Viral metagenomic investigation of two Caribbean echinoderms, Diadema antillarum (Echinoidea) and Holothuria floridana (Holothuria)

Introduction

Echinoderms play significant ecological roles in coral reef ecosystems as herbivores and detritivores (Birkeland, 1989). In coral reef ecosystems, echinoids, which are one group of echinoderms, contribute to the balance between macroalgae and coral cover by removing algae and opening space for coral recruits (Edmunds & Carpenter, 2001) and reducing dissolved organic matter inputs, disfavoring coral pathogens (Haas et al., 2016; Nelson, Wegley Kelley & Haas, 2023). Another group of echinoderms, holothurians, contribute to biogeochemical cycling through detritivory and deposit feeding, influencing remineralization of sediment-bound organic material (Uthicke, 2001a; Uthicke, 2001b; Williamson et al., 2021). Both echinoids and holothurians face considerable threats to their population densities through disease (Lessios et al., 1984; Delroisse et al., 2020) and harvesting (Purcell et al., 2013; Wolfe & Byrne, 2022), which may alter coral reef ecosystem structures in the future. Hence, there is a great need to understand factors potentially influencing their mortality, including identifying potential pathogens and other stressors that may contribute to host-microbe interactions.

The long-spined sea urchin (Diadema antillarum) is a crucial constituent of Caribbean reefs that has experienced two distinct mass mortality events in the past 50 years. Between 1983 and 1984, around 98% of D. antillarum across the Caribbean died (Lessios, 1988; Lessios, 1995; Lessios, 2016), which contributed to reef degradation through alleviated herbivore pressure on macroalgae. Macroalgae compete with corals for space, light and nutrients and their proliferation severely inhibited coral recruitment (Mumby, Hastings & Edwards, 2007). Unfortunately, the etiology of the early 1980s die-off was never determined, and no abnormal specimens exist today. In early 2022, another mass mortality event affected D. antillarum, beginning on the western side of St. Thomas (U.S. Virgin Islands) and progressing across most of the windward islands before dissipating in late summer and fall 2023 (Hylkema et al., 2023). The 2022 mass mortality was caused by a scuticociliate most closely related to Philaster apodigitiformis (Vilanova-Cuevas et al., 2023). This pathogen colonized spines, body wall, tube feet, and eventually the coelomic cavity of affected specimens (Hewson et al., 2023). This highly deleterious pathogen was first identified through host tissue transcriptomics (Hewson et al., 2023).

The sea cucumber Holothuria floridana (cf mexicana) inhabits Caribbean coral reefs. This species is amongst the most abundant holothurians on reef flats (Guzmán & Guevara, 2002), where individuals are less mobile than predicted by random walk models, suggesting they are constrained by habitat heterogeneity (Hammond, 1982). H. floridana consumes more sediment than sympatric species and contributes to nitrogen cycling through release of ammonium at rates higher than co-occurring holothurian taxa (Munger et al., 2023). Its fecal pellets may persist for longer than other species (Conde, Díaz & Adriana, 1991). As with most coastal holothurians, H. floridana hosts a diverse assemblage of epibionts, including crustacea, annelids, and mollusks (Rogers et al., 2024). The virome composition of H. floridana has not been previously assessed. However, holothurians generally are known to harbor diverse viruses (Hewson, Johnson & Tibbetts, 2020; Jackson et al., 2022; Wang et al., 2021a; Wang et al., 2021b), some of which may be associated with disease processes (e.g., skin ulceration disease and visceral ejection syndrome; Deng et al., 2008). Most viral metagenomic approaches have not been followed by study of their pathology, and as such, it is not possible to assess their roles in holothurian health. Recently, we described a novel flavivirus infecting Apostichopus californicus, and anticipated that these may be common amongst sea cucumbers (Hewson, Johnson & Tibbetts, 2020). However, surveys across habitats and between species revealed a tight association between this virus and its host, and the virus was only found in the geographic area where it was originally observed. This suggests holothurian viruses may have restricted ranges (Crandell et al., 2023). Furthermore, the flavivirus did not appear related to host mass or tissue protein content, and only weakly and positively with tissue lipid content, so its role in host health remains unclear (Crandell et al., 2023).

Viruses infect every phylum of life and play crucial roles in marine food webs by releasing dissolved organic materials, regulating population densities of dominant resource competitors, and facilitating gene exchange (Breitbart, 2012; Suttle, 2005; Suttle, 2007; Wilhelm & Suttle, 1999). Viral metagenomics has elucidated an astonishing array of viral diversity both as free agents in seawater and sediments, and in association with hosts (Breitbart et al., 2004; Breitbart et al., 2002; Jackson et al., 2022). In addition, mining of invertebrate transcriptomes has expanded the host range of many viral taxa and allowed identification of putative pathogenic agents (Shi et al., 2016). Most viruses identified by genomic surveys of animal tissues are not believed to be pathogenic in these hosts, yet investigations provide important information for downstream autecological study (Gudenkauf et al., 2014; Hewson et al., 2014).

This study surveilled viruses of H. floridana and D. antillarum as part of a wider effort to understand the roles of their associated microbiomes in animal health. The impetus for this survey was to identify candidate viral agents which could be used, downstream, in quantitative assessment of their association with tissue health. We surveyed viral metagenomes of D. antillarum that were grossly normal and those affected by D. antillarum scuticociliatosis (DaSc) collected from St. Thomas, St. John (U.S. Virgin Islands) and Saba (Caribbean Netherlands) at reference and DaSc-affected sites. We also surveyed viral metagenomes from grossly normal H. floridana near Marathon, Florida. Both species were targeted because of their important roles in coral reef ecology.

Materials & Methods

Results

Diadema antillarum viral metagenomes and viral surveillance of transcriptomes

BLASTn/BLASTx comparison of the D. antillarum contigs against viral genomes and proteins yielded only three contigs from individual libraries (all matching the microvirus Sinsheimervirus phiX174, which is used as an internal spike during library preparation and therefore a contaminant), whereas in the global assembly, five contigs matched Caudoviricetes (42–100% similarity to sequences in Genbank by BLASTn) and one contig matched a totivirus infecting the fungus Malassezia restricta (100% similarity). Annotation of D. antillarum viral metagenomic reads and host transcriptomes (Hewson et al., 2023) via Kraken2 revealed between 0.105–0.646% of viral metagenomes and 0.037–2.993% of transcriptome reads matching viral genomes. Amongst these matches were well-represented bacteriophage, archaeal viruses, and unclassified viruses, including both dsDNA and ssDNA viruses (Fig. 1). Retroviruses were also well represented in both viral metagenomes and transcriptomes but may represent endogenized retroviruses or expressed reverse transcriptases that are homologous between viruses and cellular organisms. Amongst RNA viruses, Pisoniviricetes (positive sense ssRNA viruses, including picornaviruses) were well represented in transcriptomes, but were less well represented in viral metagenomes. Furthermore, Ellioviricetes (negative sense ssRNA viruses, including bunyaviruses) were also more prevalent in both transcriptomes and viral metagenomes than other viral families. The detection of reads matching RNA viruses, but not contigs, suggests that the low number of viral reads may have prevented their assembly and representation amongst contig spectra.

Holothuria floridana RNA viral metagenomes

Contig spectra from H. floridana were dominated by Monodnaviria (Cressdnaviricota, Petitvirales; 66.8% in individual libraries; 67.2% in global assembly), followed by double stranded DNA bacteriophage (25% in individual libraries; 24.7% in global assembly). The remaining sequences matched one Mouviricetes (global assembly), a Magsaviricetes (1 each in FKSCJan and global assembly; Fig. 2), and Pisoniviricetes (7.7% of individual and 6.5 in global assembly; Fig. 3).

Ribosomal RNAs in viral metagenomes

Comparison of contig sequences to the SILVA database (accessed May 2023, release 138.1) revealed that 736 contigs matched eukaryotic and bacterial ribosomal RNAs. These contigs matched several putative sequencing contaminants including Pseudomonas, Klebsiella, and Meiothermus. Marine bacteria that were well represented in SILVA hits included Vibrio spp., Halarcobacter, and Terasakiella and Halodesulfovibrio. Eukaryotic 18S and 28S rRNA genes yielded four contigs from both global and individual assemblies matching the 18S rRNA gene of the DaSc-associated Philaster clade, which was an expected constituent of the microbiomes of DaSc-affected urchins, the ciliate Euplotes, as well as basidio- and ascomycete fungi (Fig. 4), including Malassezia restricta, which may constitute a sequencing contaminant.

Discussion

The overarching goal of the present study was to surveil viruses associated with grossly normal and abnormal D. antillarum, in an effort to identify candidate pathogenic agents associated with mass mortality, which preceded conclusion that D. antillarum scuticociliatosis (DaSc) was caused by the DaSc Philaster clade ciliate (Hewson et al., 2023; Vilanova-Cuevas et al., 2023). Our results provide several useful observations for future investigation of echinoderm viruses and perhaps in the context of investigating future marine mass mortality events. First, our data suggest that the echinoid D. antillarum, whether or not affected by DaSc, and across several tissue types, does not have the same detectability of viruses as H. floridana or other echinoderms that have been examined in prior work (Gudenkauf et al., 2014; Hewson et al., 2018a; Hewson et al., 2014; Jackson et al., 2022). Second, our data show that H. floridana is inhabited by viral orders that have been observed previously in other holothurian species (Hewson, Johnson & Tibbetts, 2020). Finally, our work suggests that viral metagenome preparation strategy may influence the detection of RNA viruses, and that few viruses detected by viral metagenomics are found in transcriptomes prepared from the same specimens.

We expected to discover prominent and common RNA viruses in D. antillarum viral metagenomes based on prior surveys of sea stars (Gudenkauf et al., 2014; Hewson, Aquino & De Rito, 2020; Hewson et al., 2014; Hewson & Sewell, 2021; Jackson et al., 2020), sea cucumbers (Hewson, Johnson & Tibbetts, 2020) and those recovered from transcriptomes (Jackson et al., 2022). Surprisingly, we did not recover any RNA viral contigs, and only a handful of DNA viral contigs, in both DaSc-affected and grossly normal urchin tissues. While we observed sequencing reads within libraries matching RNA viral genomes, they were less pronounced in viral metagenomes than in transcriptomes. Given that the large number of host sequence matches in read libraries was similar between both D. antillarum and H. floridana, the lack of viral sequences was not due to poor sequence coverage alone.

Ribosomal RNAs may be copurified with virus particles using the approach employed in this study, which have also been used to identify common protists associated with sea stars (Hewson & Sewell, 2021). We observed marine bacteria, marine ciliate and fungal rRNAs amongst viral metagenomes (Fig. 4). These results further emphasize that the low representation of viruses within D. antillarum viral metagenome contig spectra may be partially due to the extensive cellular microbial RNA and DNA present in material used to prepare viral metagenomes, possibly due to coextracted compounds that interfere with nucleases used to prepare viral metagenomes.

In contrast to D. antillarum viral metagenomes, H. floridana bore viral taxa that have been previously observed in sea cucumbers (Hewson, Johnson & Tibbetts, 2020) and more widely in echinoderm viral metagenomes (Hewson, Aquino & De Rito, 2020; Hewson et al., 2018b; Jackson et al., 2022; Shi et al., 2016). Pisoniviricetes sequences from H. floridana were most similar to picornaviral genomes recovered from transcriptome surveys of aquatic invertebrates (Shi et al., 2016). When compared to the TSA archive at NCBI, sequences matched at 37–67% nucleotide identity to transcriptome assembled fragments from sea cucumbers (Holothuria scabra, Holothuria glabberina, Patiria miniata, and Apostichopus japonicus) (Fig. 5; Table S2). Within the Pisoniviricetes, one sequence placed closest to Labyrnavirus, while others were most similar to viruses recovered from crustacean transcriptomes (Fig. 3). Pisoniviricetes have been recovered from a variety of habitats including free in plankton (Culley, Lang & Suttle, 2003; Culley, Lang & Suttle, 2006; Moniruzzaman et al., 2017; Solomon & Hewson, 2022), sediments (Zhang et al., 2024) and associated with a myriad of invertebrates (Shi et al., 2016). Their roles in disease are unclear since they are typically recovered from grossly normal specimens. Comparison of positive-sense ssRNA viruses recovered in this study against assembled shotgun transcriptomes revealed similarly to those recovered from other sea cucumbers, but at least two contigs matched viruses recovered from the chytrid Schizotrichium sp. Hence, while most sea cucumber ssRNA viruses recovered in this study likely inhabited echinoderm tissues, some may also infect co-occurring eukaryotic microorganisms that form part of the echinoderm microbiome.

The Magsaviricetes sequences recovered from H. floridana were most closely related to sequences retrieved from surveys of uncultivated crustacean viruses. The Magsaviricetes genome fragments recovered in the H. floridana survey fell within a group of viruses discovered through metagenomics, including those from grasshoppers, mosquitoes, and ants, as well as those recovered from environment surveys of estuarine (riverine) sediments and water (Fig. 2). Magsaviricetes infecting shrimp have been reported in the sea cucumber Apostichopus japonicus, suggesting that they are a secondary host (Wang et al., 2021b), and cause significant histopathologic changes in internal organs during infection (Wang et al., 2021a). However, as no gross abnormalities were noted in the surveyed specimens, it is unclear whether Magsaviricetes in H. floridana are infectious.

Double and single-stranded DNA viruses were also detected in this study, which may have resulted from active transcription of viral genes, or because ssDNA is not effectively digested by nucleases used to prepare viral metagenomes. Amongst Monodnaviria, Petitvirales (Malgrandaviricetes; Microviridae) infect bacteria and comprised a large number of contigs. It is worthwhile to note that their large representation in the contig spectra was not a consequence of amplification biases as noted in previous works (Dunlap et al., 2013; Hewson et al., 2013) since our approach did not employ rolling circle amplification via phi29 polymerase. These viruses likely infect microbiome constituents, since all currently described microviruses infect intracellular parasitic bacteria, Enterobacteriaceae, and Spiroplasma (Kirchberger & Ochman, 2023; Roux et al., 2012). In addition to Petitvirales, Cressdnaviricota were recovered from both individual and global assemblies of H. floridana. Cressdnaviricota have been reported to occur in a variety of metazoan hosts (Rosario, Duffy & Breitbart, 2012; Rosario et al., 2015), including echinoderms (Jackson et al., 2016) and may represent metazoan infections. However, to date there remains only scant evidence that any Cressdnaviricota cause disease in invertebrates.

While many dsDNA phage sequences (n = 98 in H. floridana and 5 in D. antillarum) were recovered, few could be assigned taxonomically beyond Caudoviricetes. In D. antillarum, all phage fragments could not be confidently distinguished from putative contaminants in the Enterobacteriaceae. In H. floridana, bacteriophage sequences matched most closely to those infecting marine hosts. The wide detection of phage sequences both within contig spectra and in individual read libraries may indicate that viruses of microbiome constituents are common, and that their large proportion of sequence space may prevent observation of less abundant metazoan viruses.

The lack of overlap between contigs assembled from viral metagenomes and transcriptomes may indicate that viral particle purification can affect overall viral composition when compared to bulk RNA extracts. Hence, transcriptome mining studies (e.g., Jackson et al., 2022) may reflect different viral taxa than those recovered by purifying and sequencing virus particles.

It is unclear why Pisoniviricetes (and other ssRNA viruses) were recovered by the Sigma Aldrich TransPlex WTA2 kit across several specimens, but not the Sigma Aldrich SeqPlex kit, since both employ similar chemistries. The Seqplex kit first reverse transcribes RNA using manufacturer-proprietary semi-degenerate 3′- and universal 5′-ends, whereas the TransPlex WTA2 kit uses complementary primers composed of a quasi-random 3′ end and a universal 5′ end. The manufacturer supplied semi-degenerate 3′end may bias towards cellular organisms (or against viruses) through unknown specificity. While amplification with both chemistries was performed for only one D. antillarum sample and one H. floridana sample (where both SeqPlex amplified libraries yielded no viral contigs), it is also possible that variation in viral detectability was due to different tissues used for other samples that we surveilled in this study. It is recommended that future studies of RNA viruses following this approach carefully evaluate amplification strategy to ensure efficient viral recovery from tissue specimens.

Conclusions

This study is the first to report viruses associated with two common coral reef invertebrates, Diadema antillarum and Holothuria floridana. While we were successful in recovering viral contigs typical of those identified in marine invertebrates from H. floridana, very few viral reads were identified in D. antillarum. We recommend careful choice of RNA amplification strategy to reduce biases associated with library preparation when considering viral metagenomic surveys of echinoderms. Viral metagenomic survey of viruses may be of utility to coral reef conservation by identification of potential agents affecting animal health, especially when followed by studies examining their pathogenicity. For example, observation of viruses associated with specific disease processes may be isolated and challenged against naïve hosts to satisfy e.g., Koch (1893) postulates. Hence, we recommend wide viral metagenomic surveillance of critical coral reef species to discern between viruses associated with normal specimens and those associated with mass mortality events.

Supplemental Information