A novel Actinidia cytorhabdovirus characterized using genomic and viral protein interaction features

Abstract A novel cytorhabdovirus, tentatively named Actinidia virus D (AcVD), was identified from kiwifruit (Actinidia chinensis) in China using high‐throughput sequencing technology. The genome of AcVD consists of 13,589 nucleotides and is organized into seven open reading frames (ORFs) in its antisense strand, coding for proteins in the order N‐P‐P3‐M‐G‐P6‐L. The ORFs were flanked by a 3′ leader sequence and a 5′ trailer sequence and are separated by conserved intergenic junctions. The genome sequence of AcVD was 44.6%–51.5% identical to those of reported cytorhabdoviruses. The proteins encoded by AcVD shared the highest sequence identities, ranging from 27.3% (P6) to 44.5% (L), with the respective proteins encoded by reported cytorhabdoviruses. Phylogenetic analysis revealed that AcVD clustered together with the cytorhabdovirus Wuhan insect virus 4. The subcellular locations of the viral proteins N, P, P3, M, G, and P6 in epidermal cells of Nicotiana benthamiana leaves were determined. The M protein of AcVD uniquely formed filament structures and was associated with microtubules. Bimolecular fluorescence complementation assays showed that three proteins, N, P, and M, self‐interact, protein N plays a role in the formation of cytoplasm viroplasm, and protein M recruits N, P, P3, and G to microtubules. In addition, numerous paired proteins interact in the nucleus. This study presents the first evidence of a cytorhabdovirus infecting kiwifruit plants and full location and interaction maps to gain insight into viral protein functions.

Cytorhabdoviruses encode at least six canonical proteins, including nucleocapsid protein (N), phosphoprotein (P), movement protein (P3), matrix protein (M), glycoprotein (G), and RNA-dependent RNA polymerase (L), in the order 3′-N-P-P3-M-G-L-5′ Walker et al., 2018). In the enveloped particles of viruses in the family Rhabdoviridae, protein N encapsulates the viral genomic RNA to generate N-RNA complexes, and each complex together with proteins P and L forms a ribonucleoprotein (RNP) complex, which is the minimal infectious unit. Protein M plays roles in the condensation of RNP complexes into a skeleton-like structure (RNP-M core) during virion assembly, and protein G forms transmembrane spikes (Assenberg et al., 2010;Jackson et al., 2005). In addition, some cytorhabdoviruses encode one or more accessory proteins between the coding regions of P and M and/or G and L (Walker et al., 2011). Understanding the interaction and location maps of viral proteins is necessary for revealing the functions of viral genes and the components involved in the assembly of virions. To date, there are only two cytorhabdoviruses, lettuce necrotic yellows virus (LNYV) and alfalfa dwarf virus (ADV), for which the subcellular localizations and interactions between proteins have been described Martin et al., 2012).
Previously, in order to obtain a complete overview of the kiwifruit virome, we constructed six RNA sequencing (RNA-seq) libraries from 69 kiwifruit samples . During a re-examination for viruses in these HTS data, we identified one contig that partially matched the genome sequences of some cytorhabdoviruses. Then, the genome of the virus was determined by Sanger sequencing. The virus has a genomic structure that is typical of viruses in the family Rhabdoviridae and is phylogenetically related to cytorhabdoviruses. Furthermore, the intracellular localization and interaction maps of the viral proteins were determined.

| Identification and genomic organization of Actinidia virus D
During the BLASTX searches using the assembled contig sequences from six RNA-seq libraries of kiwifruit samples against the NCBI data set, we identified a large contig (ID: contig100_2) of 13,579 nucleotides (nt) from library JS with significant amino acid (aa) sequence identity of 21.5%-44.8% to proteins encoded by cytorhabdoviruses (Table 1). The identified contig was further aligned along the best matched genomic RNA of Wuhan insect virus 4 (WhIV4, accession no. KM817650) to generate a preliminary genomic organization of a potential novel cytorhabdovirus, provisionally named Actinidia virus D (AcVD) following the previously reported kiwifruit virus names. Total RNA was extracted from each of the four samples (JS27, JS29, JS30, and JS45) included in the RNA-seq library (JS). Reverse transcription (RT)-PCR using the virus-specific primers NF/NR (Table S1) and sequencing of amplified products confirmed the presence of AcVD in two samples (JS27 and JS29). The 1,363-bp amplicons generated from the two samples shared 99.4% sequence identity with each other and the corresponding sequence of contig100_2. Then, the sample JS27 was used for further viral genome characterization.
The full-length genome of AcVD consists of 13,589 nt (GenBank accession no. MW550041), and seven open reading frames (ORFs) were identified in its anti-sense strand (Figure 1a). The proteins encoded by ORF1-5 and ORF7 matched well with the nucleocapsid protein (N), phosphoprotein (P), movement protein (P3), matrix protein (M), glycoprotein (G), and RNA-dependent RNA polymerase (L) of reported cytorhabdoviruses, and protein P6 (encoded by ORF6) had no hit with proteins deposited in the NCBI database (Table 1).
Similar to all plant viruses in the family Rhabdoviridae, the AcVD ORFs are separated by a highly conserved gene junction with the consensus sequence 3′-AUUAUUUUGAUCUG-5′, which is composed of a 3′ polyadenylation signal of the preceding gene (element I), an intergenic spacer (element II), and a transcription initiation sequence of the following gene (element III) (Figure 1b), which is one of the typical features of the genomes of viruses in the family Rhabdoviridae (Jackson et al., 2005). The sequences of elements I and II of the AcVD gene junction are similar to those of tomato yellow mottle-associated virus (TYMaV), while element III is almost identical to those of LNYV and LYMoV ( Figure 1b).
ORF1 is 1,437 nt in length and encodes a putative N protein of 468 aa with a predicted molecular weight (MW) of 53.8 kDa and an isoelectric point (pI) of 5.40 (Table 1). Pairwise comparisons revealed sequence identities ranging from 39.2% to 52.9% (nt) and from 17.4% to 40.1% (aa) between ORF1 and orthologues of reported cytorhabdoviruses (Table 2). Protein N contains a predicted bipartite nuclear localization signal (NLS) (score = 2.8, aa position 21-52) at the N-terminal region (Table 1).
ORF2 is 915 nt in length and encodes a putative P protein of 304 aa with a predicted MW of 34.1 kDa (pI = 4.12) ( Table 1).
Protein P of AcVD shares identities ranging from 32.6% to 48.7% (nt) and from 11.6% to 28.6% (aa) with orthologues of reported cytorhabdoviruses (Table 2). Two putative nuclear export signals (NESs) at aa positions 240-243 and 246 were found in protein P of AcVD (Table 1). A small overlapping ORF corresponding to P′ was predicted for AcVD. The P′ ORF encodes a small 6.3-kDa protein (51 aa) that contains a predicted transmembrane domain (aa positions 13-35).
Additionally, a putative bipartite NLS (score = 3.2) at the C-terminal region (aa positions 152-183) was found in protein M of AcVD (Table 1).
ORF5 is 1,659 nt in length and encodes a putative G protein of 552 aa with a predicted MW of 62.5 kDa (pI = 6.61) ( ORF7 is 6,318 nt in length and encodes a putative L protein of 2,115 aa with a predicted MW of 243.2 kDa (pI = 7.18) ( Table 1).
The L protein of AcVD shows sequence identities ranging from 45.9% to 54.6% at the nt level and from 26.3% to 44.5% at the aa level with those of other cytorhabdoviruses (Table 2). Three  (Table 1).

| Subcellular localization of AcVD proteins in Nicotiana benthamiana leaf cells
When the viral proteins N, P, P3, M, G, and P6 were individually fused to the N-terminus of an enhanced yellow fluorescent protein (eYFP), at 2 days post-agroinfiltration (dpi), the fluorescence signals of P-eYFP, G-eYFP, N-eYFP, and P6-eYFP were observed in the periphery of agroinfiltrated Nicotiana benthamiana leaf cells and colocalized well with the fused endoplasmic reticulum (ER) marker protein mCherry-HDEL ( Figure 3a). In addition, P-eYFP and G-eYFP also localized in the nucleus, and weak signals of N-eYFP and P6-eYFP were observed in the nucleus, as validated using the nuclear marker protein H2B-mCherry, and P6-eYFP also TA B L E 2 The nucleotide and amino acid sequence identities (%) between Actinidia virus D (AcVD) and reported cytorhabdoviruses   Figure S4).

| Protein M of AcVD is associated with microtubules
Given that the filamentous structures of M-eYFP in the cytoplasm were similar to microtubules, we tested the potential association   Table S2 combinations BiFC assays were confirmed by MYTH assays (Figure 7a). Sequence analysis showed that the 847-bp fragment of the viral ORF1 amplified using primer set 1F/1R from these samples shared 99.3%-99.8% nt sequence identity and 98.6%-100% aa sequence identity, and the 544-bp fragment of the viral ORF7 amplified using primer set 7F/7R showed 98.9%-100% sequence identities at both nt and aa levels.

| D ISCUSS I ON
In recent years, four new viruses belonging to families Fimoviridae cation. We also found that AcVD proteins P, P3, M, and G accumulated in the nucleus, although no NLS was predicted in proteins P and G. Likewise, no NLS was detected in protein P of the cytorhabdovirus ADV and the alphanucleorhabdovirus potato yellow dwarf virus, but the cognate proteins from the two viruses were located in the nucleus (Bandyopadhyay et al., 2010;Bejerman et al., 2015).
The location of AcVD protein P in both the cell periphery and the nucleus is similar to that of two beta-and a gammanucleorhabdovirus, namely datura yellow vein virus (DYVV), sonchus yellow net virus (SYNV), and maize fine streak virus Goodin et al., 2001;Tsai et al., 2005), and different from the exclusive cytoplasmic localization of LNYV protein P and/or the nuclear localization of ADV protein P Martin et al., 2012). Previous studies showed that neither ADV protein P nor SYNV protein P could enter the nucleus by interacting with importinα, and P proteins of ADV and SYNV have RNA silencing suppressor activity Deng et al., 2007;Jackson et al., 2005;Krichevsky et al., 2006;Martin et al., 2009). Elucidation of the nuclear import pathway and the specific function of AcVD protein P requires further study. The P3 protein of AcVD has an aspartic acid (D) residue in the conserved LxD/ are also localized in the nucleus of N. benthamiana cells Dietzgen et al., 2015;Martin et al., 2012;Ramalho et al., 2014). The AcVD M protein is uniquely associated with microtubules, in contrast to the reported nuclear and/or ER localization of M proteins of some plant and animal viruses in the family Rhabdoviridae (Bandyopadhyay et al., 2010;Bejerman et al., 2015;Dietzgen et al., 2015;Martin et al., 2012;Ramalho et al., 2014;Tsai et al., 2005). The AcVD P6 protein is located on the ER network and forms inclusion-like bodies along the cell membrane or in the cytoplasm, which was different from the reported ER or nuclear localization of P6/P9 of the cytorhabdoviruses ADV and barley yellow striate mosaic virus Yan et al., 2015).
We also noticed that the ER localization observed for the G-eYFP fusion protein was not clear for eYFP-G, possibly because the different fusion orientations could affect the accession or interaction of the viral protein with plant targets, as reported for ADV, MMV, and SYNV Goodin et al., 2007;.
Previous studies showed that the N-P and M-M interactions are common to some analysed viruses in the family Rhabdoviridae F I G U R E 6 Split-ubiquitin yeast two-hybrid assays for Actinidia virus D (AcVD) proteins. Proteins N, P, P3, G, and P6 were individually fused to the C-terminal half of ubiquitin (Cub) of vector pDHB1, and proteins N, P, P3, and M were individually fused to the N-terminal half of ubiquitin (NubG) of vector pPR3-N. Yeast cotransformed with pDHB1-largeT/pDSL-P53 was used as a positive control, and yeast cotransformed with pDHB1-largeT/pPR3-N was used as a negative control. Yeast growth on synthetic dropout medium lacking leucine and tryptophan (SD−LW) was used to confirm the presence of both plasmids. Medium lacking leucine, tryptophan, histidine, and adenine (SD− LWHA) was used to screen for positive interactions. A series of dilutions (10 −1 , 10 −2 , and 10 −3 ) is shown (Bandyopadhyay et al., 2010;Ge et al., 2010;Graham et al., 2008;Martin et al., 2012). The N-P and M-M interactions are also conserved for AcVD. Our study showed a complex interaction map of the AcVD-encoded proteins in BiFC assays. The interaction combinations N-P3, N-G, and M-G identified in BiFC assays were not detected in the MYTH assays (Figure 7a). During in vivo BiFC assays, plant proteins might serve as bridges in the interaction between two exogenous proteins Min et al., 2010;Paape et al., 2006). In addition, different vectors used in the yeast two-hybrid assay could also lead to different results because the fusion directions may affect protein folding or function (Zhou et al., 2019a). Previous studies revealed that some cellular factors are involved in cell-to-cell movement of SYNV, LNYV, and ADV by mediating viral protein interactions Min et al., 2010). Interestingly, the interaction of N with P, P3, or M resulted in the formation of numerous aggregates in the cytoplasm, and some interaction complexes were found to be closely adjacent to the nucleus. Similar perinuclear bodies have been identified as viral replication complexes (VRCs) (Barton et al., 2017). The N-P interaction of LNYV also results in the formation of aggregations in the cytoplasm (Martin et al., 2012). Considering the potential function of P3 in viral movement, the N-P3 interaction bodies formed in the BiFC assay might be involved in viral movement. Cytorhabdoviruses and nucleorhabdoviruses are distinguished based on their viral replication and particle assembly sites in the cytoplasm or the nucleus, respectively (Jackson et al., 2005). The N protein functions in the encapsidation of viral genomic RNA and is a component of the viroplasm (Jackson et al., 2005). Probably, the N protein of AcVD recruits the two struc- , and some orthotospoviruses (Diwaker et al., 2015;Leastro et al., 2015;Leastro et al., 2017;Tripathi et al., 2015;Widana & Dietzgen, 2017). The MP of tomato spotted wilt virus can facilitate the movement of the RNP complex through interaction with N protein (Kormelink et al., 1994;Soellick et al., 2000). The microtubule cytoskeleton is involved in the transport of materials within cells. Many plant viruses use the plant cytoskeleton for virion and VRC movement from the replication site to PDs (Naghavi & Walsh, 2017). The association of viral MPs with microtubules is necessary for the intracellular movement of plant viruses (Boyko et al., 2000;Nicolas & Manfred, 2018;Niehl et al., 2013). The disruption of microtubules can prevent TMV movement (Boutant et al., 2009;Ferralli et al., 2006;Niehl et al., 2013). The M protein of the alphanucleorhabdovirus RYSV interacts with leafhopper tubulin to facilitate viral transport . The MPs of LNYV and SYNV could interact with a microtubule-associated protein, NbVOZ1, implicating that NbVOZ1 might aid viral movement Min et al., 2010).  (Figure 7b), which might be necessary for virion assembly and intracellular trafficking of movement complexes (Assenberg et al., 2010;Fondong, 2013;Jackson et al., 2005;Zhou et al., 2019a). The nuclear location of protein M of viruses in the family Rhabdoviridae might play a role in blocking the export of host mRNAs in the infected cells through interactions with nuclear export proteins (Faria et al., 2005) and limiting competition of resources for the viral proteins (Carroll & Wagner, 1979;Clinton et al., 1978;De et al., 1982;Finke & Conzelmann, 2003;Von Kobbe et al., 2000;Zhou et al., 2019b). In the nucleus of plant cells, the AcVD protein M interacts with five nuclear-localized proteins (N, P, P3, G, and P6) (Figure 7b). This localization may be linked to the recruitment of host transcription factors, as reported for SYNV (Min et al., 2010), or some other specific functions. The function of P6 proteins of viruses in the family Rhabdoviridae is poorly understood. RYSV P6 is thought to be associated with virions in viruliferous aphids and to function as an RNA silencing suppressor (Guo et al., 2013;Jackson et al., 2005). The P6 protein of AcVD harbours the structural characteristics of class 1a viroporins, which are commonly present in mammalian viruses in the family Rhabdoviridae (Walker et al., 2011), and interacts with the P and M proteins. Viroporins participate in virus particle assembly and promote viral particle release from cells (Nieva et al., 2012;Walker et al., 2015).
Thus, the functions of P6-M and P6-P interactions could be linked with these functions.
In conclusion, here we present the complete genome sequence of AcVD, which has the same genomic organization as the viruses in the genus Cytorhabdovirus and shows low genomic sequence identities with reported cytorhabdoviruses. The virus has a low occurrence frequency in kiwifruit. A full subcellular location map and the interaction network of six AcVD proteins are reported.
Our BiFC assays provide biological evidence that the viral protein N interacts with three other viral proteins (P, P3, and G) and forms viroplasms in the cytoplasm of N. benthamiana leaf cells.
Moreover, we report for the first time that the M protein of a virus in the family Rhabdoviridae shows unique microtubule localization and that it could recruit four viral proteins (N, P, P3, and G) to microtubules.

| Virus source
To obtain a complete overview of the kiwifruit virome in China, six libraries prepared from total RNA of leaf samples from 69 kiwifruit plants were subjected to next-generation sequencing in our previous work . During a re-examination for other potential viruses in the library data sets (SRA ID: PRJNA681158), a large contig (ID: contig100_2) sharing limited amino acid sequence identities with the proteins of reported cytorhabdoviruses was identified from the JS library. Thus, the four samples (JS27, JS29, JS30, and JS45) included in the JS library were tested for the presence of the potential cytorhabdovirus using the primer set NF/NR (Table S1), which was designed based on the sequence of contig100_2. After a preliminary identification of a candidate cytorhabdovirus (tentatively named AcVD), sample JS27, which was positive for the virus, was used for amplification of the viral genome.

| Amplification of the viral genome
Seven primer sets, including the primer set NF/NR, were designed based on the sequences of contig100_2 and employed to amplify the remaining viral sequence (Table S1). Commercial kits for rapid amplification of cDNA ends (RACE) (Takara) were used to determine the 5′ and 3′ terminal sequences of the viral genome. The 5′RACE reaction was conducted according to the manufacturer's protocol. For 3′RACE, poly(A) tails were added to the 3′ ends of the total RNAs using the poly(A) polymerase kit (Takara), and cDNA was generated using the oligo(dT) primer provided in the 3′RACE kit.
Total RNA was extracted from leaf samples (100 mg) using a cetyltrimethylammonium bromide (CTAB) method as described previously (Li et al., 2008). The resulting RNAs were used as templates for the synthesis of first-strand cDNA using M-MLV reverse transcriptase (Promega) and random hexamer primers pd(N) 6 (Takara). The solutions and conditions of PCRs were similar to those previously described (Wang et al., 2016), except that the extension time and annealing temperature varied depending on the sizes of the amplicons and primer pairs used.
PCR products were gel-purified and ligated into the pMD18-T

| Sequence analysis
The potential ORFs were predicted using the ORFfinder tool on  (Kumar et al., 2018). ML trees were constructed based on the aa sequences of proteins L and N using the best-fit models LG+I+G+F and LG+G, respectively, with 1,000 bootstrap replications. GenBank accession numbers of virus sequences used for the phylogenetic analyses are given in Table S2.

| Subcellular localization and BiFC assays for viral proteins in planta
A binary vector pCNY containing an eYFP gene was used for subcellular localization experiments. Six ORFs (missing stop codons) of AcVD were fused to the N-terminus of eYFP between the XbaI and BamHI digestion sites and to the C-terminus of eYFP between the SalI and SmaI digestion sites.
For BiFC assays, the primers used for the amplification of ORF1-6 of AcVD were flanked with AttB recombination sites at their 5′ ends to facilitate subsequent Gateway vector construction (Table S3). Amplicons were gel-purified and cloned into vector pDONR/ZEO (Invitrogen) using BP Clonase II Enzyme Mix (Invitrogen). Sequence-validated entry clones (without a stop codon) of these ORFs were individually recombined into binary destination vectors pEarleygate201-YN and pEarleygate202-YC (Lu et al., 2010) using LR Clonase II Enzyme Mix (Invitrogen).

| DUAL membrane yeast two-hybrid assay
The DUAL MYTH system was used following the manufacturer's protocol (Dualsystems Biotech AG). Four AcVD proteins (N, P, P3, and M) were each cloned into the vector pPR3-N with the N-terminal half of ubiquitin (NubG), and five proteins (N, P, P3, G, and P6) were separately cloned into the vector pDHB1 with the C-terminal half of ubiquitin (Cub). The recombined pPR3-N and pDHB1 plasmids were cotransformed into yeast strain NMY51 (Weidi Bio). The transformed yeast cells were grown on plates containing synthetic dropout medium lacking tryptophan and leucine (SD−LW) (Clontech) at 30 °C for 2-4 days. The surviving clones were transferred to SD−LWHA (Clontech). The plasmid pair pDHB1-largeT/pDSL-P53 served as a positive control, and the plasmid pair pDHB1-largeT/ pPR3-N was used as a negative control. RT-PCR using the primer pair 1F/1R (Table S1), which is used to amplify an 847-bp segment including ORF1, and primer pair 7F/7R (Table S1), which is used to amplify a 544-bp fragment including ORF7.

ACK N OWLED G EM ENTS
We are grateful to Prof. L. Cai, Huazhong Agricultural University, China, for help during sampling. This work was financially supported by the programme for Key International S&T Cooperation (grant 2017YFE0110900), the Key National Project (grant 2019YFD1001800), and the Science and Technology Project for the Xinjiang Uygur Autonomous Region (grant 2018E02026).

CO N FLI C T O F I NTE R E S T
The authors have no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.