) (Fig. 4A and Dataset S7), which is consistent with observations that stress increases the likelihood of infection by some nidoviruses, such as serpentoviruses (80), as well as in seemingly healthy individuals (43, 49, 50) (Fig. 4A and Datasets S6 and S7). Similarly, replicating Megarnavirus gigas viruses were only detected in a subset of diseased oysters from British Columbia. This does not demonstrate that Megarnavirus gigas does not cause disease in oysters, as nidoviruses that cause disease have been observed in both diseased and healthy animals, including pythons (81), lizards (82), and turtles (83). The presence of nidoviruses that cause disease in seemingly healthy individuals can occur if animals are sampled at an early stage of infection or during recovery from infection (81, 84).

Another complication is that nidoviruses are not necessarily detected in all individuals during mortality events in which they are involved, such as observed in captive veiled chameleons (85), Bellinger River snapping turtles (83), and farmed Trionyx sinensis turtles (86). This can be the result of coinfection, or other factors (25, 87, 88), and is consistent with our observation that in 13 out of 33 diseased farmed oysters, Megarnavirus gigas was not detected (SI Appendix, Fig. S3). Although we may have missed the timing of nidovirus replication in some of the oysters, it is also possible that multiple factors contributed to the mass-mortality of Pacific oysters, a concern highlighted by other researchers (25) and supported by evidence from studies on Pacific (19) and Hong Kong (89) oysters. Thus, the detection of PONV1 and PONV1-like virus 12 through 14 in oysters during a mass-mortality event across three farm sites in British Columbia, does not demonstrate that Megarnavirus gigas causes disease. However, the results are provocative and highlight the need to investigate the relationship between these viruses and disease in Pacific oysters, and emphasize the potential risk to domesticated and wild oyster populations from the routine practice of shipping oyster spat internationally from hatcheries to oyster farmers.

Additionally, to determine whether other RNA viruses were linked to the mortality event, we examined another 188 RdRp-containing RNA-virus OTUs identified from all 96 oyster samples collected in British Columbia, including those in which PONV1-like viruses were detected. While multiple RNA viruses were detected, most were related to viruses associated with dietary or environmental sources (e.g., algal viruses) rather than oyster-infecting viruses. Notably, we detected two birnaviruses closely related to Pinctada birnavirus TM-2022 (NCBI accession: LC765996), which has been associated with summer atrophy (90) and potential mass mortality (91) in Japanese Pearl oysters. However, these birnaviruses occurred in low abundance (<4 RPKM) and were only found in two diseased oyster samples from British Columbia; in contrast, PONV1 was consistently detected in 25 samples from 20 diseased oysters and exhibited a strong correlation with mortality patterns (Fisher’s exact test: p = 0.001). These findings support PONV1 as a plausible cause of mortality. While causality remains to be demonstrated, our results highlight the need for future studies to investigate the physiological impacts of PONV1 infection, its transmission dynamics, and interactions with environmental stressors. International collaboration, particularly with regions experiencing similar oyster mortality events, will be essential to assess the broader distribution, host range, and potential economic impact of this virus, and, if necessary, inform global aquaculture biosecurity strategies.

In conclusion, PONV1 and its conspecific viruses including CGNV are the only nidoviruses reported to replicate in bivalves. This finding broadens the known evolutionary range of nidoviruses, arguably one of the most important groups of viruses associated with disease in animals. Moreover, its association with mass mortality in Pacific oysters, a keystone marine species (92), and the most widely cultivated and economically important shellfish (1), emphasizes the need for laboratory challenge studies to determine whether Megarnavirus gigas viruses cause disease in Pacific oysters. The association of PONV1 with oyster mortality, and the detection of closely related viruses in oysters from Europe, Asia, and North America emphasizes the need to consider a precautionary approach (93) when translocating shellfish internationally or across large geographic distances, particularly when the causative agents of disease in shellfish and their impact on wild populations remain largely unknown.

The identification of PONV1, with one of the largest genomes reported for an RNA virus, heralds a previously unknown branch of nidovirus evolution. As shown by a distinct genetic architecture and composition that encodes SH3, LRR, L domain-like, and PDZ-like domains, which are either unknown or very rarely observed in nidoviruses, as well as phylogenetic and ViPTree analyses, PONV1 represents a highly divergent group of nidoviruses. This branch of nidovirus evolution appears to be restricted to oysters, and defines the proposed family, Megarnaviridae, in the Nidovirales.

Materials and Methods

Sample Collection.

During July and August 2020, two- to three-year-old adult Pacific oysters, Crassostrea (Magallana) gigas, with most showing symptoms of disease (e.g., gaping shells and tissue necrosis) were sourced from two aquaculture farms (Station I on Farm 1 and Stations L and M on Farm 2) operating in the Baynes Sound area of Vancouver Island, BC, Canada (Fig. 1A and Dataset S1). As well, beach oysters without evidence of disease were collected from ten locations (Station A−H, J, and K) at various distances from Baynes Sound. At each farm site or location, between 2 and 17 oysters were collected (SI Appendix, Fig. S1); each oyster was dissected on-site; mantle tissue and, when possible, hemolymph were collected using sterile instruments (SI Appendix, Extended Materials and Methods). Samples were flash-frozen in liquid nitrogen and stored at -80 °C until processing.

RNA Extraction and Sequencing Library Construction.

Total RNA extraction was performed using the Direct-zol RNA Miniprep Plus kit (Zymo Research) by following manufacturer instructions. Then, ribosomal RNA (rRNA) was reduced in the samples with the Illumina Ribo-Zero Plus rRNA Depletion Kit (Illumina), and the sequencing library was constructed using the NEBNext Ultra RNA Library Prep Kit (NEB) (SI Appendix, Extended Materials and Methods). Sequencing libraries were pooled and sequenced at Genome Québec using the Illumina NovaSeq 6000 platform (S4 flow cell, and 2 × 150 bp PE mode).

Sequence assembly and viral sequence detection.

Raw reads for RNA-Seq samples were quality-controlled and trimmed using Trimmomatic v.0.38 (94), followed by de novo assembly using SPAdes v.3.15.2 (95) and MEGAHIT v.1.2.9 (96), respectively (SI Appendix, Extended Materials and Methods). The contigs from both the SPAdes and MEGAHIT assembly pipelines were combined and deduplicated using the dedupe.sh script from BBTools (https://jgi.doe.gov/data-and-tools/software-tools/bbtools/), generating 10,373,790 deduplicated contigs (>1 kb) across all samples. RNA virus sequences were identified using the RdRp-scan pipeline (41) (https://github.com/JustineCharon/RdRp-scan), yielding 213 deduplicated RdRp-containing viral contigs (>1 kb). These viral contigs were then clustered at >90% average nucleotide identity (ANI) across >75% of the alignment fragment (AF) using CheckV scripts (40) (https://bitbucket.org/berkeleylab/checkv/src/master/) to generate a nonredundant set of species-level viral populations (vOTUs). This yielded 189 RNA-virus vOTUs with genome sizes between 1,007 and 41,709 bp (N50 = 3,294 bp) that were checked for completeness with CheckV v.0.7.0 (40). The contigs, which were present in diseased farmed oysters but absent in healthy wild oysters, included a 41,709 bp nidovirus-like sequence that was analyzed in depth.

Designated PONV1 segment 1, this 41,709-bp contig encodes an RNA-dependent RNA polymerase (RdRp) and other replication-related proteins (e.g., NiRAN, helicase, ExoN) at the 3′-proximal region, but lacks identifiable structural domains. These features, and the discovery of the closely related CGNV (34), suggested that the genome of PONV1 is segmented, with structural components encoded on a separate segment, as shown for bisegmented coronaviruses in fishes (35, 36). To identify potential structural segments, we analyzed 81,030 nonredundant contigs (>1 kb) de novo assembled from the same sample (L10; ENA accession: ERR13334412) from which PONV1 segment 1 was recovered, using a similar approach as described previously (34) (SI Appendix, Extended Materials and Methods). This led to the identification of a 22,622-bp contig (PONV1 segment 2) that significantly co-occurred with segment 1 across all 96 oyster samples (Mantel test: r = 0.912, p = 0.001). PONV1 segment 2, sharing high similarity (96.6% ANI) with CGNV segment 2, contains 10 ORFs, including one with homology to a viral fusion glycoprotein, supporting its structural role. Accordingly, both segments (PONV1 segment 1 and 2) were subjected to in-depth analysis.

Virus Annotation.

Putative coding sequences (CDSs) of PONV1 (both segments) were identified using ORFfinder v.1.8 (parameters: gene code = 1; https://www.ncbi.nlm.nih.gov/orffinder/) for fragments exceeding 600 bp, and analyzed in Geneious Prime® 2023.2.1. Annotation of these CDSs, including the identification of conserved domains within the polyprotein, was carried out using NCBI Conserved Domain Search (NCBI CD-Search) and InterProScan version 5.59-91.0 (97), integrating multiple protein domain databases (SI Appendix, Extended Materials and Methods). HHpred searches (https://toolkit.tuebingen.mpg.de/tools/hhpred) against a suite of structural protein databases were used to resolve putative structural protein features. Potential ribosomal frameshifting sites (RFS) in PONV1 were identified using KnotInFrame (98). RNA secondary structure of the UTR of PONV1 at the 3’ and 5’ end, respectively, were predicted using RNAfold (99).

Profiling Abundance of Megarnavirus gigas Viruses Detected in Transcriptomes of Pacific Oysters.

To profile the “abundance” (the normalized transcript representation) of Megarnavirus gigas viruses across oyster samples collected in British Columbia (BC), the rRNA-depleted and quality-controlled (QC) transcriptomic reads were mapped to the PONV1 genome (both segments) and sequences of other oyster-associated RNA viruses (as shown above) using BBmap v.38.86 (100) (parameter: nodisk pigz = t usejni = t minid = 0.98). Reads that mapped to PONV1, but not to other oyster-associated RNA viruses, were considered as sequences from Megarnavirus gigas viruses. Sequencing coverage or depth of PONV1 was calculated using the jgi_summarize_bam_contig_depths script from the MetaBAT2 pipeline v.2.15 (101) (parameters: --minContigDepth 2). For each of the analyzed transcriptomic samples, the reads per kilobase million (RPKM) normalization of these PONV1-mapped reads was performed using CoverM v.0.6.1 (https://github.com/wwood/CoverM) with default parameters. This RPKM-normalized abundance allows relative virus loads to be estimated and, thus, facilitates the distribution of Megarnavirus gigas viruses to be compared across oyster samples.

The presence of Megarnavirus gigas is determined based on the detection of reads mapped to both segments of the PONV1 genome, with RPKM-normalized abundance values greater than zero, showing signs of virus detection. Although sequencing depth can vary across samples, potentially affecting RPKM-normalized abundances, this variability does not influence the assessment of the presence of Megarnavirus gigas. Our assessment is based not only on mapping reads to the PONV1 genome (ANI > 98%), but also on the assembly of contigs very similar to sequences found in PONV1 (>1 kb; see following section ‘Exploration of Megarnavirus gigas across locations’). These contigs are further validated through phylogenetic analysis (see below section “Phylogenetic analysis”), which confirms their similarity to PONV1.

Exploration of Megarnavirus gigas Across Locations.

To explore the global distribution of Megarnavirus gigas, sequences (>1 kb) with high similarity to PONV1 were examined across transcriptomic databases (Dataset S6) containing 848 transcriptomes from 44 studies, involving 13 oyster species from 54 locations in 14 countries on five continents. Briefly, post-QC reads were assembled using MEGAHIT v.1.2.9 (96), and RdRp-containing contigs (>1 kb) were identified using the RdRp-scan pipeline, as described above (SI Appendix, Extended Materials and Methods). Contigs assigned to Nidovirales were considered “nidovirus-like contigs”, and were extracted using Seqkit v.2.2.0 (102). After a clustering analysis using the CheckV scripts (40), those clustered with PONV1 segment 1 (>90% ANI across >75% AF) were considered to be from the same species as PONV1, Megarnavirus gigas (tentative).

To retrieve segment 2 of Megarnavirus gigas viruses, the genomic sequence of PONV1 segment 2 was used as a query in a BLASTn search (p < 1e-5) against all nonredundant contigs de novo assembled from each metatranscriptome. Contigs showing >90% sequence identity to PONV1 segment 2 were retained. Within each metatranscriptome, contigs were deduplicated using the dedupe.sh script from BBTools (https://jgi.doe.gov/data-and-tools/software-tools/bbtools/) and subsequently passed through the same CheckV clustering scripts (40), as above, to identify a representative sequence similar (>90% ANI across >75% AF) to PONV1 segment 2 per sample. Gene prediction and functional annotation of the contigs were performed as outlined above under “Virus annotation”.

The presence/absence and “abundance” (the normalized transcript representation) of Megarnavirus gigas viruses across oyster samples and/or locations were assessed as described above under “Profiling abundance of Megarnavirus gigas viruses detected in transcriptomes of Pacific oysters” (SI Appendix, Extended Materials and Methods). The presence of Megarnavirus gigas viruses was inferred when reads mapped (>98% ANI) to both segments of the PONV1 genome, with RPKM-normalized abundance values greater than zero for each segment, indicating the presence of the virus. Sequencing coverage or depth of the Megarnavirus gigas segments was calculated using the jgi_summarize_bam_contig_depths script from the MetaBAT2 pipeline v.2.15 (101), as above. The overall abundance and sequencing coverage/depth of Megarnavirus gigas were calculated as the average values across its two genome segments.

Phylogenetic Analysis.

To elucidate the evolutionary relationship of PONV1 to other nidoviruses, we conducted phylogenetic analysis of the RdRp-containing polyprotein sequence from PONV1, and representative nidoviruses from the ICTV (version 37) and NCBI RefSeq (version 216) databases (n = 133; Dataset S8). The RdRp amino acid sequences were aligned using MAFFT version 7 (103) and trimmed using Clipkit v.1.4.1 (104) for phylogenetic reconstruction (SI Appendix, Extended Materials and Methods). ML phylogenies were conducted using IQ-TREE v.2.2.0.3 (105) employing the best-fit model and 1000 bootstrap replicates. The evolutionary context of PONV1-like virus 1 through 15, n = 15; Dataset S7) was also analyzed in the same way.

Comparative Analysis of Nidoviruses.

To enable comparative analysis of Megarnavirus gigas viruses (PONV1 and PONV1-like viruses 1 through 15), which possess bisegmented genomes, we concatenated the two segments of each virus into a single contiguous sequence using Geneious Prime® 2023.2.1. Segment 2 was appended to the end of segment 1, with one to three “N” nucleotides inserted at the junction. The number of inserted N’s was chosen to preserve the original reading frame of segment 2, thereby maintaining its predicted coding potential. This generated a single composite contig per virus, facilitating the following comparative analyses with other RNA viruses.

Species-level genome similarity (>90% ANI across >75% AF) among oyster-associated nidoviruses (PONV1 and 15 other viruses in the proposed species, Megarnavirus gigas) was compared with RNA viruses from different hosts or habitats cataloged in the ICTV (version 37) and NCBI RefSeq (version 216) databases, as well as those identified from publicly available RNA viromic databases [Serratus (36), RVMT (38), Tara Ocean (51), IMG/VR4 (52), and LucaProt (39)] was assessed using clustering analysis and all-versus-all BLAST (parameters: BLASTn, -max_target_seqs 10000 -perc_identity 90). This analysis, which applied a threshold of >90% average nucleotide identity (ANI) across >75% alignment fraction (AF), was performed using the CheckV script (40) as previously described.

Genome-wide translated-nucleotide sequence similarities among Megarnavirus gigas members (PONV1 and 15 other viruses in the species, Megarnavirus gigas), and other nidoviruses in the ICTV (version 37) and NCBI RefSeq (version 216) databases (n = 133; Dataset S8) were assessed using ViPTreeGen v.1.1.3 (53) with default parameters. The analysis employs an all-versus-all tBLASTx (E-value = 0.01) comparison, which translates nucleotide sequences into amino acid sequences across all six reading frames, applied to the predicted coding sequences of nidoviruses (gene code = 1). Furthermore, leveraging ViPTreeGen v.1.1.3 (53) using default settings, a viral proteomic tree (ViPTree) incorporating both the 16 viruses belonging to Megarnavirus gigas, as well as other nidoviruses from the ICTV (version 37) and NCBI RefSeq (version 216) was also generated, providing insights into their genome-wide evolutionary relationships (SI Appendix, Extended Materials and Methods).

Statistical Analysis.

To evaluate the association between the presence of viruses in the species Megarnavirus gigas and oyster mass-mortality events, we conducted a Fisher’s exact test. The test was applied to a 2x2 contingency table that compared the number of virus-positive oysters (the RPKM-normalized abundance of Megarnavirus gigas viruses > 0) and virus-negative oysters (the RPKM-normalized abundance of Megarnavirus gigas viruses = 0) in farmed oysters exhibiting signs of disease during mass mortality events (N_{Virus-positive} = 20, N_{Virus-negative} = 13) and in healthy wild oysters from nonmortality locations (N_{Virus-positive} = 0, N_{Virus-negative} = 26). The test was performed using the fisher.test() function in R version 4.2.1 (106), with p < 0.05 considered statistically significant. Fisher’s exact test was selected due to the small sample sizes and the binary nature of the data (presence or absence of Megarnavirus gigas viruses).

To determine the co-occurrence or abundance correlation between PONV1 segment 1 and other contigs (including PONV1 segment 2) across oyster samples, a Mantel test (parameters: method = “pearson”, permutations = 999) was performed on Euclidean distance matrices derived from their abundance profiles across 96 oyster samples from BC (Dataset S1), as well as global oyster samples from various locations (Dataset S6). The analysis was conducted using the R package Vegan v.2.5 (107), and results with P < 0.05 were considered statistically significant (SI Appendix, Extended Materials and Methods).

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (PDF)

Dataset S02 (PDF)

Dataset S03 (PDF)

Dataset S04 (PDF)

Dataset S05 (PDF)

Dataset S06 (PDF)

Dataset S07 (PDF)

Dataset S08 (PDF)

Data, Materials, and Software Availability

The genome sequences of PONV1 and other Megarnavirus gigas viruses (PONV1-like virus 1 through 15) were deposited in GenBank under accession numbers PQ030832 (108), PV579160 (109), BK068944 to BK068958 (110), and PV928314 to PV928328 (111), respectively. Raw sequencing reads generated by this study were deposited in the European Nucleotide Archive (ENA) under the accession numbers ERR13334386 to ERR13334481. The metadata of the raw sequence file can be found in Dataset S1. Other supplementary materials, including the 189 RdRp-containing contig sequences, phylogenetic trees, and relevant codes, can be found in Figshare via the link https://doi.org/10.6084/m9.figshare.27264522 (112). All study data are included in the article and/or supporting information.