Identification of protein interactions of grapevine fanleaf virus RNA-dependent RNA polymerase during infection of Nicotiana benthamiana by affinity purification and tandem mass spectrometry

The RNA-dependent RNA polymerase (1EPol) is involved in replication of grapevine fanleaf virus (GFLV, Nepovirus, Secoviridae) and causes vein clearing symptoms in Nicotiana benthamiana. Information on protein 1EPol interaction with other viral and host proteins is scarce. To study protein 1EPol biology, three GFLV infectious clones, i.e. GHu (a symptomatic wild-type strain), GHu-1EK802G (an asymptomatic GHu mutant) and F13 (an asymptomatic wild-type strain), were engineered with protein 1EPol fused to a V5 epitope tag at the C-terminus. Following Agrobacterium tumefaciens -mediated delivery of GFLV clones in N. benthamiana and protein extraction at seven dpi, when optimal 1EPol:V5 accumulation was detected, two viral and six plant putative interaction partners of V5-tagged protein 1EPol were identified for the three GFLV clones by affinity purification and tandem mass spectrometry. This study provides insights into the protein interactome of 1EPol during GFLV systemic infection in N. benthamiana and lays the foundation for validation work.

Replication of GFLV occurs on endoplasmic reticulumderived vesicles in infected plant cells and requires de novo lipid synthesis [3]. GFLV RNA1-encoded protein 1C VPg and RNA2-encoded protein 2A HP localize to the perinuclear replication compartments that contain double-stranded RNA molecules [3]. Little is known about the involvement of other viral proteins in replication although GFLV RNA1-encoded proteins 1E Pol and 1B Hel are suspected to be essential [3]. In addition, no information is available on the interaction of protein 1E Pol with other GFLV proteins and host proteins for replication.

ACCESS
Similarly, the molecular mechanisms underpinning GFLV symptom development remain largely unknown although recent advances have been made in model herbaceous hosts. For example, GFLV strain F13 produces a hypersensitive response triggered by the RNA2-encoded protein 2A HP on inoculated leaves of Nicotiana occidentalis [9]. In contrast, GFLV-F13 produces an asymptomatic infection in N. benthamiana, while GFLV strain GHu produces distinct vein clearing symptoms on apical leaves of this plant species [7,10]. A symptom determinant for vein clearing of GFLV-GHu was recently mapped to residue 802 of the RNA1-encoded protein 1E Pol . This residue, which is a lysine in GFLV-GHu, is necessary but not sufficient for vein clearing development [10].
While much work has been done to describe molecular events of GFLV infection [3,11], information about the molecular context of protein 1E Pol during infection, in particular its protein interactants, is lacking. Here, we built on our previous work and developed a method to isolate protein 1E Pol from protein extracts of N. benthamiana infected with GFLV and initiated a proof-of-concept study of the 1E Pol protein interactome via affinity purification coupled to tandem mass spectrometry. To this end, tagging GFLV protein 1E Pol was essential because, in our hands, efforts to generate an antibody that specifically detected 1E Pol were unsuccessful. Indeed, an antibody raised against a synthetic peptide (HVPSKTSFM-KVPDELC) designed in a conserved N-terminus sequence failed to unambiguously detect an immunoreactive product of the expected size in total soluble protein extracts of GFLVinfected N. benthamiana via SDS-PAGE and western blot detection, although the same approach was successful in a  K802G , an asymptomatic mutant of GFLV-GHu in which lysine 802 of protein 1E Pol was mutated to glycine, and their corresponding version for which protein 1E was tagged with a V5 epitope. Insets show close-ups of vein clearing symptoms (first and second panels) and asymptomatic leaves (third and fourth panels) at six dpi. (b) Detection of protein 1E Pol :V5 accumulation in N. benthamiana leaf tissue systemically infected with wild-type (-) or V5-tagged 1E Pol (+) GFLV strains GHu and F13-1E CΔ1 , and mutant GHu-1E K802G by western blot with an anti-V5 antibody. TSP from a mock-inoculated N. benthamiana were used as a control. (c) Analysis of affinity purified 1E Pol complexes in TSP from N. benthamiana leaves systemically infected with GFLV-GHu or GFLV-GHu-1E:V5 by western blot. TSP were extracted in lysis buffer 4 amended with 1X Halt protease inhibitor cocktail (Thermo Fisher Scientific) and further diluted 1 : 5 in DTT-free lysis buffer. Inputs were affinity purified with V5 polyclonal antibody (Invitrogen PA1-993) conjugated to Protein A Dynabeads (Invitrogen). The top row of lane labels indicates (e) elution products or (i) input fractions of the affinity purification. The second row indicates V5-tagged (+) or non-tagged (-) GFLV-GHu. The third row indicates the concentration of DTT included in the lysis buffer. The fourth row indicates whether the V5 polyclonal antibody used for affinity purification was untreated (-) or cross-absorbed against TSP from healthy N. benthamiana tissue (+). Proteins were probed with a polyclonal anti-V5 antibody (Invitrogen PA1-993, non-cross-absorbed) and a goat anti-rabbit alkaline phosphataseconjugated secondary antibody (Invitrogen T2191) and developed with 1-Step NBT/BCIP solution. The predicted molecular weight of (i) V5-tagged 1E Pol (93 kDa) is indicated with a red arrowhead, and (ii) a putative V5-tagged 1D Pro E Pol (117 kDa) or 1C VPg 1D Pro E Pol (120 kDa) precursor is indicated with a green arrowhead. Bottom panels show Ponceau staining of the RuBisCO large subunit. Molecular weight standards (kDa) are shown by black arrowheads.
previous study [7]. Thus, we decided to tag protein 1E Pol as an alternative to producing an antibody against 1E Pol .
For GFLV strain F13, a similar epitope tagging approach abolished infectivity in planta. Only mutants GFLV-F13-1E CΔ1 :V5 and GFLV-F13-1E CΔ4 :V5, in which a single or four residues were truncated at the C-terminus of protein 1E Pol prior to the addition of the V5 epitope tag, established a systemic infection in N. benthamiana, as shown by DAS-ELISA using specific antibodies (Table S3). Infection of GFLV-F13-1E CΔ1 :V5 and GFLV-F13-1E CΔ4 :V5 was asymptomatic, consistent with the phenotype of the untagged virus [7,10,17]. Less 1E Pol :V5 accumulated in N. benthamiana infected with GFLV-F13-1E CΔ1 :V5 compared to GFLV-GHu:V5 and GFLV-GHu1E K802G: V5 following infection (Fig. 1b). More work is needed to explain this differential accumulation of protein 1E Pol :V5 between GFLV strains GHu and F13. Of the two GFLV-F13 tagged mutants only F13-1E CΔ1 :V5 reliably showed an anti-V5 immunoreactive 1E Pol :V5 signal in protein extracts from apical N. benthamiana leaves tested by western blot detection (Table  S3). Thus, this mutant was further used in a comparative proteomics analysis of tagged and untagged virus isolates.  #The ion score for an MS/MS match in Mascot is based on the calculated probability, P, that the observed match between the experimental data and the database sequence is a random event.
The reported score is −10Log(P). Ion scores that are higher than the Mascot identity threshold indicate high quality matches. The Mascot peptide identity threshold is the ion score with expected significance threshold to be 0.05.
GFLV recombinants carrying protein 1E Pol with the epitope tags FLAG, 3XFLAG, HA or myc repetitively failed to establish systemic infection in N. benthamiana (Table S3). Interestingly, of the five epitope tags tested, V5 has the lowest proportion of acidic residues. The V5 epitope (1.4 kDa) has 7 % acidic residues in contrast to the FLAG, 3XFLAG, HA and myc tags, which have 63, 50, 40 and 22 % acidic residues, respectively (Table S1). A low proportion of acidic residues is consistent with the residue composition of the C-terminal 53 amino acids of protein 1E Pol of GFLV strains GHu (1.9 % acidic residues) and F13 (3.7 % acidic residues), suggesting the functionality of protein 1E Pol may depend on the charge of its C-terminus. It is also possible that epitope tags, except V5, might have affected the stability of protein 1E Pol or its capacity to be properly translated. More work is needed to address these issues.
A replicated time course experiment in N. benthamiana via mechanical inoculation with GFLV-GHu-1E Pol :V5 showed optimal accumulation of immunoreactive 1E Pol :V5 in TSP of apical leaves at 6-9 dpi by SDS-PAGE and western blotting with an anti-V5 antibody. This optimal period corresponds with noticeable GFLV-GHu vein clearing symptoms at 4-6 dpi (Fig. 1a) and symptoms fading at 9-10 dpi.
Of the four lysis buffers [18][19][20][21] (Table S4) tested for the extraction of 1E Pol :V5 at seven dpi, a reducing agent such as DTT was necessary to confidently detect an immunoreactive peptide (Fig. 2a). It may be that a reducing environment is necessary to dissociate protein 1E Pol from membranes, nucleic acids or other macromolecules. In support of this hypothesis, GFLV 1E Pol likely associates with endoplasmic reticulumderived membranes for replication [3,11,22], although it might not bind to membranes by itself when expressed ectopically in N. benthamiana, suggesting a probable dependence on another viral protein, likely protein 1B Hel , to anchor to the replication complex, as previously discussed [3]. Similarly, the polymerase domain-containing precursor protein of tomato ringspot virus, another nepovirus, associates with endoplasmic reticulum-derived membranes [23].
Affinity purifications of the 1E Pol :V5 protein complexes were performed on TSP extracted from N. benthamiana systemically infected with GFLV-GHu-1E:V5, GFLV-F13-1E CΔ1 :V5 or GFLV-GHu-1E K802G :V5 collected at seven dpi. Untagged GFLV strains GHu and F13 were used as negative controls. Three biological replicates were performed for each virus treatment. Affinity purification conditions were as previously described [18] with the following modifications. The anti-V5 tag polyclonal antibody was bound to Dynabeads Protein A (Life Technologies, Invitrogen) at a concentration of 10 µg antibody per 1 mg of beads and 5 mg of beads in 5 ml of diluted plant cell lysate were used. For protein extraction, lysis buffer 4 (Table S4), a HEPES-based lysis buffer, was supplemented with 2.5 mM DTT, the lowest concentration of DTT required for sufficient extraction of 1E Pol :V5, and lysates were further diluted 1 : 5 in lysis buffer without DTT (Fig. 2) to minimize loss of 1E Pol -plant protein interactions and inhibition of bait capture by the V5 antibody, respectively. Washes of beads were performed with DTT-free lysis buffer and the full bead volume was subjected to on-bead trypsin digestion, as described [18]. Additionally, to enrich for 1E Pol :V5 and reduce background binding during affinity purifications, the anti-V5 antibody was cross-absorbed by sodium sulphate precipitation [24] with proteins from a blend of N. benthamiana tissue infected with untagged GFLV-GHu, F13 and GHu-1E K802G in equal proportions. The antibody fraction was recovered by overnight dialysis, as previously described [18]. Cross-absorption of the anti-V5 antibody resulted in the loss of some anti-V5 immunoreactive protein bands of higher molecular mass in western blot detection (Fig. 1c).
Sample cleaned for mass spectrometry was performed using OMIX C18 100 µl tips (Agilent, A57003100). After trypsin digestion, peptides were eluted from beads in 100 µl 0.1 % formic acid. Protein complexes were analysed on an Orbitrap Fusion Lumos mass spectrometer (Thermo Scientific) equipped with a nano-ion spray source coupled to an EASY-nLC 1200 system (Thermo Scientific). The liquid chromatography system was configured with a self-pack PicoFrit 75 -µm analytical column with an 8 -µm emitter (New Objective, Woburn, MA) packed to 25 cm with ReproSil-Pur C18-AQ, 1.9 µm material (Dr. Maish HPLC, GmbH).
Thermo RAW files were converted to Mascot generic files (.mgf) using ProteoWizard [25]. The protein search database was generated from amino acid sequences corresponding to all coding gene sequences from version 1.01 of the N. benthamiana genome assembly downloaded from the Sol Genomics Network [26], amino acid sequences from nepoviruses including mutant and natural strains of GFLV, plus common mammalian affinity purification contaminant proteins downloaded from NCBI. The Mascot v. 2.5.1 software used to search the mgf files identified an average of 722 protein groups per replicate. Search parameters included one fixed modification (cysteine: carbamidomethyl), two variable modifications (deamidation of asparagine and glutamine and/or methionine oxidation), trypsin enzyme specificity, one missed cleavage, a peptide mass tolerance of ±20 ppm, fragment mass tolerances of ±0.5 Da, and ion charge=2+, 3+ or 4+.
Data were imported into Scaffold Q+4.80.4 (Proteome Software Inc., Portland, OR) for spectrum counting analysis. Peptide and protein false discovery rates were set at <1.0 % using the Peptide Prophet algorithm [27] with delta-mass correction. Protein identifications were accepted if they contained at least two identified peptides. Protein probabilities were assigned by the Protein Prophet algorithm [27]. Proteins that contained similar peptides and could not be differentiated based on tandem mass spectrometry analysis alone were grouped to satisfy the principles of parsimony.
Spectral counts for proteins in the tagged (n=9 biological replicates, each from independent plants) and untagged (n=5 biological replicates, each from independent plants) samples were compared using a Student's t-test to identify proteins that may bind to protein 1E Pol , but not in a GFLV strain specific manner, because no host proteins were found to be enriched when each virus strain was tested individually. Log 2 fold-change enrichment (tagged/untagged) was calculated to determine enrichment in the experiments with the tagged 1E Pol .
Three RNA1-encoded GFLV proteins were significantly enriched in the affinity purification experiments: 1E Pol , 1B Hel and 1D Pro (Table 1). A single peptide spectral matching to protein 1E Pol was detected at low levels in three of the five untagged replicates (Table 1), consistent with minimal levels of non-specific binding of 1E Pol to the beads or the anti-V5 antibody used in the affinity purifications. In contrast, 1E Pol was abundantly detected in the tagged replicates, as expected, and proteins 1B Hel and 1D Pro were only detected in experiments with tagged protein 1E Pol (Table 1). A total of six plant proteins were identified to co-purify with GFLV protein 1E Pol at low levels ( Table 1). Among the six plant proteins, three proteins were specifically found in the tagged 1E Pol samples: plastid transcriptionally active 14 Set domain protein (pTAC14), splicing factor 3B subunit one protein and dynamin-related protein 5A. The other three plant proteins, WD40 domain-containing protein, translation initiation factor IF-2 protein and P-type ATPase (PMA1), were detected in both the tagged and untagged samples but were found to be enriched (1.2 to 2.8 Log 2 -fold) in the tagged samples with a P-value <0.05 ( Table 1).
Three of the plant proteins identified in complex with 1E Pol have been shown to play a role in plant or animal virus infection. For example, protein pTAC14 is localized in the chloroplast and regulates plastid gene expression [28]; several virus proteins have been shown to be involved in a plasma membrane and chloroplast signalling pathway to suppress salicylic aciddependent plant defenses [29]. In rice, the splicing factor 3B subunit one protein regulates the expression of genes involved in cell death and resistance responses [30]. Splicing factor 3B subunit one has also been characterized to interact with animal viruses and plays a critical role in the replication of human immunodeficiency virus [31]. Dynamin-related protein 5A is critical for plant infection by soybean mosaic virus (SMV, genus Potyvirus, family Potyviridae) [32]. Dynamin-related protein 5A was identified in purifications of SMV virions using proteomics and knock-down of dynamin-related protein 5A in plants inhibited SMV infection in soybean [32]. Most of the plant proteins reported in this study were also described in the interactome of purified potato leafroll virus (genus Polevirus, family Luteovidae) virions during infection of N. benthamiana [18]. Additionally, recent evidence supports a role for PMA1 in the induction of an immune response manifested by cell death in N. benthamiana, as shown by silencing and overexpression assays [33].
Our proof-of-concept study based on affinity purifications of V5-tagged 1E Pol coupled with protein identification by tandem mass spectrometry provided a snapshot of the putative protein interaction network of protein 1E Pol during systemic GFLV infection of N. benthamiana. Methods to optimize the extraction of protein 1E Pol from plant tissue may increase the depth of proteomic coverage and lead to a more thorough characterization of the plant-virus interactome. GFLV protein 1D Pro , the viral protease which processes in cis and in trans the two viral polyproteins into individual mature peptides [1,3,5,6,11], was enriched in affinity purifications with V5-tagged 1E Pol , suggesting that 1D Pro complexes with 1E Pol during virus infection. The 1B Hel helicase protein [1,3,5,11] was also enriched as a protein interacting in complex with 1E Pol during virus infection. It is also possible that protein 1D Pro peptides are enriched in affinity purifications with 1E Pol due to the presence of putative 1D Pro 1E Pol or 1C VPg 1D Pro 1E Pol precursors of polyprotein processing that were suggested by western blot analyses (Figs 1  and 2). More research is needed to test whether interactions between GFLV proteins 1E Pol , 1D Pro and 1B Hel are direct or indirect, to understand the stoichiometry of binding and the significance of these interactions during virus infection, in particular in replication.
Orthologous of most plant host proteins enriched in our data set were previously documented to be involved in virus-host interactions with other plant or animal viruses [18,[28][29][30][31][32]. These host protein candidates will need to be validated in follow-up genetic approaches to verify whether they have a role in GFLV infection in N. benthamiana.