Soybean RNA interference lines silenced for eIF4E show broad potyvirus resistance

Abstract Soybean mosaic virus (SMV), a potyvirus, is the most prevalent and destructive viral pathogen in soybean‐planting regions of China. Moreover, other potyviruses, including bean common mosaic virus (BCMV) and watermelon mosaic virus (WMV), also threaten soybean farming. The eukaryotic translation initiation factor 4E (eIF4E) plays a critical role in controlling resistance/susceptibility to potyviruses in plants. In the present study, much higher SMV‐induced eIF4E1 expression levels were detected in a susceptible soybean cultivar when compared with a resistant cultivar, suggesting the involvement of eIF4E1 in the response to SMV by the susceptible cultivar. Yeast two‐hybrid and bimolecular fluorescence complementation assays showed that soybean eIF4E1 interacted with SMV VPg in the nucleus and with SMV NIa‐Pro/NIb in the cytoplasm, revealing the involvement of VPg, NIa‐Pro, and NIb in SMV infection and multiplication. Furthermore, transgenic soybeans silenced for eIF4E were produced using an RNA interference approach. Through monitoring for viral symptoms and viral titers, robust and broad‐spectrum resistance was confirmed against five SMV strains (SC3/7/15/18 and SMV‐R), BCMV, and WMV in the transgenic plants. Our findings represent fresh insights for investigating the mechanism underlying eIF4E‐mediated resistance in soybean and also suggest an effective alternative for breeding soybean with broad‐spectrum viral resistance.

and viruses, which are responsible for significant economic losses annually (Liu et al., 2016;Whitham et al., 2016). Among these, soybean mosaic virus (SMV) is the most widespread and devastating viral pathogen in soybean-growing areas, resulting in serious yield reductions and seed quality deterioration (Hill and Whitham, 2014;Hajimorad et al., 2018). Yield losses are usually reported to be approximately 8-35% (Hill and Whitham, 2014); however, losses of more than 50% and even total crop failure have been documented during severe outbreaks (Liao et al., 2002). SMV originates from SMV-infected seeds and is nonpersistently transmitted by more than 30 different migratory aphid species, within and among soybean fields (Steinlage et al., 2002). Symptoms induced by SMV infection include mosaic patterns, chlorosis, rugosity, curling, and necrosis of soybean leaves, subsequently leading to plant dwarfing and seed discoloration (seed coat mottling), which significantly reduces the commercial value of soybean seeds (Kim et al., 2016;Zhang et al., 2011). The tremendous damage suffered from SMV necessitates the introduction of viral resistance in soybean crops for improving soybean production and productivity in China (Gao et al., 2015b(Gao et al., , 2018.
Soybean mosaic virus is a member of the largest and most successful genus of plant pathogenic viruses, Potyvirus, within the family Potyviridae (Adams et al., 2005;Luan et al., 2016). Similar to other potyviruses, the genome of SMV is a monopartite, single-stranded, positive-sense RNA molecule of approximately 10 kb, harboring a viral genome-linked protein (VPg) covalently attached to the 5′ terminus and a poly(A) tail at the 3′ end (Gagarinova et al., 2008;Hajimorad et al., 2018). The viral genome contains two open reading frames (ORF) encoding 11 mature multifunctional proteins, namely protein 1 (P1), helper component-proteinase (HC-Pro), protein 3 (P3), pretty interesting Potyviridae ORF (P3N-PIPO), six kilodalton 1 (6K1), cylindrical inclusion protein (CI), six kilodalton 2 (6K2), VPg, nuclear inclusion a-proteinase (NIa-Pro), nuclear inclusion b (NIb), and coat protein (CP) (Chung et al., 2008;Gagarinova et al., 2008). Furthermore, based on their differential responses and pathogenicity to soybean plants, numerous SMV isolates have been grouped into seven strains (G1-G7) in the United States (Cho and Goodman, 1979) and into 22 strains (SC1-SC22) in China (Li et al., 2010). Additionally, a novel recombinant SMV strain (SMV-R), which likely originated from an interspecific recombination event between SMV and bean common mosaic virus (BCMV) or a BCMV-like virus, has been identified in China (Yang et al., 2011(Yang et al., , 2014. The use of naturally occurring host resistance is the most economical, effective, and eco-friendly approach for protecting against plant pathogens and preventing crop yield losses in agricultural practices (Kang et al., 2005;Maule et al., 2007). Resistance genes can be categorized as dominant or recessive, based on their inheritance; interestingly, dominant resistance genes predominantly confer resistance against bacteria and fungi, while recessive resistance appears to be more frequently found for viruses than for other plant pathogens (Diaz-Pendon et al., 2004;Truniger and Aranda, 2009;Wang and Krishnaswamy, 2012;Chandrasekaran et al., 2016). More specifically, genes conferring recessive resistance against potyviruses are much more frequent than those against other viruses, and potyviral resistance is often not restricted to a single potyvirus (Provvidenti and Hampton, 1992;Ruffel et al., 2002).
Previous studies have shown that eIF4E and its isoform eIF(iso)4E can be selectively recruited in various plant-potyvirus pairs (Duprat et al., 2002;Lellis et al., 2002;Sato et al., 2005;Estevan et al., 2014). eIF4E belongs to a multigene family, of which four genes, that is, eIF4E1 (accession no. EU912426), eIF4E2 (accession no. XM_003546012), eIF(iso)4E1 (accession no. XM_003535948), and eIF(iso)4E2 (accession no. BT098172), have been reported in soybean (Wang et al., 2013;Xu et al., 2017). Our previous research (Zhang, 2012) focused on eIF4E1 and eIF(iso)4E1, with a total of 208 soybean cultivars being used for SMV resistance assessment and 17 cultivars being identified as SMV resistant (Table S7). Further analyses on these 17 resistant cultivars proved that, compared with the soybean cultivar Nannong 1138-2 (highly susceptible host), five resistant cultivars harbored mutated eIF4E1s (Table S8 and Text S1), of which four were unable to interact with SMV VPg in the yeast two-hybrid (Y2H) screen system (Table S8). Furthermore, all eIF(iso)4E1s from the 17 resistant cultivars were the same and identical to that of Nannong 1138-2 (Text S1). Consequently, we speculated that eIF4E, rather than eIF(iso)4E, might play the leading role in the soybean-SMV pathosystem. Thus, in the present study, we focused on eIF4E1.
Considering the unique status of eIF4E, both as a crucial regulator of cellular metabolism and a controller of resistance/ susceptibility to potyviruses, we conducted experiments to identify spatiotemporal expression patterns of eIF4E1 in soybean, to analyse subcellular localization in Nicotiana benthamiana, and to determine its protein-protein interactions with SMV. Furthermore, using RNAi via a cotyledonary node-Agrobacterium-mediated transformation system, transgenic soybean plants expressing the transgene construct of inverted repeat-eIF4E1i fragments, which were able to form the RNA hairpin structure inducing specific post-transcriptional gene silencing of eIF4E1, were developed. Robust and broad-spectrum resistance against multiple SMV strains and two additional potyviruses, namely BCMV and watermelon mosaic virus (WMV), was observed in transgenic soybeans and was confirmed by monitoring for viral symptoms and viral titers. Results from this study provide fresh insights for investigating the molecular basis of eIF4E-mediated resistance in soybean, and may indicate an alternative strategy for breeding soybean resistant to SMV and other potyviruses.

| Spatiotemporal expression analysis of soybean eIF4E1
In case of temporal responses of eIF4E1 to SMV infection, the relative expression levels in Tianlong 1 (SMV susceptible) showed obvious up-and down-regulation patterns before and after 4 hr post-inoculation (hpi), respectively, exhibiting maximum expres-   (Wang et al., 2013;Xu et al., 2017).

| Subcellular localization of soybean eIF4E1 and analysis of protein-protein interaction with SMV
To examine the intracellular distribution of soybean eIF4E1 in planta, eIF4E1 was fused with green fluorescent protein (GFP) and transiently expressed in N. benthamiana. The results suggested that eIF4E1 was present in both the nucleus and cytoplasm ( Figure 2a Yeast two-hybrid screen system. Yeast co-transformants were identified on selective quadruple dropout medium SD/−Leu/−Trp/−Ade/−His/+X-α-Gal with blue color staining. Yeast containing pBT3-STE + pPR3, pBT3-STE-eIF4E1 + pPR3, or pBT3-STE + pPR3-SMV served as negative controls. Yeast cells co-transformed with pPR3-P3N-PIPO + pBT3-STE-GOS12 were used as positive control. (c) Bimolecular fluorescence complementation assay. eIF4E1-YN and SMV-YC were co-agroinfiltrated into leaves of 4-week-old N. benthamiana. Interactions between YN and YC, YN and SMV-YC, and eIF4E1-YN and YC were used as negative controls. Scale bars = 20 μm cytoplasm, revealing the involvement of VPg, NIa-Pro, and NIb in SMV infection and multiplication.

| Generation of transgenic soybean plants silenced for eIF4E1
An RNAi strategy was employed to determine the role of soybean eIF4E1 in SMV infection, and 31 positive T 0 plants were developed (Table S2). The silencing effect was assessed by quantitative realtime reverse transcription polymerase chain reaction (RT-qPCR) analysis of eIF4E1 (primer 4 in Table S1) transcript levels in T 0 plants.
Significant reductions (approximately 80-90%) in eIF4E1 transcript accumulation were detected in six randomly selected T 0 plants when compared with that in nontransformed plants (Figure 3a), indicating that the silencing strategy was efficient.
Southern blot analysis was performed, and 10 T 1 plants derived from T 0 line 1 (Table 1) exhibited the same integration pattern (single copy of T-DNA) in the soybean genome. As expected, all bands were greater than 3.66 kb in size (Figure 3b), which was greater than the fragment between the left border and the unique HindIII site ( Figure S1), and the hybridization signal was not detected in nontransformed plants. The single T-DNA insertion strongly suggested stable heritability, and two of these 10 T 1 plants (Table 1) were selected for propagating homozygous progenies for further analyses.

| Robust SMV resistance in T 1 and T 2 generations
One hundred and forty-eight T 1 soybean plants from 18 independent T 0 lines and 42 T 2 plants from T 0 line 1 were inoculated with SMV strain   (Table 1). Of all the T 0 lines, T 0 line 1 presented the best SMV resistance, with all T 1 progenies being highly resistant (Table 1). Hence, two T 1 plants (nos. 1-1 and 1-16, Table 1) derived from T 0 line 1 were selected for generating T 2 -T 4 progenies for further analyses. In the T 2 generation, 33 highly resistant plants were confirmed, with a percentage of up to 78.6%, and no susceptible plants were found (Table 1). Following the SMV challenge, nontransformed and negative T 1 plants exhibited typical mosaic leaves, remarkably dwarf plant phenotypes, and severe seed discoloration ( Figure 4a). However, resistant T 1 plants were symptomless, exhibited healthy growth, and produced clean seeds, similar to those of the mock control ( Figure 4a). Moreover, unlike nontransformed plants, which produced 84.65% mottled seeds, only 30.89% of the seeds harvested from T 1 lines were mottled, and seed coat mottling in T 2 -T 4 lines was almost completely eliminated (Table S3).
Furthermore, six highly resistant T 2 plants were randomly se- In the DAS-ELISA analysis, only three T 2 plants were identified as SMV susceptible, and viral titers of the other T 2 plants were below the detection limits (Table S4).
These results proved that robust SMV resistance can be achieved by silencing soybean eIF4E1 using RNAi, implying that soybean eIF4E1 acted as a susceptibility factor for SMV infection.

| Broad-spectrum resistance against multiple potyviruses in T 3 and T 4 generations
As shown in Table 2, highly resistant plants were the most numerous, and no susceptible plants were found in homozygous T 3 /T 4 generations inoculated with the seven potyviruses (SMV, BCMV, and WMV). However, all T 3 /T 4 plants were found to be susceptible to bean pod mottle virus (BPMV) ( Table 2), indicating that eIF4E1mediated resistance was nonfunctional against BPMV, which may be due to its generic position (genus Comovirus; family Secoviridae).
As shown in Figure 5a, compared with the leaves of nontransformed  Figure S2). Although the virus content in T 2 /T 4 plants inoculated with SMV strain SC3 was far less than that in nontransformed plants (Figures 4b and 5b), it was still more than that of Kefeng 1 to a certain extent. We speculated that this resulted from the remaining low transcript levels of eIF4E1 in transgenic plants (Figure 3a), which could sustain multiplication for a small amount of virus. DAS-ELISA was performed with T 2 /T 3 lines at 3 and 5 weeks post-inoculation (wpi), and viral titers of T 2 /T 3 lines separately challenged with the seven potyviruses were below 2.0, at both 3 and 5 wpi, demonstrating robust resistance to these viruses (Tables S5 and S6). However, consistent with the results of resistance evaluation (Table 2 and Figure 5a) and RT-qPCR (Figure 5b), both nontransformed plants and transgenic lines were susceptible to BPMV (Tables S5 and S6).
In summary, these experiments provided evidence of the involvement of soybean eIF4E1 in broad-spectrum potyvirus resistance, suggesting that soybean eIF4E1 is the susceptibility factor, not only for SMV, but also for BCMV and WMV.

| D ISCUSS I ON
The cap-binding protein eIF4E participates in initiating mRNA translation and in controlling resistance/susceptibility to potyviruses.
Subcellular localization showed that soybean eIF4E1 was simulta- Chinese SMV strains have been fine-mapped to soybean chromosomes 2, 6, 13, and 14 (MLG-D1b, C2, F, and B2) (Hajimorad et al., 2018). Although Rsv and Rsc loci are located in close proximity to each other, the allelic relationship between them remains unclear, and none of these genes have been cloned thus far, therefore it is impossible to simply transform the resistance genes for generating transgenic SMV resistance (Liu et al., 2016;Hajimorad et al., 2018). In addition, the resistance spectrum of the Rsv and Rsc loci is limited or late-susceptible, making it difficult to cultivate soybean varieties with multistrain SMV resistance through traditional breeding programmes, which is a labour-intensive and timeconsuming process, and is always accompanied by the generation of undesirable traits (Gao et al., 2015a). Furthermore, strong selection pressure resulting from the extensive use of dominant genes is an important driving force for the frequent emergence of resistance-breaking SMV strains/isolates (Steinlage et al., 2002;Gagarinova et al., 2008). In comparison with dominant resistance, recessive resistance is often broader and more durable because of its lower selective pressure on the viruses (Pyott et al., 2016;Gal-On et al., 2017;Hajimorad et al., 2018).

Introduction of viral segments into plants might raise public concern
and generate new viral variants through recombination between the introduced viral segments and other infecting viruses (Wang et al., 2013). Moreover, RNAi targeting viral genes may be hindered by the continuously evolving SMV population, possessing high variability along with error-prone replication, mutation, and recombination; as a result, the specificity of the RNAi sequence would gradually be attenuated. Hence, silencing the soybean eIF4E1, as shown in the present study, can be an effective alternative for controlling SMV infections.
Functional redundancy has been observed between eIF4E and eIF(iso)4E in plant growth, and tobacco plants exhibited the semidwarf phenotype only when eIF4E and eIF(iso)4E genes were simultaneously silenced (Combe et al., 2005). Previous studies using RNAi targeting eIF4E factors to generate viral resistance have shown differential developmental phenotypes in diverse crop species (Mazier et al., 2011;Rodríguez-Hernández et al., 2012;Wang et al., 2013;Xu et al., 2017). Transgenic tomato lines silenced for eIF4E showed slightly impaired growth and fertility, while no obvious vegetative defects were observed in lines silenced for eIF(iso)4E; however, the F 1 hybrid resulting from these two lines exhibited a pronounced semi-dwarf phenotype, suggesting a cumulative effect of the silencing of eIF4E and eIF(iso)4E genes (Mazier et al., 2011). Eight transgenic melon lines silenced for eIF4E were obtained and self-pollinated, of which only one T 0 line produced abundant T 2 seeds, as transgenesis often affected growth and fertility of the resulting plants (Rodríguez-Hernández et al., 2012). Transgenic plum lines lacking either eIF4E or eIF(iso)4E did not show any phenotypic alterations, compared with the wild-type plants, indicating a complementary effect of the two isoforms (Wang et al., 2013). Transgenic peanut plants silenced for eIF4E and/or eIF(iso)4E did not phenotypically differ from the control plants (Xu et al., 2017). In the present study, no apparent developmental defects were observed in the transgenic soybean plants silenced for eIF4E1, which might be due to the silencing effect not being thorough and the compensatory functions of other genes.

Previous research has confirmed that both eIF4E1 and eIF4E2
are involved in viral resistance in tomato (Mazier et al., 2011). In the present study, many mildly resistant (38.5%) and susceptible (27.7%) plants were identified in the T 1 generation (Table 1), implying that most T 0 lines did not trigger much SMV resistance, although they exhibited a strong reduction in eIF4E1 transcript accumulation ( Figure 3a). Interestingly, only one (T 0 line 1) of the 18 T 0 lines showed significant resistance and all its T 1 progenies were highly resistant to SMV (Table 1). Hence, we speculated that in T 0 line 1, soybean eIF4E2 was also silenced, which enhanced the viral resistance. To verify this hypothesis, 24 T 5 plants derived from T 0 line 1 were randomly selected for RT-qPCR analysis of the eIF4E1 and eIF4E2 (primer 5 in Table S1) transcript levels. As shown in Figure S3, a significant decrease in transcript accumulation was observed in T 5 plants, not only in eIF4E1 (more than 90% decrease), but also in eIF4E2 (60-90% decrease), when compared with nontransformed plants. This demonstrated that the enhanced viral resistance in the T 0 line 1 could be attributed to the simultaneous silencing of soybean eIF4E1 and eIF4E2, which is consistent with the fact that both eFI4E1 and eIF4E2 have to be down-regulated for viral resistance in tomato (Mazier et al., 2011). We can therefore conclude that soybean eIF4E1 and eIF4E2 play overlapping or redundant roles in the virus multiplication cycle.
SMV, BCMV, and WMV can infect soybean crops, resulting in yield reductions, and mixed infections and synergistic interactions are common among these viruses in Chinese field-grown soybean plants (Zhou et al., 2014;Yang et al., 2017Yang et al., , 2018. Furthermore, genetic exchanges among SMV, BCMV, and WMV occur frequently, and recombinant SMV variants have been reported prevalent in Chinese soybean fields, presenting a complicated and severe challenge to soybean farming in China (Yang et al., 2011(Yang et al., , 2014Zhou et al., 2015;Chen et al., 2017;Jiang et al., 2017). Hence, it is imperative to confer soybean plants with resistance, not only against SMV, but also against BCMV and WMV. In this study, a high level of broad-spectrum resistance to five SMV strains (SC3/7/15/18 and SMV-R), BCMV, and WMV was developed in transgenic soybean (Tables 1 and 2, Figures 4 and 5, and Tables S4-S6). Our results suggest that eIF4E-mediated resistance to potyviruses, based on RNAi, is effective and broad-spectrum, providing an efficient strategy for combatting viral pathogens in soybean.

| Expression analysis of soybean eIF4E1 using RT-qPCR
Spatiotemporal expression profiles of eIF4E1 were explored in soy- 12, 24, 48, and 72 hpi). Inoculation was performed as previously described (Li et al., 2010), and the relative expression levels were calibrated using mock-inoculated (inoculated with PBS) controls. To determine the spatial expression patterns of eIF4E1, samples were collected from various healthy soybean tissues, including roots, stems, leaves, flowers, immature pods, and mature seeds, from Tianlong 1 and Kefeng 1. Roots, stems, and leaves were collected at the V2 stage, flowers were collected at the R2 stage, and immature pods were collected at the R5 stage. All samples were stored at −80 °C until RT-qPCR analysis.
Gene-specific primers for RT-qPCR were designed targeting soybean eIF4E1 (primer 3 in Table S1), using Primer Premier 5.0 software, and the gene Tubulin (accession no. AY907703; primer 6 in Table S1) was used as an internal reference control. Total RNA extractions and first-strand cDNA syntheses were performed using an RNA Simple Total RNA Kit (Tiangen) and PrimeScript RT Master Mix (Takara), respectively, according to the manufacturer's instructions. RT-qPCR was performed in a 20-μL final volume, containing 2 μL of template cDNA (approximately 50 ng), 0.4 μL of each primer (10 μM), 10 μL of 2 × SYBR Premix Ex Taq (Takara), and 7.2 μL of sterilized double-distilled water. Thermal conditions were set as follows: 95 °C for 30 s; followed by 40 cycles at 95 °C for 5 s, 55 °C for 30 s, and 72 °C for 30 s. Reactions were analysed in triplicate, in 96-well plates, on a LightCycler 480 II (Roche). Transcript levels were quantified using the relative quantification (2 -ΔΔCt ) method (Livak and Schmittgen, 2001) and data were compared with internal controls.

| Subcellular localization
The 711-bp full-length coding sequence of eIF4E1 (primer 1 in Table S1) without its stop codon was amplified from Tianlong 1 by RT-PCR using KOD FX (Toyobo). According to the manufacturer's manual for the

| Y2H and BiFC assays
Y2H screening was performed using the Matchmaker DUAL membrane system (Dualsystems Biotech) according to the manufacturer's protocols. The eIF4E1 of Tianlong 1 and 11 genes of SMV strain SC3 (primers 8-18 in Table S1) were amplified by RT-PCR using

| Western blot analysis
The expression of fusion proteins in subcellular localization ( Figure   S4a) and BiFC ( Figure S4b,c) was verified by western blot analysis.

| Vector construction, soybean transformation, and confirmation of transgenepositive plants
The 348-bp RNAi fragment eIF4E1i (primer 2 in Table S1) was amplified from the eIF4E1 coding sequence (nucleotide sites 267-614) of Tianlong 1 by RT-PCR and recombined into the vector pB7GWIWG2(II) using the Gateway system. The resulting recombinant construct ( Figure S1) contained the phosphinothricin acetyltransferase (bar) gene conferring resistance to the herbicide phosphinothricin and was introduced into A. tumefaciens EHA105.
Tianlong 1 was used in the cotyledonary node-Agrobacteriummediated transformation system and putative transformants were simultaneously verified by leaf-painting, PCR, and LibertyLink strip.
Soybean transformation and confirmation of transgene-positive plants were performed as previously described (Gao et al., 2015a).

| Southern blot hybridization analysis
Total genomic DNA (c.30 μg) was digested completely with the HindIII restriction endonuclease (Thermo), which recognizes a unique site within the T-DNA region ( Figure S1). Digested DNA was separated on 0.8% agarose gel and transferred to Hybond-N + nylon membrane (Amersham). A PCR-generated bar gene fragment (primer 7 in Table S1) labelled with digoxigenin (DIG)-High Prime (Roche) was used as a probe ( Figure S1). Prehybridization, hybridization, membrane washing, and signal detection were carried out using DIG-High Prime DNA Labeling and Detection Starter Kit II (Roche), according to the manufacturer's protocols.
Mechanical inoculation was carried out in an insect-proof greenhouse as previously described (Li et al., 2010), and plants were regularly sprayed with pesticides to prevent cross-infection via aphids.

| Molecular detection of virus accumulation in transgenic soybeans
At the transcriptional level, virus accumulation in T 2 /T 4 generations was detected by RT-qPCR analysis of the viral CP genes (primers 19-22 in Table S1), and the gene Tubulin was used as an internal ref-

CO N FLI C T O F I NTE R E S T S
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.

S U PP O RTI N G I N FO R M ATI O N
Additional supporting information may be found online in the Supporting Information section.