Alphaflexivirus Genomes in Stony Coral Tissue Loss Disease-Affected, Disease-Exposed, and Disease-Unexposed Coral Colonies in the U.S. Virgin Islands

Viral infections of endosymbiotic dinoflagellates (family Symbiodiniaceae) within coral tissues are hypothesized to play a role in stony coral tissue loss disease (SCTLD) (1), a widespread disease that affects Caribbean stony corals (2–4). Here, we present high-quality draft genome sequences for two viruses in the family Alphaflexiviridae, coral holobiont-associated alphaflexvirus (CHFV) 1 and 2 (Fig. 1A), that were assembled from metatranscriptomes from SCTLD-affected, SCTLD-exposed, and control (unexposed) coral holobionts sampled during a SCTLD transmission experiment (5). The field collections were authorized by the Department of Planning and Natural Resources Coastal Zone Management under permit number DFW19057U.

Tissue samples were harvested from 12 frozen fragments of three coral species (Montastraea cavernosa, Porites astreoides, and Pseudodiploria strigosa) collected from St. Thomas, U.S. Virgin Islands (Table 1). Total RNA was extracted using the RNAqueous-4PCR total RNA isolation kit (Invitrogen, Life Technologies AM1914). Tissues were lysed using a refrigerated Qiagen TissueLyser II microcentrifuge at 30 oscillations per second for 30 s. The elution stage consisted of two consecutive 30-μL elutions. Contaminating DNA and chromatin were removed from the total RNA using the Ambion DNase I (RNase-free) kit (Invitrogen, Life Technologies AM2222). Samples were preprocessed by Novogene Co., Ltd. (Davis, CA, USA) for mRNA enrichment using polyA tail capture; the mRNA libraries underwent 150-bp, paired-end sequencing on an Illumina NovaSeq 6000 instrument using the NEBNext Ultra II RNA library prep kit.

All bioinformatic tools were run using default parameters unless otherwise specified. BBSplit (BBMap v38.90) was used to map quality-filtered (fastp v0.20.1 [16]) reads to coral or Symbiodiniaceae transcriptomes (9) and generate three read files: (i) coral, (ii) Symbiodiniaceae, and (iii) noncoral/non-Symbiodiniaceae. Noncoral/non-Symbiodiniaceae reads were combined and normalized using BBnorm.sh within BBMap (Table 1). Normalized reads were assembled using the program TransPi (17). Multiple assemblies were generated using rnaSPADES v3.14.0 (kmer: 75,85,91,107 nucleotides) (18), Trans-ABySS v2.0.1 (kmer: 25,35,55,75,85 nucleotides) (19), SOAPdenovo-Trans v1.03 (kmer: 25,35,55,75,85 nucleotides) (20), Trinity v2.9.1 (kmer: 35 nucleotides) (11), and Velvet v1.2.12/Oases v0.2.09 (kmer: 65,71,81,87,91,97,101 nucleotides) (21, 22). The multiple assemblies were concatenated into a single file, and the EvidentialGene tr2aacds pipeline v2019.05.14 (23, 24) was used to collapse duplicates and remove misassembled contigs from the assembly file. VirSorter2 (25) was used to detect RNA viruses from the nonredundant metatranscriptome-assembly file (minimum length, 300 nucleotides). Viral genomes similar to known members of the Alphaflexiviridae were identified by aligning translated open reading frames (ORFs) to the proteic version of the Reference Virus Database (26, 27) with DIAMOND BLASTx v2.0.11.149 in “ultra-sensitive” mode (28, 29). Cenote-Taker 2 (30) was used to annotate identified viral genomes with similarity to the Alphaflexiviridae and calculate the genome coverage using the normalized reads. The alphaflexivirus read count per sample library was estimated by mapping nonnormalized reads to the nonredundant assembly using bowtie2 (31) with the align_and_estimate_abundance.pl script (11; Table 1).

The CHFV1 and CHFV2 genomes are linear, share 85.9% genome-wide nucleotide identity, and are 6,228 and 6,227 nucleotides long with 42.4% and 42.0% G+C content, respectively. Coverages for the CHFV1 and CHFV2 assemblies are estimated at 334.9× and 123.4×, respectively. CheckV (32) was used to identify the genomes as high quality with 90% completeness (average amino acid identity-based [medium-confidence]). Visualization of a tBLASTx (33) pairwise alignment between the CHFV genomes was conducted using Easyfig (34) and depicted the genomes’ three shared ORFs (Fig. 1A). The closest relative of the CHFV genomes, as determined using Cenote-Taker 2, is strawberry mild yellow edge virus (NCBI protein accession number NP_620642.1) (35), sharing ∼33.5% amino acid similarity for the RNA-dependent RNA polymerase (RdRp) (ORF1).

A phylogenetic tree was generated from translated RdRp sequences from the two CHFVs and 16 plant-associated alphaflexiviruses (Fig. 1B). The CHFV replicase sequences formed a clade with the RdRp sequence of an unclassified alphaflexivirus that infects Cymodocea nodosa seagrass (Fig. 1B).

The CHFV genomes reported here constitute genomic-based evidence of filamentous viruses from coral colonies. Quantitative PCR primer sets can be developed from these genome assemblies to support the critical next step of characterizing the presence/absence and abundance of coral holobiont-associated alphaflexiviruses across coral colonies, to further clarify the potential role of viruses in SCTLD.

This work was supported by the National Science Foundation (Biological Oceanography) award numbers 1928753 to M.E.B. and T.B.S., 1928609 to A.M.S.C., 1928817 to E.M.M., 19228771 to L.D.M., 1927277 to D.M.H., and 1928761 to A.A., as well as by VI EPSCoR (NSF numbers 0814417 and 1946412).

This work could not have been completed without the hours of lab and fieldwork dedicated by Adam Glahn, Danielle Lasseigne, Amanda Long, Bradley Arrington, Daniel Mele, Kathryn Cobleigh, Bradford Dimos, Nicholas Macknight, Naomi Huntley, and Alexandra Gutting. We also thank Samantha Coy for her input regarding CHFV transcripts, as well as Nikoloas Schizas and the Marine Genomic Biodiversity Lab of the Department of Marine Sciences at the University of Puerto Rico—Mayaguez for providing us with access to computational resources used to conduct this study.

This is contribution number 229 from the Center for Marine and Environmental Studies at the University of the Virgin Islands.