Binding between elongation factor 1A and the 3ʹ‐UTR of Chinese wheat mosaic virus is crucial for virus infection

Abstract The Chinese wheat mosaic virus (CWMV) genome consists of two positive‐strand RNAs that are required for CWMV replication and translation. The eukaryotic translation elongation factor (eEF1A) is crucial for the elongation of protein translation in eukaryotes. Here, we show that silencing eEF1A expression in Nicotiana benthamiana plants by performing virus‐induced gene silencing can greatly reduce the accumulation of CWMV genomic RNAs, whereas overexpression of eEF1A in plants increases the accumulation of CWMV genomic RNAs. In vivo and in vitro assays showed that eEF1A does not interact with CWMV RNA‐dependent RNA polymerase. Electrophoretic mobility shift assays revealed that eEF1A can specifically bind to the 3ʹ‐untranslated region (UTR) of CWMV genomic RNAs. By performing mutational analyses, we determined that the conserved region in the 3ʹ‐UTR of CWMV genomic RNAs is necessary for CWMV replication and translation, and that the sixth stem‐loop (SL‐6) in the 3ʹ‐UTR of CWMV genomic RNAs plays a key role in CWMV infection. We conclude that eEF1A is an essential host factor for CWMV infection. This finding should help us to develop new strategies for managing CWMV infections in host plants.

cellular processes, including protein translation, cytoskeleton organization, nuclear export, and ubiquitin-dependent protein degradation (Andersen et al., 2003;Vera et al., 2014). During protein synthesis, the ternary complex of eEF1A binds to and delivers aa-tRNAs to the ribosome. When the aa-tRNA anticodon in the ribosome matches the mRNA codon bound to the ribosome, GTP is hydrolysed to GDP and then combined with eEF1A. eEF1A is restored to active GTP under the action of the nucleotide exchange factor eEF1B (EF-Ts).
Chinese wheat mosaic virus (CWMV) has two single-strand positive-sense genomic RNAs. Chinese wheat mosaic virus is a member of the genus Furovirus, family Virgaviridae. CWMV does not contain a poly(A) sequence at the RNA 3ʹ end of the genome (Adams et al., 2009). CWMV RNA1 consists of 7,147 nucleotides (nt) and encodes three proteins: a 153-kDa replication-associated protein, a 212-kDa RdRp, and a 37-kDa movement protein (MP). CWMV RNA2 comprises 3,564 nt and encodes four proteins: a 19-kDa major capsid protein (CP), two minor CP-related proteins (a 23-kDa N-CP and an 84-kDa CP-RT, produced through an initiation of translation at the noncanonical CUG start codon or through occasional readthrough of the UGA termination codon, respectively), and a 19-kDa RNA silencing suppressor Diao et al., 1999;Sun et al., 2013;Yang et al., 2001). Both the 5ʹ and 3ʹ termini of RNA1 and RNA2 contain UTRs. The 3ʹ-UTR of RNA1 and RNA2 includes a highly conserved tRNA-like structure, but the function of this structure is unclear. Full-length CWMV infectious clones have recently been developed and used successfully to infect Triticum aestivum and Nicotiana benthamiana (Yang et al., 2016). Very few host factors have been identified to participate in CWMV infections or in other furovirus infections in plants.
Because eEF1A has been shown to regulate the replication of several RNA viruses, we decided to investigate the function of eEF1A in CWMV infection. We used in vivo assays to investigate whether eEF1A can positively influence CWMV infection in plants, as well as in vitro and reverse-genetic assays to determine whether eEF1A can facilitate CWMV infection in plants via its binding to the 3ʹ-UTR of CWMV genomic RNAs. homology-based analysis revealed that the amino acid sequences of eEF1As from Capsicum annuum, Solanum lycopersicum, Solanum pennellii, Solanum tuberosum, N. benthamiana, Ipomoea nil, Oryza sativa, Tarenaya hassleriana, and T. aestivum share approximately 98% sequence identity ( Figure S1). It has been reported that NbeEF1A is required both for TMV and TuMV infection Zeenko et al., 2002). Thus, we decided to analyse the expression of Western blot and northern blot assays further showed that this correlated with increases in CWMV CP and CWMV genomic RNAs concentrations (Figure 1d,e). Thus, because N. benthamiana is an excellent experimental host plant for CWMV studies, and because an efficient and reliable agroinfiltration method has been developed for reverse-genetic assays for CWMV infection (Yang et al., 2016(Yang et al., , 2017, we decided to focus our research on the role of NbeEF1A in CWMV infection in subsequent experiments.

| Silencing of NbeEF1A expression inhibits CWMV infection
A 400 nt sequence fragment of NbeEF1A was RT-PCR amplified and cloned into a tobacco rattle virus (TRV)-based vector to produce TRV:NbeEF1A. This vector was used to perform virus-induced gene silencing (VIGS) in N. benthamiana plants via agroinfiltration. At 7 dpi, five assayed plants were sampled and analysed by performing RT-qPCR. The results showed that NbeEF1A transcript levels in assayed plants #1, #3, and #4 were approximately 0.4to 0.26-fold lower than that in TRV:00-inoculated control plants ( Figure S2). To investigate the role of NbeEF1A in CWMV infection, we inoculated assayed plants #1, #3, and #4 with CWMV. Plants inoculated with TRV:00 and then CWMV (TRV:00 + CWMV) served as controls. At 21 dpi with CWMV, NbeEF1A-silenced plants showed milder CWMV symptoms than TRV:00 + CWMVinoculated plants (Figure 2a). At 7 days after infection with CWMV, western blot and northern blot assays showed that significantly lower levels of CWMV CP and genomic RNAs were detected in CWMV-inoculated #1, #3, and #4 plants than in TRV:00 + CWMVinoculated plants (Figure 2b,c). In addition, RT-qPCR analyses also showed that the replication level of CWMV RNA1 and RNA2 in CWMV-inoculated #1, #3, and #4 plants was significantly lower than that in leaves of TRV:00 + CWMV plants (Figure 2d). Due to the critical role of eEF1A in promoting mRNA translation, the effect of eEF1A on translation efficiency of CWMV genes was evaluated by RT-qPCR analysis. The results showed that the translation efficiency of CWMV CP in CWMV-inoculated NbeEF1Asilenced plants was 0.54-fold lower than that in TRV:00+CWMV plants ( Figure 2e). benthamiana plants at 7 to 28 dpi. Plants mock infiltrated with the empty vector were used as a control. (d) Western blot assay for NbeEF1A and CWMV capsid protein (CP) expression at 7 to 28 dpi in CWMV-infected N. benthamiana leaves using eEF1A-or a CP-specific antibody, respectively. Coomassie Brilliant blue (CBB)-stained RuBisCO gel was used to show protein loadings. (e) Northern blot assay for CWMV genomic RNAs (gRNAs) in CWMV-infected N. benthamiana plants at 7 to 28 dpi. Ethidium bromide-stained gel was used to visualize RNA loadings. The expression of eEF or CWMV CP genes was measured by quantitative reverse transcription PCR (RT-qPCR) using gene-specific primers. The relative expression level of these genes was calculated using the 2 −ΔΔCt method. Data shown are the mean ± SD of three biological samples. Each biological sample had three technical replicates. Significant differences between treatments were determined using Student's t test (*p < 0.05; n.s., no significant difference). Viral RNAs and proteins were quantified using ImageJ software

| Overexpression of NbeEF1A in plants enhances CWMV accumulation
To further elucidate the role of NbeEF1A during CWMV infection, we constructed a p35S:NbeEF1A-GFP expression vector to transiently overexpress NbeEF1A in N. benthamiana leaves. Leaves were coinfiltrated with Agrobacterium containing this expression vector and CWMV using agroinfiltration. Plants coinoculated with a green fluorescent protein (GFP) expression vector (p35S:GFP) and CWMV acted as controls. The expression of GFP and NbeEF1A-GFP in infiltrated plants was confirmed at 3 dpi by performing a western blot assay using an anti-GFP antibody ( Figure S3). Total protein and RNA were extracted from infiltrated leaves at 7 dpi and accumulation of CWMV CP and genomic RNAs was analysed by performing western blot and northern blot assays, respectively. Accumulation of CWMV CP and genomic RNAs was higher in NbeEF1A-GFP + CWMVinoculated plants than in control plants (Figure 3a,b). Moreover, RT-qPCR analysis also showed that the replication level of CWMV RNA1 and RNA2 in the NbeEF1A-GFP + CWMV-inoculated plants was 5.7-to 6.3-fold higher than that in leaves of plants coinoculated with GFP and CWMV ( Figure 3c). As expected, the translation efficiency of CWMV CP in the NbeEF1A-GFP + CWMV-inoculated plants was 1.45-fold higher than that in control plants ( Figure 3d).

| eEF1A binds to the 3ʹ-UTR of CWMV genomic RNAs but not CWMV RdRp
The successful infection of a plant by an RNA virus and disease symptom development are dependent on complicated molecular interactions between viral and host factors. Numerous reports have indicated that eEFs can interact with viral RdRps and/or viral genomic RNAs Li et al., 2010Li et al., , 2015Thivierge et al., 2008;Yamaji et al., 2006). To determine whether CWMV replication-associated proteins can interact with NbeEF1A, we performed yeast two-hybrid assays, coimmunoprecipitation (Co-IP) assays, and pull-down assays. Because the fulllength replication-associated protein has a large molecular weight, the CWMV replication-associated protein (Rep) was divided into NbeEF1A-GST was also observed by pull-down assays. The VPg-Pro protein of TuMV was used as a positive control in these assays.
Our results showed that although the VPg-Pro protein interacted with NbeEF1A, none of the CWMV replication-associated protein domains interacted with NbeEF1A (Figure 4b-d). Further analyses also showed that these three domains did not interact with TaeEF1A in yeast cells ( Figure S4). eEF1A has been reported to interact with viral RNAs to ensure RNA virus replication in cells (Blackwell & Brinton, 1997;Davis et al., 2007). Here, we decided to explore whether NbeEF1A can bind to CWMV genomic RNAs by performing electrophoretic mobility shift assays (EMSA). We prepared in vitro transcripts of the CWMV (+/−) 3ʹ-UTR and (+/−) 5ʹ-UTR of RNA1 and RNA2, which were biotin labelled (BL) at their 3ʹ ends (+/−). In F I G U R E 3 Transient overexpression of NbeEF1A in Nicotiana benthamiana plants enhances CWMV RNA accumulation. (a) Western blot assay for CWMV coat protein (CP) expression in plants coinoculated with CWMV and p35S:NbeEF1A-GFP (NbeEF1A-GFP + CWMV) through agroinfiltration using a CP-specific antibody at 7 days postinoculation (dpi). Plants coinoculated with CWMV and p35S:GFP (GFP + CWMV) acted as controls. Proteins were resolved using sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) stained with Coomassie Brilliant blue (CBB) to visualize protein loadings. Viral protein was quantified using ImageJ software. (b) Northern blot assay for CWMV RNA accumulation. Ethidium bromide-stained gel was used to visualize RNA loadings. Three plants from the same treatment were used for the assay. Viral RNA was quantified using ImageJ software. (c) Relative transcript level of viral RNA in NbeEF1A-GFP + CWMV-inoculated plants and GFP + CWMV-inoculated plants acted as controls. (d) Translation efficiency assay for CWMV CP in GFP + CWMV-inoculated plants and NbeEF1A-GFP + CWMV-inoculated plants served as controls. Translation efficiency calculated as the relative expression in polysomal/total RNA fractions. Relative transcripts levels are the mean ± SD of three biological samples; each biological sample had three technical replicates. Significant differences between treatments were determined using Student's t test (*p < 0.05) The assays also showed no effect on the formation of the 3ʹ-UTR RNA/NbeEF1A complex when unlabelled CWMV 301-781 RNA1 (UL-

| SL-6 in 3ʹ-UTR is crucial for CWMV accumulation
To investigate the function of SL-6 in eEF1A-regulated CWMV accumulation, we changed the ACCGGCC of SL-6 to ugaacau for producing the mutant m3ʹ-UTR ugaacau , and then complemen- Data presented are the mean ± SD from three biological samples per treatment; each biological sample had three technical replicates. Significant differences between treatments were determined using Student's t test (*p < 0.05) 0.63-fold lower while that in CWMV Ins(auguuca) -inoculated plants was similar than that in WT CWMV-inoculated plants (Figure 8f). Thus, we conclude that the SL-6 in the 3ʹ-UTR is important for CWMV multiplication in infected cells. Our analyses showed that TaeEF1A is also an important host factor for CWMV, BSMV, or WYMV infection in wheat plants (Figure 1a).

| D ISCUSS I ON
In addition, the expression of NbeEF1A was upregulated under CWMV, TMV, or TuMV infection in N. benthamiana (Figure 1b).
These results suggest that eEF1A may be a general host factor required for infection by different viruses. eEF1A is known to be highly conserved throughout the eukaryotes. Our amino acid sequence analysis also showed that eEF1As in N. benthamiana and T. aestivum are highly conserved ( Figure S1 (Figure 3a,b). We also found that the replication and translation efficiency of CWMV exhibited significant changes when NbeEF1A was silenced or overexpressed in N. benthamiana plants (Figures 2d,e and 3c,d). Based on these findings, we conclude that eEF1A is required for CWMV infection in plants. To our knowledge, this is the first report to show that eEF1A regulates CWMV infection in plants.
Virus replication complexes (VRC) are sites where viral RNAs are synthesized in cells (Cotton et al., 2009). Previous studies have reported that eEF1A is a component of VRCs and can control the replication of different viruses via interactions with viral RdRps Yamaji et al., 2006). In this study, we found that NbeEF1A cannot interact with CWMV RdRp in vivo or in vitro (Figure 4b-d). VRCs are known to contain ribosomes, tubulin-like structures, endoplasmic reticulum, RdRp and other viral proteins, and viral RNAs (Asurmendi et al., 2004;Liu et al., 2005). Moreover, eEF1A has been shown to form complexes with RdRp and other proteins encoded by TuMV   Translation efficiency calculated as the relative expression in polysomal/total RNA fractions. Data presented are the mean ± SD of three biological samples per treatment. Each biological sample had three technical replicates. Significant differences between treatments were determined using Student's t test (*p < 0.05) eEF1A can bind to aminoacylated tRNAs and other mammalian RNA species (Shamovsky et al., 2006). In a separate report, Zeenko et al. have shown that eEF1A can interact with the 3ʹ-UTR of TMV genomic RNA to regulate virus replication in N. benthamiana (Zeenko et al., 2002). TYMV and BMV also have TLSs in their RNAs and these TLSs can all bind to eEF1A to promote viral replication (Annamalai & Rao, 2007;. These previous reports encouraged us to investigate whether NbeEF1A can bind to the TLSs in UTRs of CWMV RNAs. Our results show that NbeEF1A is indeed capable of interacting with the 3ʹ-UTR of CWMV genomic RNAs (Figure 4e). Several studies have also shown that eEF1A can bind to RNA structures outside 3ʹ-UTRs. For example, eEF1A has been shown to bind to a region in the Hepatitis delta virus genomic RNA to promote HDAg mRNA transcription and the synthesis of its negative-strand RNA (Sikora et al., 2009). The 5ʹ-UTR of HIV-1 genomic RNA has been shown to interact with eEF1A and this interaction is important for late DNA synthesis during reverse transcription (Li et al., 2015). It has been demonstrated that for some plant RNA viruses the VRC could specifically recognize the 3ʹterminal portion of the viral genomic RNA, which contains a unique promoter to enhance viral replication (Osman et al., 2000;Vlot et al., 2001). The specific binding of NbeEF1A to the 3ʹ-UTR of CWMV genomic RNAs ( Figure 5) suggests a role for eEF1A-binding in both assembly of the CWMV VRC and template recognition for viral replication. However, other potential roles of the eEF1A interaction to CWMV RNA 3ʹ-UTR, such as anchoring the CWMV VRC to a specific membrane or cytoskeleton in a host cell or participation in cell-tocell spread of viral RNA, should also be examined in future studies.
Taken together, our results show that the 3ʹ-UTR of CWMV genomic RNAs is required for CWMV RNA replication.
The 3ʹ-UTRs often are composed of SLs, pseudoknots (PKs), or TLSs, and tend to be conserved among different RNA viruses and within virus groups. Here, we also used the RNA structure Web to predict the PKs in the 3ʹ-UTR of RNA1 and RNA2 of CWMV. The results showed that the 3ʹ-UTR of RNA1, but not of RNA2, contains a potential PK ( Figure S6c). Several studies have also shown that PKs in the TLS of turnip yellow mosaic virus, BMV, and TMV are involved in infectivity of these viruses (Chandrika et al., 2000;Deiman et al., 1998;Dreher et al., 1996). The PK region and TL regions interact with eEF1A independently or simultaneously during TMV infection (Zeenko et al., 2002). Thus, we speculate that the potential PKs may be responsible for the weak infectivity of CWMV ΔR1 (Figure 6c, d), which also requires further experimental analysis in the future. MST analysis also showed that the TLS in the CWMV 3ʹ-UTR is essential for the binding between NbeEF1A and CWMV genomic RNAs An early investigation of turnip yellow mosaic virus showed that the TLS at the 3ʹ end of viral RNA participates in virus replication . Here, six TSL structures were predicted in the CWMV RNA2 3ʹ-UTR ( Figure S6). Our mutational analyses showed that SL-6 was important for binding eEF1A and the CWMV 3ʹ-UTR (Figure 7). Because the affinity of eEF1A for viral RNA is independent of GTP and other RNA binding sites that are specific for binding to aa-tRNA (Slobin, 1983), we considered that the binding of eEF1A to the SL-6 of the CWMV 3ʹ-UTR plays an important role in CWMV infection. To confirm this idea, we generated a series of CWMV mutants with an altered SL-6 (CWMV ugaacau ), a deleted SL-6 (CWMV Δ3561-3567 ), and a compensatory SL-6 (CWMV InsACCGGCC ). in the TLS can affect RNA affinity for eEF1A . This finding suggests that the mutation in SL-6 of CWMV 3ʹ-UTR interferes with its ability to bind to eEF1A and disrupts the synthesis of the CWMV negative-strand, thereby impeding the initiation of viral RNA replication. eEF1A is one of the most abundant proteins in eukaryotic cells and is one of the most characterized proteins of the translational machinery (Andersen et al., 2003). Our results have demonstrated that eEF1A is involved in the replication and translation of CWMV through binding to the CWMV 3ʹ-UTR. Thus, it seems likely that CWMV might have evolved to use it in various ways. Future studies should reveal the relevance of eEF1A not only for CWMV replication but also for translation or other CWMV infection steps, which should contribute to our understanding of host-virus interactions, propagation strategies, and the adaptive evolution of RNA viruses.

| Plant growth and virus inoculation
N. benthamiana plants were grown inside a growth chamber maintained at 25 °C and a 14 hr light/10 hr dark cycle. Agrobacterium tumefaciens GV3101 harbouring CWMV RNA1 (pCB-35S-R1) or RNA2 (pCB-35S-R2) infectious clones were obtained from a previously reported source (Yang et al., 2016). The mutant CWMV RNA2 constructs produced in this study were also transformed individually into A. tumefaciens GV3101, and the Agrobacterium cultures were grown individually overnight at 28 °C in a yeast extract peptone medium containing kanamycin (50 µg/ml) and rifampicin (50 µg/ ml). The resulting Agrobacterium cultures were pelleted and then resuspended in an infiltration buffer (100 mM MES, pH 5.2, 10 mM MgCl 2 , 200 mM acetosyringone) to obtain an OD 600 of 0.6 followed by >2 hr incubation at 25 °C. Agrobacterium harbouring pCB-35S-R1 was mixed with Agrobacterium harbouring pCB-35S-R2 in a 1:1 ratio or with one of its derivatives prior to infiltration into N. benthamiana leaves. The infiltrated plants were grown inside a growth chamber maintained at 15 ± 2 °C, with a 14 hr/10 hr (light/dark) cycle and 70% relative humidity.
Wheat cv. Yangmai 158 plants were grown to the two-leaf stage in a greenhouse and then inoculated with in vitro transcripts of CWMV, BSMV, or WYMV after adding 1 vol of FES buffer (1% wt/vol sodium pyrophosphate, 1% wt/vol macaloid, 1% wt/vol celite, 0.5 M glycine, 0.3 M K 2 HPO 4 , pH 8.5, with phosphoric acid). Inoculum was applied by gently rubbing it on the surface of leaves that had been abraded with carborundum. To determine the effect of SL structures in the CWMV 3ʹ-UTR on viral replication, we deleted the SL-6 from the 3ʹ-UTR of CWMV RNA2 or altered its SL structure by replacing ACCGGCC with ugaacau to produce mutants 3ʹ-UTR Δ3561-3565 or m3ʹ-UTR ugaacau , respectively. The mutant 3ʹ-UTR Δ3561-3565 was made through RT-PCR amplification using specific primers from a CWMV-infected N. benthamiana leaf sample, and mutants m3ʹ-UTR ugaacau and m3ʹ- The construction of these three activation domain vectors has been described previously (Yang et al., 2017). The primers used in this study are listed in Table S1.

| Silencing NbeEF1A expression through virusinduced gene silencing
To silence NbeEF1A expression in N. benthamiana plants, leaves were infiltrated with Agrobacterium harbouring TRV:NbeEF1A. The VIGS procedure used in this study was similar to that reported previously (Ratcliff et al., 2001). The TRV:NbeEF1A-or TRV:00-infected (control) plants were grown at 25 °C for 7 days and then inoculated again with CWMV through agroinfiltration and maintained at 15 °C.

| RNA extraction and RT-qPCR analysis
Total RNA was extracted from T. aestivum and N. benthamiana leaf samples with a HiPure plant RNA mini kit (Magen). First-strand cDNA was synthesized using a first-strand cDNA synthesis kit (TOYOBO) followed by quantitative PCR using an Applied Biosystems QuantStudio 6 Flex system (Applied Biosystems) and a SYBR Green Master Mix kit (Vazyme). Each treatment had three biological replicates and each biological replicate had four technical replicates. The relative expression levels of assayed genes or the CWMV CP gene were calculated using the 2 −ΔΔCt method (Livak & Schmittgen, 2001).
The primers used in this study are listed in Table S1.

| Northern blot assays
Northern blot assays were performed as previously described (Yang et al., 2016). Briefly, total RNA (c.5 µg per sample) was separated in a 2% formaldehyde agarose gel through electrophoresis (60 V for 1.5 hr). The separated RNA bands were transferred onto a Hybond-N + membrane (Amersham Biosciences) followed by a 10 min crosslink under a UV light. The blot was hybridized with a digoxigenin-labelled probe specific for the 3ʹ end of CWMV genomic RNAs. The probe was produced using a Detection Starter Kit II following the manufacturer's instructions (Roche).

| Western blot assays
Western blot assays were conducted as previously described (Yang et al., 2016). Protein samples were separated using SDS-PAGE and transferred onto nitrocellulose membranes. Blots were blocked using phosphate-buffered saline (PBS) containing 5% skimmed milk, rinsed several times with PBS, and then probed with a specific primary mouse antibody (at a 1:5,000 dilution) followed by an anti-mouse (at a 1:10,000 dilution) alkaline phosphatase conjugate. The detection signal was visualized by incubating blots in an enhanced chemiluminescence solution according to the manufacturer's instructions (Invitrogen) and scanned using an Amersham Imager 680 machine (GE Healthcare BioSciences).

| Translation efficiency assay
Translation efficiency was assayed as previously described with minor modifications (Merchante et al., 2015). First, 10 g of N. benthamiana leaves was ground into fine powder in liquid nitrogen. Then, 2 g of sample was used to extract the total RNA with a HiPure plant

| Yeast two-hybrid assay
The yeast two-hybrid assay was performed using a Matchmaker Gold yeast two-hybrid system and a Yeastmaker yeast transformation system 2 according to the manufacturer's instructions (Clontech Laboratories).

| Electrophoretic mobility shift assay
Biotin was integrated at the 3ʹ end of the UL RNA probes using a Pierce RNA 3ʹ end biotinylation kit (Thermo Fisher Scientific) to produce BL RNA probes. The EMSA binding reaction and chemiluminescence detection were conducted using a LightShift chemiluminescent RNA EMSA (REMSA) kit (Thermo Fisher Scientific). Briefly, a BL RNA probe was incubated with a purified protein sample for 30 min at 25 °C. The mixture was analysed in a 5% native polyacrlamide gel by performing electrophoresis, transferred onto a Hybond-N + membrane, and crosslinked for 45 s under an UV light. The detection signal was scanned using an Amersham Imager 680 machine.

| Microscale thermophoresis assay
The affinity of the purified NbeEF1A for RNA was determined using Monolith NT.115 (NanoTemper Technologies). Microscale thermophoresis (MST) labelling of NbeEF1A was conducted in PBS solution containing a Monolith NT protein labelling kit RED according to the manufacturer's instructions (NanoTemper Technologies). Samples were then loaded into NanoTemper hydrophilic-treated capillaries.
The resulting samples were analysed by the manufacturer using NanoTemper analytical software to estimate their equilibrium dissociation constant (K D ) values.

| Pull-down
Purified GST-NbeEF1A protein was incubated with purified His, TuMV VPg-Pro-His, CWMV Met-His, CWMV Hel-His, or CWMV RdRp-His at room temperature for 15 min. Next, 25 μl of GST-Trap agarose (ChromoTek) was added to each reaction system, which was then incubated at 4 °C for 2 hr. After beads were collected and washed three times with Tris-buffered saline solution (TBS; 10 mM Tris-HCl pH 8.0, 150 mM NaCl), the reaction mixtures were run using SDS-PAGE and immunoblotted with an anti-GFP antibody (TransGene) and an anti-His antibody (TransGene).

| Co-immunoprecipitation assay
Proteins were transiently coexpressed in leaves of N. benthamiana by Agrobacterium infiltration. Coimmunoprecipitation assays were performed on N. benthamiana leaves that were harvested 2 days after infiltration, pooled, and ground in liquid nitrogen. Total protein was extracted with an extraction buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 10 mM MgCl 2 , 5 mM DTT, 0.1% Triton X-100). Protein extracts were incubated with 25 μl GFP-Trap agarose (ChromoTek) for immunoprecipitation ranging from 2 hr to overnight at 4 °C.
Finally, beads were collected and washed three times with TBS solution, then the reaction mixtures were run using SDS-PAGE and immunoblotted with an anti-GFP antibody and an anti-His antibody. Dr Xinshun Ding for his help during the preparation of this manuscript.

CO N FLI C T O F I NTE R E S T
The authors declare that there is no conflict of interest.

AUTH O R CO NTR I B UTI O N S
J.Y., H.Z., and J.C. conceived the project and designed the experiments. X.C. carried out the experiments with assistance from M.X., L.H., J.L., T.Z., J.Y., and Q.L. All authors analysed and discussed the results, and J.Y. wrote the manuscript. All authors confirm that they have no conflict of interest to declare.

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.