The Eukaryotic Translation Initiation Factor 4E (eIF4E) and its isoform [eIF(iso)4E] are known as susceptibility factors for potyvirus infection in a variety of plant species such as melon, tomato and pepper (Wang and Krishnaswamy, 2012). The eIF4E protein interacts with the 5′ 7-methylguanosine (m⁷G) cap of eukaryotic mRNA to recruit the translation complex for protein biosynthesis. Potyviruses are single stranded (+)RNA viruses that mimic the structure of the eukaryotic mRNA’s m⁷G cap by their genome-linked viral protein (VPg) cap, thus taking advantage of host metabolism for translation of their RNA (Sanfaçon, 2015). The plasmodiophorid Polymyxa graminis serves as a vector of Barley yellow mosaic virus (BaYMV) and Barley mild mosaic virus (BaMMV), both of which are bymoviruses of the Potyvirideae family. P. graminis transmits these viruses mainly to winter barley seedlings, particularly under cool and moist weather conditions in autumn, which results in high yield losses and renders bymoviruses major pathogens in Europe and Asia. Barley was among the first species in which resistance-mediating alleles of the EIF4E gene were described and utilized in breeding (Kanyuka et al., 2005; Stein et al., 2005). The rym4 and rym5 alleles of HvEIF4E carry mutations that result in amino acid changes in the binding domain of the encoded protein. Consequently, the interaction between the host’s eIF4E and the viral RNA cap-like structure formed by the genome-linked viral protein (VPg), is hampered, thereby preventing the translation of viral RNA and thus the replication and spread of the virus in the plant (Li and Shirako, 2015). In total, seven allelic variants of the EIF4E gene have been shown to confer resistance to different isolates of the BaMMV/BaYMV complex [reviewed in Jiang et al. (2020)]. Based upon this principle, barley cultivars resistant to BaYMV and BaMMV have been bred and widely grown in temperate climate regions. However, the comprehensively used rym4- and rym5-based resistances have already been overcome by some strains such as BaYMV-2 (overcoming rym4) as well as BaMMV-Sil and BaMMV-Teik (overcoming rym5) (Habekuss et al., 2008; Jiang et al., 2020). Consequently, there is an urgent need for new resistance-conferring HvEIF4E alleles or novel resistance mechanisms, especially in winter barley breeding. Besides the EIF4E alleles rym4 and rym5, several other bymovirus resistance loci have been described in barley [reviewed in Jiang et al. (2020)]. Most of these are recessive and were found in Asian landraces. However, only for rym1/rym11, the responsible gene, i.e., the Protein Disulfide Isomerase-Like 5-1 (PDIL5-1), has been identified (Yang et al., 2014). Nevertheless, winter barley breeding still relies on rym4 and rym5, since the introduction of new resistance sources into current elite lines requires time-consuming backcrosses. Taking advantage of the CRISPR-associated (Cas) endonuclease technology (Koeppel et al., 2019), targeted knock-outs of EIF4E were reported to be associated with Potyvirus resistance in other plant species such as cucumber and cassava (Chandrasekaran et al., 2016; Gomez et al., 2019). In barley, however, only non-synonymous single nucleotide polymorphisms (SNPs) have been described that lead to amino acid changes in the mRNA cap-binding domain of eIF4E. Therefore, it is assumed that the loss-of-function of EIF4E might be lethal in barley even though HvEIF(iso)4E does exist as a conserved paralog (Yang et al., 2016). Here, we addressed four positions in the EIF4E gene previously described to carry different SNPs in the rym4 and rym5 resistance alleles of the virus-susceptible winter barley cv. “Igri” using Cas9 and accordingly customized guide RNAs (gRNAs). Via Agrobacterium-mediated DNA transfer to embryogenic pollen cultures, target motif-specific mutant plants were generated and their progeny were tested for resistance against BaMMV.

After cloning pInt_EIF4E_TM1, pInt_EIF4E_TM2, pInt_EIF4E_TM3, and pInt_EIF4E_TM4 bearing cas9 and target motif-specified gRNA expression units, a barley protoplast-based test assay was performed to pre-validate the construct performance in terms of mutagenesis efficiency. After transformation and incubation of protoplasts (see Figure 3A), their genomic DNA was extracted and PCR amplicons of target regions were deep-sequenced. For each target motif, several hundred reads were obtained. The alignment of the sequencing reads to the “Morex” reference sequence revealed that the proportions of mutated sequence reads related to the total read numbers (including wild-type) were as high as 17.1%, with the best results being achieved in target motif 1 (see Figure 1B). The most frequent mutations observed in protoplasts were small deletions of 1–5 base pairs or insertions of one base pair (A or T) at the expected Cas9 cleavage site residing three to four base pairs upstream of the protospacer-adjacent motif (PAM) (Jinek et al., 2012) as shown in Figure 3B. Unfortunately, only after sequencing, the PAM of target motif 4 was found to be unsuitable, because it features a single nucleotide polymorphism in “Igri” as compared to the barley genomic reference sequences of cv. “Morex”.

After cloning the binary vectors pBin_EIF4E_TM1, pBin_EIF4E_TM2, and pBin_EIF4E_TM3 with the expression cassettes for cas9 and the gRNAs against target motifs 1, 2, and 3, Agrobacterium-mediated DNA transfer to embryogenic pollen cultures of winter barley cv. “Igri” was used to produce transgenic plants. Addressing target motifs 1, 2, and 3, four, twenty-seven and fifty primary transgenic plants were obtained from one, five and one transformation experiments, respectively. All regenerated plants were PCR-positive for the cas9 gene and were subjected to Sanger sequencing of the addressed target motifs. Whereas independent 1-bp insertions of T or A were found three base pairs upstream of the PAM in each of the four plants carrying pBin_EIF4E_TM1 (see Figure 3C), no mutations whatsoever were detected in plants transformed using pBin_EIF4E_TM2 and pBin_EIF4E_TM3. The undisturbed chromatograms obtained for the four primary mutant plants were taken as indication for homozygosity of the induced mutations. Progenies of three independent primary mutants were genotyped by PCR screening for the cas9 transgene and by Sanger sequencing of the target region. All individuals carried the same mutations known from their respective parent, thereby confirming its homozygous state. In addition, these M2 plants were mechanically inoculated with BaMMV, whereby all of them proved resistant to the infection (see Figure 2C).

The insertion of one base pair at position 169 of the EIF4E coding sequence results in a shift of the reading frame during translation of the mRNA, which itself causes a non-sense amino acid sequence downstream of this insertion site including a premature stop codon at position 306 (after approx. one third) of the coding sequence (see Supplementary Material).

Yield was assessed based on thousand-grain weight and grain number per plant (GNP) of the three analyzed M2 families. The thousand-grain weight of the mutant families P1 (mean 46 g, ±3.5), P3 (47 g ± 3.1), and P4 (45.7 g, ±2.4) was on a par with that of non-mutated, transgenic segregants from a similar experiment (50.7 g, ±4.9) and of “Igri” wild-type (48.9 g, ±3.2) grown under the same glasshouse conditions (see Figure 4A; Supplementary Table S2). According to a one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test, no significant differences in the thousand-grain weight were observed (p ≥ 0.05). However, the total grain yield per plant was decreased owing to a reduction in the number of grains produced per plant. This phenomenon was consistent across all three analyzed mutant lines (P1: 72 grains, ±29; P3: 79 grains, ±25; P4: 72 grains, ±30) which were compared to wild-type (220 grains, ±59) and non-mutated segregants (198, ±37) (p < 0.001; see Figure 4B; Supplementary Table S2).

The pre-validation of cas9/gRNA constructs via protoplast assay proved conclusive for targeted mutagenesis at the whole-plant level. Notably, comparatively low mutation rates were expected, because targeting specifically those regions known for resistance-conferring mutations required trade-offs with respect to some efficiency-related criteria for the selection of target motifs (Kumlehn et al., 2018). Target motifs 2 and 3 showed zero to very low mutation rates in protoplasts and so they did in cas9/gRNA transgenic plants. Positional effects might have played a role, since both target motifs mutated with low efficiencies are located close to each other on exons 2 and 3. Furthermore, the protospacer-adjacent motif (PAM) of target motif 4 was found to be unsuitable only after sequencing, because it features a single nucleotide polymorphism in “Igri” as compared to the barley genomic reference sequences that were used for target motif selection (see also Supplementary Material). For target motif 1, the insertions found in transgenic plants were also amongst the most frequent mutations detected in the respective protoplast assay (see Figure 3B). The mutations proved instantly homozygous in all four primary transgenic plants, most likely due to the particularity of the applied transformation method which involves gene transfer to haploid cells followed by whole-genome duplication (Kumlehn et al., 2006; see Figures 2C, 3C). To the best of our knowledge, this represents the first example of generating homozygous mutations via transfer of cas9/gRNA-encoding DNA to haploid cells that are capable of giving rise to (doubled haploid) plants. This enables us to evaluate phenotypes of homozygous mutants already in early generations after transformation. As a trade-off, these plants are most likely also homozygous for the cas9/gRNA-transgene. However, this disadvantage is considered irrelevant in the scientific exemplification demonstrated in the present study, in which also a rather outdated barley cultivar was used to take advantage of its amenability to genetic transformation. Moreover, there is the option of crossing these mutants with wild-type plants, followed by selfing of F1 plants, which would readily result in T-DNA-free segregants that still carry the mutated eif4e allele in the homozygous state. Also, the mutation at the target motif itself prevents the occurrence of new mutation events by disturbing the gRNA to bind to the target sequence. To exclude any effects of the transgene itself, five non-mutated plants carrying events of the same transgene except from the 20 bp target-specific gRNA sequence were used as control and their grain yield was on a par with that of “Igri” wild-type plants.

The one-base pair insertion between position 169 and 170 in the EIF4E coding sequence (position 400-401 in the genomic sequence) causes a frameshift in the translational reading frame, resulting not only in a nonsense amino acid sequence downstream of the mutation, but also in a premature stop codon at position 265 bp of the coding sequence (see Supplementary Material). Consequently, the resultant protein sequence is non-functional with very high certainty. Owing to the loss-of-function of the eIF4E protein, the plant lacks one of its most important interaction partners for bymoviruses (Wang and Krishnaswamy, 2012). That mechanism has been reported to confer broad resistance to several bymoviruses in a number of dicotyledonous plants as, for instance, cucumber, tomato and Arabidopsis (Chandrasekaran et al., 2016; Pyott et al., 2016; Atarashi et al., 2020; Moury et al., 2020). However, in a temperate cereal such as barley, the full knock-out of EIF4E was rather expected to be lethal; Yang et al. (2016) screened over 2,900 wild and domesticated barley accessions and identified 65 haplotypes for HvEIF4E, of which 19 were associated with resistance to bymoviruses. The allelic diversity included various non-synonymous point mutations, whereas knockout alleles were not present. Hence, it was concluded that a knockout of HvEIF4E cannot be complemented by its paralogue HvEIF(iso)4E without disadvantage for the plants. Hofinger et al. (2011) described the haplotype diversity of EIF4E in barley in comparison to its paralogue EIF(iso)4E. While a rather high diversity was described for the first, most likely due to positive selection pressure by co-evolution with bymoviruses, rather low haplotype diversity was found for the latter, suggesting different functions of EIF4E and EIF(iso)4E in barley. Nevertheless, during our experiments in climate chamber and glasshouse, HvEIF4E knockout plants grew vigorously and produced grains. However, the total number of grains per plant was significantly affected. Previous publications on targeted EIF4E knockouts with resulting resistance to Potyviruses in other plant species unfortunately did not contain yield data to refer to here (Chandrasekaran et al., 2016; Gomez et al., 2019). Since isogenic wild-type lines for barley accessions with rym4- or rym5-mediated resistance are not yet available, it is not even known whether these widely used resistances are associated with yield reduction. The results of our study need confirmation under field conditions to elucidate whether the eif4e knockout represents a viable novel source of resistance to bymoviruses in barley, which most likely, in contrast to rym4 and rym5 and the other alleles known at this locus, may be effective to all strains of BaMMV and BaYMV. In the event that the yield penalty seen in the present investigation will be confirmed under field conditions, the option remains for future biotechnological approaches to generate mutant alleles with sufficient functionality for the plant by inducing in-frame mutations, such as via base editing, so that a recruitment of the corresponding gene products by bymoviruses is no longer possible.

The availability of eif4e knockout mutants further offers the opportunity to dissect the effects of the individual SNPs jointly present in the rym4- and rym5-alleles as well as to elucidate the impact of various EIF4E alleles on yield, which could be achieved via complementation using synthetic alleles each carrying only one of those SNPs or various combinations thereof. The identification of effective SNPs will facilitate further efforts to generate novel resistance-conferring alleles by precise genome editing approaches.