Variability in eukaryotic initiation factor iso4E in Brassica rapa influences interactions with the viral protein linked to the genome of Turnip mosaic virus

Plant potyviruses require eukaryotic translation initiation factors (eIFs) such as eIF4E and eIF(iso)4E to replicate and spread. When Turnip mosaic virus (TuMV) infects a host plant, its viral protein linked to the genome (VPg) needs to interact with eIF4E or eIF(iso)4E to initiate translation. TuMV utilizes BraA.eIF4E.a, BraA.eIF4E.c, BraA.eIF(iso)4E.a, and BraA.eIF(iso)4E.c of Brassica rapa to initiate translation in Arabidopsis thaliana. In this study, the BraA.eIF4E.a, BraA.eIF4E.c, BraA.eIF(iso)4E.a, and BraA.eIF(iso)4E.c genes were cloned and sequenced from eight B. rapa lines, namely, two BraA.eIF4E.a alleles, four BraA.eIF4E.c alleles, four BraA.eIF(iso)4E.a alleles, and two BraA.eIF(iso)4E.c alleles. Yeast two-hybrid (Y2H) and bimolecular fluorescence complementation (BiFC) analyses indicated that TuMV VPg could not interact with eIF4E, but only with eIF(iso)4E of B. rapa. In addition, the VPgs of the different TuMV isolates interacted with various eIF(iso)4E copies in B. rapa. In particular, TuMV-UK1/CDN1 VPg only interacted with BraA.eIF(iso)4E.c, not with BraA.eIF(iso)4E.a. Some single nucleotide polymorphisms (SNPs) were identified that may have affected the interaction between eIF(iso)4E and VPg such as the SNP T106C in BraA.eIF(iso)4E.c and the SNP A154C in VPg. Furthermore, a three-dimensional structural model of the BraA.eIF(iso)4E.c-1 protein was constructed to identify the specific conformation of the variable amino acids from BraA.eIF(iso)4E.c. The 36th amino acid in BraA.eIF(iso)4E.c is highly conserved and may play an important role in establishing protein structural stability. The findings of the present study may lay the foundation for future investigations on the co-evolution of TuMV and eIF(iso)4E.


Results
Variability in eIF4E and its isoform eIF(iso)4E in B. rapa. Two copies of eIF4E (BraA.eIF4E.a and BraA.eIF4E.c) and two copies of eIF(iso)4E [BraA.eIF(iso)4E.a and BraA.eIF(iso) 4E.c] could potentially complement Col-0::dSpm [possessing a transposon knock-out of eIF(iso)4E]. Thus, primers were designed to amplify the cDNAs of these copies in the eight B. rapa lines, followed by sequencing. A sequence analysis identified several variants within the BraA.eIF4E.a of the eight lines (the lines 80122, 80186, 80124, Chiifu, 2079, and BP058 are resistant to TuMV C4, whereas 80425 and R-o-18 are susceptible). Most of the variations were non-synonymous, although some were synonymous (nts 300, 444, 525, and 564) (Table S1). There were no differences in nucleotide/amino acid sequences among lines 80124, BP058, and 2079, which included only three differing bases/ amino acids (nts 98, 164, and 638; aas 33, 55, and 213) compared to line 80186 (Tables S1 and 1). The amino acid sequences of 80122, 80425, Chiifu, and R-o-18 were the same, although there were also some synonymous nucleotide variations. Numerous variations were also identified in BraA.eIF4E.c, although most were synonymous (Tables S2 and 2). Notably, the nucleotide/amino acid sequences of 80122 and Chiifu were the same, as were those of 80425 and 2079. In addition, the nucleotide/amino acid sequences of 80186, BP058, and R-o-18 were the same, and differed in two bases/amino acids from 80124.

Specific single nucleotide polymorphisms (SNPs) affect the interaction between eIF(iso)4E
and VPg. Some amino acid changes seem to be involved in strain-specific interactions between eIF(iso)4E and TuMV VPg. The Y2H and BiFC analysis indicated that TuMV-UK1/CDN1 VPg could interact with BraA. eIF(iso)4E.c-1 (Chiifu) but not with BraA.eIF(iso)4E.c-2 (R-o-18). Between BraA.eIF(iso)4E.c-1 (Chiifu) and BraA.eIF(iso)4E.c-2 (R-o-18), five differing bases (nt T 106 C, C 155 T, T 239 C, C 449 A, and C 546 T) were identified (Table S4), which were predicted to result in the substitution of four amino acids (F 36 L, A 52 V, I 80 T, and P 150 Q) ( Table 4). Thus, primers were designed at the four loci, and site-directed mutagenesis (using eIF(iso)4E of Chiifu as template) was successfully implemented based on the overlap-extension PCR to detect which locus was essential for the interactions. Y2H and BiFC analyses indicated that the amino acid substitution F 36 L (nt T 106 C) in BraA. eIF(iso)4E.c played a critical role in the interaction between TuMV-UK1 VPg and BraA.eIF(iso)4E.c-1 (Chiifu), and the amino acids A 52 V (nt C 155 T), I 80 T (nt T 239 C), and P 150 Q had little influence (Fig. 3A,B). Compared to TuMV-UK1 VPg, TuMV-CDN1 VPg in the Y2H and BiFC assays performed differently in the interaction: amino acids F 36 L (nt T 106 C), A 52 V (nt C 155 T), and I 80 T (nt T 239 C) in BraA.eIF(iso)4E.c were the key elements; and P 150 Q did not play an essential role in the observed interaction (Fig. 3A,B). Taken together, the amino acid F 36 L (nt T 106 C) in BraA.eIF(iso)4E.c is a key site of the protein that generally affects the interaction between BraA. eIF(iso)4E.c-1 and TuMV-UK1/CDN1 VPg.
TuMV-C4 VPg could interact with BraA.eIF(iso)4E.a-3 (80425); while TuMV-CDN1 VPg could not. Sequence analysis of TuMV-C4 VPg and TuMV-CDN1 VPg identified five base/amino acid changes (nts A 154 C, G 289 A, A 301 G, A 313 G, and A 349 G; aas I 52 L, E 97 K, N 101 D, N 105 D, and I 117 V) (Fig. 3C). Five mutants (using TuMV-C4 VPg as a template) were obtained by overlap-extension PCR. The Y2H and BiFC analyses indicated that the interaction between TuMV-C4 VPg and BraA.eIF(iso)4E.a-3 (80425) was mediated by the amino acid substitutions I 52 L, E 97 K, and N 105 D of TuMV VPg, and the amino acid substitution N 101 D and I 117 V did not affect this particular interaction (Fig. 3D,E). Compared to BraA.eIF(iso)4E.a-3 (80425), BraA.eIF(iso)4E.c-1 (Chiifu) exhibited differences in this interaction. In the trials, the amino acid substitution I 52 L was essential for the interaction, whereas the other four sites did not affect the interaction (Fig. 3D,E). Taken together, the amino acid substitution I 52 L in TuMV-C4 VPg plays a critical role in the interaction between BraA.eIF(iso)4E.a and TuMV VPg. Therefore, various amino acids in eIF(iso)4E were essential to the interaction, whereas a few amino acids in TuMV VPg were also significant in the interaction.
Analysis of the mechanism underlying the interaction of TuMV VPg-eIF(iso)4E. Amino acid sequence alignment of BraA.eIF(iso)4E.c-1 (Chiifu) and BraA.eIF(iso)4E.c-2 (R-o-18) identified four substitutions: Phe/Leu-36, Ala/Val-52, Ile/Thr-80, and Pro/Gln-150 (Table 4). Furthermore, a three-dimensional (3D) structural model of the BraA.eIF(iso)4E.c-1 protein was constructed (Fig. 4A) to identify the special conformation of the four sites. The 36 th amino acid in BraA.eIF(iso)4E.c is highly conserved and is pivotal in establishing the structural stability of the protein. The amino acid Phe is highly conserved in plant eIF4Es and eIF(iso)4Es, whereas in human and yeast, this site is occupied by a Leu residue 54 . The 36 th amino acid in BraA.eIF(iso)4E.c changed from Phe to Leu, and Phe is an aromatic amino acid that contains a benzene ring. However, Leu is an aliphatic amino acid that is linear and does not contain a benzene ring (Fig. 4B). Thus, the two amino acid substitutions may influence the biochemical functions and the structure of the protein. Compared to the 36 th amino acid, the 52 th , 80 th , and 150 th amino acids are not conserved in the protein. The 52 th amino acids changed from Val to Ala, the 80 th changed from Ile to Thr, and the 150 th changed from Pro to Gln, which are all aliphatic amino acids (Fig. 4B). The structure of the BraA.eIF(iso)4E protein includes eight β-strands, three α-helices, and three extended loops. There is a large cavity at its cap-binding site that undergoes conformational changes in the cap-binding loops. The 36 th amino acid is located in the middle of the first β-strands region and may play an essential role in protein structure and function (Fig. 4A). In addition, the 36 th amino acid is located in the cap-free structure of BraA.eIF(iso)4E protein, thereby suggesting that it plays an essential role in the allosteric regulation of the BraA. eIF(iso)4E protein.

Discussion
In a previous study, the BraA.eIF4Es and BraA.eIF(iso)4Es from the B. rapa 'RLR22' line could not interact with the TuMV isolates 51 . The two copies of eIF4E (BraA.eIF4E.a and BraA.eIF4E.c) and two copies of eIF(iso)4E [BraA.eIF(iso)4E.a and BraA.eIF(iso)4E.c] were transformed into Col-0::dSpm, which had a transposon knocked out of the eIF(iso)4E gene. However, this resulted in a change from complete susceptibility to complete resistance to TuMV, and all four Brassica transgenes complemented the A. thaliana eIF(iso)4E knockout. These changes conferred susceptibility to both mechanical and aphid challenge with TuMV 51,52 . In this study, the Y2H and BiFC assays also showed that the TuMV-C4/UK1/CDN1 isolates did not interact with BraA.eIF4Es, but rather with BraA.eIF(iso)4Es. The interaction of the eIFs and TuMV VPgs differed between B. rapa and A. thaliana. BraA.eIF4Es and BraA.eIF(iso)4Es from the B. rapa 'RLR22' line could not interact with the TuMV isolates in vitro, but could interact with the TuMV isolates in the Col-0::dSpm, which was somewhat misleading. Genomic analyses of diploid B. rapa have indicated that it evolved from a hexaploid ancestor and then underwent a whole genome triplication event 55 , which resulted in more gene copies in B. rapa than in A. thaliana. These genomic changes have formed more complex compounds that are related to the TuMV infection process, and the same genes induce different results in B. rapa and A. thaliana.
Thus, different parts of eIF(iso)4E and various amino acids in VPg influence this particular interaction. Sequence comparison between BraA.eIF(iso)4E.a and BraA.eIF(iso)4E.c identified various amino acid substitutions that may affect the interactions (Fig. S1). Comparison of the TuMV-C4 and TuMV-UK1 VPgs identified four amino acids substitutions, namely, F 89 L, N 105 D, P 114 S, and M 119 V (Fig. 3C). These amino acid changes, particularly those involving residues 89 and 114, may influence the interaction. Thus, it is possible that different strains of TuMV interact with various eIF(iso)4E proteins to influence protein translation. Some amino acids indicated evidence of positive selection, which may have contributed to virus resistance in eIF4E and eIF(iso)4E. For example, in wheat, the G 107 R substitution in the cap-binding pocket plays a key role in both VPg interactions and cap-binding, whereas the L 79 R change that is located within an external loop influences VPg, but not cap-binding 56 .
BraA.eIF(iso)4E.a interacted with the TuMV-C4 VPg, whereas BraA.eIF(iso)4E.a-1 and −2 in 80122 showed loss of function and could not interact with the TuMV-C4 VPg. Thus, BraA.eIF(iso)4E.a-1 and −2 are resistant alleles for TuMV-C4, and the deleted parts of their proteins (from 70th to 200th amino acids) are essential to this   56 . The co-evolution of the VPg of TuMV and eIF(iso)4E in B. rapa may have resulted in variations in both proteins. Further investigation into the co-evolutionary relationship between TuMV and B. rapa suggests that the amino acids of eIF(iso)4E that influence its interaction with the VPgs of different TuMV isolates are highly variable.
Accordingly, the TuMV isolate resistance loci could be screened to identify resistance genes and could be cloned to generate plants with broad-spectrum resistance. Furthermore, the 3D structural model of the BraA. eIF(iso)4E.c protein indicates that the 36 th amino acid is highly conserved and plays an important role in stabilizing the protein structure. However, the main purpose of building a structural protein model was to provide the location of the mutation in a 3D structural context. This may provide information on whether the mutation is buried within the hydrophobic core, lying on the surface, located near an active site, proximal to an interface, or close to a site of post-translational modification. This model may also be used as a guide in predicting the effect of specific mutations on protein function.
In Arabidopsis, TuMV VPg can only interact with LSP [eIF(iso)4E]. Mutations involving LSP (i.e., lsp mutants) exhibit premature termination, thereby conferring loss of function and resistance to TuMV 18,48 . In B. rapa, retr02 [BraA.eIF(iso)4E.a] has been identified as a recessive resistance gene for TuMV C4 16,59 . The product of the resistance gene retr02 is also involved in premature protein termination and is thus unable to interact with TuMV C4 VPg, thereby resulting in resistance to TuMV C4 52 . Thus, the introduction of mutations in the BraA.eIF(iso)4E.a gene may confer plant resistance to TuMV C4, e.g., Retr02, and mutations in the BraA.eIF(iso)4E.c gene induce resistance to TuMV UK1. In addition, mutations in both BraA.eIF(iso)4E.a and BraA.eIF(iso)4E.c confer resistance to both TuMV C4 and TuMV UK1. It is also possible to determine whether TuMV isolates affect such mutants; and when these are resistant, the range of activity (broad resistance). This strategy could be used to create new varieties that may be used in breeding as well as in marker-assisted selection.
Identification of eIF4E and its isoform eIF(iso)4E in B. rapa. The B. rapa genome has been sequenced, and the Brassica database (BRAD) includes predicted genes and associated annotations (InterPro, KEGG2, SWISS-PROT), B. rapa genes orthologous to those in A. thaliana, and genetic markers and maps for B. rapa.
Sequences representing the complete set of eIF4E and eIF(iso)4E genes in A. thaliana were acquired from The Arabidopsis Information Resource (TAIR) database (www.arabidopsis.org) and were used to search the data from the B. rapa ssp. pekinensis cv. Chiifu genome V1.5 and its set of annotated genes (http://Brassicadb.org) for homologous genes. Estimation of the number of eIF4E and eIF(iso)4E genes in the genome of B. rapa was conducted by analysis of expressed sequence tag (EST) data downloaded from the NCBI EST database and mRNA sequencing data (unpublished data). Full-length eIF4E and eIF(iso)4E protein sequences of A. thaliana genes were retrieved from the TAIR database and the UniProt protein database (www.uniprot.org).
Cloning and sequencing of eIF4E and its isoform eIF(iso)4E. Total RNAs were extracted from the leaves of the B. rapa lines (80122, 80425, 80186, 80124, Chiifu, 2079, R-o-18, and BP058) using TRIzol reagent (Invitrogen, Carlsbad, CA, USA), and first-strand cDNA was synthesized with a polydT primer using a Prime Script TM RT-PCR kit (TaKaRa, Dalian, China) according to the manufacturer's instructions. The cDNAs were used as templates for PCR.
Generic primers (Table S5) were designed using the eIF4E gene reference sequence from the B. rapa genome, which encompassed the majority of the ORF. PCR was performed on the cDNAs using KOD Hot Start DNA polymerase (TOYOBO, Osaka, Japan). The PCR products were sequenced to analyze the allelic variability of the genes.

Y2H.
Interactions between proteins were assayed with a Gal4-based Y2H system, as described by the manufacturer (Clontech, Mountain View, CA, USA). Yeast strains and plasmid vectors were obtained from Clontech Laboratories (Clontech). A bait plasmid, pGBKT7, was used to fuse the VPg to the DNA-binding domain of BD. A prey plasmid, pGADT7, was used to express the eIF4E genes (Clontech). Gene-specific primers were designed to introduce restriction enzyme sites (Table S6). The DNA sequences encoding VPg from TuMV-C4 and TuMV-UK1 were amplified using forward primer Bio120213 (NdeI site) and reverse primer Bio120214 (XmaI site). BraA.eIF4E.a was amplified using forward primer Bio120850 (EcoRI site) and reverse primer Bio120851 (XhoI site), BraA.eIF4E.c was amplified using forward primer Bio120852 (EcoRI site) and reverse primer Bio120853 (XhoI site), BraA.eIF(iso)4E.c was amplified using forward primer Bio120854 (EcoRI site) and reverse primer Bio120855 (XhoI site), and BraA.eIF(iso)4E.a was amplified using forward primer Bio120075 (EcoRI site) and reverse primer Bio120076 (XhoI site). The eIF(iso)4E sequences from Arabidopsis Col-0 were amplified using forward primer Bio120582 (EcoRI site) and reverse primer Bio120583 (XhoI site), and cloned into pGADT7 as a positive control for the interaction in the yeast two-hybrid assay. The amplified fragments were digested with EcoRI/XhoI and NdeI/XmaI and cloned into the corresponding restriction sites of pGADT7 and pGBKT7, respectively. All constructs were confirmed by sequencing. The Matchmaker GAL4 two-hybrid system (Clontech) was used according to the manufacturer's protocols. pGADT7:4E and pGBKT7:VPg constructs were transformed into AH109 yeast strains. After yeast transformation, colonies were grown on various selective media lacking leucine, tryptophan, histidine, and adenine (SD-LW/ SD-LWH/SD-LWHA). Plates were incubated at 30 °C and growth was checked 3-5 days after inoculation. Each assay was performed in triplicate. The empty vectors pGADT7 and pGBKT7 were used as negative controls; the interaction between murine p53 and SV40 large T antigen (controls from the Matchmaker GAL4 two-hybrid system 3) was used as a positive control; and TuMV-VPg and Arabidopsis eIF(iso)4E (LSP) were also used as positive controls. In addition, each partner and empty vectors were used as assay controls.
BiFC. Molecular techniques were performed using standard protocols [60][61][62] . The eIF(iso)4E genes were amplified using the primers above, which contained the BamHI and XhoI sites, and the eIF(iso)4E genes PCR products and pSPYNE empty vector were digested by BamHI and XhoI. Then, the recombinant vector eIF(iso)4E-pSPYNE was constructed using T 4 ligase. Similarly, the TuMV VPg genes were amplified using specific primers (Table S7), which contained the ClaI and XhoI sites, and the TuMV VPg gene PCR products and pSPYCE empty vector were digested by the ClaI and XhoI sites. The primer pair Bio120901/Bio120902 was used to amplify BraA.eIF4E.a, Bio120903/Bio120904 for BraA.eIF4E.c, Bio120905/Bio120906 for BraA.eIF(iso)4E.c, Bio120907/Bio120908 for BraA.eIF(iso)4E.a, Bio120909/Bio120910 for LSP, and Bio120911/Bio120912 for VPg. The recombinant vector TuMV VPg-pSPYCE was constructed using T 4 DNA ligase. The recombinant vectors were confirmed by sequencing. Each experiment was performed in triplicate. The positive controls included the combination of bZIP63YN and bZIP63YC, and the negative controls were the YNE-empty and YCE-empty vectors. In addition, each partner and empty vector was used as controls.
The B. rapa protoplasts were prepared and cultured as described elsewhere 62 . The fresh leaves were obtained from Chinese cabbage plants at the four-leaf stage. The reagents used in the assays were 1.5% cellulase R10; 0.4% mecerozyme R10; 0.4 M D-mannitol; 20 mM KCl; 20 mM MES (pH 5.7); 10 mM CaCl 2 ; 0.1% BSA; 5 mM β-mercaptoethanol, and the assays were conducted as indicated in the protocol 62 . The recombinant vectors [eIF(iso)4E-pSPYNE, TuMV VPg-pSPYCE, bZIP63YN, bZIP63YC, YNE-empty and YCE-empty, constructed in BiFC assays] were transfected into the protoplasts, the plant cells were cultured in the dark, and the fluorescence signals were assessed using a laser confocal scanning microscope after 24-h. Equipment and settings. The three-dimensional (3D) structural of BraA.eIF(iso)4E.c-1 protein was modeled by Phyre2 63 , and the model was displayed by Swiss-PdbViewer 64 .