PROTEIN DISULFIDE ISOMERASE LIKE 5-1 is a susceptibility factor to plant viruses

The resistance gene rym11 was previously located to a 0.074-cM region on chromosome 4HL (18–20). BAC contigs were identified by sequence comparison of flanking markers and BAC library screening. Sequence analysis of the entire BAC contigs revealed an overlap of ∼18 kb between BAC clones HVVMRXALLhA0173N06 and HVVMRXALLhA0172k23 residing on the original finger-printed contigs (FPCs) ctg551 and ctg1996 (SI Appendix, Fig. S1), respectively. Thus, a physical contig harboring rym11 and its flanking markers was established. New markers were then developed and the genetic interval of rym11 narrowed to 0.0196 cM between markers C5B4C and C2B18S2 (Fig. 1A). This target interval was represented by 12 overlapping BAC clones (SI Appendix, Fig. S1 and Fig. 1A) spanning a distance of about 1.25 Mbp (Fig. 1A). Annotation of the nonrepetitive sequences of this interval identified four ORFs (Fig. 1A). This included two genes with sequence homology to ELONGATION FACTOR 1-α and one homolog of β3-GLUCURONYLTRANSFERASE 43. Resequencing of these three genes revealed no polymorphisms between susceptible (Rym11) and resistant (rym11) genotypes. The fourth ORF encoded a homolog of PROTEIN DISULFIDE ISOMERASE LIKE 5–1 (HvPDIL5-1). This gene contained a 1,375-bp deletion of the putative promoter region and the first three exons in the resistant genotypes W757/112 and W757/612 (SI Appendix, Fig. S2A). Due to this large deletion, no transcript from HvPDIL5-1 could be detected (SI Appendix, Fig. S2B). HvPDIL5-1 was therefore a likely candidate for rym11-based resistance. The allele carrying the 1,375-bp deletion was named rym11-a.

We used three independent approaches to validate the hypothesis that HvPDIL5-1 is a susceptibility factor for bymovirus infection. First, we screened an EMS-induced mutant population of barley by targeting induced local lesions in genomes (TILLING) (21). Two independent mutants 9699 and 10253 were identified each carrying a premature STOP codon in the coding sequence (CDS) of HvPDIL5-1 (SI Appendix, Table S1 and Fig. 1B). M₃ progeny of the originally heterozygous M₂ plants segregated for virus resistance. After phenotyping, the plants were genotyped for the presence of the mutation. All homozygous mutant M₃ plants were found to be completely resistant. Homozygous wild-type and heterozygous M₃ plants were predominantly susceptible with infection rates comparable to the susceptible controls (Table 1 and SI Appendix, Table S2). Resistance tests were replicated twice in M₄ progeny of a homozygous mutant and a wild-type M₃ plant of each of the mutant lines 9699 and 10253, respectively. All homozygous mutant M₄ plants were tested resistant. Homozygous wild-type M₄ progeny were susceptible, with infection rates between 66% and 100%, similar to susceptible controls (Table 1 and SI Appendix, Table S2 and Fig. S3). Mechanically infected homozygous wild-type HvPDIL5-1 M₄ progenies showed the characteristic yellow discoloration during vegetative growth stage (Fig. 1C), stunted growth at heading stage (Fig. 1D), as well as late maturation (Fig. 1E). Mock inoculated homozygous mutant M₄ individuals did not show any detectable difference in overall growth habit at heading and maturation stages compared with control plants (SI Appendix, Fig. S4). We conclude that HvPDIL5-1 loss-of-function mutants exhibit no deleterious growth effects while acquiring resistance to bymovirus infection.

Second, we amplified a 576-bp cDNA containing the 456-bp full-length ORF of wild-type HvPDIL5-1 from barley cv. Naturel, and used this to complement the resistant genotype W757/612 by Agrobacterium-mediated transformation with the intention to induce susceptibility. W757/612 contains the recessive resistance allele rym11-a. Two independent transgenic events (T0) were obtained. T₁ plants were mechanically infected, phenotyped, and subsequently genotyped for the presence of the transgene. RT-PCR amplification revealed a 3:1 segregation (presence/absence) of the transgene in the respective T₁ families, indicating the presence of a single transgene locus (Table 1). Plants carrying the transgene were observed to be predominantly susceptible to BaMMV infection (Table 1 and SI Appendix, Table S2) whereas T₁ null segregants were consistently resistant (Table 1 and SI Appendix, Table S2). Therefore, complementation of rym11-a with wild-type HvPDIL5-1 converted a resistant genotype into a susceptible one.

Third we performed allelism tests between naturally occurring rym11 alleles. The breeding line Russia57 was previously reported to carry the recessive monogenic resistance locus rym11 (18). Resequencing HvPDIL5-1 from Russia57 revealed a 17-bp deletion causing a frameshift (henceforth named rym11-b, Fig. 1B). F₁ plants obtained by crossing W757/112 (rym11-a) and Russia57 (rym11-b) showed complete resistance to mechanical inoculation with BaMMV (Table 1 and SI Appendix, Table S2). Thus, both rym11-a and -b confer bymovirus resistance.

Natural variation of HvPDIL5-1 was surveyed by resequencing the entire ORF in a geographically referenced collection of 365 wild (Hordeum vulgare ssp. spontaneum) and 1,452 domesticated barley accessions (H. vulgare ssp. vulgare) (SI Appendix, Table S3). High-quality sequence was obtained from 1,732 accessions for 1,974 bp of genomic sequence including the complete ORF. We used the 456 bp of assembled CDS for calling exon-based sequence diversity. Overall, four deletions and 25 SNPs defined 28 independent exon haplotypes (SI Appendix, Table S4). Relative to haplotype I (i.e., wild-type HvPDIL5-1 of cv. Naturel), nine haplotypes contained exclusively synonymous variants, whereas the remaining 18 contained deletions or nonsynonymous variants, sometimes in combination with additional synonymous SNPs (SI Appendix, Table S4). Twenty-three haplotypes were exclusively observed in accessions originating from the Near East (SI Appendix, Fig. S5). Three haplotypes (I, III, and IV) were shared between wild and domesticated barleys (Fig. 2). Haplotype I was found in 1,607 of the 1,732 accessions (92.8%) (SI Appendix, Table S3) and was present in all geographical regions (Fig. 3A). Median-joining (MJ) network analysis revealed that HvPDIL5-1 homologs from Brachypodium distachyon, Secale cereale, Triticum aestivum, and Hordeum bulbosum forming the outgroup, were most closely related to haplotype I in the H. vulgare lineage (SI Appendix, Fig. S6). Haplotype I is therefore most closely related to or represents the ancestral HvPDIL5-1 haplotype. The remaining haplotypes in barley, therefore, likely evolved by natural selection (Fig. 2).

Among these 1,732 accessions, only a single genotype, HOR1363, carried the rym11-a resistance allele (haplotype XXVIII). Twenty-seven accessions carried rym11-b (haplotype II) (SI Appendix, Table S4). All 28 accessions showed resistance upon natural and artificial virus infection (SI Appendix, Table S5). Two additional haplotypes (VII and VIII) encoded defective PDIL5-1 proteins. Haplotype VII carried a 1-bp deletion and haplotype VIII exhibited a nonsynonymous substitution at base-pair position 182 of CDS (SI Appendix, Table S4 and Fig. 1B). Accessions carrying either of these haplotypes conferred resistance to BaMMV and BaYMV (SI Appendix, Table S5) and were named rym11-c and -d, respectively (Fig. 1B). F₁ hybrids carrying rym11-a and either rym11-c or -d were resistant to BaMMV (rym11-a + c, rym11-a + d, Table 1). All naturally occurring resistance alleles (rym11-a, -b, -c, and -d) were derived directly by single mutation events (deletion or substitution) of haplotype I (Fig. 2). Four haplotype-I–containing accessions were tested for resistance by mechanical infection with BaMMV and all were susceptible (SI Appendix, Table S5). Overall haplotype I predominates among accessions of a worldwide diversity collection. We conclude that susceptibility to Bymovirus is the ancestral state. We propose that recessive rym11 resistance alleles most likely arose by selection of mutations in HvPDIL5-1 during the migration of domesticated barleys into new environments where naturally occurring Bymoviruses were prevalent. This proposal is supported by our observation that no resistance-conferring alleles were found in wild barleys (Fig. 2). Furthermore, only a single rym11-a carrying barley landrace HOR1363 originated from within the accepted natural distribution range of the wild species (Turkey). All 32 accessions carrying recessive resistance alleles (rym11-b, -c, and -d) were collected from Eastern Asian countries (SI Appendix, Fig. S5 and Fig. 3B) where high BaMMV/BaYMV diversity has previously been observed (4, 22).

In many plants, recessive resistance to Potyviruses is controlled by natural variation in the highly conserved EUKARYOTIC TRANSLATION INITIATION FACTOR 4E (eIF4E) (9). To determine the level of PDIL5-1 conservation, we conducted a sequence comparison and phylogenetic analysis among putative functional orthologs from different plant and animal species. HvPDIL5-1 is a specific member of the PDI gene family and encodes 151 amino acids. This particular member carries a single functional thioredoxin (TRX) domain (SI Appendix, Table S6) putatively catalyzing the formation, reduction, and isomerization of disulfide bonds. It contains the C′-terminal tetrapeptide EDEL, which controls the retention of the protein within the endoplasmic reticulum (ER), as could be demonstrated for the homologous human protein ERp16 (also called ERp18, ERp19, or hLTP19) (23). Eighteen homologous genes in plants and 30 genes in animals were identified (SI Appendix, Table S6). A phylogenetic analysis for 17 PDI proteins of species from all evolutionary and organizational stages (e.g., unicellulate and multicellulate organisms, algae/primitive land plants/higher vascular plants, etc.) revealed a single rooted tree with a split between the animal and plant branch but otherwise consistent sorting according to the evolutionary level or organization (Fig. 4A). Sequence similarity inside the TRX domain separated the plant from the animal branch and diversity in the C′-terminal region revealed differentiation between monocotyledonous and dicotyledonous plant species (SI Appendix, Figs. S7 and S8). Furthermore, the majority of homologous proteins are predicted to be secreted to the ER (SI Appendix, Table S6). The 3D structure of the homologous proteins of a protist (Capsaspora owczarzaki), of a unicellulate algae (Chlamydomonas reinhardtii) and of barley were simulated in reference to the human protein ERp18 (24). The predicted 3D structure was highly conserved across kingdoms (Fig. 4B). As PDIL5-1 homologs share structure, function, and subcellular localization between organisms, it is tempting to speculate that some may similarly function as dominant virus susceptibility factors in organisms other than barley.

We identified a susceptibility factor to bymovirus disease in barley encoded by HvPDIL5-1, a member of the PROTEIN DISULFIDE ISOMERASE (PDI) gene family. The PDI gene family usually contains more than 10 members in different eukaryotic species; e.g., there are 20 genes in Homo sapiens and 18 in rice (25–27). The major function of PDIs in humans is to contribute to the formation of disulfide bonds by induction, oxidization, and isomerization; all catalyzed by the active TRX domain (25, 27). PDIs also play a role as chaperones in the quality check system for correct protein folding (28). Based on its high amino acid sequence conservation to human ERp16, which encodes for an ER-resident thioldisulfide oxidoreductase fulfilling chaperone activities (23), we conclude that plant homologs have retained this function. Thus, HvPDIL5-1 could have been recruited by Bymoviruses in barley to act as cellular chaperone (protein folding, stabilization, or facilitating transport) during virus infection. Other cellular chaperones like HSP70, HSP90, and DNAJ-like proteins have been shown to perform similar functions in other plant species in the interactions with several genera of plant viruses (12). Both functional motifs and tertiary structures of HvPDIL5-1 and its homologs in other plants and animals are highly conserved. By identifying HvPDIL5-1 as susceptibility factor to the potyvirus-related Bymoviruses in barley, it seems plausible that other members of the PDI gene family may play a similar role in other plant–virus interactions. Given that the highly conserved eukaryotic translation initiation factor gene family has been repeatedly identified as susceptibility factors in numerous pathosystems (9), we consider this a realistic possibility.

Members of plant PDIs have not been reported so far as sensu stricto susceptibility factors to any plant viruses and little is known about plant PDI/virus interactions. However, virus-induced gene silencing (VIGS) of two PDI family members in tobacco, NtERp57 and NtP5, partially decreased N-gene mediated resistance to tobacco mosaic virus (TMV) (29). Whereas this hinted at an interaction between plant PDIs and viruses, it did not indicate a role as a susceptibility factor. The situation is different in animals. PDI family members, particularly cell-surface PDI proteins, may participate in the infection process of multiple human and animal viruses, e.g., HIV (HIV) (30–36). Nonspecific inhibition of PDI activity suppressed the PDI-mediated redox environment of plasma membranes (33) and therefore interfered with HIV envelope protein-directed cell fusions (32). Knockdown of specific PDI members could influence the infectivity of several viruses, e.g., HIV, Newcastle disease virus (NDV) and mouse polyomavirus (31, 32, 35). In human cell cultures, knockdown of the PDI family members, PDI, ERp29, ERp57, and ERp72 inhibited the accumulation of viral particles (30, 34, 36). Our data support and extend the role of PDIs as important host components required for the successful infection or replication of viruses in animals to plants.

Resequencing HvPDIL5-1 in a large georeferenced collection of wild and domesticated barleys revealed less haplotype diversity in domesticated (12 haplotypes) compared with wild barley (18 haplotypes) (although almost fourfold more accessions of domesticated barley were surveyed). This is in strong contrast to the four naturally occurring loss-of-function alleles of HvPDIL5-1. All four confer resistance to Bymoviruses, and all four were exclusively observed in domesticated barleys. Geographically, resistant alleles were overrepresented in winter-type barley from Eastern Asia (SI Appendix, Fig. S9), suggesting selection for resistance during range extension as barley cultivation spread east from its center of origin. This is consistent with the higher diversity of Bymoviruses in Eastern Asia (4, 22) where a nonfunctional HvPDIL5-1 may have provided an adaptative advantage.

Barley yellow mosaic virus disease is one of the most important diseases in winter barley (15). Its soil-borne transmission precludes plant protection by use of agrochemicals. In European barley breeding this stimulated the extensive use of rym4/rym5, a single natural recessive resistance conferred by alleles of the EUKARYOTIC TRANSLATION INITIATION FACTOR 4E (37). However, rym4/rym5 resistance-breaking isolates have recently been observed (16, 17, 38). rym11 is effective against all European strains of BaMMV/BaYMV (16–18). We have shown here that resistance is conferred by loss-of-function mutations in HvPDIL5-1. HvPDIL5-1 must therefore provide a host component that is essential for virus establishment and/or replication in planta. To overcome rym11-based resistance the virus would therefore need to establish an interaction with an alternative and functionally redundant host component. Today, rym11 plays a minor role in European barley breeding. Its implementation may, however, reduce the potential for the emergence of resistance-breaking viral isolates. Pyramiding several resistance genes in a single genotype is considered to be a promising strategy to obtain durable virus resistance (39). With the identification of rym11 as loss-of-function alleles of HvPDIL5-1, it is now possible to combine two recessive resistance loci (rym11 and rym4/rym5) on the basis of perfectly linked and diagnostic molecular markers.

High-resolution genetic mapping of rym11 was conducted in 5,102 F₂ individuals of the cross cv. Naturel (Rym11/Rym11) × resistant genotype W757/112 and W757/612 (both rym11/rym11) (20). Eighteen recombinants of this population were used for further genetic and physical delimitation of rym11. For allele mining, 1,816 georeferenced accessions from 70 countries including 365 wild and 1,451 domesticated barleys were selected (SI Appendix, Table S3). F₁ hybrids were obtained by crossing the resistant genotypes W757/112 and/or W757/612 with the newly identified gene haplotypes.

BAC contigs were identified either by sequence comparison of flanking markers (20) to sequence resources integrated to the physical map (40) or by PCR-based screening of cv. Morex BAC library (41). Presence of previously unobserved overlaps was surveyed by (i) reassembly of all BACs belonging to the identified contigs with software FPC (42) at Sulston cutoff e-10 as well as (ii) sequencing a minimum tiling path (MTP) of BACs of the respective physical contigs. BAC sequencing was performed on Roche/454 GS FLX Titanium (Roche Applied Science) and 454-shotgun reads were assembled using the software MIRA version 3.2.1 essentially as described previously (40). Sequence overlaps between BACs were confirmed by PCR amplification and inspected by sequence comparison (BLASTn) (43) and dot-plot analysis (Dotter) (44).

Repetitive elements of BAC clone sequences were masked by applying K-mer statistics (45). Nonmasked sequences were compared by BLASTn analysis to barley high-confidence genes (40), barley full-length cDNAs (46), HarvEST_U35 (47), as well as annotated gene sets of the sequenced grass species rice (48), Brachypodium (49) and sorghum (50). Putative ORFs of potential candidate genes were amplified. The amplicons were purified by using NucleoFast 96 PCR kit (Macherey-Nagel) and then sequenced by using ABI-3730xl technology (Applied Biosystems).

RNA extraction, first-strand cDNA synthesis, and RT-PCR were performed as described previously (51). The young leaves of seedlings at three-leaf stage or BaMMV-inoculated leaves 5 wk postinoculation were sampled for total RNA extraction. The barley UBIQUITIN CARRIER gene (NCBI accession no. AK361071, HvUBC) was used as endogenous constitutively expressed control (51).

A large ethyl methanesulfonate (EMS)-induced mutant population of BaMMV-susceptible cv. Barke, comprising 10,279 M₂ families (21), was screened for induced mutations in the gene HvPDIL5-1 by TILLING (21). Two amplicons separately covered the first three and the last exon of the gene (SI Appendix, Table S7), respectively. To identify potential mutants, heteroduplex analysis of PCR amplicons was performed applying the Mutation Discovery kit and gel-dsDNA reagent kit, and samples were subsequently run on an Advance FS96 system (Advanced Analytical). To verify the identified mutations, the gene was subsequently amplified from identified M₂ individuals and PCR amplicons were Sanger sequenced. All M₂ individuals with nonsynonymous mutations were propagated (SI Appendix, Table S1) for resistance tests.

For the identification of naturally occurring allelic diversity, the full-length coding sequence of HvPDIL5-1 was analyzed by PCR and amplicon sequencing. Four primer pairs were designed to cover the entire length of 1,974-bp of HvPDIL5-1 genomic sequence (SI Appendix, Table S7). PCR products were amplified, purified, and sequenced as described above. Sequence alignment and assembly was performed with Sequencher 4.7 (Gene Codes), and allelic haplotypes and polymorphic loci were defined by DNASP 5.10.01 (52). A MJ network was constructed with DNA Alignment 1.3.1.1, Network 4.6.1.1, and Network Publisher 1.3.0.0 software (Fluxus Technology).

A 576-bp cDNA fragment containing the entire ORF of wild-type HvPDIL5-1 amplified from cv. Naturel was cloned into the binary vector pIPKb002 (53) to generate ZmUbi_PDIL5-1_ox. The obtained destination plasmid was transformed into Agrobacterium tumefaciens strain AGL-1 (54). Isolation and culture of immature embryos, Agrobacterium-mediated transformation and regeneration of transgenic barley plants were conducted on the resistance parent W757/612 as previously described (54), with minor modifications. Prior to inoculation with Agrobacterium immature embryos were cultured on BCIM with dicamba at a final concentration of 5 mg/L for five days at 24 °C in the dark. Coculture of pre-treated immature embryos took place on stacks on moistened (300 μL BCCM) filter paper in 5.5 cm diameter petri dishes. Due to absence of transcript of the endogenous copy of HvPDIL5-1 in W757/612, transcription of the transgene could be monitored by RT-PCR.

Homology search of HvPDIL5-1 in other species was performed by BLASTx and tBLASTx according to the protein similarity. All homologs were requested to contain the conserved functional domain (TRX domain) and active center (Cys-x-x-Cys) by Conserved Domain Database (55). Multiple sequence alignments of all homologous proteins in monocots, dicots, and mammals were performed by the online software tool ClustalW2 (56). A neighbor-joining phylogenetic tree was produced based on deduced amino acid sequences by MEGA 5.05 (57).

Sequence conservation of the active center and the C′-terminal tetrapeptide of the homologous proteins were graphically represented by WebLogo (58). A 3D model of HvPDIL5-1 and the homologous proteins was simulated on SWISS-MODEL workspace (59) by using the 3D structure of human ERp18 (PDB accession no. 2k8vA) (24) as a reference.

Tests for resistance to BaMMV and BaYMV were performed under controlled growth chamber and field conditions as described previously (17, 20). Tilling mutants (M₃ families, M₄ progeny of homozygous wild-type and mutant M₃ plants), transgenic T₁ plants, and F₁ hybrids were only tested under growth chamber conditions. In brief, plants at three-leaves stage were mechanically inoculated twice with isolate BaMMV-ASL1 (60) at an interval of 5–7 d. Five weeks after first inoculation, infected plants were scored for the presence of mosaic symptoms and virus particles were detected by DAS-ELISA with BaMMV-specific antibodies. Genotype ‘Maris Otter’ was used as susceptible control to monitor the infection rate in each experiment and showed infection rates between 70% and higher than 90%, as reported previously (17). Accessions from the diversity panel showing nonfunctional HvPDIL5-1 alleles were analyzed by mechanical inoculation of BaMMV. In parallel, these accessions (30 plants each) were cultivated in Quedlinburg, Germany, under field conditions providing natural infection with BaMMV and BaYMV. The pathological responses were subsequently determined by scoring of symptoms. Furthermore, leaf samples pooled from 15 plants per accession were tested by DAS-ELISA diagnostics with BaMMV and BaYMV-specific antibodies.