Detection and discovery of plant viruses in soybean by metagenomic sequencing

A total of 135 soybean samples displaying virus-like or unusual disease symptoms were collected from eight different states in the USA during the 2016–2019 growing seasons and stored at −80 °C. Images of leaves from the 78 virus-containing samples that possessed one or more complete or nearly complete viral genome(s) are shown in Fig. 1 and Additional file 1: Fig. S1. A list of the 78 samples with their origin, description of symptoms, and associated complete or nearly complete viral genome(s) is provided in Table 1 and Additional files 11 and 12: Tables S1 and S2.

Samples collected in 2016 were ground in liquid nitrogen and total RNA was extracted using the Pure Link™ RNA Mini-Kit (Invitrogen by Fisher Scientific, Carlsbad, CA). Approximately 100 mg of leaf tissue ground in liquid nitrogen was added to Trizol and mixed thoroughly before being processed according to the manufacturer’s directions. Total RNA was treated with DNase I (TURBO DNA-free™ Kit, ThermoFisher Scientific) to remove genomic DNA contamination. The 2017–2019 samples were ground in liquid nitrogen and total RNA was extracted using Direct-zol RNA Miniprep Plus (www.​zymoresearch.​com) that included an on-column DNAse treatment following the manufacturer’s instructions. Total RNA was quantified with a Qubit 4.0 fluorometer (Life Technologies, Carlsbad, CA, USA) and quality was assessed with a NanoDrop 2000 (ThermoFisher Scientific, CA, USA). Samples with high quality total RNA had A260/280 and A260/230 ratios between 1.8 and 2.1. Samples with low quality total RNA had A260/280 and A260/230 ratios between 1.0 and 1.8.

The DNA-free total RNA was ribosome depleted (Ribo-Zero rRNA Removal Kit (Plant Leaf), Illumina, San Diego, CA, USA; RiboMinus Plant kit, ThermoFisher Scientific, CA, USA). Strand-specific cDNA libraries were prepared from ribosome-depleted RNA using the NEBNext Ultra II RNA library prep kit (New England Biolabs, Ipswich, MA, USA) [45]. Paired-end Illumina sequencing was performed on the dual-indexed cDNA libraries (New England Biolabs, Ipswich, MA) using HiSeq3000 (150 bp from each end; 2 lanes per library; Illumina, San Diego, CA; Additional file 2: Fig. S2A).

FastQC v.0.11.2 was used to assess the quality of the raw sequence reads to determine the necessity of trimming low-quality reads [46]. All paired-end FASTQ files were processed with Trimmomatic v0.36 [47] using sliding window 4:15 and excluding reads below a minimal length of 36. The trimmed paired-end FASTQ sets were examined by FastQC again to confirm improvement of read quality. Trimmed paired-end RNA-Seq reads were mapped against the soybean genome v2.1 (https://​plants.​ensembl.​org/​Glycine_​max) [48] using the HISAT2 alignment program [49] (–un-conc –al-conc). The reads with sequence similarities greater than 60% to the soybean genome sequences were removed. The remaining unmapped reads were mapped to plant virus reference genome sequences using Bowtie2 aligner [50] (–sensitive-local –un-conc –al-conc). The plant virus reference sequences were composed of 20,433 virus genome sequences obtained from the NCBI GenBank (NCBI: National Center for Biotechnology Information; https://​www.​ncbi.​nlm.​nih.​gov/​genome/​viruses). The mapped viral reads and the unmapped non-soybean non-viral reads were de novo assembled separately using Trinity v2.6.6 [51] with a default k-mer of 25 to obtain potential viral contigs. All assembled contigs were queried against the all-organism NCBI nucleotide and protein database through BLASTN (Basic Local Alignment Search Tool) and BLASTX searches with default parameters using Blastplus v2.7.1 (Additional file 2: Fig. S2B) [52, 53]. A virus was determined as present in each sample if at least one contig had significant hits (E-value < 1e-10 for BLASTN, and E-value < 1e-5 for BLASTX as cutoff) to NCBI sequences with the virus description and with a query coverage greater than 80% and identity greater than 95%. Contigs matching bacteria, fungi or insect sequences with high query coverage and high sequence identity were removed. Contigs less than 1000 bp matching viruses with lower query coverage and sequence identity were not considered.

A phylogenetic tree was constructed from the alignment of two ClYVV IA isolates (IA-2016 and IA-2017), and other potyvirus species using PhyML 3.0 software [54]. In brief, the amino acid sequences of the polyprotein from the two identified ClYVV IA isolates, nine known ClYVV isolates from GenBank, and as an outgroup, two bean yellow mosaic virus (BYMV) isolates were aligned using MUSCLE alignment software [55]. Additionally, the phylogenetic analysis of SIlV1 was conducted using the amino acid sequences of movement protein (MP), coat protein (CP), RNA dependent RNA polymerase (RdRp) and replicase obtained in this study and other established ilarvirus species using PhyML 3.0 software. The full-length genome sequences of SIlV1 were submitted to ORF finder [56] to determine the predicted amino acid sequences. Alignments were performed using MUSCLE 3.8.31 alignment software [55]. To construct phylogenetic trees for both ClYVV and SIlV1 isolates, maximum likelihood analysis was performed using PhyML 3.0 phylogeny online server, with the bootstrap replication (1000 replicates) used to assess the statistical support of the groups on the tree. The best-fit substitution model used in the analysis was determined with the automatic model selection by Smart Model Selection (SMS [57],) on PhyML 3.0, using the Akaike Information Criterion (AIC, [58]) with the + G, + I, and + F parameters. FigTree v.1.4.4 (http://​tree.​bio.​ed.​ac.​uk/​software/​figtree/​) was used to visualize the phylogenetic tree.

N. benthamiana were grown in a growth chamber set to 22/18 °C (day/night) with a 16/8 h light/dark photoperiod. Broad bean (Vicia faba cv. Broad Windsor and cv. Robin Hood; https://​territorialseed.​com) and soybean (Acre Edge 22R269, Williams 82, and 41 Nested Association Mapping (NAM) parents) seeds were germinated in growth chambers set to 28/24 °C (day/night) with a 16/8 h light/dark photoperiod. After ten days, seedlings were moved to a greenhouse kept at 28/24 °C (day/night) with a 16/8 h light/dark photoperiod. Three to four leaves of four-week-old N. benthamiana, or the primary leaves of ten-day old broad bean and soybean were lightly dusted with carborundum powder. Inoculum was prepared by grinding frozen tissue with a mortar and pestle in 10 mM phosphate buffer, pH 7.5. Leaves were rub-inoculated using a cotton swab. For mock-inoculated control plants, leaves were lightly dusted with carborundum and rub-inoculated with 10 mM phosphate buffer, pH 7.5. Three biological replicates were performed, and for ClYVV experiments, each replicate had seven plants (N. benthamiana), six plants (broad bean) or 20 plants (soybean). For TRSV experiments, each replicate included 20 plants. Symptomatic and mock-inoculated N. benthamiana, broad bean (cv. Broad Windsor and cv. Robin Hood), and soybean leaves were collected at 21 dpi and stored at −80 °C. Symptomatic soybean leaves from ClYVV IA-2016 plants and mock-inoculated leaves were collected at 35 dpi and stored at −80 °C.

Reverse transcription-polymerase chain reaction (RT-PCR) was performed for the following samples: Field-grown soybean samples (S1-S78); greenhouse-grown N. benthamiana leaf tissues, broad bean leaf tissues, and soybean leaf tissues infected with ClYVV-IA-2016 or ClYVV-IA-2017; and greenhouse-grown N. benthamiana leaf tissues and soybean leaf tissues infected with TRSV. For greenhouse experiments, mock-inoculated N. benthamiana leaves, broad bean leaves, and soybean leaf tissues were used as negative controls. In brief, total RNA was extracted using Direct-zol RNA Miniprep Plus kit (www.​zymoresearch.​com) that included an on-column DNase treatment. Total RNA quality was assessed with a NanoDrop 2000 (ThermoFisher Scientific, CA, USA) and quantified with a Qubit 4.0 fluorometer (Life Technologies, Carlsbad, CA, USA). All RNA samples were reverse-transcribed using the RevertAid First Strand cDNA Synthesis Kit (ThermoFisher Scientific, CA, USA). The first-strand cDNA was quantified using Qubit 4.0 fluorometer. One µl of cDNA template was used in PCR carried out in 2X GoTaq Green Master Mix (Promega, USA), 10uM of forward and reverse primers in a total volume of 25ul. Amplifications were in a C1000 touch thermal cycler (Bio-Rad, Hercules, CA, USA) programmed as follows: 2 min at 95 °C with 32 cycles of 30 s at 94 °C, 30 s at 60 °C, and 45 s at 72 °C, followed by 10 min at 72 °C. The PCR products were visualized by electrophoresis on a 1% agarose gel containing SYBR Safe DNA gel stain (Life Technologies, Carlsbad, CA, USA). The PCR products were cleaned using ExoSAP-IT PCR product cleanup reagent (ThermoFisher Scientific, CA, USA) and sequenced by Sanger sequencing (DNA Facility, Iowa State University, Iowa, USA). For the new ilarvirus identified in sample (S78), PCR was performed with RNA1, RNA2, and RNA3 primer pairs under the following conditions: 2 min at 94 °C with 40 cycles of 1 min at 95 °C, 1 min at 54.5 °C, and 1 min at 72 °C, followed by 10 min at 72 °C. The PCR products were visualized by electrophoresis on a 1% agarose gel and gel purified using Wizard SV gel and PCR clean-up system (Promega, USA). The purified PCR products were sequenced by Sanger sequencing (University of Wisconsin Biotechnology Center, Madison, USA). Primer sequences were designed using Primer3 plus and PrimerQuest™ (www.​idtdna.​com) from assembled contigs, and the primers [43, 59‐62] used in this study are listed in the Additional file 13: Table S3.

Seed transmissibility was examined using seeds collected from five greenhouse-grown ClYVV IA-2017-infected soybean plants (cv. Acre Edge 22R269) exhibiting disease symptoms. For controls, seeds were also collected from two mock-inoculated soybean plants (cv. Acre Edge 22R269). For seed transmission testing, 100 seeds from each ClYVV infected parent and mock-inoculated control parent plant were sown in growth chambers set to 26/24 °C (day/night) with a 16/8 h light/dark photoperiod. After fourteen days, two young leaflets from each soybean seedling were freshly harvested. A pool of leaflets from 10 soybean seedlings was harvested as an individual group. The parents prior to seed collection and the seedlings in groups of 10 for a total of 10 groups from each ClYVV-infected parent were serologically tested for presence of a potyvirus using the ImmunoStrip for Potyvirus Group as described above.

N. benthamiana leaf tissues infected with ClYVV IA-2017 and corresponding leaf tissues of mock-inoculated plants were collected at 10 dpi and examined using TEM. Infected leaves were dissected and 2 mm portions were placed into 1% paraformaldehyde, 3% glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.2 and fixed for 48 h at 4 °C. Samples were washed in 0.1 M cacodylate buffer three times/10 min each, and post fixed with 1% osmium tetroxide in 0.1 M sodium cacodylate buffer for 1 h at room temperature. Samples were washed with deionized water three times/15 min each, and en bloc stained using 2% uranyl acetate in distilled water for 1 h. Samples were washed in distilled water for 10 min and dehydrated through a graded ethanol series (25, 50, 70, 85, 95, and 100%) for 1 h each step. Samples were further dehydrated with three changes of pure acetone, 15 min each, and infiltrated with Spurr's formula (hard) epoxy resin (Electron Microscopy Sciences, Hatfield PA) with graded ratios of resin to acetone until fully infiltrated with pure epoxy resin (3:1, 1:1, 1:3, pure) for 6–12 h per step. Tissues were placed into BEEM capsules and were polymerized at 70 °C for 48 h. Thin sections were made using a Leica UC6 ultramicrotome (Leica Microsystems, Buffalo Grove, IL) at 50 nm and collected onto single slot carbon film grids. TEM images were collected using a 200 kV JEOL JSM 2100 scanning transmission electron microscope (Japan Electron Optics Laboratories, USA, Peabody, MA) with a GATAN One View 4 K camera (Gatan Incorporated, Pleasanton, CA).

Soybean samples with virus-like symptoms or disease symptoms of unknown etiology were collected for RNA sequencing during four growing seasons (2016–2019) as shown in Fig. 1, Table 1, Additional file 1: Fig. S1 and Additional files 11 and 12: Tables S1 and S2. The cDNA libraries generated from each sample were sequenced on an Illumina HiSeq 3000 platform (Additional file 2: Fig. S2A). The number of paired-end reads per sample ranged from 19,351,404 in S1 to 18,867,866 in S78, and after quality filtering the number of high quality trimmed raw reads ranged from 18,581,716 in S1 to 18,216,248 in S78 as shown in Additional file 14: Table S4.

The filtered reads were mapped against the soybean genome v2.1 to remove soybean reads (Additional file 2: Fig. S2B) from the analyses as shown in Additional file 14: Table S4. The non-soybean reads were mapped to plant virus reference genome sequences (20,433 genome sequences; Additional file 2: Fig. S2B) obtained from NCBI GenBank as shown in Additional file 14: Table S4. The mapped viral reads were de novo assembled with a default k-mer of 25 using the Trinity assembler (Additional file 2: Fig. S2B). The unmapped non-soybean non-viral reads were also de novo assembled with a default k-mer of 25 using the Trinity assembler. To identify virus-associated sequences, the de novo assembled contigs were annotated against the all-organism NCBI nucleotide and protein database through BLASTN and BLASTX searches. Of the 135 samples, 78 of them (S1–S78) contained contigs corresponding to full-length or nearly full-length viral genomes and proteins as shown in Additional file 15: Table S5. Of the remaining 57 samples, 27 samples had de novo contig length less than 1000 bp with few hits to known viruses and other pathogens such as bacteria, oomycetes, and fungi. None of these samples had partial viral genomes. Additionally, 23 samples had no hits to viral genome sequence but uncovered pathogens such as bacteria, oomycetes, and fungi while the remaining 7 samples had no pathogens detected. Due to potentially misleading interpretations of partial sequences for virus identity [63], a conservative approach of only contigs corresponding to complete or nearly complete genome sequences were used in subsequent analyses. The top GenBank accessions that had the highest query coverage (> 80%) and highest sequence identity (> 95%) to each contig is listed in Additional file 15: Table S5. These contigs share similarity to SbDV, AMV, TSV, TRSV, BPMV, SVNV, and ClYVV viruses in the following families: Luteoviridae, Bromoviridae, Secoviridae, Bunyaviridae, and Potyviridae.

A most surprising outcome from the sequence analyses was the detection of ClYVV in samples from Iowa, USA collected in 2016 (Sample S1) and 2017 (Sample S7; Table 1). Single contigs covering nearly the entire ClYVV reference genome (Gm isolate; GenBank accession number: KF975894.1) were obtained from each sample as shown in Additional file 15: Table S5. The sequences of the 2016 ClYVV isolate (IA-2016; GenBank accession number MK292120.1) and the 2017 ClYVV isolate (IA-2017; GenBank accession number MK318185.1) were 9610 nucleotides (nt) and 9627 nt long, respectively (Table 1; Additional file 15: Table S5), and they share 96.5% nt identity with one another (Additional file 3: Fig. S3). The ClYVV IA-2016 and -IA-2017 isolates shared the greatest nt sequence identity to the ClYVV Gm isolate, 96.39% and 96.32%, respectively, from South Korea (Additional file 15: Table S5). The ClYVV-IA-2016 and -IA-2017 genomes encode predicted polyproteins of 3,072 amino acids (aa) that were > 98% identical to the ClYVV I89-1 reference sequence (GenBank accession number: BAT50981.1) (Additional file 15: Table S5). The presence of ClYVV in S1 and S7 field samples was verified by PCR amplification using detection primers designed from the assembled contigs as listed in Additional file 13: Table S3, followed by Sanger sequencing as shown in Additional file 4: Fig. S4.

To determine the relationships between the IA-2016 and IA-2017 isolates with the nine full-length ClYVV isolates in GenBank, the amino acid sequences were aligned, and a phylogenetic tree was constructed using the Maximum Likelihood algorithm. Sequences of two BYMV isolates from Japan were included as an outgroup, which only had 59.4–60.3% nt identity to ClYVV IA-2016 and ClYVV IA-2017. The ClYVV IA isolates formed a single cluster separate from China (Hefei isolate), South Korea (Dendrobium isolate), Ohio (contig 27.1), Japan (90-1 Br2 and No. 30 strains), and Australia ClYVV sequences (CYVV isolate; Fig. 2). ClYVV IA-2016 (MK292120.1) and ClYVV IA-2017 (MK318185.1) were most closely related to ClYVV I89-1 strain (BAT50981.1) and Gm isolate (AHL28796.1; Fig. 2), which is consistent with the nt sequence identity values shown in Additional file 15: Table S5. These results showed that ClYVV IA-2016 and ClYVV IA-2017 were distinct from one another, and they may have a common origin with isolates previously identified in Japan and South Korea.

To recover the ClYVV IA isolates, sap from the frozen soybean samples (S1 and S7) was used to mechanically inoculate N. benthamiana plants. At 10 dpi, systemic leaves of plants inoculated with either ClYVV IA-2016 or ClYVV IA-2017 exhibited mosaic chlorosis and epinasty (Additional file 5: Fig. S5). At 21 dpi, ClYVV IA-2017 plants were stunted, and the systemically-infected leaves showed pronounced curling and mosaic chlorosis (Fig. 3A and B). The ClYVV IA-2016 plants continued to exhibit milder symptoms (Fig. 3A and B). ClYVV IA-2016 and ClYVV IA-2017 in N. benthamiana plants was verified by PCR amplification (Fig. 3C) using detection primers designed from the assembled contigs (Additional file 13: Table S3) followed by Sanger sequencing (Additional file 6: Fig. S6).

To determine if the two ClYVV isolates systemically infect legumes, mechanical transmission was performed on broad bean and soybean plants. At 10 dpi, symptoms of mottling and hyponasty were observed on the systemic leaves of broad bean cultivars Broad Windsor and Robin Hood that were inoculated with either isolate. At 21 dpi, Broad Windsor plants infected with ClYVV IA-2016 developed more severe symptoms with pronounced mottling and leaf curling (Fig. 4A and B). In contrast, mild mottling with subtle leaf curling was observed in ClYVV IA-2017-infected systemic leaves (Fig. 4A and B). Robin Hood plants infected with either ClYVV IA-2016 or ClYVV IA-2017 were severely stunted with mild mottling patterns on systemic leaves (Fig. 4C and D). Systemic leaves infected with ClYVV IA-2016 exhibited mild hyponasty, and while ClYVV IA-2017 did not cause leaf curling, systemic leaves and stems developed necrosis and eventually wilted (Fig. 4C and D). ClYVV IA-2016 and ClYVV IA-2017 in Broad Windsor and Robin Hood infected plants was verified by PCR amplification (Fig. 4E) using detection primers designed from the assembled contigs (Additional file 13: Table S3) followed by Sanger sequencing (Additional file 6: Fig. S6).

To determine infectivity in soybean, the cultivars Acre Edge 22R269 and Williams 82, and the 41 genetically diverse soybean lines contributing to the nested association mapping (NAM) panel were mechanically inoculated with ClYVV IA-2016 or ClYVV IA-2017. Acre Edge 22R269 was used in these experiments because it was the variety in which ClYVV IA-2016 was originally found. Chlorotic spots were observed at 30 dpi on older leaves of Acre Edge 22R269 plants infected with ClYVV IA-2016, which was followed by yellowing and vein clearing in older and younger systemic leaves by 35 dpi (Fig. 5A and B). Chlorotic spots were observed at 19 dpi on older leaves of Acre Edge 22R269 plants infected with ClYVV IA-2017, which was followed by yellowing and vein clearing in the older and younger systemic leaves by 21 dpi (Fig. 5C and D). ClYVV IA-2017-infected plants were stunted in growth, whereas ClYVV IA-2016-infected plants were not stunted in comparison to the mock-inoculated control plants (Fig. 5 A and C). ClYVV IA-2016 and ClYVV IA-2017 infection was verified in Acre Edge 22R269 plants by RT-PCR (Fig. 5E) using detection primers designed from the assembled contigs (Additional file 13: Table S3) followed by Sanger sequencing (Additional file 6: Fig. S6). Neither ClYVV isolate was able to infect Williams 82 or the 41 NAM parents. Additionally, Acre Edge 22R269 infected plants were used as a source of inoculum to infect Broad Windsor and Robin Hood plants. At 21 dpi, Robin Hood plants infected with either ClYVV IA-2016 or ClYVV IA-2017 were severely stunted with mottling patterns on systemic leaves (Additional file 7: Fig. S7A). Broad Windsor plants infected with ClYVV IA-2016 or ClYVV IA-2017 developed symptoms with mottling and leaf curling (Additional file 7: Fig. S7B). ClYVV IA-2016 and ClYVV IA-2017 in both broad bean cultivars was verified by RT-PCR amplification (Additional file 7: Fig. S7C) using detection primers designed from the assembled contigs (Additional file 13: Table S3) followed by Sanger sequencing (Additional file 6: Fig. S6).

Transmission electron microscopy (TEM) was performed to visualize ClYVV IA-2017-infected N. benthamiana cells at 10 dpi. Infected N. benthamiana tissues contained flexuous filamentous virus particles of 760–780 nm in length, and cytoplasmic and nuclear inclusions typical of those caused by ClYVV and other potyviruses [64]. ClYVV was observed in the phloem parenchyma and mesophyll cells of infected tissues. The cylindrical cytoplasmic inclusions comprising laminated aggregates induced by ClYVV were observed in the phloem parenchyma cells of the vascular bundle and surrounding mesophyll cells (Fig. 6B) but were absent in corresponding cells of mock-inoculated plants (Fig. 6A). Numerous cylindrical cytoplasmic inclusions with laminated aggregates were found in the cytoplasm of two adjacent mesophyll cells (Fig. 6C) and scattered throughout the cytoplasm of the vascular parenchymal cell located above a xylem element and mesophyll cell (Fig. 6D–F). The crystalline nuclear inclusions in the form of cuboidal or rhomboidal crystals induced by ClYVV was strongly associated within the nucleolus of the vascular parenchymal cell located above a xylem element and mesophyll cell (Fig. 6D). Additionally, several cuboid or rhomboid crystalline inclusions were observed in the cytoplasm of two adjacent mesophyll cells (Fig. 6C), phloem parenchyma cell 0 (Fig. 6B), and the vascular parenchymal cell located above a xylem element and mesophyll cell (Fig. 6D–F). Large areas containing virions aligned in parallel were observed in the cytoplasm of an infected vascular parenchymal cell (Fig. 6E). These results confirmed the presence of viable ClYVV identified in field-grown soybean.

The potential for seed transmissibility of ClYVV IA-2017 was tested in Acre Edge 22R269. Seeds were collected from five ClYVV IA-2017-infected plants that were symptomatic and positive in the ImmunoStrip for Potyvirus Group assay (Additional file 8: Fig. S8A). Seeds were also collected from two mock-inoculated control plants. The 500 progeny seedlings tested from the five ClYVV-infected parents (P1 to P5) did not exhibit disease symptoms and appeared identical to the progeny seedlings from the mock-inoculated parents. For each parent, leaf samples from ten pools of ten seedlings were collected at 14 days after germination and assayed using the ImmunoStrip for Potyvirus Group (Additional file 8: Fig. S8B). Consistent with the lack of symptoms, all 50 pools tested negative for potyvirus, indicating that ClYVV-IA-2017 is either not seed transmissible, or seed transmission occurs at a frequency of less than 0.2% in Acre Edge 22R269.

Of the diverse soybean lines that were tested, Acre Edge 22R269 was unique in its ability to support systemic infection by ClYVV-IA-2016 and ClYVV-IA-2017. This led us to investigate whether Acre Edge 22R269 is generally susceptible to ClYVV. Acre Edge 22R269 seedlings were inoculated with ClYVV-No. 30 [65], soybean mosaic virus strain N (SMV-N) as a positive control, and two chimeric viruses ClYVV-No. 30 containing HC-Pro from SMV-N and SMV-N containing HC-Pro from ClYVV-No.30 [66]. The seedlings were susceptible to SMV-N and SMV-N containing the HC-Pro from ClYVV-No.30, but not to ClYVV-No. 30 or the chimeric ClYVV-No. 30 carrying the SMV-N HC-Pro (Additional file 16: Table S6). These results indicate that Acre Edge 22R269 is not generally susceptible to ClYVV, and further suggest that the host range determinant is not the HC-Pro protein [66].

Three contigs (RNA1, RNA2 and RNA3 segments; Fig. 7) identified from sample S78 collected in Iowa, USA in 2016 had high sequence identity to various ilarviruses in BLASTN searches against the GenBank nr database. The RNA1 segment was approximately 75% identical to parietaria mottle virus (PMoV) (FJ858202.1, MZ405646.1, KT005243.1, AY496068.1), tomato necrotic spot virus (MH780154.1), and bacopa chlorosis virus (FJ607140.1). The RNA2 segment had the highest sequence coverage and 76% sequence identity to PMoV strain CR8 (FJ858203.1), and the RNA3 segment had the highest query coverage and 72% sequence identity to bacopa chlorosis virus (JQ015298.1). The tripartite genome organization is consistent with other ilarviruses (Fig. 7). RNA1 encodes a 1,092 aa protein containing viral RNA methyltransferase and RNA helicase domains (ORF1a), RNA2 encodes the RNA dependent RNA polymerase (RdRp; ORF2a, 812 aa) and the overlapping ORF2b (201 aa), and RNA3 encodes the movement protein (MP; 293 aa) and coat protein (CP; 246 aa). Because the highest nucleotide identity for any RNA segment was only 76%, we suggest that this is a novel Ilarvirus, for which we propose the name, soybean ilarvirus 1(SIlV1) IA-2016 isolate. PCR primer sets (Additional file 13: Table S3) were designed to different regions of RNA1, RNA2, and RNA3 to confirm the presence of this virus in S78, and the PCR amplicons were sequenced to confirm their identity (Additional file 4: Fig. S4).

To further investigate the phylogenetic relationship of SIlV1 to 26 other ilarviruses, we performed multiple sequence alignments using the replicase (ORF1a), RdRp (ORF2a), MP, and CP aa sequences and generated phylogenetic trees (Fig. 8). Maximum-likelihood trees based on the aa alignments showed that SIlV1 is a new member of subgroup 1 that includes PMoV, blackberry chlorotic ringspot virus (BCRV), TSV, strawberry necrotic shock virus (SNSV), privet ringspot virus (PRSV), ageratum latent virus (AgLV), and cape gooseberry ilarvirus 1 (CGIV-I) (Fig. 8).

We detected six plant viruses well known to infect soybean: TSV, AMV, TRSV, SbDV, BPMV, and SVNV, and they occurred in both single and mixed infections (Tables 1, 2, and Additional files 11 and 12: Tables S1 and S2). The presence of all viruses in field samples (S1-S78) was verified by PCR amplification using detection primers designed from the assembled contigs (Additional file 13: Table S3) followed by Sanger sequencing (Additional file 4: Fig. S4).

TSV and AMV were each present in 27 of the 78 samples. In BLASTX searches, the TSV sequences from all 27 samples had > 94% amino acid identity to the reference sequences (GenBank accession numbers: RNA1: ACJ38087.1, ALF12288.1; RNA2: AMB72705.1, ALF12289.1, ACJ38088.1; RNA3: ACJ38090.1, ALF12291.1) (Additional file 15: Table S5). TSV was present in samples from Iowa, Wisconsin, Indiana, Maryland, and Alabama. Sequence analysis suggested that 10 distinct isolates were represented in these samples: IA-2016 (S3), IA-1-2017 (S9), IA-2-2017 (S10), IA-3-2017 (S11), IA-1-2018 (S22), IA-2-2018 (S23), IA-1-2019 (S27), IA-2-2019 (S28), MD-2016 (S34), and WI-2016 (S38).

A total of 27 samples collected from Iowa, Ohio, and Missouri contained full-length AMV sequences (Tables 1, 2 and Additional files 11 and 12: Tables S1 and S2). In the BLASTX analysis, AMV sequences identified in all 27 samples had > 97% amino acid identity to the reference sequences (GenBank accession numbers: RNA1: AXP79048.1; RNA2: ADO85716.1, AGV15825.1, CBJ56556.1; RNA3: AAL73140.2, AWK57376.1, AWK57376.1) (Additional file 15: Table S5). AMV was present in samples from Iowa, Ohio, and Missouri. Sequence analysis identified 8 distinct isolates: IA-2017 (S4), IA-1-2018 (S15), IA-2-2018 (S16), IA-3-2018 (S17), IA-4-2018 (S18), OH-2017 (S39), OH-2-2017 (S40), and MO-2017 (S41).

TRSV was identified in 10 samples from 4 Iowa counties, USA during the 2017 growing season (Tables 1, 2 and Additional file 12: Table S2). Sequence analysis identified three distinct isolates: IA-1-2017 (S12), IA-2-2017 (S13), and IA-3-2017 (S14) that shared > 91% sequence identity to the TRSV reference genomes previously reported in California USA, Australia, and South Korea (GenBank accession numbers: RNA1: U50869.1; and RNA2: MH427298.1, KJ556850.1) (Additional file 15: Table S5). The presence of the three distinct TRSV isolates infected in N. benthamiana and soybean was verified by PCR amplification using detection primers designed from the assembled contigs (Additional file 13: Table S3) followed by Sanger sequencing (Additional file 9: Fig. S9). SbDV was detected in 4 soybean samples from 1 Iowa county, USA collected over the 2016–2018 growing seasons (Tables 1, 2). Sequence analysis suggested 4 distinct isolates IA-2016 (S2), IA-2017 (S8), IA-1–2018 (S20), and IA-2-2018 (S21) that shared > 97% sequence identity to the SbDV reference genome previously reported in Illinois USA (GenBank accession number: KJ786321.1) (Additional file 15: Table S5). BPMV was detected in 3 samples collected from 2 counties in Iowa, USA during the 2017 and 2018 growing seasons (Tables 1, 2). Sequence analysis suggested 3 distinct isolates IA-2017 (S5), IA-2–2017 (S6), and IA-2018 (S19) that shared > 98% sequence identity to the BPMV reference genomes previously reported in Iowa, USA and Kentucky, USA (GenBank accession numbers: RNA1: GQ996948.1, GQ996951.1; and RNA2: AF394609.1, GQ996949.1) (Additional file 15: Table S5).

Finally, SVNV was detected in 12 samples collected from 5 different states (Indiana, Delaware, Maryland, Alabama, and Ohio) in 2016 and 2017 (Table 2; Additional files 11 and 12: Tables S1 and S2). These samples were specifically collected based on vein necrosis symptoms. Sequence analysis identified 8 distinct isolates: IN-1-2016 (S29), IN-2-2016 (S30), IN-3-2016 (S31), DE-2016 (S32), MD-2016 (S33), AL-1-2016 (S35), AL-2-2016 (S36), and AL-3-2016 (S37) that shared > 96% sequence identity to the SVNV reference genomes previously reported in Arkansas USA (Milan_TN isolate; GenBank accession numbers: L RNA: GU722317.1; M RNA: GU722318.1; S RNA: GU722319.1, respectively) (Additional file 15: Table S5). SVNV recently emerged as a soybean pathogen, and previous studies indicated that there is low genetic diversity in this virus. We were interested in whether there has been a change in the genetic variation of this virus over time. In order to test this, the complete sequences of the nucleocapsid protein (NP) genes of the 8 SVNV isolates identified in this study along with the NP genes of 15 previous isolates [16, 68] were obtained. Pairwise comparison of the NP gene of 8 isolates with the 15 isolates revealed identities of 98 to 99.8% at the nucleotide level and 97 to 100% at the amino acid level (Additional file 17: Table S7). Additionally, the pairwise comparison of the NP gene of the 8 new SVNV isolates with each other revealed that there is 96 – 98% nt sequence identity and 97 – 100% amino acid sequence identity. These results suggest that there continues to be low diversity in NP sequences among SVNV isolates.

Mixed infections of viruses that co-occurred in soybean were observed in 14 samples. The co-infections of AMV and TSV were detected in 4 samples (S18, S24, S25, and S26) collected from 4 different Iowa counties in 2018 (Table 2). In 2017, co-infections of AMV and TRSV, as well as AMV and BPMV, were detected in 2 samples (S12 and S13) and one sample (S6) from Iowa, respectively. Additionally, mixed infections of AMV and SbDV were identified in one sample (S20) from Iowa in 2018. Co-infections of TSV and TRSV were found in one sample (S11) collected in 2017 while TSV and SbDV were detected in one sample (S21) collected in 2018 from Iowa. Mixed infections of SVNV and TSV were identified in 3 samples (S31, S34, and S37) from 3 different states (Indiana, Maryland, and Alabama) in 2016, while co-infections of SVNV and AMV were detected in one sample (S40) from Ohio in 2017. The presence of mixed infection of viruses was verified by PCR amplification using detection primers designed from the assembled contigs (Additional file 13: Table S3) followed by Sanger sequencing (Additional file 10: Fig. S10).