Blue crabs were collected from coastal states of the United States, including Massachusetts (MA), Rhode Island (RI), New York (NY), Maryland (MD), Delaware (DE), North Carolina (NC), Texas (TX), and Louisiana (LA) between the years 2009 and 2021 (Figure 1). A portion of crabs collected prior to 2020 were also used in the analysis of CsRV1 prevalence (Zhao et al., 2020). Crab sex and carapace width (CW, measured laterally spine-to-spine), sampling date and locations were recorded during collection. Whole crab or two walking legs removed from each crab were chilled on ice at the time of harvest. For molecular analysis, frozen specimens were then shipped to the Institute of Marine and Environmental Technology (IMET) in Baltimore, MD and stored at −20°C until further analyses. Live crabs from RI and Shinnecock Bay, NY, collected in the year 2021, were shipped chilled to IMET where tissues were collected for virus purification, histology, and electron microscope observations.

Crab dissections were performed with sterile wooden rods and single-use razor blades on a bench cleaned with ELIMINase™. After the external cuticle was cleaned with ELIMINase™, approximately 50 mg of muscle and hypodermis was dissected from a walking leg and homogenized in 1.0 ml of homemade Trizol (Rodriguez-Ezpeleta et al., 2009), with a Savant FastPrep™ FP120 homogenizer (MP Biomedicals, Santa Ana, CA, United States). RNA extraction followed protocols used by Spitznagel et al. (2019). After Trizol-chloroform separation of RNA and precipitation with isopropanol, two 12,000g centrifuge washes with 500 μl 75% ethanol were carried out. Resulting RNA pellets were dissolved in 50 μl 1 mM EDTA and stored at −80°C. Process control samples (muscle from frozen smelt) were extracted before and after sets of tested crab samples to monitor for cross contamination between each sample. RNA quality and concentration were determined by NanoDrop™ spectrophotometry (Thermo Scientific, Waltham, MA, United States). The dsRNA content of RNA extractions was revealed by electrophoresis on 1.0% agarose TBE gel and stained with ethidium bromide. The isolated dsRNA was agarose gel purified with the NucleoSpin® Gel and PCR Clean-Up Kit (Takara Bio, San Jose, CA, United States).

Purified dsRNA of a single crab infected with both Callinectes sapidus toti-like virus 1 (CsTLV1) and Callinectes sapidus toti-like virus 2 (CsTLV2), collected from Agawam River, MA, United States was used for cDNA synthesis with barcoded octamers (5’-GGCGGAGCTCTGCAGATATC-NNNNNNNN-3’) with the M-MLV Reverse Transcriptase 1st-Strand cDNA Synthesis Kit (Biosearch Technologies, Hoddesdon, United Kingdom). The resulting cDNA was amplified by PCR using the barcode primers (5’-GGCGGAGCTCTGCAGATATC-3’). Amplification was achieved through 40 cycles of 95°C for 5 s (denaturation), and 60°C for 30 s (annealing), followed by 72°C for 30 s (elongation). PCR products of 250–500 bp were obtained by agarose gel purification with a NucleoSpin® Gel and PCR Clean-Up Kit (Takara Bio, San Jose, CA, United States). The DNA library was constructed using the NEBNext R Ultra™ DNA Library Prep Kit for Illumina (NEB, Ipswich, MA, United States) following manufacturer’s instructions (NEB, Ipswich, MA, United States). The library was sequenced in a 2 × 250 paired-end configuration on the Illumina MiSeq platform with a MiSeq Reagent kit v3 (Illumina, San Diego, CA, United States).

Sequencing barcodes were trimmed, and low quality and short reads were removed with CLC Genomics Workbench 9.5.2 (Qiagen, Hilden, Germany). The clean reads were collected and used for de novo assembly (Grabherr et al., 2011) with default settings (word-size = 45, Minimum contig length > =500). A preliminary set of contigs coding proteins of at least 150 amino acids were identified with ORF finder in CLC Genomics Workbench. ORFs of de novo derived contigs were used to search using the NCBI web server for non-redundant database using the BLASTp program. The conserved domains and motifs in the ORF were searched by NCBI Conserved Domain Database (CDD; http://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi). Dotknot (Huang et al., 2005) was used to search the H-type pseudoknots with estimated free energy (EFE). Predicted RNA secondary structures were visualized by Pseudoviewer 2.5 (Byun and Han, 2006).

Terminal sequences were determined using a SMARTer® Rapid Amplification of cDNA Ends (RACE) 5′/3′ Kit (Takara Bio, San Jose, CA, United States) with purified dsRNA as the initial template. The 3′ poly(A) tailing of RNA was performed at 37°C for 30 min using E. coli Poly(A) Polymerase (NEB, Ipswich, MA, United States). Poly (A)-tailed dsRNA was used for RACE First-strand cDNA synthesis of each terminus using 5′- or 3′- CDS primers as described by the manufacturer (Takara Bio, San Jose, CA, United States). Then, 5′-RACE and 3′-RACE PCR amplification was performed with viral gene specific primers (GSP) and universal primers (UMP), and then a nested PCR was performed with gene specific primers short (GSPS) and universal primer short (UPS; Supplementary Table 1) using Advantage 2 polymerase Mix (Takara Bio, San Jose, CA, United States). The conditions for amplification were 30 cycles at 94°C for 30 s, annealing at 68°C for 30 s, and elongation at 72°C for 2 min, with a final extension at 72°C for 10 min. Amplicons were then purified from the gel using NucleoSpin® Gel and PCR Clean-Up Kit (Takara Bio, San Jose, CA, United States), cloned into pGEM®-T Vector Systems (Promega Corporation, Madison, WI, United States), and sequenced. Sequence verification and filling of gaps between contigs were achieved with Sanger sequencing with primers in Supplementary Table 1. PCR conditions were 30 cycles at 94°C for 30 s, annealing at 56°C for 30 s, and elongation at 72°C for 90 s, with a final extension at 72°C for 10 min.

Predicted RdRps from all reference sequences, and closest homologues from NCBI were aligned with RdRps of newly identified viruses in this study using MAFFT 7.0 (Katoh and Standley, 2013) with an accurate option (L-INS-i). The alignment was used for constructing the Maximum Likelihood (ML) phylogenetic tree with protein substitution model JTT in CLC Genomics Workbench. RdRp amino acid sequence of Helminthosporium victoriae virus 145S (HvV145S; Refseq. YP-052858) was used for the outgroup of the ML tree. Branch support values greater than 0.5 were shown in the tree. The accession numbers of the proteins and the corresponding virus names and acronyms are shown in Supplementary Table 2.

To screen CsTLV1 and CsTLV2 infections in C. sapidus, a probe-based RT-qPCR assay was developed with primer pairs designed to detect a 193-bp region of CsTLV1 genome and a 183-bp region of CsTLV2 (Table 1) simultaneously. Probes designed for detecting CsTLV1 and CsTLV2 were also shown in Table 1. DsRNA standards were created by in vitro RNA synthesis. In brief, PCR products amplified by the primer pairs mentioned above were purified and cloned into pGEM®-T Vector Systems (Promega Corporation, Madison, WI, United States). Plasmids containing the targeted region were used as templates to synthesize each strand of the viral RNA standards by T7 or Sp6 RNA polymerase, respectively (Sigma-Aldrich, St. Louis, MO, United States). Viral RNAs were then quantified and annealed into dsRNA on ice, and serially diluted in 25 ng per μl yeast tRNA carrier. Standard curves were generated by RT-qPCR amplifications of a 10-fold dilution series of synthesized dsRNA containing 10–10e6 genome copies per μl. The qPCR cycling contained qScript® Virus 1-Step ToughMix® (Quantabio, Beverly, MA, United States) in 10 μl reactions with 0.25 μM each primer and 0.25 μM each probe for both genomes. To anneal PCR primers to dsRNA, primers and extracted RNA were combined, heated to 95°C for 5 min, and then cooled to 4°C prior to being added to the reverse transcriptase and Taq polymerase reaction mixture. Reverse transcription and amplification conditions were 50°C for 10 min (reverse transcription) followed by 1 min at 95°C (reverse transcriptase inactivation and template denaturation). Amplification was achieved through 40 cycles of 95°C for 10 s, and 61°C for 30 s. Gene target copies were then calculated as copies per mg of crab muscle, and samples with greater than 100 copies per mg were recorded as CsTLV1 and CsTLV2 positive, which was based on empirical observations of cross contamination in process control RNA extractions.

All statistical tests were conducted using RStudio 1.1.456 (R Core Team, 2019). Significant correlations were defined as those where p ≤ 0.05. To determine whether CsTLV1 and CsTLV2 infections were correlated with sex, crab size or latitude, binomial (infected vs. non-infected) generalized logistic regression models (GLM) were conducted (alpha = 0.05). Akaike’s information criterion (AIC) was used to choose the best GLM to determine, which factors best correlate with CsTLV1 and CsTLV2 infection (Aho et al., 2014). The Pearson correlation was used to test the correlation between the variables (Kirch, 2008).

Callinectes sapidus collected from RI and Long Island (NY) with CsTLV1 and CsTLV2 confirmed and quantified by RT-qPCR were dissected with muscle, gill and hepatopancreas tissues removed for further examination for virus presence. These crabs were also tested by RT-qPCR (methods in Zhao et al., 2020) to confirm they were not infected by CsRV1. For histological analyses, the tissues were fixed in Bouin’s solution at 4°C overnight, and then placed into 75% ethanol for long-term storage. Preserved tissues were processed according to the standard operating procedures for embedding, sectioning and Hematoxylin and Eosin (H&E) staining (Luna, 1968). Slides were then observed with an Echo Revolve Microscope (San Diego, CA, United States).

For electron microscopy examination, crab tissues were immersion fixed in fixative buffer (2% paraformaldehyde, 2.5% glutaraldehyde, 2 mM CaCl2 in 0.1 M PIPES Buffer, and pH 7.35) at 4°C overnight. Tissue fragments were then trimmed into ~1 mm³ cubes, post-fixed with 1% osmium tetroxide, washed in water and stained en bloc with 1% (w/v) uranyl acetate for 1 h. Specimens were then washed and dehydrated using 30, 50, 70, 90, and 100% ethanol in series. After dehydration, specimens were embedded in Araldite-Epoxy resin (Araldite, Embed 812, Electron Microscopy Sciences, Hatfield, PA, United States; Vorimore et al., 2013). Ultrathin sections (~ 70 nm) were cut and examined in a Tecnai T12 TEM (FEI, Hillsboro, OR, United States) operated at 80 KV. Digital images were acquired using an AMT bottom mount CCD camera and AMT600 software (Advanced Microscopy Techniques, Woburn, MA, United States). Crab samples infected with only CsTLV2 were not preserved well enough to be included in TEM and histology examinations.

In a search for viral dsRNA in blue crabs, RNA extracted from leg muscle analyzed on agarose gels. The RNA of 31% of sampled of crabs (9 of 29) harvested from the Agawam River, MA in the summer of 2008, showed two prominent dsRNA bands, termed dsRNA-S and dsRNA-L. The apparent molecular weights of the two dsRNA segments were ∼6.5 and ∼7.5 kbp, respectively (Figure 2).

Total dsRNA, containing both dsRNA-S and dsRNA-L was sequenced with NGS. Assembly of trimmed and quality filtered reads (16,650 reads: 42.4% of the total reads) resulted in two contigs for each CsTLV1 and CsTLV2, at 383- and 120-fold average coverage, respectively. The longest contig was 4,016 nucleotides (nt) in length. The 5′ and 3′ untranslated regions (UTRs) of both genomes were obtained by RACE PCR and Sanger sequencing, to reveal two toti-like genome sequences of 6,444 and 7,421 nt, and designated as CsTLV1 and CsTLV2, respectively. Each genome contained two ORFs (ORF1 and ORF2) encoding Cp and RdRp proteins, respectively (Figure 3). Genomic sequences were not detected for any other virus in the NGS library.

The predicted Cp and RdRp proteins of CsTLV1 and CsTLV2 showed limited amino acid sequence identity with each other (21% for Cp and 27% for RdRp, respectively). CsTLV1 RdRp amino acid sequences showed >33% identity with the corresponding predicted RdRp of Beihai barnacle virus 15, Ahus virus, Parry’s Creek toti-like virus 1, and Diatom colony associated dsRNA virus 17 genome type A (Shi et al., 2016; Urayama et al., 2016; Pettersson et al., 2019; Williams et al., 2020). CsTLV2 RdRp amino acid sequence showed >40% identity with the RdRp encoded in Plasmopara viticola lesion associated totivirus-like 5, Hubei toti-like virus 5, and Beihai sesarmid crab virus 7 (Shi et al., 2016; Chiapello et al., 2020; Supplementary Table 3). A search of the CDD and multiple protein alignment confirmed that the predicted RdRp domains of CsTLV1 and CsTLV2 contain eight conserved motifs (I–VIII), including the GDD motif, which are the typical characteristics of virus RdRps (Figure 4). Sequence analyses of CsLTV1 and CsTLV2 indicated that there is an overlap region between ORFs 1 and 2 (Figure 3), that allows ORF 2 to be translated as a fusion protein with ORF 1 through a-1 ribosomal frameshift motif “GGAUUUU” at 3,199–3,205 nt positions in CsTLV1, and “AAGAAAA” at positions of 2,685–2,691 nt in CsTLV2. An H-type pseudoknot structure was predicted in the downstream of each putative slippery site at positions 3,227–3,259 nt of CsTLV1 genome and 2,697–2,760 nt of CsTLV2 (Figure 3).

A maximum likelihood phylogenetic tree was used to show the relationships between CsTLV1 and CsTLV2 and other selected totivirids. As shown in Figure 5, RdRp amino acid sequence multiple alignments of CsTLV1 and CsTLV2 and the corresponding toti and toti-like viral sequences revealed that CsTLVs is most closely related to, but distinct from, Totivirus, and Artivirus which is a proposed genus that includes IMNV and IMNV-like viruses (Zhai et al., 2010). CsTLV1 and the five toti-like viruses with the highest identity from GenBank formed a cluster in the tree (bootstrap value = 77%), adjacent to but different from the cluster of CsTLV2 and its close toti-like virus species (bootstrap value = 100%).

The probe-based RT-qPCR assay consistently detected as few as 10 copies of the target when tested on a dilution series of synthesized dsRNA standards (Table 2). Efficiency and sensitivity of the assay were evaluated by running10 RT-qPCR standard curves. The mean slope was 3.29 with a SD of 0.06 for CsTLV1, and the mean slope was 3.25 with a SD of 0.07 for CsTLV2. There is an average efficiency of 100.1 and 100.3% for CsTLV1 and CsTLV2 respectively, under typical use with the synthesized dsRNA standard.

The prevalence of CsTLV1 and CsTLV2 was investigated by RT-qPCR in 875 crabs from the US Atlantic and Gulf of Mexico coasts. CsTLV1 and two infections were detected in the northern states (MA, RI, and NY, n = 496) but not the lower latitudinal Atlantic states (DE, MD, and NC, n = 299) and Gulf states (LA and TX, n = 80). CsTLV1 and CsTLV2 infections in C. sapidus were detected in all three estuaries sampled in MA: Agawam River, Acushnet River, and Falmouth (Figure 1; Table 3). CsTLV1 RNA prevalence in C. sapidus sampled from MA (n = 198) varied from 11.8% (Agawam River; n = 127) to 30.6% (Acushnet River; n = 49), and CsTLV2 RNA prevalence ranged from 12.6 to 36.7%. In crabs from RI, CsTLV1 prevalence was 27.6% and CsTLV2 prevalence was 32.8%. Viral RNA was detected in crab specimens collected from Moriches Bay, Shinnecock Bay, and Napeague Bay in Long Island, NY (n = 133) with an average prevalence of 42.1 and 43.6% for CsTLV1 and CsTLV2, respectively. CsTLV1 RNA was detected in Georgica Pond, NY, with low prevalence in 2012 (5.5%; n = 18), but not in 2013, 2020, or 2021, and CsTLV2 RNA was detected in 2012 (5.5%; n = 18) and 2013 (5.2%; n = 19), but not in 2020 (n = 33) or 2021 (n = 37).

Overall, CsTLV1 was never observed in the absence of CsTLV2, and co-infections of CsTLV1 were detected in 91.5% (107/117) of CsTLV2 positive specimens (Table 3). The dsRNA copy number per mg muscle ranged from 6.5 × 10e2 to 1.2 × 10e8 for CsTLV1, and 1.3 × 10e2 to 6.3 × 10e8 for CsTLV2.

A binomial (infected vs. non-infected) generalized linear model (GLM) was used to test whether latitude, crab size, or sex could predict CsTLV1 and CsTLV2 infection status (Table 4). The 415 male and 140 female specimens, from 20 to 196 mm in carapace width (CW), were used for GLM analysis. Specimens that were PCR-positive for CsTLV1 and CsTLV2 ranged from 29 to 150 mm in CW. Prevalence for male and female crabs was 12.0 and 10.7%, respectively. Pearson correlation tests showed no significant correlation between latitude and crab size or sex. The full model analyzing the association between CsTLV1 and CsTLV2 infections and latitude, crab sex, and carapace width (CW), differed significantly from null models (p < 0.01), in which latitude and CW were significant fixed effects (p < 0.01). The reduced model, including only the association between CsTLV1 and CsTLV2 infection and latitude, sex, or CW, reinforced that latitude and CW were the significant factors correlated with CsTLV1 and CsTLV2 prevalence (p < 0.01; Table 4). CsTLV1 and CsTLV2 prevalence was positively related to latitude in the reduced model (slope is 2.33 for CsTLV1 and 2.38 for CsTLV2; p < 0.01), and CW showed a negative association with CsTLV1 and CsTLV2 prevalence (slope = −1.00 for CsTLV1 and slope = −1.13 for CsTLV2; p < 0.01). In both full and reduced models, the association between CsTLV1/CsTLV2 prevalence and crab sex was not significant (p > 0.1).

Crabs assessed to be infected with CsTLV1 and CsTLV2 by RT-qPCR were selected for TEMs observation (Supplementary Figure 1). TEM revealed the presence of isometric virus particles, with a diameter of ~40 nm in C. sapidus muscle, gill, and hepatopancreas tissues (Figure 6). Completed virions were present in the connective tissue and hemocytes of these tissues. We observed putative viroplasm in the gill of CsTLV1 and CsTLV2 infected crab and packed arrays of mature virions in the hepatopancreas of infected crab.

Histological analysis of muscle, hepatopancreas and gills tissues of crabs naturally infected with CsTLV1 and CsTLV2 showed necrosis and hemocyte infiltration. Skeletal muscle in normal uninfected crabs is generally smooth, striated and with few circulating hemocytes (Figure 7A). Infected skeletal muscle had general necrosis and showed vacuolated areas with increased numbers of circulating hemocytes (Figure 7B). Hepatopancreas tubules in normal uninfected crabs have defined outer membranes and moderate numbers of circulating hemocytes circulating within hemal spaces between tubules (Figure 7D). Infected hepatopancreas often showed massive hemocytic infiltration (Figure 7E). Gills of normal uninfected crabs have moderate numbers of circulating hemocytes in hemal spaces (Figure 7G). Infected gills had considerably increased numbers of circulating hemocytes within necrotic areas (Figure 7H). At higher magnification, infected hemocytes in muscle, hepatopancreas, and gills often had pyknotic or karyorrhectic nuclei (magenta arrows) as well as opaque, slightly eosinophilic intracytoplasmic inclusion bodies (blue arrows; Figures 7C,F,I).