The eukaryotic translation initiation factor 4E confers multiallelic recessive Bymovirus resistance in Hordeum vulgare (L.)

Authors

  • Nils Stein,

    1. Institute of Plant Genetics and Crop Plant Research (IPK), Corrensstr. 3, D-06466 Gatersleben, Germany,
    Search for more papers by this author
    • N.S. and D.P. contributed equally to this work.

  • Dragan Perovic,

    1. Institute of Plant Genetics and Crop Plant Research (IPK), Corrensstr. 3, D-06466 Gatersleben, Germany,
    Search for more papers by this author
    • N.S. and D.P. contributed equally to this work.

  • Jochen Kumlehn,

    1. Institute of Plant Genetics and Crop Plant Research (IPK), Corrensstr. 3, D-06466 Gatersleben, Germany,
    Search for more papers by this author
  • Bettina Pellio,

    1. Institute of Crop Science and Plant Breeding I, Justus-Liebig-University, Heinrich-Buff-Ring 26-32, D-35392 Gießen, Germany, and
    Search for more papers by this author
  • Silke Stracke,

    1. Institute of Plant Genetics and Crop Plant Research (IPK), Corrensstr. 3, D-06466 Gatersleben, Germany,
    Search for more papers by this author
  • Stefan Streng,

    1. Institute of Plant Genetics and Crop Plant Research (IPK), Corrensstr. 3, D-06466 Gatersleben, Germany,
    Search for more papers by this author
  • Frank Ordon,

    1. Institute of Crop Science and Plant Breeding I, Justus-Liebig-University, Heinrich-Buff-Ring 26-32, D-35392 Gießen, Germany, and
    2. Institute of Epidemiology and Resistance, Federal Center for Breeding Research on Cultivated Plants, Theodor Roemer Weg 4, D-06449 Aschersleben, Germany
    Search for more papers by this author
  • Andreas Graner

    Corresponding author
    1. Institute of Plant Genetics and Crop Plant Research (IPK), Corrensstr. 3, D-06466 Gatersleben, Germany,
    Search for more papers by this author

(fax +49 39482 5155; e-mail graner@ipk-gatersleben.de).

Summary

Virus diseases are widespread threats for crop production, which can, in many cases, be controlled efficiently by exploiting naturally occurring resistance. Barley, an important cereal species of the Triticeae, carries two genes, rym4 and rym5, which are located in the telomeric region of chromosome 3HL and confer recessive resistance to various strains of the Barley yellow mosaic virus complex. The barley ‘eukaryotic translation initiation factor 4E’ (Hv-eIF4E) was identified as a candidate for resistance gene function by physical mapping on a 650 kb contig. It is located in a chromosomal region characterized by suppressed recombination, in a position collinear to its homologue on rice chromosome 1L. Sequence diversity in the coding region of Hv-eIF4E, as calculated from a collection of unrelated barley accessions, revealed non-silent single nucleotide polymorphisms (SNPs) in four of its five exons. Stable transformation of a resistant barley genotype with a genomic fragment or a full-length cDNA of Hv-eIF4E derived from susceptible cultivars induced susceptibility to Barley mild mosaic virus. Moreover, the identification of SNPs diagnostic for rym4 and rym5 provides evidence that these are two alleles, which confer different resistance specificities. These findings demonstrate that variants of Hv-eIF4E confer multiallelic recessive virus resistance in a monocot species. The identification of eIF4E as the causal host factor for bymovirus resistance illustrates that mutations in this basic component of the eukaryotic translation complex form a seminal mechanism for recessive virus resistance in both dicot and monocot plants.

Introduction

In large parts of Europe and East Asia winter barley production is threatened by the occurrence of Barley yellow mosaic and Barley mild mosaic virus (BaYMV, BaMMV), which can cause yield losses up to 50%. Under field conditions the virus is transmitted via the soil-borne Plasmodiophorid Polymyxa graminis (for review see Kanyuka et al., 2003). Because of the ubiquitous occurrence of P. graminis down to a soil depth of 60 cm, chemical plant protection is neither economic nor practicable and the main control of the disease relies on breeding of resistant cultivars. Up to now, eight independent genetic loci distributed over the barley genome have been identified that confer mostly recessive resistance to either one or several strains of this virus complex (summarized in Werner et al., 2003). In European cultivars resistance is based on two recessive genes (rym4, rym5). The gene rym4 confers resistance to BaMMV and BaYMV while rym5 additionally confers resistance to another strain of BaYMV, BaYMV-2. Both genes reside at high genetic resolution within the same marker interval on barley chromosome 3HL (Pellio et al., 2005) indicating that rym4 and rym5 may represent two very closely linked genes or alleles of the same gene. The mechanism of resistance conferred by rym4/rym5 is unknown. Under field conditions, infection occurs via the root system, however, the genes are also effective after mechanical inoculation of virus to leaves (Ordon and Friedt, 1993). In any case, rym4/rym5-based resistance results in the absence of virus particles at the sensitivity of enzyme-linked immunosorbent assays (ELISA).

Recently, a number of recessive potyvirus resistance genes have been isolated from dicotyledonous species identifying the ‘eukaryotic initiation factor 4E (eIF4E)’ or its isoform eIF(iso)4E as the host determinant. Interaction between either of the above two proteins with the ‘viral genome-linked protein (VPg)’ could be of general importance in recessive potyvirus resistance in plants (Diaz-Pendon et al., 2004). As Bymoviruses belong to the family Potyviridae and possess a VPg-gene it has been speculated that the homologous genes in barley might play a role in recessive resistance to BaMMV and BaYMV (Kanyuka et al., 2003).

To investigate the principle of the rym4/5-based recessive Bymovirus resistance in barley map-based isolation of the resistance locus was initiated based on two high resolution maps and a marker located 0.034 cM proximal to the rym5 gene (Pellio et al., 2005). In this study we describe cloning and structural characterization of the rym4 resistance gene from barley, the functional proof via complementation analysis and the determination of the allelic diversity present at this locus. Our results demonstrate that eIF4E is able to confer multiallelic, and multispecific virus resistance in a monocotyledonous plant species.

Results

Chromosome walking at the Rym4/5 locus

As a starting point for the construction of a BAC contig covering the Rym4/5 region the closely linked EcoRI/MseI AFLP fragment E31M41 (Pellio et al., 2005) was selected. Subsequently, eight steps of chromosome walking were performed to construct a 650 kb physical contig at the Rym4/5 locus (Figure 1, Table S2). About two-third of the contig co-segregated with the resistance gene and no recombination event distally to Rym4/5 was observed in the two high-resolution populations (Figure 1).

Figure 1.

Chromosome walking at the Rym4/5 locus.
A BAC contig of the Rym4/5 region on barley chromosome 3HL was established from a proximal flanking marker (E31M41) and was genetically anchored to the AxW (rym5) mapping population (resolution = 0.017 cM).
(a) The number of recombination events between markers is indicated.
(b) High resolution genetic map of the interval MWG838-Bmac29.
(c) Schematic representation of the physical map of the Rym4/5 locus by overlapping BAC clones.
(d) Physical map deduced from the BAC contig giving the relative positions of the mapped markers as determined based on fingerprinting analysis. * = marker produced a presence (W)/absence (I, F, A) polymorphism, + = marker produced a presence (I, F, A)/absence (W) polymorphism, $ = STS-marker not mapped but used for screening the BAC library and extending the contig.

Genetic mapping of BAC-derived markers was mainly restricted to the rym5 high-resolution mapping population (AxW), as most markers were non-polymorphic between the parents of the rym4 population (IxF). Both populations differed significantly in the overall detected sequence polymorphism as deduced from almost 20 kb of sequence information obtained from comparative sequencing of 18 independent STS marker loci: only two SNPs were determined between Igri and Franka as compared with 97 SNPs and 26 insertion/deletion polymorphisms between Alraune and W122.37-1. However, this difference in polymorphism had no substantial effect on recombination frequencies observed in the MWG838-Bmac29 interval in both populations (2.1 and 1.5 cM, respectively).

The complete contig spans eight recombination events between marker No51 and the gene rym5 corresponding to 0.13 cM (Figure 1). Recombination frequency was unevenly distributed. In the proximal quarter, between markers No51 and B451U, the ratio of physical to genetic distances was in the range between 0.8 and 2.3 Mb cM−1, whereas it increased to over 30 Mb cM−1 for the remaining part of the contig (Figure 1).

Identification of a candidate gene from the Rym4/5 contig

The BAC clones 778N10, 519J04, 801A11, 204C04, 793G02, 700P01, and 199E23 covered the part of the contig co-segregating with the resistance locus. Therefore, all possible genes present in this region could serve at first approximation as candidate genes for virus resistance. Six BACs 778N10-700P01 were sequenced to full-length (GenBank AY661558) and a detailed annotation of this 439 kb of contiguous sequence is described elsewhere (Wicker et al., 2005). Only two open reading frames (ORFs) were detected, located on a 10 kb stretch of BAC 519J04 and both represented intact genes oriented tail to tail on opposite strands at a distance of 1048 bp between the two STOP-codons. The first gene consisted of five exons with a total length of 645 bp (Figure 2a,b). Its entire coding sequence (CDS) exhibited strong nucleotide sequence identity (BLASTN identities = 93%, E-value ≤0.0) to wheat eIF4E, and therefore, was named eIF4E of Hordeum vulgare (Hv-eIF4E). Comparisons of the predicted barley protein to homologous proteins from other plant species revealed 93 and 77% amino acid (AA) identity over its entire length with wheat and rice, and 72% AA identity with Arabidopsis between AA 40 and 214. The second gene consisted of eight exons with a total length of 543 bp. On the protein level it was 81% identical to a predicted rice protein (MCT-1-like protein) and the gene was therefore named ‘Hordeum vulgare MCT-1-like protein-like’ (Hv-MLL). The exon/intron structure of the two genes was supported by EST sequences (tentative unigene consensi TC122413 and TC122721) available from ‘The Institute for Genome Research (TIGR)’ barley gene index (http://www.tigr.org/tigr-scripts/tgi/T_index.cgi?species=barley, release 8.0, January 9, 2004). Nineteen and thirteen cDNA clones represented Hv-eIF4E and Hv-MLL out of 341 924 public barley ESTs, respectively. Both genes (marker No519 is part of the gene Hv-MLL) co-segregated with the Rym4/5 locus (Figure 1b). Interestingly, only Hv-eIF4E was in collinear position between barley and rice since Os-eIF4E was present as CDS 32 or CDS 1 of the overlapping rice BACs AP003448 and AP003277 (Figure 3). Regarding Hv-MLL, the homologous rice sequence (BLASTN identities = 83%, E-value ≤4.4 × e−42) was detected on rice chromosome 1S (BAC AP003722) and thus was in a non-collinear position in both species. In rice, the micro-collinearity was further disturbed by inversions of gene order proximally and distally to Hv-eIF4E, including at least seven predicted CDS of rice BAC P0518C01 being involved (Figure 3). The following analyses were focused on the characterization of Hv-eIF4E, as homologous genes had been shown to be involved in recessive Potyvirus resistance in dicotyledonous plant species.

Figure 2.

Sequence analysis of the barley gene Hv-eIF4E.
(a) The coding sequence of the gene has a length of 645 bp. Exons are indicated by alternating presence/absence of underlining. Grey boxes indicate positions of single nucleotide polymorphisms (SNP) between independent genotypes.
(b) Graphical representation of the intron/exon structure of the gene. An arrow indicates the start codon.
(c) Allelic sequencing of all exons of the gene in a collection of 56 barley cultivars revealed distinct haplotypes for genotypes either carrying rym4 or rym5 or for susceptible genotypes, respectively.
(d) SNP-haplotypes were translated to highlight the amino acid (AA) exchanges in the deduced protein sequence.

Figure 3.

The gene eIF4E is present at orthologous loci in barley and rice.
Comparative mapping of two single copy rice cDNA-probes RZ783 and C112 (NCBI AA231661 and C97914) and indirect mapping of rice marker E60152 (NCBI AU101489) with the orthologous barley EST-based STS marker GBR1425 (EMBL CA018793) led to identification of the orthologous region on rice chromosome 1L represented by two overlapping BAC/PAC clones (NCBI AP003448 and AP003277, corresponding to 270 kb). The gene eIF4E was present at collinear position whereas the second gene MLL of the Rym4/5 contig was not present in this interval. Microcollinearity was disturbed between barley and rice represented by two inversions – one proximal and the other distal to the resistance locus flanked by markers GBR1425/GBR1834 and GBS1019/GBSW1, respectively. (*The physical distance in rice between CDSs P0518C01.10 and P0518C01.12 equals 12 kb.)

Hv-eIF4E exhibits sequence polymorphisms associated with rym4/5-based resistance

To ascertain whether allelic differences in the CDS of Hv-eIF4E could provide the basis for resistance to BaMMV and BaYMV comparative sequencing of the gene was performed in a set of 56 barley accessions. These included (i) European and Asian cultivars containing rym4, (ii) European and Asian cultivars and landraces containing rym5, (iii) resistant cultivars and genotypes without information on allelism to rym4/5, and (iv) susceptible cultivars (Table S3). Overall, nine single nucleotide polymorphisms (SNPs), organized in seven different haplotypes, were detected in four of the five exons (Figure 2a,c) and all SNPs led to AA changes in the deduced protein sequence (Figure 2d). However, not all of them change the physicochemical properties of the involved AA (Table 1). The three-dimensional (3D) structure was modeled to ascertain the potential position of these AA in the putative Hv-eIF4E protein. All observed AA changes appeared on the surface of the protein in close proximity to the cap-binding domain (Figure 4).

Table 1.  Amino acid changes in the gene Hv-eIF4E of rym4, rym5 or susceptible barley cultivars
Bp positionbAA positioncAA changesa
rym4rym5Other
  1. aThe reference sequence originated from the cultivar Morex (susceptible to BaMMV, BaYMV).

  2. bThe total length of the CDS is 645 bp.

  3. cThe length of the protein is 215 AA.

  4. dAA changes in European cultivars containing the rym4 specificity.

  5. eAA changes in Asian cultivars containing the rym4 specificity.

  6. fAA changes present only in European cultivars containing the rym4 specificity.

17057Ser (hydrophilic) → Phe (hydrophobic)  
353118Lys (polar, basic) → Thr (hydrophilic)d
Lys (polar, basic) → Ile (hydrophobic)e
  
359120 Thr (hydrophilic) → Ser (hydrophilic) 
478160 Asn (uncharged) → Asp (polar, acidic) 
481161 Gln (uncharged) → Lys (polar, basic) 
614205Ser (hydrophilic) → Phe (hydrophobic)f  
617206Asp (polar, acidic) → Gly (neutral)  
622208  Gly (neutral) → Ala (neutral)
623   Gly (neutral) → Ser (hydrophilic)
Figure 4.

Three-dimensional model of barley eIF4E.
Simulation of the putative 3D surface structure of Hv-eIF4E was achieved based on sequence homology to mouse eIF4E (PDB code: 1ej1A GI = 7546551). (a) A view on the cap-binding domain of the protein (cap-interacting residues are highlighted in green color, Marcotrigiano et al., 1997) is given. AA-residues involved in polymorphism of rym4 (Ser57, Lys118, Ser205, Asp206) and rym5 (Thr120, Asn160, Gln161) genotypes are indicated in red and blue color, respectively. Gly208, which is affected by SNPs in bp-position 622 and 623 of susceptible cultivars is highlighted in yellow color. All AA polymorphisms are located exclusively in the neighborhood of the cap-binding domain. No AA change affected the dorsal side of the protein (b).

European and Asian cultivars carrying the rym4 allele differed from the other accessions analyzed in bp-positions 170, 353, and 617 of the CDS with positions 170 and 617 being diagnostic for all analyzed rym4 accessions (Figure 2c). Interestingly, the SNP detected at bp-position 353 was tri-allelic with two different alleles characteristic for accessions from Europe and Asia. A fourth SNP exclusively restricted to European rym4-cultivars was detected at bp-position 614. Three bp-changes at positions 359, 478, and 481 were determined for genotypes containing rym5 (Figure 2c; Table S3). While the rym5 resistance present in all but one of these accessions can be traced back to the Chinese landrace Mokusekko 3 (Graner et al., 1999b; Konishi et al., 1997) an unrelated rym5 containing Chinese landrace (Hsingwuke 2) shared the identical haplotype. The allele of the SNP at bp-position 481 of confirmed rym5 genotypes was shared by another resistant cultivar (Yane Hadaka 44). Here, resistance was found to be allelic to rym4 regarding resistance to BaMMV (Götz and Friedt, 1993) but allelism with respect to BaYMV still remains to be determined. It cannot be excluded that Yane Hadaka 44 harbors a further functional allele of the Rym4/5 locus. Based on these findings, only the SNPs at bp-position 359 and 478 can be regarded as rym5 diagnostic SNPs. Additional polymorphisms observed at bp-positions 622 and 623 were independent of the status of resistance/susceptibility to BaMMV/BaYMV.

Complementation with Hv-eIF4E induces susceptibility in rym4 plants

In order to induce susceptibility, both a full-length cDNA and a gDNA construct of Hv-eIF4E from susceptible genotypes were transformed separately into rym4 (resistant) plants. Five to 21 progeny plants of 27 randomly selected T1 families (two obtained with the cDNA, 25 with the gDNA construct) were analyzed for resistance to BaMMV. In total 433 plants were tested. Twenty-one families segregated either in a 3:1 or 15:1 ratio (χ2 < 0.05) for susceptibility/resistance or exhibited complete susceptibility. These ratios indicate the presence of one, two or, in the latter case, even more independent transgene loci. Completely susceptible families may also be due to homozygosity of the transgene at the T0 plant level, which is not unexpected, as haploid microspores were transformed.

Four individuals of each T1-family were selected for genotyping of the transgene. These were either the first four plants from non-segregating families or two resistant and two susceptible individuals from segregating families (Table 2). In 17 of the 27 analyzed T1 families presence of the transgene coincided completely with susceptibility to BaMMV. Expression of the transgene was determined by RT-PCR in the four genotyped individuals of four (nos 12, 41, 77 and 103) of these 17 families and was completely correlated to susceptibility (Figure 5a,b).

Table 2.  Summary of complementation analysis of T1 progenya
 T1 familiesControls
cDNA(2)b gDNA (25)bIgri (Rym4)NIL4.1-6 (rym4)
ResSusResSusResSusResSus
  1. aOn average 16 plants of 27 T1 families were analyzed by DAS-ELISA for determining presence/absence of virus infection; results are based on four plants per T1 family selected for genotyping.

  2. bThe number of analyzed families per construct (cDNA, gDNA) is given in brackets.

  3. cTG = presence of the transgene, WT = absence of the transgene.

  4. dTransgenic plants showing a resistant phenotype can be explained by escapes from the artificial infection procedure, post-transcriptional gene silencing (PTGS), post-translational effects or insertions of truncated transgenes.

  5. eResistant Igri indicate escapes from artificial infection.

TGc619d49    
WTc2304e2711
Figure 5.

Transgenic expression of Hv-eIF4E induces susceptibility to BaMMV.
T1 families were analyzed for presence/absence as well as expression of the transgene by PCR and RT-PCR, respectively. Plants carrying (a) the cDNA construct of Hv-eIF4E from the susceptible cultivar Barke can be distinguished from the wild-type (WT) rym4 by PCR due to the presence of an intron-2-less extra amplicon (marked by an asterisk), whereas (b) presence of the gDNA transcript from the susceptible cultivar Morex is determined after CAPS analysis of the amplicons with RsaI (arrowhead). Expression of the transgene was always determined by RsaI-CAPS analysis of the RT-PCR amplicons (double arrowheads). Transgene expression leads to susceptibility to BaMMV with either of the two constructs. Absence of wild-type amplicon in some susceptible transgenic progeny is likely due to transgene overexpression since the transgene is under control of the constitutive Ubiquitin promoter.

Two (family nos 33 and 52) of the remaining 10 families, which stayed either completely resistant in presence of the transgene or which included resistant transgenic plants, represented truncated insertions of the transgene. Transgenic members of these two families were lacking the 3′-part of the transgene as well as the selection marker gene hpt (Hygromycin phosphotransferase), which is supposed to integrate after the transgene of interest due to its position in the binary vector. The lack of transgene expression was furthermore confirmed by RT-PCR in family no 33 (data not shown). Three additional of the ambiguous 10 families were analyzed for transgene expression: no expression could be observed in family no 55, family no 38 showed none or only weak transgene expression and in family no 63 the transgene was expressed in all individuals tested (data not shown). Interestingly, the latter family segregated for susceptibility, compared to the other two, which were uniformly resistant. Thus, in the initial two cases silencing of the transgene is the most likely explanation for maintained resistance, whereas in family no 63, the observed cases of resistance may be either due to escapes from artificial infection or to post-translational modifications.

Altogether, the analysis of transgenic T1 families provides a clear proof for the function of Hv-eIF4E in rym4-mediated resistance in barley.

Discussion

Hv-eIF4E mediates rym4/5-based recessive Bymovirus resistance in barley

The ‘eukaryotic translation initiation factor 4E’ of barley (Hv-eIF4E) has been identified as a candidate gene for rym4-mediated Bymovirus resistance by a classical attempt of positional cloning. Transformation of the allele from a susceptible genotype into resistant genetic background provided the functional proof for the role of Hv-eIF4E in the establishment of resistance/susceptibility to BaMMV in barley. Furthermore, mutations in eIF4E confer recessive resistance to plant viruses belonging to the Potyvirideae in monocotyledonous and dicotyledonous plant species and therefore represent a seminal mechanism in the establishment of recessive resistance.

In planta, the translation initiation factor 4E (cap-binding protein) and its isoform eIF(iso)4E, bind with different affinities to mono- or hyper-methylated m7G-caps of mature eukaryotic mRNAs. This process triggers the assembly of the whole translation initiation complex and finally leads to the formation of the functional ribosome and to translation (for review see Gingras et al., 1999). Plant viral mRNAs are translated by the host translation system and compete with host mRNA for the involved components like the ‘cap-binding proteins’. Although potyviral RNAs lack an m7G cap, they encode a ‘virus genome-linked protein’ (VPg) with binding affinity to eIF4E, which attaches to the cap-free 5′ end of the viral mRNA. VPg likely mediates the interaction with eIF4E and thus with the host translation machinery or alternatively with the translocation machinery. A model that was supported by direct interaction between VPg of Turnip mosaic potyvirus with Arabidopsis eIF4E and eIF(iso)4E (Leonard et al., 2000; Wittmann et al., 1997) or eIF4E from Brassica perviridis (Leonard et al., 2004), and between VPg of Tobacco etch virus and tobacco eIF4E (Schaad et al., 2000). On the other hand, direct interaction between pea eIF4E and VPg of Pea seedborne mosaic virus could not be demonstrated (Gao et al., 2004). Mutations in the VPg of Pea seed borne potyvirus and of Potato virus Y allow overcoming recessive resistance in pea, pepper and potato, respectively (Borgstrom and Johansen, 2001; Keller et al., 1998; Moury et al., 2004). In analogy, mutations in the VPg of BaMMV and BaYMV-1 were correlated to breaking of rym4/5-mediated resistance in barley (Kanyuka et al., 2004; Kuhne et al., 2003). Hence, VPg provides the virulence determinant in several plant–Potyvirus/Bymovirus interactions. Together with our results showing induced susceptibility upon stable transformation with an Hv-eIF4E allele from susceptible cultivars these data suggest that a direct interaction between VPg of BaMMV and eIF4E maybe required for establishing susceptibility in barley. However, an experimental proof for this hypothesis by either yeast-two-hybrid analysis, in vivo co-localization or ELISA-based protein–protein interaction assays still has to be shown for the barley interaction with BaMMV/BaYMV.

Also in pepper, lettuce and pea AA changes in the eIF4E protein caused recessive resistance. Alignment of the conserved part of the deduced protein sequence revealed that the rym4-specific change of Ser-57-Phe in Hv-eIF4E was located only 2 and 3 AA downstream of the Ala-70-Pro mutation in lettuce (Nicaise et al., 2003), the Val-67-Glu in pepper (Ruffel et al., 2002) and three to seven AA upstream of polymorphisms in Ala73, Ala74 and Ser78 in pea (Gao et al., 2004) (Figure S1) that are likely to be involved in causing resistance and tolerance, respectively. Furthermore, all AA variability found in genotypes carrying rym4 or rym5 are in proximity of the cap-binding domain as was determined by modeling of the Hv-eIF4E 3D-structure using mouse eIF4E as a model (Marcotrigiano et al., 1997).

The putative mechanism of interaction between VPg and eIF4E involved in conferring susceptibility is unclear. It either may allow the virus to utilize directly the host cap-binding complex to achieve translation of its own genome. Consequently, mutations in Hv-eIF4E changing the properties of the interacting domain would interfere with plant–virus interaction eventually leading to resistance. Alternatively eIF4E, via binding to eif4G, may be involved in virus transport. This was postulated by Lellis et al. (2002) and was recently supported by the functional analysis of the gene sbm1 in pea. Here mutations in eIF4E exhibited impaired cell-to-cell movement of virus particles (Gao et al., 2004). Moreover, free VPg could block eIF4E for host translation by occupying the cap-binding pocket at higher affinity than m7G caps of plant mRNA. This would resemble a sequestration of eIF4E by 4E-binding proteins (4E-BP), which bind to the eIF4G binding domain of eIF4E and thus interfere with cap-binding complex assembly as reported for animal/Picornavirus systems (for review Gale et al., 2000). However, in the latter case, mutations in eIF4E leading to resistance would follow a dominant or semi-dominant inheritance, as upon infection of a heterozygote the mutant eIF4E still would sustain host translation and thus prevent infection. This is in conflict with the recessive mode of inheritance of rym4-mediated resistance as well as with the results of our complementation experiments where transgenic plants possess both an endogenous resistance allele (present in the near isogenic line) as well as the susceptibility allele (present as the transgene) and showed full susceptibility. We assume VPg/eIF4E interaction to be more likely required for virus translation/replication or cell-to-cell trafficking.

Rym4/5-mediated resistance is based on multiple alleles of Hv-eIF4E

Here we provide evidence that recessive mutations in the gene Hv-eIF4E confer rym4-mediated resistance to BaMMV. It is very likely that the same allele also confers resistance to BaYMV-1 as no intralocus recombination leading to altered resistance specificities has been observed (Graner et al., 1999b; Pellio et al., 2005). Moreover, the presence of rym5-specific sequence polymorphisms indicates that resistance to the more virulent strain BaYMV-2 is conferred by a second allele of Hv-eIF4E. Diversity in eIF4E conferring multiallelic recessive virus resistance to plant viruses seems to be a widespread mechanism, as different pathogen specificities were recognized also by independent alleles of eIF4E at the Pvr2, Mo1 and Sbm1 loci in pepper, lettuce and pea (Gao et al., 2004; Nicaise et al., 2003; Ruffel et al., 2002). Thus screening for natural diversity of Hv-eIF4E could provide an option for the discovery of new resistance alleles for breeding programs.

Keeping in mind the lack of recombination in the rym4/5 region, rym5-associated SNPs still might represent the result of linkage disequilibrium (LD) of Hv-eIF4E with a physically closely linked gene conferring rym5-specific resistance. This possibility, however, seems to be unlikely, since (i) rym4- and rym5-specific SNPs in Hv-eIF4E have been confirmed for barley accessions not related by descent, (ii) rym4 and rym5 were shown to be allelic at least with respect to BaMMV (Götz and Friedt, 1993) and (iii), genetic analysis did not reveal any evidence for the presence of additional genes at the Rym4/5 locus (Pellio et al., 2005). The observation that the rym4 and rym5 alleles are characterized each by a different set of mutations suggests an independent origin of the two alleles.

Hv-eIF4E is present at orthologous positions in the barley and rice genomes

The current study revealed strong macro-collinearity of the Rym4/5 locus of barley to a telomeric region of rice chromosome 1L but confirmed also previous observations about disturbed micro-collinearity between rice and other grass genomes (Bennetzen and Ramakrishna, 2002). Interestingly, the resistance gene Hv-eIF4E, which fulfills a basic function in planta, is present at orthologous loci in both genomes. It seems possible that collinearity between cereal genomes will be found more frequently at loci carrying genes fulfilling basic non-species specific functions as was also shown for the genes Vrn1 (Yan et al., 2003) and ror2 (Collins et al., 2003) which are at collinear positions between wheat, barley, and rice. Hence, the rice genome information might be further exploited for the isolation of additional recessive Bymovirus resistance genes. In this regard, a homologue of eIF(iso)4E, which confers susceptibility to Potyviruses in Arabidopsis (Duprat et al., 2002; Lellis et al., 2002) is located on rice chromosome 10 (data not shown) and thus may represent a potential ortholog of the barley gene rym7 located on the syntenic barley chromosome 1H (Graner et al., 1999a). Other components of the translation initiation complex like eIF4G for which a mutant (cum2) is known to interfere with virus multiplication in Arabidopsis (Yoshii et al., 2004) may also represent candidate genes for Bymovirus resistance in barley. Therefore, mapping of homologues to eIF4G and further components of the barley translation initiation complex may represent an efficient strategy for the identification of further resistance genes in barley.

The Rym4/5 locus is located in a cold spot of recombination

The ratio of genetic versus physical distance changed considerably within the established contig. This is a well-known phenomenon in cereal genomes. Almost 75% of the contig co-segregated with Rym4/5 in nearly 8000 meiotic events (Pellio et al., 2005) resulting in a ratio of around 30 Mb cM−1 distal to the gene. The reasons for a low frequency of recombination distal to Rym4/5 remain to be determined. It has been shown that recombination occurs preferentially in or nearby genes and may be missing from regions void of genes but consisting of complex arrangements of repetitive DNA (Schnable et al., 1998). This is in accordance with our data from sequencing of the Rym4 region, which provided no evidence for the presence of additional non-transposon-based genes on the established contig (Wicker et al., 2005). On the other hand, reduced recombination could be due to a high level of sequence diversity at orthologous loci preventing chromosome pairing during meiosis. However, the parental genotypes of the rym4 mapping population revealed only 2% of polymorphisms detected for those of the rym5 population without a significant difference in recombination frequencies in both populations. Therefore, the most likely explanation for the lack of recombination distal to the gene may be simply the lack of genes in the characterized region.

The presented data reveals that variants in the gene Hv-eIF4E provide a seminal mechanism for recessive virus resistance in a broad range of plant species. The identification of two alleles showing different resistance specificities opens new perspectives for the exploitation of natural diversity present in ex situ or in situ collections to identify further alleles with broad-spectrum resistance to Bymoviruses. Moreover, our findings tempt us to speculate that variants of other genes of the translation machinery represent the functional basis of other recessive resistance loci in the barley genome.

Experimental procedures

Plant material

Genetic mapping was performed in either one of the two high-resolution populations Igri (Rym4) × Franka (rym4) (45 recombinant inbred lines, RIL, out of 1040 F2 plants, subsequently referred to as I × F) and Alraune (Rym5) × W122.37-1 (rym5) (85 RILs out of 2846 F2 plants, subsequently referred to as A × W) (Pellio et al., 2005). Markers that were non-polymorphic in the rym4/5 populations were mapped in the ‘Oregon Wolfe Barley’ population (Costa et al., 2001). Disomic and ditelosomic wheat–barley addition lines (Islam et al., 1981) were utilized for chromosomal assignments.

Resistance tests

Resistance tests were carried out in a growth chamber (16°C day/12°C night) on T1-progeny according to Friedt (1983). The presence of virus particles was assessed by DAS-ELISA according to Clark and Adams (1977) using specific antiserum against BaMMV, kindly provided by Dr F. Rabenstein, Federal Centre for Breeding Research on Cultivated Plants (BAZ), Aschersleben, Germany. Plant samples leading to absorption of below 0.1 at 405 nm were scored as resistant.

Marker analysis and genetic mapping

Procedures for genomic DNA isolation, Southern hybridization, and radioactive labeling were essentially performed as described earlier (Graner et al., 1991). Post-hybridization washes of rice cDNA-based markers (kindly provided by T. Sasaki, NIAS, Tsukuba, Japan) were performed at reduced stringency (Perovic et al., 2004). Triticeae ESTs for rice/barley synteny-based marker saturation were selected as previously described (Perovic et al., 2004). STS-PCR primers for bacterial artificial chromosome (BAC)-clone or synteny-based markers (summarized in Table S1) were designed with Primer3 software (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi). For conversion of RFLP-based markers (designated as GBR) into PCR-based CAPS markers the SNP2CAPS software (Thiel et al., 2004) was utilized. Linkage analysis was performed using mapmaker (Lander et al., 1987), mapmanager (Manly, 1993).

Chromosome walking

Hybridization-based screening of the Morex BAC library (Yu et al., 2000) was performed according to Woo et al. (1994) except for higher stringency hybridization and washes at 70°C for repetitive probes. DNA pools of the same library were employed for PCR-screening (GeneAMP Thermocycler System 9700; Applied Biosystems, Darmstadt, Germany; cycling conditions: 96°C, 5 min initial denaturation, followed by 50 cycles 96°C/30 sec, primer-specific annealing temperature/30 sec, 72°C/30 sec). Eighty-two super-pools comprising 10 individual 384-multiwell plates each (total library = 816 plates) provided the template for the initial round of PCR followed by screening of the individual positive DNA plate-pools. Positive 384-well multiplates were subsequently screened by colony PCR. The insert size of BAC clones was determined by pulsed-field gel electrophoresis (PFGE; CHEF-DRII; Bio-Rad, Munich, Germany) after digestion with NotI at pulse time ramping from 5 to 15 sec, temperature 14°C, 6 V cm−1, 14 h, on 1% agarose gels in 0.5 x TBE buffer. Sequence information for the development of sequence tagged site (STS) markers has been generated either by direct BAC end sequencing (Stein et al., 2000) or after (i) BAC end rescue (Ripoll et al., 2000), (ii) random and selected subcloning of restriction fragments (Brueggeman et al., 2002), (iii) subcloning of unique restriction fragments from overlapping BACs, or (iv) low-pass sequencing of random subclones of BAC inserts (Stein et al., 2000). Overlapping BACs were identified either by Southern hybridization to fingerprint blots obtained after single or double restriction of BAC DNA or by comparative sequencing of amplicons from the target sequence of the corresponding BACs.

Sequence analysis of the candidate gene

All five exons of the gene Hv-eIF4E were amplified (primers exon1: 317s = GTTTGCGTCCGCTCCTCCATCTTCCT, 1170as = CTGCCCGCCTAGATCTATCAACAACC; exon2/3: BF2 = ACTGCGGATCTAGATATGGC, F1R = GAATACATACCTGAGCAGTTTC; exon4/5: 4857s = ACCTGGGTAAATGCTATCACG, 5396as = TTATGGAGTACAACGACGACAAACAC), purified and subjected to cycle sequencing with the DYEnamic ET Dye terminator chemistry (Amersham Bioscience, Freiburg, Germany) capillary automatic sequencing device. Sequence handling and analysis was performed with Lasergene v5.0 (DNASTAR Inc., Madison, WI, USA) or Sequencher v3.0 (Gene Codes Corporation, Ann Arbor, MI, USA) software. Modeling of the putative 3D structure of Hv-eIF4E was performed as comparative analysis to the crystal structure of mouse eIF4E (PDB code = 1ej1A) on the SWISS-MODEL server (Schwede et al., 2003, http://swissmodel.expasy.org//SWISS-MODEL.html). The predicted structure was visualized with the software PyMol 0.97 in ‘surface’ mode (DeLano, 2004).

Agrobacterium-mediated transformation of barley

A 5.6 kb NheI/EcoRI genomic fragment from the susceptible cultivar Morex, originating from BAC 519J04 and containing the entire Hv-eIF4E coding region, 63 bp upstream and 1028 bp downstream sequence, was inserted into the multiple cloning site of the vector pUbi-ABM (DNA Cloning Service, Hamburg, Germany) downstream of the maize ubiquitin promoter (including the 5′ untranslated region and the first intron) to generate the plasmid pUbi_eIF4E_bac. The resulting expression cassette was transferred as an SfiI-fragment into the respective restriction site of the binary vector p6U (DNA Cloning Service) to generate p6U_Ubi_eIF4E_bac. A further plasmid p6U_Ubi_eIF4E_cds, which contains a 645 bp full-length cDNA from the susceptible cultivar Barke, with 62 bp upstream and 33 bp downstream sequence was generated by an analogous cloning strategy. Isolation and culture of immature pollen, Agrobacterium-mediated transformation of pollen cultures as well as generation of transgenic barley plants was conducted as described by Kumlehn et al. (J. Kumlehn, L. Serazetdinova, G. Hensel, D. Becker and H. Loerz, IPK Gatersleben, unpublished data) on a near isogenic line (NIL4.1-6) carrying the Rym4-locus of the resistant cultivar Franka in a genetic background of the susceptible cultivar Igri.

Molecular characterization of transgenic plants

Presence of the transgene was determined in T0- and T1-transgenic plants by Southern hybridization and/or PCR analysis (primers exon2/3: E2F = CTAGCAAGTTGAATGTTGGAGC, F2R = CTGGTTCTTACGCACGCTGACG and exon4/5: 4857s/5396as). Genomic DNA (gDNA) derived amplicons originating from the cDNA transgene were discriminated directly from the wild-type amplicon through a size polymorphism due to the absence of intron 2 of Hv-eIF4E (product size of 220 bp instead of 308 bp, respectively) in the transgene. The genomic transgene fragment and the endogene were differentiated in transgenic plants by CAPS analysis using RsaI, which cleaves the endogene (resistance allele) due to a rym4-specific SNP in exon 2 at bp-position 353 of the coding sequence. Expression of the transgene was monitored by RT-PCR on cDNA (iSript cDNA Synthesis Kit; Bio-Rad) obtained from DNaseI-treated (Fermentas St. Leon-Rot, Germany) total RNA (Purescript RNA Isolation Kit; Gentra Systems, MN, USA) followed by RsaI CAPS analysis.

Acknowledgements

We are grateful to Dr A. Habekuss for performing resistance tests, and J. Perovic, U. Krajewski and S. Stegmann for excellent technical assistance. The work was supported by grants of the ‘Deutsche Forschungsgemeinschaft (DFG)’ to A. G. (Gr 1317/3-2) and F. O. (Or 72/2-2), and the ‘Bundesministerium für Bildung und Forschung (BMBF)’ to A. G. (GABI-0312280A).

Supplementary Material

The following material is available from http://www.blackwellpublishing.com/products/journals/suppmat/TPJ/TPJ2424/TPJ2424sm.htm

Table S1 Primer-information for STS markers derived from the rym4/5 contig

Table S2 Overview on the individual steps of chromosome walking

Table S3 SNP-haplotypes of the exons of the gene Hv-eIF4E

Figure S1 Sequence alignment of eIF4E from barley (Hv), pepper (Ca), pea (Ps), lettuce (Ls) and mouse (Mm).

The software ClustalW 1.8 (Thompson et al., 1994) was used to align the deduced protein sequences (Genbank ID or PDB code are given in the sequence identifier). ○ = AA involved in cap-binding in mouse eIF4E as deduced from the co-crystal structure with 7-methyl-GDP (Marcotrigiano et al., 1997), bsl00079 = polymorphic AA residues in rym4 and rym5 genotypes of barley (this work), • = polymorphic AA residues in pvr21 and pvr22 genotypes of pepper (Ruffel et al., 2002), ◆ = polymorphic AA residues in sbm1 and sbm11 genotypes of pea (Gao et al., 2004), bsl00066 = polymorphic AA residues in mo11 and mo12 genotypes of lettuce (Nicaise et al., 2003).

Ancillary