Sample collection

A total of twenty tea leaf samples exhibiting viral disease symptoms were collected in 2021 and 2022 from different tea plantations in the Guilan and Mazandaran provinces in the north of Iran. The collected plants showed a range of virus-like symptoms, including necrotic ring blotch, discoloration (albino and chlorine), yellowing, ring spots, and mosaic. This study is based on integrated analyses of two original transcriptomic datasets generated by our group [14]. The first dataset was produced from an RNA-seq experiment conducted in 2021, and the second from a separate RNA-seq experiment conducted in 2022. The methodology for library preparation and sequencing was consistent across both projects and is detailed in the Materials and Methods section below.

Total RNA extraction, library construction and sequencing

High-throughput sequencing (HTS) was performed on 20 tea leaf samples. Total RNA was extracted from approximately 100 mg of leaf tissue from each sample using the RiboEX Kit (GeneAll, South Korea) according to the manufacturer’s instructions. The quality and concentration of the extracted RNA were assessed using agarose gel electrophoresis and Nanodrop spectrophotometer (Thermo Fisher Scientific, USA). Equal amounts of RNA from each sample were pooled. Integrity and purity of the total RNA were also analyzed using a Bioanalyzer 2100 based on an RIN number > 7. RNA-Seq libraries were then prepared using the TruSeq Stranded Total RNA Preparation Kit with Ribo-Zero depletion according to the manufacturer’s instructions. The libraries were sequenced by Novogene (China) on an Illumina HiSeq 2000 platform, generating 150 bp paired-end reads.

Bioinformatic analysis

Raw reads produced from the HTS were processed using CLC Genomics Workbench (version 20, QIAGEN, Venlo, The Netherlands) to remove adapter sequences and low-quality sequences with more than two ambiguous nucleotides. The cleaned and trimmed reads were then mapped to the tea genome (Camellia sinensis genome GCF_004153795.1_AHAU_CSS) using CLC Genomics Workbench 20 with default parameters and a length fraction and similarity fraction set to 0.8. From the clean reads, 27,244,799 and 78,269,958 reads were mapped to the tea genome. After removing host-derived reads, the remaining unmapped reads were de novo assembled into contigs using the ‘De Novo Assembly’ tool in CLC Genomics Workbench v20. The assembly was performed with a word size (for read overlap) of 20 and a minimum contig length of 200 nucleotides. The resulting contigs were compared with sequences available in the NCBI viral reference database (https://www.ncbi.nlm.nih.gov/genome/viruses/) using the BLASTn tool within CLC Genomics Workbench. The search parameters were set with an E-value cutoff of 1e-5. For enhanced confidence in virus detection, viral hits were filtered using the following criteria: contig length ≥ 700 nt and Open Blast Output results. Contigs for which the Greatest HSP length in the BLAST results was ≥ 700 nucleotides were considered highly likely to be of viral origin. The nucleotide identity of these selected contigs to their closest viral matches was high (e.g., > 84% for TPNRBV and >78% for CoAV1 isolates identified in this study). This length filter was applied to prioritize the assembly and analysis of conventional plant viruses.

To ensure our analysis was also sensitive to very small viral agents such as viroids, we conducted an additional, BLASTn search of all contigs against the viroid database. No significant hits to small agents (e.g., viroids) were identified in this dataset.

Verification of HTS data

To confirm the presence of TPNRBV in the individual samples used for the HTS pool, total RNA was reverse transcribed using the Easy cDNA Synthesis Kit (Parstous, Iran) with random hexamer primers, according to the manufacturer’s protocol. TPNRBV presence was confirmed by PCR using two specific primer pairs, TPNRBV3-F/TPNRBV3-R (5’-TTCGCCACTCACAAAGACAACAAACT-3’/5’-GTAGCGGAGCGGAAAGAAAAGACT-3’) and TPNRBV-MP-F/TPNRBV-MP-R (5’- GCGGATCCATGTCTATAGC-3’/5’- ACAAGCTTAAAAGTAATAGGTGGTACAGC −3’) [1,4]. PCR amplifications were performed in a final volume of 25 μL, containing 3 μL of template cDNA, 1.5 μL of each primer, 12.5 μL of 2 × PCR Master Mix, and deionized water to a final volume of 25 μL. The PCR program conditions were as follows: initial denaturation at 95 °C for 5 min, followed by 35 cycles at 95 °C for 30 s, 54 (TPNRBV3-F/R) and 50 (TPNRBV-MP-F/) for 30 s, and 72 °C for 40 s (for the RNA3 segment) or 60 s (for the MP), with a final extension at 72°C for 7 min. The amplified PCR products were electrophoresed in a 1% agarose gel stained with SYBR Safe (SinaClon, Iran) to verify the appropriate fragment size. The obtained amplicons were then subjected to Sanger sequencing by Sinuhe Biotech Company (Iran).

De novo assembly of the TPNRBV genome and SNV prediction

To assemble the complete genome sequences of TPNRBV from the RNA-Seq data, an initial sensitive mapping step was implemented. Given the low abundance of viral RNA within the host transcriptome and to accommodate potential genetic divergence of the local isolate, permissive alignment parameters were used to maximize viral read recovery. Specifically, trimmed reads were mapped to the TPNRBV reference sequences (NC_040401-NC_040404) using CLC Genomics Workbench 20 with a length fraction and similarity fraction set to 0.5 and 0.8, respectively. To identify SNVs in the assembled virus genomes, the raw RNA-Seq data were mapped to the reference genomes using CLC Genomics Workbench 20 with the following parameters: minimum coverage of 2, minimum variant frequency of 0.01, and maximum variant p-value of 10⁻⁶. After filtering out synonymous SNVs, the frequency and distribution of polymorphisms across the four segments of the TPNRBV genome were analyzed. SNV prediction was performed by detecting nucleotides that differed from the reference sequences. The positions of identified single-nucleotide variations (SNVs) were visualized on the 3D protein structures by integrating the Protein Data Bank (PDB) from the RCSB PDB database with CLC Genomics Workbench. It is important to note that, as the RNA-seq data originated from a pool of 20 plants, the identified SNVs and their frequencies represent a composite of viral sequences from the entire sampled population. Therefore, a low-frequency variant could represent either a true minor variant within a single host’s viral population or a fixed variant in a virus infecting only a small subset of the plants in the pool.

TPNRBV gene expression using RNA‑Seq data

Analysis of transcript expression levels for each coding DNA sequence (CDS) in the TPNRBV genome (TPNRBV ORFs) was performed by mapping the trimmed reads to the TPNRBV Refseq using CLC Genomics Workbench with the following parameters: length fraction of 0.8, similarity fraction of 0.8, costs of 2 for mismatches and deletions, and 3 for insertions. The mapping results were used to calculate the transcripts per kilobase million (TPM) for each ORF. TPM normalization was performed using only the total number of reads mapped to the TPNRBV reference genome, thereby enabling comparison of relative expression levels among viral ORFs independent of host-derived reads

Phylogenetic analysis/ construction of phylogenetic trees

The consensus sequences obtained from the preceding analyses (for the four segments of the TPNRBV genome) were subjected to a BLASTn search to identify the most similar sequences, which were then used to construct phylogenetic trees. The phylogenetic relationships among the Iranian TPNRBV isolate and genome sequences retrieved from GenBank were analyzed using MEGA version 11 [15]. To determine the phylogenetic position of the Iranian TPNRBV isolates, four phylogenetic trees were constructed using the maximum likelihood (ML) method with 1000 bootstrap replicates. The nucleotide substitution models TN93 + G + I, GTR + G, HKY + G, and HKY + G were applied to RNA1-RNA4. Additionally, the nucleotide sequence identity percentages for the four segments of the TPNRBV genome were calculated using the Sequence Demarcation Tool (SDT v1.2) software [16], which was used to generate pairwise nucleotide sequence identity matrices.

Recombination analysis

Potential recombination events were identified in the four segments of the TPNRBV genome. The analysis was conducted using the Recombination Detection Program (RDP v4.97) [17] with multiple algorithms, including GENECONV, Bootscan, Chimaera, MaxChi, SiScan, 3Seq, and RDP, all run with default settings. Only recombination events detected by at least four of these algorithms were considered reliable and included in the analysis.

Collection of tea leaves for tea virome study and library preparation

A total of 20 leaf samples were collected from 16 different geographical locations in 2021 and 2022. As shown in Fig 1, the collected tea leaves exhibited symptoms of ring blotch and foliage discoloration (Fig 1).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Viral symptoms were observed on tea plants in the tea plantations of Guilan and Mazandaran provinces.

https://doi.org/10.1371/journal.pone.0354634.g001

Total RNAs were extracted from each plant sample, and two composite (pooled) samples were subjected to RNA-Seq. The samples were pooled based on the collection years. Two separate libraries were prepared, and paired-end sequencing was performed using an Illumina HiSeq 2000 system.

De novo transcriptome assembly

High-throughput sequencing of the two libraries yielded a total of 38,125,708 and 103,830,264 raw paired-end reads for the 2021 and 2022 samples, respectively. After removing adapter sequences, low-quality reads, and host-derived reads, 10,861,237 and 25,603,934 clean reads were obtained from each library, respectively. De novo transcriptome assembly of the unmapped reads resulted in 11,638 contigs for library 2021 and 34,454 contigs for library 2022. The approximately threefold increase in the total number of contigs assembled from the 2022 library compared to the 2021 library (34,454 vs. 11,638) is consistent with the proportional increase in clean sequencing data available for assembly (~2.7-fold higher), reflecting expected technical variations between sequencing runs. The number of TPNRBV-specific contigs was similar in both libraries (36 vs. 35), confirming the consistent and dominant presence of the virus in the sampled pools. BLASTn analysis showed that the majority of the contigs were associated with TPNRBV in both libraries. The remaining non-TPNRBV contigs were examined using stringent criteria (contig length ≥ 700 nucleotides and meaningful coverage of viral genomic regions in BLASTn). While some showed short, partial homology to other plant viruses or to non-plant viral sequences (e.g., phages or mycoviruses likely associated with the plant microbiome), none met the thresholds for confident identification of an additional plant virus. Therefore, TPNRBV was confirmed as the dominant and robustly detected plant virus in these transcriptome libraries.

A total of 35 TPNRBV-related contigs were identified from library 2021, and 36 from library 2022. Despite the larger total dataset in 2022, the key difference was in the sequencing depth supporting the viral genomes, which was 4- to 5-fold higher (e.g., ~ 3.8 million vs. ~ 0.8 million mapped reads for RNA1). This increased depth provided greater confidence in the genome assemblies and enhanced sensitivity for variant detection. The mapped reads were sufficient to cover the full viral genome, as confirmed by the high-coverage maps (Fig 2a, b). Importantly, no region of the TPNRBV genome in either library showed coverage below the minimum threshold of 2 × , and no systematically low-coverage regions were observed, indicating that the assemblies are unlikely to be affected by coverage-related errors.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2.

a, b. Reads coverage based on alignment of sequenced reads on the reference TPNRBV genome (RNA-seq reads mapping to the reference TPNRBV genome).

https://doi.org/10.1371/journal.pone.0354634.g002

The depth of coverage across the four genomic segments was not uniform, a common observation in transcriptomic studies of multipartite RNA viruses. Despite this variation, the coverage maps (Fig 2a, 2b) confirm that the entire length of all segments was covered, with no regions falling below the minimum coverage threshold of 2X required for reliable SNV detection. This ensures the completeness of the genome assemblies and the robustness of the variant analysis

The complete genome sequences of the Iranian TPNRBV isolates (designated TPB-Iran and TPNRBV-Ir) were found to be 5922 nt for RNA1, 4107 nt for RNA2, 2678 nt for RNA3, and 2272 nt for RNA4, with poly-A tails. These sequences have been deposited in GenBank under accession numbers ON475416–ON475419 and PV700595- PV700598. The assembled genome sequences showed the highest pairwise nucleotide identity to isolates from China. BLASTn analysis revealed high-confidence matches (E-value = 0.0, Query Coverage = 100%) with the following percent identities for each segment: RNA1: 96.13% to a Fujian isolate (OQ948458); RNA2: 96.38%; RNA3: 97.27%; and RNA4: 94.89% to Hangzhou isolates (NC_040402-NC_040404). Furthermore, BLASTx analysis confirmed the presence of conserved domains characteristic of TPNRBV, including methyltransferase-helicase (RNA1), helicase-RNA polymerase (RNA2), a virion membrane protein (RNA3-P24), and a movement protein (RNA4). A comprehensive pairwise identity matrix is provided in Fig 5.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Frequency, type, and positions of identified SNVs on the TPNRBV genome. a, b, c: RNA seq1; d,e,f: RNA seq2.

https://doi.org/10.1371/journal.pone.0354634.g003

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Expression levels (transcript per million, TPM) of Tea plant necrotic ring blotch virus (TPNRBV) in RNA-seq data of infected plants.

https://doi.org/10.1371/journal.pone.0354634.g004

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5.

(left) Phylogenetic position of the two Iranian TPNRBV isolates based on the assembled consensus sequences for four segments of TPNRBV. Phylogenetic trees for four TPB-Iran and TPNRBV-Ir segments were constructed using MEGA 11 using the maximum likelihood (ML) method with 1,000 bootstrap replicates. right) Pairwise nucleotide sequence identity matrix was determined using SDT v1.2.

https://doi.org/10.1371/journal.pone.0354634.g005

Identification of sequence variants within the TPNRBV population

To assess the genetic diversity of the TPNRBV population across the sampled tea plants, intra-population single-nucleotide variants (SNVs) were called from the pooled transcriptomic data. By applying a 1% threshold for variant calling, 93 non-synonymous SNVs were identified, with frequencies ranging from 0.26% to 21.47% (Fig 3; S1 Table). The distribution of these frequencies was highly skewed: 90 out of 93 SNVs (96.8%) occurred at frequencies below 5%, consistent with a background of transient mutations. Only three variants were found at frequencies above 5% (5.76%, 14.44%, and 21.47%).

These non-synonymous SNVs were located exclusively in the open reading frame (ORF) regions, with none detected in the untranslated regions (UTRs). The details of the identified SNVs are as follows: 44 SNVs in RNA1, 37 in RNA2, 1 in RNA3, and 1 in RNA4. This uneven distribution resulted in a high density of non-synonymous variants in the replicase genes encoded by RNA1 and RNA2, while the movement protein (RNA4) and most ORFs of RNA3, including the highly expressed P22 gene, showed remarkably low variant counts. The distribution of these 93 non-synonymous SNVs was highly uneven, with the vast majority concentrated in RNA1 and RNA2. In contrast, RNA3 and RNA4 each contained only one SNV, and notably, no SNVs were identified in the P22 gene located in RNA3 at the 1% frequency threshold in this pooled dataset. Furthermore, 15 SNVs were shared between both samples from the 2021 and 2022 collections. The 15 SNVs conserved between the 2021 and 2022 pools are of particular interest, as their persistence may indicate a potential selective advantage or founder effect in the local TPNRBV population; their functional impact merits further investigation. Additionally, 9 insertions and one multi-nucleotide variant (MNV) were found (S1 Table). The distribution of SNVs across the four RNA segments of the TPNRBV genome is shown in Fig 3.

While linking these SNVs to the tertiary structure of the protein using the PDB database, no visible changes were detected in the 3D protein structure. This structural comparison provided a preliminary, qualitative assessment based on visual inspection of the superimposed models. Quantitative analyses of biophysical properties—such as changes in binding pocket volume, electrostatic surface potential, or protein stability—were not performed, as they require specialized molecular dynamics simulations or energy calculations that were beyond the primary genomic focus of this study. It is therefore possible that the identified mutations could affect functional properties not discernible in a static structural alignment.

Population-level expression profile of TPNRBV ORFs

The relative expression levels of TPNRBV ORFs across the infected plant population were assessed using the pooled RNA-seq data. Comparing the transcripts per million (TPM) values between the two pools, the highest expression was observed for P22, followed by P24, P14, and the movement protein. In contrast, helicase-RNA polymerase and methyltransferase-helicase were expressed at the lowest levels. The expression patterns of TPNRBV ORFs are illustrated in Fig 4. To further assess whether the high expression of the P22 gene could be influenced by technical artifacts, read coverage across the P22 ORF was examined. The P22 region showed continuous read coverage along its entire length in both RNA-seq libraries, with no abrupt drops or localized peaks indicative of assembly errors or post-transcriptional processing. Importantly, coverage across the P22 ORF did not fall below the minimum threshold required for reliable variant detection. The absence of detectable SNVs within P22 further supports the accuracy of its assembly and suggests that its elevated TPM values reflect true high transcript abundance rather than technical bias.

At the population level, our analysis showed no significant difference in viral transcript expression between the two datasets, with P22 being the most abundantly expressed gene and having the highest mRNA levels within the viral transcriptome.

RT-PCR for validation of the TPNRBV in HTS data

RT-PCR confirmed the presence of TPB-Iran and TPNRBV-Ir in the individual samples used for the pools using two specific primer pairs for TPNRBV. Amplicons of the expected sizes (459 and 948 bp) were obtained and sequenced using Sanger sequencing. BLASTn analysis of the resulting fragments confirmed the presence of TPNRBV in these plants.

RT-PCR analysis of the individual plant RNA extracts that made up the HTS pools confirmed the high prevalence of TPNRBV. The virus was detected in 19 out of 20 while CoAV1 detected in 5 out of the 20 samples (four of them were co-infected with TPNRBV). A single infection of CoAV1 was detected in an individual sample (S2 Table).

Phylogenetic analysis

The recovered consensus sequences of TPNRBV from the HTS data enabled us to compare the phylogenetic relationships of the TPB-Iran and TPNRBV-Ir isolates with other known TPNRBV isolates. BLASTn searches of the four TPNRBV genome segments showed high nucleotide sequence identity (94.89% to 97.27%) with the TPNRBV isolate from China (TPNRBV_HZ; NC_040401-NC_040404).

To understand the phylogenetic position of the Iranian TPNRBV isolates and gain insights into the virus’s genetic diversity, the complete genome sequence of the TPNRBV isolates obtained in this study were analyzed alongside 12 previously reported TPNRBV isolates. Phylogenetic trees were constructed using the Maximum Likelihood method with 1000 bootstrap replicates in MEGA11. The two major phylogenetic groups in the trees were designated Clade I and Clade II. These clades represent well-supported and consistent divisions, with bootstrap values ≥ 50% at the defining nodes (Fig 5). Detailed information, including accession numbers for the nucleotide sequences used in the phylogenetic analysis, is provided in S3 Table. Since the TPNRBV genome consists of four RNA segments, four independent phylogenetic trees were generated. The analysis of RNA1-RNA2 nucleotide sequences grouped the TPNRBV isolates into two main clades, which were further divided into sub-clades. In the trees constructed from the RNA2, RNA3, and RNA4 sequences, the two Iranian isolates (TPB-Iran and TPNRBV-Ir) were placed in clade II together with isolates from Hangzhou, China (TPNRBV_HZ). However, in the RNA1-based tree, the Iranian isolates were grouped in clade I with those from Fujian (Fig 5).

The high sequence similarity with the TPNRBV isolate from China suggests a potential connection between viral populations in these two geographically distinct locations. Although the phylogenetic tree showed the presence of main clusters, no clear separation by country of origin was observed. Recombination analysis using RDP4 did not detect any recombination events in the analyzed TPNRBV sequences. A pairwise nucleotide sequence identity matrix generated using SDT v1.2 revealed percentage identities of 93.90%–100% for RNA1, 94.10%–100% for RNA2, 95.80%–100% for RNA3, and 95.30%–100% for RNA4 (Fig 5).

The presence or absence of conserved nucleotide motifs at the 5’ termini of TPB-Iran and TPNRBV-Ir genome segments, as described by Maruyama et al. (2022), was also examined. Similar to isolates from Hangzhou, TPB-Iran and TPNRBV-Ir lacked these conserved sequences in RNA2, RNA3, and RNA4. However, such as TPNRBV_HZ, RNA1 of TPB-Iran and TPNRBV-Ir contained the sequence 5’-ATTAACGA-3,’ along with five additional nucleotides before the conserved motif (5’-TGGGGATTACGA-3’).

This shared, specific pattern of motif loss in RNA2-RNA4 coupled with its conserved (and extended) presence in RNA1 constitutes a notable synapomorphy (a derived trait indicative of common ancestry) between the Iranian and Hangzhou isolates. While prior studies suggest such terminal motifs are significant for viral replication and packaging in segmented RNA viruses [11,18], this precise, multi-segment deletion is unlikely to have arisen independently. Therefore, this structural signature provides stronger evidence for a direct evolutionary link between these isolates than nucleotide identity percentages alone.

Consistent with this link, both isolates also share a unique 110-nt insertion in the 3′ UTR of RNA4. In a study by Chen et al. (2023), the variability in the length of the 3’ non-coding regions of TPNRBV RNA4 was attributed to extensive deletions and insertions. A 110-nucleotide insertion, previously identified only in TPNRBV_HZ by Chen et al., was also found in both TPB-Iran and TPNRBV-Ir isolates. This insertion suggests that these isolates may have resulted from a single gene recombination/insertion event.

Considering the high sequence similarity, the congruent reassortment pattern in both Iranian isolates, and their shared unique 110-nt insertion, a marker unlikely to arise independently, the most parsimonious hypothesis for the emergence of TPNRBV in Iran is a single introduction event. This likely involved infected plant material carrying a virus population with mixed ancestry (or a pre-existing reassortant) from China, followed by its local establishment and spread. The evidence does not strongly support a scenario of multiple, independent introductions of different genomic segments.

Considerations on the pooled-sample design

The interpretations of this study are framed by its pooled-sample design. While this approach efficiently captures population-level diversity and expression profiles, it has inherent limitations. First, the SNV frequencies reported represent a composite from multiple plants; a variant at 1% frequency could be a rare mutation in one host’s viral population or a fixed difference in a virus infecting 1 out of 20 plants. Consequently, our data describe the genetic structure of the TPNRBV population at the field level but cannot delineate within-host evolutionary dynamics. Second, the TPM values reflect average viral gene expression across all infected tissue in the pool and should not be interpreted as temporal expression changes within a single infection. Furthermore, the SNVs reported here were identified bioinformatically; their independent validation by methods such as Sanger sequencing remains a necessary step for future work. Despite these constraints, our study provides valuable insights into the population genomics and expressed repertoire of TPNRBV in Iranian tea plantations, establishing a foundation for more detailed within-host studies in the future.

Limitations and future perspectives

A critical next step is the functional prioritization of the identified SNVs. Specifically, the three high-frequency non-synonymous variants (at 5.76%, 14.44%, and 21.47%) represent prime candidates for in silico predictive analysis using tools such as SIFT or PROVEAN to assess whether their amino acid changes are deleterious, neutral, or potentially advantageous. Subsequent experimental validation will be crucial to determine their actual impact on viral protein function, fitness, and pathogenicity.