Diverse Novel Viruses Coinfecting the Tropical Ornamental Plant Polyscias balfouriana in China

The viromic profile of Polyscias balfouriana cv. Marginata, a perennial woody and ornamental plant, was determined using ribosomal RNA-depleted total RNA (rRNA-depleted totRNA) sequencing. Five viruses (i.e., polyscias mosaic virus, PoMV; one potential novel rhabdovirus; and three novel viruses of Betaflexiviridae and Closteroviridae) were detected and prevalence-surveyed in Hainan province, China. The genomes of polyscias capillovirus 1 (PCaV-1) and polyscias citrivirus 1 (PCiV-1) of family Betaflexiviridae were completed, and the genomes of polyscias crinivirus 1 (PCrV-1) of Closteroviridae were nearly completed lacking the 5′ and 3′ termini. PCaV-1 shares 68% genome nucleotide (nt) identity and 66% replicase (Rep) amino acid (aa) identity with homologues in apple stem grooving virus (ASGV). PCiV-1 shares 65% genome nt identity and 64% Rep aa identity with homologs in citrus leaf blotch virus (CLBV). Meeting the species demarcation criteria, PCaV-1 and PCiV-1 were considered to be new species in genera Capillovirus and Citrivirus, respectively. PCrV-1 shares high genome nt identity (62%), heat shock protein 70-like protein (HSP70h) and RNA-dependent RNA polymerase (RdRp) aa identity (78–80%) with homologues in tomato chlorosis virus (ToCV). We tentatively consider PCrV-1 to be an unclassified member of the Crinivirus genus. PoMV, PCaV-1, PCiV-1, and PCrV-1 are the prevalent viruses with >73% occurrence in the Xinglong Tropical Botanical Garden, Hainan, China.

Polyscias balfouriana, also known as Aralia balfouriana or balfour aralia, is a perennial woody plant usually cultivated as an ornamental species in gardens or indoors. This plant is attractive and highly valued for its beautiful foliage. The main cultivar is P. balfouriana was purified from the sample that pooled P. balfouriana Marginata and other plants and sequenced on an Illumina platform with a paired-end 150 bp read length [7]. We obtained a total of 103,474 clean reads assembled into 132 contigs, 45 of which were viral contigs by BLASTx analysis. Thirteen contigs with lengths ranging from 250 to 604 nucleotides (nt) hit the viruses of the Capillovirus and Citrivirus genera in the Betaflexiviridae family by sharing 54.6% to 92.6% amino acid (aa) identity with different viral proteins. Similarly, 32 contigs with lengths ranging from 288 to 1343 nt hit the tomato chlorosis virus (ToCV) of the Crinivirus genus in the Closteroviridae family by sharing 32.1% to 85.9% aa identity. We thus preliminarily predicted the presence of multiple new viruses from the Betaflexiviridae and Closteroviridae families in the balfour aralia sample.
On the basis of dsRNA sequencing results, to further determine the virome of P. balfouriana, we conducted rRNA-depleted totRNA sequencing to identify novel viruses, molecularly characterise them, and assess their occurrence in Hainan province, China.

Plant Materials and HTS
The original P. balfouriana sample showing no viral symptoms compared to the healthy plants was collected from Xinglong Tropical Botanical Garden, Hainan province, China, in 2019 ( Figure 1A). dsRNA was purified from 0.75 g of fresh leaf material and sequenced (2 × 150 bp) on an Illumina Novaseq 6000 instrument for the preliminary screening of viruses using previously described methods. In 2021, we conducted rRNA-depleted totRNA sequencing on the same individual P. balfouriana plant for detailed viromic screening. Total RNA extraction, rRNA depletion, and subsequent library preparation were conducted at LC-Bio Technology Co., Ltd., Hangzhou, China using previously described methods [9]. Briefly, 2 µg of total RNA was treated for the removal of rRNA using the Epicentre Ribo-Zero Gold Kit (Illumina, San Diego, CA, USA). The rRNA-depleted RNAs were then fragmented using the Magnesium RNA Fragmentation Module (New England Biolabs, Ipswich, MA, USA) at 94 • C for 5 to 7 min. The cleaved RNA fragments were reversetranscribed into cDNA using SuperScript™ II Reverse Transcriptase with random hexamer primer (Invitrogen, Carlsbad, CA, USA). PCR products were sequenced (2 × 150 bp) on an Illumina Novaseq 6000 instrument following the manufacturer's procedure [9].

Data Processing and Virus Annotation
After the adapter trimming and filtering of raw reads, clean reads obtained from dsRNA sequencing were de novo assembled into contigs using IDBA-UD assembler v1.1 [10]. For rRNA depleted totRNA sequencing, since the reference genome of Polyscias balfouriana is not currently available, the genomic sequence of Panax japonicus from the same family Araliaceae (https://ftp.ncbi.nlm.nih.gov/genomes/all/GCA/020/ 205/505/GCA_020205505.1_ASM2020550v1/, accessed on 4 October 2021), was used for host-derived read elimination using HISAT v2.1.0 [11] to reduce the size of the dataset. The remaining reads were de novo assembled into contigs using Trinity v2.13.3 [12], and annotated using BLASTn and BLASTx [13] against the nonredundant nucleotide (nt) and protein (nr) GenBank databases [14] with a conservative e-value cut-off of 10 −4 .

Completion of Genomes of Viruses Discovered from HTS
The genomes of the newly identified viruses were reconstructed by reverse-transcription polymerase chain reaction (RT-PCR) with several pairs of specific primes designed on the basis of the generated viral contigs (Table S1) using TransScript ® One-Step gDNA Removal and cDNA Synthesis SuperMix (Transgen, Beijing, China) and SuperStar Max DNA Polymerase (GenStar, Beijing, China). The full genomes of the two new viruses of Betaflexiviridae were obtained with 5 and 3 untranslated regions (UTRs) completed, whereas the genome of the novel crinivirus of Closteroviridae lacked the 5 and 3 termini. The LD-oligo(dT) primer (5 -TGTGTTGGGTGTGTTTGGTTTTTTTTTTTTTTT-3 ) was used to synthesise the cDNAs of RNA viruses with poly(A) tails, and the dt-rev primer (5 -TGTGTTGGGTGTGTTTGG-3 ) paired with gene-specific primers was used to complete the 3 -terminal sequences. The 5 -terminal sequences were amplified using a SMARTer RACE 5 /3 Kit (Takara Bio, Mountain View, CA, USA; Table S1). The amplified PCR products were either directly sequenced or cloned into the pGEM-T Easy vector (Promega, Madison, WI, USA) before sequencing. The near-complete genomes of the crinivirus were determined on the basis of assemble scaffolds from rRNA-depleted totRNA sequencing.

Analysis of the Occurrence of the Three Novel Viruses
To further investigate the occurrence of the three newly identified viruses, we collected 22 P. balfouriana samples (three different cultivars) mainly exhibiting four different leaf morphologies (Figure 1A-G; detailed pictures of each leaf are shown in Figure S2) from Xinglong Tropical Botanical Garden (18 • 44 06.6 N 110 • 11 36.1 E), Hainan province, China in 2021. RT-PCR was conducted to test the occurrence of five discovered viruses from HTS, with detection primers listed in Table S1.

Results of rRNA-Depleted totRNA Sequencing and Discovery of Viruses
The rRNA-depleted totRNA sequencing experiment yielded 76,048,918 raw reads (72,691,030 clean reads) with Q20 and Q30 values greater than 93%, and 30,349,892 (41.8%) reads mapped to the host reference genome, while 42,341,138 (58.2%) unmapped reads were de novo assembled into 286,133 contigs. Similar to the preliminary dsRNA sequencing results, BLASTn and BLASTx analyses revealed that 1738 contigs corresponded to viruses belonging to the Betaflexiviridae (Capillovirus and Citrivirus genera), Closteroviridae (Crinivirus genus), and Caulimoviridae and Rhabdoviridae (Table 1) families. Of the 1738 contigs, 113 were long (1000-6219 nt) and had good coverage of the viral genomes. Three long contigs of 6219, 5344, and 3704 nt shared~90% nt identity (95% coverage) with the complete genome of the polyscias mosaic virus (PoMV, NC_055562), a recently identified badnavirus from Polyscias fruticosa L. (Ming aralia) [8]. Considering the 7.14% nt divergence of RT + RNAse H sequences, lower than the species demarcation criteria of 20% in the Badnavirus genus, the badnavirus found in this study was considered to be PoMV, which was for the first time discovered from P. balfouriana in China. One 5855 nt contig shared 60.57% aa identity (96% coverage) with the RdRp of cabbage cytorhabdovirus 1 (YP_010086794), which indicated the presence of a potential novel cytorhabdovirus in the Rhabdoviridae family. No further characterisation was conducted for the viruses of Caulimoviridae and Rhabdoviridae. Thirteen long contigs (1000 to 4873 nt) corresponded to criniviruses with good coverage, and were used to assemble the near-complete scaffolds of polyscias crinivirus RNA1 and RNA2 genomes in this study.
Phylogenetic trees based on the Rep and CP aa sequences of representative members of the Betaflexiviridae family showed that these capilloviruses were separated into two clades, and PCaV-1 clustered with ASGV separate from MuVA, CVA, and CuVA (Figure 2), while PCiV-1 clustered with other citriviruses in a single clade. The trees were constructed using the neighbour-joining method, and the statistical significance of branches was evaluated by bootstrap analysis (1000 replicates). Bootstrap values above 50% are shown. Scale bar represents 50% aa sequence divergence. Newly discovered viruses are indicated by red circles (polyscias citrivirus 1, PCrV-1) and blue squares (polyscias capillovirus 1, PCaV-1). Genera are indicated on the right. Capillovirus genus (containing PCaV-1), pink background; Citrivirus, blue background.
On the basis of the accepted species demarcation molecular criteria for the family Betaflexiviridae, which are of 72% nt identity (or 80% aa identity) for Rep and CP genes [16], polyscias capillovirus 1 and polyscias citrivirus 1 were tentatively considered to be novel species in the Capillovirus and Citrivirus genera, respectively. Recently, several tentative novel capilloviruses have been identified from perennial woody plants, including loquat [17], birch [18] and rubber [19]. Additionally, some new citriviruses were identified from Nandina domestica [20] and ornamental peony [21]. Most capilloviruses and citriviruses have been originally isolated from woody plants, in this case, P. balfouriana.
On the basis of the accepted species demarcation molecular criteria for the family Betaflexiviridae, which are of 72% nt identity (or 80% aa identity) for Rep and CP genes [16], polyscias capillovirus 1 and polyscias citrivirus 1 were tentatively considered to be novel species in the Capillovirus and Citrivirus genera, respectively. Recently, several tentative novel capilloviruses have been identified from perennial woody plants, including loquat [17], birch [18] and rubber [19]. Additionally, some new citriviruses were identified from Nandina domestica [20] and ornamental peony [21]. Most capilloviruses and citriviruses have been originally isolated from woody plants, in this case, P. balfouriana.
A close molecular relationship was found among PCrV-1, ToCV, and SSVV on the basis of the identities of encoded proteins (Table S3) and phylogenetic analyses ( Figure S1). The accepted species demarcation molecular criteria for the Closteroviridae family are 75% aa identity for HSP70h, RdRp, and CP genes. The HSP70h and RdRp genes of PCrV-1 share higher aa identities (78% and 79%) with those of ToCV, while only the CP gene shares 71% aa identity satisfying the criteria.
To further clarify the taxonomy of PCrV-1, we compared the aa identities of HSP70h and Rep between the 13 reported formal species in the genus Crinivirus (Table S4). We discovered higher HSP70h and Rep identities (>75%, the criteria) among four viruses: bean yellow disorder virus (BYDV), lettuce chlorosis virus (LCV), tetterwort vein chlorosis virus (TwVCV), and cucurbit yellow stunting disorder virus (CYSDV), and between strawberry pallidosis-associated virus (SPaV) and diodia vein chlorosis virus (DVCV ; Table S4). Thus, with no biological data for vector species or serological specificity, we could not determine PCrV-1 to be a new species in the Crinivirus genus only on the basis of the sequence identities in this study, although its host is a different perennial woody plant from the herbaceous hosts of ToCV [22]. Therefore, we tentatively considered polyscias crinivirus 1 to be an unclassified member of the Crinivirus genus. Since the hosts of criniviruses are generally herbaceous plants [23], P. balfouriana is the second known woody host for criniviruses alongside mulberry [24].

Occurrence of Five Discovered Viruses in P. balfouriana in Hainan Province, China
The most prevalent virus in P. balfouriana in Xinglong Tropical Botanical Garden in Hainan province is badnavirus PoMV with occurrence of 86%, followed by PCrV-1 (18/22, 82%), PCaV-1 and PCiV-1 (16/22, 73%), and the uncharacterised novel rhabdovirus (32%, 7/22). Only two samples had no target viruses ( Figure S2). Although various morphological samples were tested, coinfection was still common, with 14 out of 22 (63.6%) samples being positive for the four target viruses excluding the novel rhabdovirus. Inspired by the phenomenon of the tulip breaking virus being one of five plant viruses causing the colour breaking of tulip flowers, it would be interesting to explore whether the coinfected viruses identified in this study contributed to the appearance of balfour aralia. However, no significant correlations could be concluded between the morphologies and symptoms, and the identified viruses on the basis of the limited data in this study. The high prevalence of the four popular viruses (PoMV, PCaV-1, PCiV-1, PCrV-1) could be due to vegetation propagation from mother plants. Some undiscovered viruses present in the samples may also contribute to the appearances various leaves. Further investigations are necessary to assess the virome, and evaluate the distribution, prevalence, and potential pathogenicity of these new viruses.
Supplementary Materials: The following supporting information can be downloaded at https: //www.mdpi.com/article/10.3390/v14061120/s1: Table S1: primers used for the completion of genomes of PCaV-1 and PCiV-1, and for the detection of the three viruses; Table S2: amino acid identities of Rep and CP between PCaV-1 and PCiV-1, and other members in genera Capillovirus and Citrivirus; Table S3: amino acid identities of HSP70h, RdRp, and CP between PCrV-1, and other formal and unclassified members in the genus Crinivirus; Table S4: amino acid identities of HSP70h (lower left) and RdRp (upper right) between formal species in genus Crinivirus; Figure S1: phylogenetic trees constructed on the basis of HSP70h and CP amino acid sequences of formal and tentative members in the Crinivirus genus; Figure

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.