Multiple copies of eukaryotic translation initiation factors in Brassica rapa facilitate redundancy, enabling diversification through variation in splicing and broad-spectrum virus resistance

Authors


Summary

Recessive strain-specific resistance to a number of plant viruses in the Potyvirus genus has been found to be based on mutations in the eukaryotic translation initiation factor 4E (eIF4E) and its isoform, eIF(iso)4E. We identified three copies of eIF(iso)4E in a number of Brassica rapa lines. Here we report broad-spectrum resistance to the potyvirus Turnip mosaic virus (TuMV) due to a natural mechanism based on the mis-splicing of the eIF(iso)4E allele in some TuMV-resistant B. rapa var. pekinensis lines. Of the splice variants, the most common results in a stop codon in intron 1 and a much truncated, non-functional protein. The existence of multiple copies has enabled redundancy in the host plant's translational machinery, resulting in diversification and emergence of the resistance. Deployment of the resistance is complicated by the presence of multiple copies of the gene. Our data suggest that in the B. rapa subspecies trilocularis, TuMV appears to be able to use copies of eIF(iso)4E at two loci. Transformation of different copies of eIF(iso)4E from a resistant B. rapa line into an eIF(iso)4E knockout line of Arabidopsis thaliana proved misleading because it showed that, when expressed ectopically, TuMV could use multiple copies which was not the case in the resistant B. rapa line. The inability of TuMV to access multiple copies of eIF(iso)4E in B. rapa and the broad spectrum of the resistance suggest it may be durable.

Introduction

Recruitment of the eukaryotic translation machinery to the 5′ end of mRNA is a crucial regulatory step in initiation of cap-dependent translation. Investigation of recessive resistance to plant viruses identified the involvement of a group of proteins involved in mRNA translation, particularly eukaryotic translation initiation factor 4E (eIF4E), its isoform eIF(iso)4E (Robaglia and Caranta, 2006) and to a lesser extent eIF4G (Le Gall et al., 2011). This was particularly true for recessive resistance to members of the Potyviridae, the largest group of plant viruses. The potyvirus ‘viral protein genome-linked’ (VPg) and Arabidopsis thaliana eIF(iso)4E have been shown to interact in yeast two-hybrid binding assays (Wittmann et al., 1997). Studies on pea eIF4E supported the view that the binding site for the VPg of a potyvirus overlapped with the mRNA cap-binding site (Ashby et al., 2011). Caliciviruses also possess a VPg and some have been shown to utilise eIF4E to translate their genome in mammals (Goodfellow and Roberts, 2008). Mutations in eIF4E in a range of plant species have been shown to confer resistance to a range of potyviruses (Robaglia and Caranta, 2006), and knocking out eIF(iso)4E in Arabidopsis resulted in resistance to the potyviruses Turnip mosaic virus (TuMV) and Lettuce mosaic virus (Duprat et al., 2002; Lellis et al., 2002).

The TuMV-resistant RLR22 Brassica rapa var. pekinensis (Chinese cabbage), derived from an accession identified in a screen of more than 3000 lines (Liu et al., 1996) has broad-spectrum TuMV resistance (Walsh et al., 2002). Artificial (mechanical) inoculation of RLR22 using leaf sap from TuMV-infected plants resulted in chlorotic spots in inoculated leaves, with no detectable systemic spread of the virus (Rusholme et al., 2007; Figure S1 in Supporting Information). Following natural aphid TuMV challenge, no symptoms were seen and no virus was detected in any leaves (Rusholme et al., 2007), indicating that RLR22 plants would be completely resistant (immune) in the field. Segregation following a cross between RLR22 and the TuMV-susceptible line R-o-18 of the closely related B. rapa ssp. trilocularis (Roxb.) Hanelt. (yellow sarson; Figure S1), revealed that the resistance was due to a recessive gene, retr01 (coincident with the copy of eIF(iso)4E on chromosome A4) that was epistatic to a dominant gene, ConTR01 (coincident with one of the other copies of eIF(iso)4E, or one of the copies of eIF4E, both on chromosome A8; Rusholme et al., 2007). Crosses between a further TuMV-resistant Chinese cabbage line (BP8407) derived from the screened accessions (Liu et al., 1996) and a susceptible Chinese cabbage line, revealed that this resistance was due to a single recessive gene, retr02, that mapped to the same locus as retr01 (Qian et al., 2013).

Results

Three copies of eIF4E and of eIF(iso)4E identified in RLR22 B. rapa

Three copies of eIF4E and three copies of eIF(iso)4E were identified in a genomic library of the B. rapa line R-o-18 (Jenner et al., 2010) and a genomic library of RLR22. This was consistent with genomic analyses of Brassica diploid species, which indicated that they evolved from genome triplication of an ancestor with a genome similar to A. thaliana (Town et al., 2006). Sequencing of the genes and comparison of the RLR22 sequence with a sequence-based genetic map of B. rapa (Wang et al., 2011) confirmed the location of retr01 (BraA.eIF(iso)4E.a) on chromosome A4 in B. rapa. Molecular markers located retr02 on a scaffold also from chromosome A4 (Qian et al., 2013). The location of the ConTR01 candidates BraA.eIF4E.c and BraA.eIF(iso)4E.c from RLR22 (Rusholme et al., 2007) was confirmed on chromosome A8, again by comparison with the sequence-based genetic map (Wang et al., 2011).

retr01 is mis-spliced in resistant plant lines

To identify the mechanism of broad-spectrum resistance to TuMV in B. rapa, several plant lines were studied. BraA.eIF(iso)4E.a was sequenced from four lines known to have broad-spectrum resistance, or derived from lines with broad-spectrum resistance to TuMV [RLR22 (Walsh et al., 2002; possessing retr01 and ConTR01; Rusholme et al., 2007), BP058 (Walsh et al., 2002), Jong Bai No. 2 (Hughes et al., 2002)], BP8407 (possessing retr02; Qian et al., 2013) and four lines known to be susceptible to all TuMV isolates tested [R-o-18 (Rusholme et al., 2007), Ji Zao Chun (Qian et al., 2013), CK 2 and CK 1 (not susceptible to UK 1 TuMV)]. The genomic sequence of BraA.eIF(iso)4E.a on chromosome A4 of RLR22 and BP8407 was identical, showing that retr02 is retr01, rather than a different allele. All the TuMV-resistant lines were found to have an extra G (indel), adjacent to the splice site of BraA.eIF(iso)4E.a exon 1 and intron 1 (at position 201 nucleotides from the ATG), relative to alleles in all susceptible lines.

To study the expression of BraA.eIF(iso)4E.a, reverse-transcriptase (RT) PCR was carried out on all the lines and the products were sequenced. Plants to be genotyped were challenged with TuMV isolate CDN 1, or TuMV-C4, to verify resistance/susceptibility. All TuMV-susceptible lines (lacking the indel) produced a single product approximately 600 nucleotides (nt) in size (Figure 1a), corresponding to the correctly spliced version of the gene (Figure 1b). All the lines possessing the indel were resistant to TuMV and had a larger major product of approximately 664 nt, plus a smaller minor product of approximately 600 nt (Figure 1a). The larger product retained the extra G and the whole of intron 1 (Figure 1c) and the smaller product was of a similar size to the correctly spliced version. Sequencing of RT-PCR products of BraA.eIF(iso)4E.a showed that introns 2–4 were correctly spliced in all resistant and susceptible plants. The retention of intron 1 by all lines possessing the indel resulted in the introduction of a premature stop codon in intron 1 at the 234 nt position (Figure 1c). Cloning RT-PCR products from RLR22 revealed a further four less common variants. These included a variant possessing the last 15 nt of intron 1, resulting in a slightly elongated, in-frame mRNA sequence (Figure 1d), one with intron 1 excised along with the last 3 nt of exon 1, resulting in a slightly truncated, in-frame mRNA with a substitution (Figure 1e), one with an extra G at the end of exon 1, resulting in a premature stop codon (Figure S2b), and one that was correctly spliced (Figure 1b). There was a clear association between lines possessing the indel, resulting in mis-splicing of BraA.eIF(iso)4E.a (retr01), and broad-spectrum resistance.

Figure 1.

BraA.eIF(iso)4E.a is mis-spliced in resistant Brassica rapa and in transgenic Arabidopsis thaliana. (a) Detection of a single transcript in susceptible plants, corresponding to correctly spliced BraA.eIF(iso)4E.a, splice variants in resistant B. rapa plants and splice variants from an Arabidopsis Col-0::dSpm plant complemented with the RLR22 allele of BraA.eIF(iso)4E.a, separated using an Agilent 2100 Bioanalyser. (b) Correctly spliced BraA.eIF(iso)4E.a detected in susceptible plants (a). (c) The most common mis-spliced variant retained the extra G and the whole of intron 1 resulting in a premature stop codon at the 234 nucleotide (nt) position, detected in resistant lines and Arabidopsis Col-0::dSpm complemented by the RLR22 allele of BraA.eIF(iso)4E.a. (d) Variant retaining the last 15 nt of intron 1 (in-frame), detected in RLR22 and Col-0::dSpm complemented by BraA.eIF(iso)4E.a (a). (e) Variant lacking the last 3 nt of exon 1 (in-frame) and with a substitution, detected in RLR22 and some F2 plants from the RLR22, CK 1 cross.

The most common mis-spliced variant of BraA.eIF(iso)4E.a is non-functional for TuMV

To establish whether the most common splice variant of BraA.eIF(iso)4E.a (intron 1 retained, Figure 1c) could be functional for TuMV, cDNA of the genes from the TuMV-susceptible line Ji Zao Chun and TuMV-resistant line BP8407 were expressed in Escherichia coli Rosetta(DE3)pLysS with a 6× Histidine tag. The deduced sizes of the Ji Zao Chun and BP8407 proteins (minus the 6× Histidine tag, approximately 18 kDa) appeared to be consistent with the predicted sizes (22.50 and 8.67 kDa, respectively; Figure 2a), confirming that the mis-splicing resulted in a truncated protein. Yeast two-hybrid assays confirmed the physical interaction between the VPg of TuMV-C4 and the BraA.eIF(iso)4E.a protein from the susceptible Ji Zao Chun line and the lack of interaction with truncated BraA.eIF(iso)4E.a proteins [BraA.eIF(iso)4E.a mRNA with an extra G at the end of exon 1 (Figure S2b), or intron 1 retained (Figure 1c)] from BP8407 (Figure 2b and Table S1), indicating that these truncated proteins are unlikely to be functional for the virus. Expression of the intron 1-retained construct in the yeast was confirmed by RT-PCR.

Figure 2.

The most common mis-spliced variant of BraA.eIF(iso)4E.a produced a truncated eIF(iso)4E protein that did not interact with Turnip mosaic virus (TuMV) ‘viral protein genome-linked’ (VPg) and two less common splice variants did interact with the VPg of TuMV-C4, in yeast two-hybrid assays. (a) Brassica rapa BP8407 (intron 1 retained) and Ji Zao Chun BraA.eIF(iso)4E.a recombinant proteins with a 6× histidine tag expressed in transgenic Escherichia coli Rosetta(DE3)pLysS after induction with isopropyl β-d-1-thiogalactopyranoside. BraA.eIF(iso)4E.a from the resistant BP8407 was truncated compared with that from the susceptible Ji Zao Chun and their deduced molecular weights were consistent with the polypeptide sequences predicted from the mRNAs. pET-32a empty vector. (b) Yeast two-hybrid assay (i) TuMV-C4 VPg + RLR22 BraA.eIF(iso)4E.a (from a splice variant lacking last 3 nucleotide (nt) of exon 1, resulting in loss of an amino acid and a substitution) from a F2 plant from the RLR22, CK 1 cross, (ii) TuMV-C4 VPg + RLR22 BraA.eIF(iso)4E.a (from splice variant retaining last 15 nt of intron 1, resulting in an extra five amino acids) from Arabidopsis thaliana Col-0::dSpm transformed with the RLR22 allele of BraA.eIF(iso)4E.a, (iii) TuMV-C4 VPg + BP8407 BraA.eIF(iso)4E.a (from splice variant retaining whole of intron 1, resulting in a stop codon in intron 1), (iv) TuMV-C4 VPg + BP8407 BraA.eIF(iso)4E.a (from splice variant retaining an extra G at the end of exon 1, resulting in a stop codon in exon 2), (v) TuMV-C4 VPg + Ji Zao Chun BraA.eIF(iso)4E.a, (vi) TuMV-C4 VPg + Arabidopsis eIF(iso)4E (positive control), (vii) TuMV-C4 VPg + empty prey, (viii) Ji Zao Chun + empty bait.

BraA.eIF(iso)4E.c is the only candidate for ConTR01

To identify the second gene (ConTR01) involved in the TuMV resistance in the progeny of the cross between RLR22 and R-o-18 (Rusholme et al., 2007), following genotyping of B1S1 families, a B1S2 population homozygous for retr01 (RLR22 allele of BraA.eIF(iso)4E.a), but segregating for BraA.eIF(iso)4E.c and homozygous for the RLR22 allele of BraA.eIF4E.c (the ConTR01 candidates; Rusholme et al., 2007), was produced. This family segregated for resistance and susceptibility. At the BraA.eIF(iso)4E.c locus, all plants homozygous for the RLR22 allele were resistant, whereas the heterozygotes and the plants homozygous for the R-o-18 allele segregated for resistance. All plants from a family segregating for BraA.eIF4E.c (homozygous for the RLR22 alleles of BraA.eIF(iso)4E.a and BraA.eIF(iso)4E.c) were resistant to TuMV. These results ruled out BraA.eIF4E.c as a candidate for ConTR01, leaving BraA.eIF(iso)4E.c as the only candidate. Sequencing BraA.eIF(iso)4E.c from R-o-18 and RLR22 showed four amino acid differences (L36F, V52A, T80I and Q150P). Attempts to detect an interaction between the R-o-18 or RLR22 alleles of BraA.eIF(iso)4E.c and the VPg of TuMV-C4 in yeast two-hybrid experiments were unsuccessful (Table S1). So it is not clear whether any of the amino acid differences affect interaction with the VPg.

Whole transcriptome sequencing of R-o-18 and RLR22 plants revealed qualitative and quantitative differences in BraA.eIF(iso)4E.a expression between the two lines. It was correctly spliced in R-o-18 (Figure 1b), whereas in RLR22, only the variant retaining intron 1 (Figure 1c) was detected. Expression in RLR22 was significantly lower than in R-o-18. There were no significant differences between the levels of expression of BraA.eIF(iso)4E.c and BraA.eIF4E.c between the two lines (Table 1).

Table 1. Expression levels of copies of eIF4E and eIF(iso)4E from Turnip mosaic virus (TuMV)-susceptible Brassica rapa ssp. trilocularis R-o-18 and TuMV-resistant B. rapa ssp. pekinensis RLR22 plants measured in fragments per kilobase of exon per million fragments mapped (FPKM) from transcriptome analysis. The data show that all genes except BraA.eIF(iso)4E.b were expressed and the only gene where there was a significant difference in expression between the two lines was BraA.eIF(iso)4E.a
Plant lineGene
BraA.eIF(iso)4E.a BraA.eIF(iso)4E.b BraA.eIF(iso)4E.c BraA.eIF4E.a BraA.eIF4E.c
  1. BraA.eIF4E.b is a pseudogene lacking exons 2 and 3 (Jenner et al., 2010).

  2. a

    Statistically significant ( 0.005).

R-o-187.70.0510.53.713.5
RLR222.00.0712.62.911.8
Q-value0.0047a10.78100.80400.8573

In crosses between Chinese cabbage, resistance is inherited as a single recessive gene

As the susceptible parent in the original cross, R-o-18, was a different subspecies from RLR22 (Chinese cabbage), the genetic inheritance of resistance was also investigated in crosses with Chinese cabbage lines. Segregation ratios of genotypes in F2 generations from crosses between Ji Zao Chun and BP8407 (Qian et al., 2013) and between RLR22 and two Chinese cabbage lines (CK 1 and CK 2) were consistent with those predicted for a single recessive gene (χ2 = 2.15, 2.89 and 4.40, respectively, all > 0.05). CK 1 and CK 2 crosses demonstrated that most F2 plants that were homozygous for retr01 were resistant following artificial (mechanical) inoculation; however, a small number of plants showed an occasional chlorotic spot on uninoculated leaves. Reverse transcriptase PCR of the latter plants detected mis-spliced variants of BraA.eIF(iso)4E.a, including one with the last 3 nt of exon 1 missing (Figure 1e), resulting in the loss of one amino acid and a substitution. This mis-spliced variant interacted with the VPg of TuMV-C4 in yeast two-hybrid assays (Figure 2b), indicating that it could have been functional for TuMV.

As the sequence of BraA.eIF(iso)4E.c from the CK 1, CK 2, BP8407 and Ji Zao Chun lines was identical to that in RLR22, all the susceptible Chinese cabbage parents already had the ConTR01 candidate, explaining why resistance segregated monogenically in these crosses. This result, along with those for the segregation of resistance in the offspring of the RLR22 R-o-18 cross showed that the presence of three copies of eIF(iso)4E can lead to different segregation patterns in offspring, depending upon which copies of eIF(iso)4E from the parental lines can be used and accessed by TuMV. Interestingly, from the CK 1 and CK 2 crosses with RLR22, F2 plants heterozygous at the retr01 locus were less susceptible to TuMV than those plants that were homozygous for the allele from the susceptible parent (Figure 3a). Enzyme linked immunosorbent assay (ELISA) confirmed that the heterozygotes [mean A405 from ELISA for CK 1 offspring 0.81 (±0.074) and for CK 2 offspring 0.28 (±0.04)] accumulated significantly less virus than homozygotes [mean A405 for CK 1 offspring 2.09 (±0.132) and for CK 2 offspring 1.47 (±0.159), both < 0.001].

Figure 3.

Susceptibility of Brassica rapa plants to Turnip mosaic virus (TuMV) CDN 1 and Arabidopsis thaliana Col-0::dSpm plants complemented by copies of BraA.eIF(iso)4E and BraA.eIF4E from two RLR22 loci of each gene. (a) F2 plants derived from a cross between the Chinese cabbage lines RLR22 and CK 1, a plant heterozygous for BraA.eIF(iso)4E.a with limited systemic infection (left), a plant homozygous for the CK 1 copy of BraA.eIF(iso)4E.a with severe systemic infection (middle) and a plant homozygous for the RLR22 copy of BraA.eIF(iso)4E.a, resistant to TuMV (right). (b) Arabidopsis Col-0 transformed with copies of eIF4E and eIF(iso)4E from B. rapa RLR22, following artificial (mechanical) inoculation with TuMV, 20 days post-inoculation. (c) Complementation of Arabidopsis Col-0::dSpm plants with copies of eIF4E and eIF(iso)4E from B. rapa RLR22, following artificial (mechanical) inoculation with TuMV, 20 days post-inoculation. The dSpm transposon insertion is in At5 g35620, inactivating At.eIF(iso)4E.

Ectopic expression of copies of BraA.eIF4E and BraA.eIF(iso)4E from RLR22 B. rapa complemented the TuMV-resistant Arabidopsis eIF(iso)4E knockout line Col-0::dSpm

We investigated the ability of TuMV to use different copies of RLR22 eIF4E and eIF(iso)4E by transforming BraA.eIF4E.a, BraA.eIF4E.c, BraA.eIF(iso)4E.a and BraA.eIF(iso)4E.c into the Arabidopsis Col-0::dSpm line possessing a transposon knock-out of eIF(iso)4E that conferred resistance to TuMV (Duprat et al., 2002). Reverse transcriptase PCR confirmed expression of all these genes in the transgenic Col-0 and Col-0::dSpm lines. Artificial (mechanical; Figure 3b) and aphid challenge of both transformed Col-0 controls and Col-0::dSpm showed that all four copies of eIF4E/eIF(iso)4E from RLR22 complemented the eIF(iso)4E knockout (Figure 3c), indicating that TuMV could use all copies investigated. The ectopic expression was misleading in that it indicated that these genes from R-o-18 (Jenner et al., 2010) and RLR22 were functional for TuMV. Segregation of the phenotypes and genotypes in B1S1 (Rusholme et al., 2007) and B1S2 plants from the cross between these two lines clearly demonstrated that either these genes were not available or were not functional for the virus in RLR22. The segregation also demonstrated that the two copies of eIF4E were either not available or were not functional for the virus in R-o-18. The RT-PCR and transcriptome analyses of transformed Arabidopsis plants did not reveal any correctly spliced copies of RLR22 BraA.eIF(iso)4E.a, but did detect eight mis-spliced variants (Figures 1 and S2). Of these, the one possessing the last 15 nt of intron 1 (Figure 1d; in-frame, with an additional five amino acids), could potentially be functional for TuMV. Yeast two-hybrid assays confirmed that this mis-spliced variant interacted with the VPg of TuMV-C4 (Figure 2b and Table S1), and this may explain why the knockout line was complemented by the RLR22 allele of BraA.eIF(iso)4E.a.

Discussion

Amino acid sequences of some mis-spliced variants of BraA.eIF(iso)4E.a suggested they could be functional. Comparing the sequences of these with those of correctly spliced BraA.eIF(iso)4E.a and eIF4E and eIF(iso)4E in a number of plant species (Monzingo et al., 2007; German-Retana et al., 2008), indicated that a number of variants had the correct amino acids at important sites (Figure S3). This suggests that some of the mis-spliced variants could be functional for TuMV. The ability of two of these mis-spliced variants to bind the TuMV VPg in yeast two-hybrid assays further supports this possibility. The existence and low frequency of these potentially functional variants, combined with artificial inoculation of large amounts of virus, might explain why unexpected limited infection was seen in some plants. For example, the limited infection of RLR22-inoculated leaves (which did not spread systemically) and the occasional very limited systemic infection of plants homozygous for the RLR22 allele of BraA.eIF(iso)4E.a derived from the crosses we made.

Our confirmation of three copies of eIF(iso)4E in RLR22 revealed how it was possible for the copy that TuMV would normally use in Chinese cabbage to be non-functional for both plant and virus, without apparently disadvantaging the plant. The durability of the resistance will be dependent upon the virus not mutating to be able to utilise/access other copies in planta. It may also be possible that TuMV could evolve to be capable of effective cap-independent translation (Basso et al., 1994). The situation could be further complicated by the variability of the mis-splicing of eIF(iso)4E resulting in sufficient quantities of splice variants that could be functional for the virus. The difficulty we experienced in detecting correctly spliced variants in RLR22 and variants with minor amino acid changes indicates that this is highly unlikely. Evidence to date suggests that the VPg protein of potyviruses competes with host plant mRNA cap for eIF4E binding (Gao et al., 2004) and eIF(iso)4E binding (Plante et al., 2004). The fact that it was possible to induce limited infection in inoculated leaves (which was not able to spread systemically) only by artificially inoculating excessive amounts of TuMV into resistant plants (homozygous for retr01), indicates that there was very little eIF(iso)4E protein present that was functional for the virus, and/or the virus was less efficient than mRNA cap in competing for the protein. As plants homozygous for the ConTR01 candidate that were heterozygous for retr01 were clearly less susceptible to TuMV than plants homozygous for the retr01 allele from susceptible plants, the amount of eIF(iso)4E protein in these brassica plants must be limiting for virus replication.

In this study we have demonstrated a mechanism of translation factor-based resistance, which, unlike previous examples that are strain-specific, is broad spectrum. Mis-splicing has also been induced artificially by targeting induced local lesions in genomes (TILLING) of tomato (Piron et al., 2010). This resulted in resistance to the potyviruses Pepper mottle virus (PepMoV) and strain-specific resistance to Potato virus Y (PVY) through a substitution in the splice site of eIF4E (G1485A), causing the deletion of exons 2 and 3 in the mRNA (Piron et al., 2010). At least one PVY isolate was able to overcome the resistance (Piron et al., 2010). It was suggested that the resistance-breaking PVY isolate was able to use a different copy of eIF4E. We have shown that TuMV can use at least two copies of eIF4E and two of eIF(iso)4E from both susceptible (Jenner et al., 2010) and resistant plants, when expressed ectopically in Arabidopsis and that all these copies are expressed in B. rapa RLR22. This suggests that copies of eIF4E and eIF(iso)4E other than BraA.eIF(iso)4E.a are inaccessible to TuMV in RLR22, possibly because of where in the cell they are expressed. This, along with the inability of a wide range of virus isolates from around the world, representing different genotypes, pathotypes and serotypes to overcome the resistance in RLR22 (Walsh et al., 2002), indicates that the broad-spectrum resistance mechanism could provide durable potyvirus resistance in a range of plant species. The presence of multiple copies of eukaryotic translation initiation factors in B. rapa has facilitated redundancy. The redundancy arising from the mis-splicing has enabled diversification and resulted in the plant being able to evade pathogen infection.

Experimental Procedures

Plant materials and virus isolates

Brassica rapa RLR22, R-o-18 (Rusholme et al., 2007), BP058 (Walsh et al., 2002), Jong Bai No. 2 (Hughes et al., 2002), BP8407, Ji Zao Chun (Qian et al., 2013) and inbred Syngenta lines CK 1 and CK 2 plants were grown in insect-free glasshouses at 18°C. Arabidopsis thaliana Col-0::dSpm (Duprat et al., 2002) and Col-0 plants were grown in a growth room at 20°C with a 9/15-h day/night light cycle. The TuMV isolates used in this study were CDN 1 (Jenner and Walsh, 1996) and TuMV-C4 (Qian et al., 2013).

Plant inoculation assays

Brassica and Arabidopsis plants were artificially (mechanically) inoculated with TuMV isolates or healthy plant sap, assessed and tested for the presence and quantity of TuMV by ELISA (Rusholme et al., 2007; Jenner et al., 2010). Aphid transmission of TuMV to Arabidopsis was performed using Myzus persicae aphids (Jenner et al., 2010).

Construction of a genomic library, identification and cloning of copies of eIF4E and eIF(iso)4E

A genomic library was prepared from the B. rapa line RLR22 with broad-spectrum resistance to TuMV by Warwick Plant Genomic Libraries Limited using the pCC1FOS fosmid vector obtained under licence from Epicentre Technologies (http://www.epibio.com/). Copies of eIF4E and eIF(iso)4E were identified in the same manner as copies were identified in R-o-18 (Jenner et al., 2010) and sequenced.

Expression of BraA.eIF(iso)4E.a

The RNA was extracted from brassica or A. thaliana leaves using the Ambion RNAqueous kit (Life Technologies, http://www.lifetechnologies.com/). Reverse-transcription reactions were carried out with 1 μg of total RNA for 15 min at 70°C using either CN4 antisense primer 5′-AGAAAGCTGGGTTCAGACAGTGAACCTAGTTCTTC-3′ (including an attB site, underlined) for R-o-18, or CN5 antisense primer 5′-AGAAAGCTGGGTTCAGACAGTGAACCGAGTTCTTC-3′ (including an attB site underlined) for Chinese cabbage lines, except Ji Zao Chun. For PCR, 4 μl of the reverse-transcription reaction was used as a template in 50-μl reactions, with 5 U of Taq polymerase (Invitrogen, http://www.invitrogen.com/) per reaction. The PCR was run for 30 cycles of 30 sec at 95°C, 30 sec at 58°C and 80 sec at 72°C. We used the following primers: CN3 sense 5′-AAAAAGCAGGCTCGATGGCGACAGAGGATG-3′ (including an attB site underlined) and either CN4 or CN5 for amplifying BraA.eIF(iso)4E.a from all B. rapa lines except BP8407 and Ji Zao Chun where Bio11535F sense 5′-ATGGCGACAGAGGATGT-3′ and Bio11536R antisense 5′-TCAGACAGTGAACCGA-3′ were used. The RT-PCR products were separated by agarose gel electrophoresis and detected by staining with GelRed (Bioline, http://www.bioline.com/), or GoldView (SBS Genetech Ltd, http://www.sbsbio.com/). Products were also ran on an Agilent 2100 Bioanalyser (Agilent Technologies, http://www.home.agilent.com/) to view the different sized products. Bands were excised from gels and DNA purified using Gel Extraction Kits (Qiagen, http://www.qiagen.com/). The products were cloned for sequencing using the TOPO TA Cloning Kit for Sequencing (Invitrogen).

Protein analysis

BraA.eIF(iso)4E.a was amplified from BP8407 and Ji Zao Chun by first-strand cDNA synthesis with a poly dT primer using a Prime Script® RT-PCR kit (TaKaRa, http://www.takara-bio.com/) according to the manufacturer's instructions. Gene-specific PCR was carried out using Bio11537 sense 5′-GGAATTCCATGGCGACAGAGGATGTG-3′ (EcoRI site underlined) and Bio11538 antisense 5′-CCGCTCGAGTCAGACAGTGAACCGAG-3′ (XhoI site underlined) primers. The amplified fragments were then digested and cloned into pET-32a (Novagen, http://www.novagen.com/) with a 6× histidine tag. The constructs were transformed into E. coli Rosetta(DE3)pLysS. Expression of recombinant proteins was induced at 20°C and 28°C for 20 h by the addition of 1 mmol isopropyl β-d-1-thiogalactopyranoside. Molecular weights were determined using SDS-PAGE (Murphy and Kyle, 1994).

Yeast two-hybrid assays

Protein interactions were tested using the Matchmaker GAL 4 yeast two-hybrid system (Clontech). BraA.eIF(iso)4E.a was amplified from BP8407 and Ji Zao Chun RNA by RT-PCR with Bio11538 antisense primer followed by PCR with Bio11537 and Bio11538 primers. The VPg was amplified by RT-PCR from the RNA of brassica plants infected by TuMV-C4, with Bio120214 antisense primer 5′-CCCCGGGTCACTCGTGGTCCACTGGGA-3′ (XmaI site underlined), followed by PCR with Bio120213 sense 5′-CCCATATGATGGCGAAAGGTAAGAGGC-3′ (NdeI site underlined) and Bio120214 antisense primers. The TuMV VPg cDNA was cloned into the pGBKT7 plasmid by the NdeI (5′-end) and XmaI (3′-end) sites to generate a bait plasmid. BraA.eIF(iso)4E.a cDNAs were cloned into the pGADT7 plasmid by the EcoRI (5′-end) and XhoI (3′-end) sites to generate prey plasmids. Bait and prey constructs were transformed into the yeast strain AH109; the reporter genes were HIS3 and ADE2. Empty vectors pGADT7 and pGBKT7 were used as negative controls along with the manufacturer's positive control.

Genotyping of B. rapa plants at BraA.eIF(iso)4E.a, BraA.eIF(iso)4E.c and BraA.eIF4E.c loci, production of B1S2 lines and analysis of virus susceptibility

We extracted DNA from leaves of B1S1 B. rapa plants using the DNeasy kit (Qiagen). The primers BR2 sense 5′-TCTCCTCCACTTCTTCCCAATAC-3′ and BR14 antisense 5′-TAGACAAGGCTTGGCTTGAAACTG-3′ were used to genotype relevant eIF(iso)4E copies from RLR22 and R-o-18, giving different sized products for BraA.eIF(iso)4E.a (larger for R-o-18) and BraA.eIF(iso)4E.c (larger for R-o-18). To genotype BraA.eIF4E.c, primers CN55 sense 5′-TCTTTGTTGGTGGGTTAGATTCCG-3′ and CN56 antisense 5′-ATCAACGCAAGCAACTACATCGAG-3′ amplified the R-o-18 allele and primers CN44 sense 5′-TTTCTTGTTGGGTTAAGTGAAG-3′ and CN45 antisense 5′-CAAGCAACTACATGGAAAAAAC-3′ amplified the RLR22 allele. Two plants from B1S1 seed (Rusholme et al., 2007), one homozygous for the RLR22 alleles of BraA.eIF(iso)4E.a and BraA.eIF4E.c but heterozygous for BraA.eIF(iso)4E.c and one homozygous for the RLR22 alleles of BraA.eIF(iso)4E.a and BraA.eIF(iso)4E.c but heterozygous for BraA.eIF4E.c, were identified. These were then vernalised, grown on to flower and selfed to produce B1S2 seed, segregating for the respective genes. A single nucleotide polymorphism marker designed by Syngenta was used for genotyping F2 populations from the RLR22 and CK 1 and CK 2 crosses at the BraA.eIF(iso)4E.a. locus. The ELISA absorbance values of plants homozygous for the R-o-18 allele of BraA.eIF(iso)4E.c and heterozygous at this locus were analysed using Student's t-test.

Transcriptome analysis

The transcriptomic expression profile of an A. thaliana Col-0::dSpm plant complemented with BraA.eIF(iso)4E.a from RLR22 was determined using total RNA extracted from a young leaf and oligo(dT) selection performed twice using Dynal magnetic beads (Invitrogen). Illumina library preparation was performed using a mRNA-TruSeq sample prep kit version five (Illumina Inc., http://www.illumina.com/), according to the manufacturer's protocol (15018818 revA). The library was sequenced using Illumina's GAIIx sequencing system. Using the Illumina CASAVA pipeline, 70-bp end sequence reads were base-called and scored for read quality.

Sequences of mRNA from total RNA extracted from young leaves from each of three R-o-18 plants and three RLR22 plants were determined by SeqWright using an Illumina HiSeq 2000 system (Illumina Inc.). Ribosomal RNA depletion was performed using the RiboMinus Eukaryote Kit for RNA-Seq (Life Technologies, A10837-08) and libraries constructed using the TruSeq RNA Sample Preparation Kit (Illumina Inc.). The sequencing runs were paired-end, 100-bp reads and were analysed using Illumina CASAVA version 1.8.

Sequence reads were aligned to the published Arabidopsis (Lamesch et al., 2012) and B. rapa (Wang et al., 2011) genome assemblies using Tophat (Trapnell et al., 2009) and Bowtie (Langmead et al., 2009) algorithms, respectively. The cufflinks algorithm (Trapnell et al., 2010) was used to calculate and compare fragments per kilobase of exon per million mapped fragments (FPKM), to estimate relative transcript abundances.

Complementation of A. thaliana Col-0::dSpm

BraA.eIF(iso)4E.a, BraA.eIF(iso)4E.c, BraA.eIF4E.a and BraA.eIF4E.c were amplified from RLR22 genomic DNA, cloned and transformed into the A. thaliana line Col-0::dSpm possessing a transposon knock-out of eIF(iso)4E (Duprat et al., 2002) as described for R-o-18 copies of the genes (Jenner et al., 2010).

RLR22 sequences have been deposited in GenBank. Full-length genomic DNA of: BraA.eIF(iso)4E.a, JA722714; BraA.eIF(iso)4E.c, JA722768; BraA.eIF4E.a, JA722747; BraA.eIF4E.c, JA722756. Mis-spliced variants of BraA.eIF(iso)4E.a: with whole of intron 1 retained, cDNA sequence, JA722715; with last 15 nt of intron 1 retained, cDNA sequence, JA722717; with last 3 nt of exon 1 missing, cDNA sequence, JA722719. cDNA sequences of: BraA.eIF(iso)4E.c, JA722769; BraA.eIF4E.a, JA722748; BraA.eIF4E.c, JA722757.

Acknowledgements

We thank J. Bambridge, L. Ward, J. Selby and A. Baker for assistance, D. Lydiate and E. Higgins for providing sequence information and seed and C. Robaglia for Arabidopsis dSpm. This work was supported by a UK BBSRC Crop Science Initiative grant (BB/E00668X/1 to JAW and GCB), a BBSRC Industrial C.A.S.E. PhD studentship (BB/G017808/1 to JAW for CFN), Key Laboratory of Biology and Genetic Improvement of Horticultural Crops, Ministry of Agriculture, China grant (to RS), Chinese 973 Program grant (2012CB113906) (to RS) and Syngenta Seeds.

Accession Numbers

JA722714, JA722768, JA722747, JA722756, JA722715, JA722717, JA722719, JA722769, JA722748, JA722757.

Ancillary