Rye Pm8 and wheat Pm3 are orthologous genes and show evolutionary conservation of resistance function against powdery mildew


  • The Pm8 sequence has been deposited in the GenBank database under the accession number KF572030 and Pm3-1B under accession number KF572031.


The improvement of wheat through breeding has relied strongly on the use of genetic material from related wild and domesticated grass species. The 1RS chromosome arm from rye was introgressed into wheat and crossed into many wheat lines, as it improves yield and fungal disease resistance. Pm8 is a powdery mildew resistance gene on 1RS which, after widespread agricultural cultivation, is now widely overcome by adapted mildew races. Here we show by homology-based cloning and subsequent physical and genetic mapping that Pm8 is the rye orthologue of the Pm3 allelic series of mildew resistance genes in wheat. The cloned gene was functionally validated as Pm8 by transient, single-cell expression analysis and stable transformation. Sequence analysis revealed a complex mosaic of ancient haplotypes among Pm3- and Pm8-like genes from different members of the Triticeae. These results show that the two genes have evolved independently after the divergence of the species 7.5 million years ago and kept their function in mildew resistance. During this long time span the co-evolving pathogens have not overcome these genes, which is in strong contrast to the breakdown of Pm8 resistance since its introduction into commercial wheat 70 years ago. Sequence comparison revealed that evolutionary pressure acted on the same subdomains and sequence features of the two orthologous genes. This suggests that they recognize directly or indirectly the same pathogen effectors that have been conserved in the powdery mildews of wheat and rye.


In the agricultural environment, plants are under permanent attack from pathogens. Resistance breeding in crop species strongly relies on resistance (R) genes. They encode proteins which directly or indirectly recognize effector molecules delivered into plant cells by pathogens (Dodds and Rathjen, 2010). In the co-evolutionary arms race of host and pathogen, the emergence of an effector gene leading to successful pathogen invasion is followed by the emergence of a new R gene in the host plant (Kanzaki et al., 2012). R genes frequently break down rapidly because there is strong selection for virulent pathogen mutants under agricultural conditions (McDonald and Linde, 2002). Therefore, resistance breeding depends on the constant identification of new resistance resources to be integrated into breeding programmes.

The elite wheat (Triticum aestivum L.) cultivars currently grown agriculturally only represent a small fraction of the wheat gene pool. In order to further improve wheat for resistance to biotic and abiotic stresses, quality and yield, breeders have been using the genetic diversity of closely related wild and domesticated species (Baum et al., 1992). The transfer of alien chromatin to wheat resulted in agronomically useful wheat–alien translocation lines that carry important functional resistance genes against several pathogens, mainly rust and powdery mildew (Friebe et al., 1996; Tyrka and Chelkowski, 2004). Despite their agronomical importance, little is known about these translocations on the molecular level and it remains to be determined whether single genes or gene clusters mediate the observed resistance traits.

In the 1930s, a wheat cultivar was developed which carried the translocated 1RS chromosome arm of the rye (Secale cereale L.) cultivar Petkus, replacing wheat chromosome arm 1BS. Cultivars with this so-called 1BL.1RS translocation (Mettin et al., 1973; Zeller, 1973) showed high yield potential, wide adaptation and disease resistance against powdery mildew, stem rust, leaf rust and stripe rust (Rabinovich, 1998; Kim et al., 2004). Due to these favourable traits, in 1998 wheat–rye translocation lines accounted for 50% of the International Maize and Wheat Improvement Center (CIMMYT) high-yielding bread wheat cultivars, covered over 5 million hectares of the wheat grown area worldwide and are still cultivated at large scale (Villareal et al., 1998; Purnhauser et al., 2011). The gene on chromosome arm 1RS conferring resistance to wheat powdery mildew (Blumeria graminis f.sp. tritici) was named Pm8 (McIntosh, 1988). Soon after its widespread use in the 1970s, increasing mildew virulence to Pm8 was reported (Bennett, 1984; Heun and Friebe, 1990).

To date, in the wheat gene pool 43 genetic loci with nearly 70 genes/alleles have been described to mediate resistance against wheat powdery mildew. More than 30 of these Pm genes/alleles were introgressed from wild relatives, demonstrating the widespread use of resistance genes from foreign species in wheat breeding (Tyrka and Chelkowski, 2004; McIntosh et al., 2012). So far, only two Pm genes have been cloned: Pm3 from wheat (Yahiaoui et al., 2004) and a key member of the Pm21 resistance locus transferred from Haynaldia villosa to hexaploid wheat (Cao et al., 2011). For Pm3, 17 functional alleles were isolated which share more than 97% nucleotide sequence identity and code for coiled-coil (CC), nucleotide-binding site, ARC1 and ARC2 (NB-ARC) and leucine-rich-repeat (LRR) domain proteins (Yahiaoui et al., 2004, 2006, 2009; Srichumpa et al., 2005; Bhullar et al., 2009, 2010).

The last common ancestor of wheat and rye lived around 7.5 million years ago (Huang et al., 2002a,b). Both wheat and rye are hosts of the cereal powdery mildews (Blumeria graminis) but due to strict host specialization they are infected by two different formae speciales, f.sp. tritici (Bgt) and f.sp. secalis (Bgs), respectively. From barley (f.sp. hordei) and wheat powdery mildew it is known that the hosts separated about 12 million years ago while the pathogen separated about 2 million years later, indicating host–pathogen co-evolution (Oberhaensli et al., 2011). This might also be true for wheat and rye powdery mildew. Analysis of individuals of a cross between wheat and rye powdery mildew demonstrated that avirulence genes from the rye powdery mildew are recognized by wheat R genes active against wheat powdery mildew (Matsumura and Tosa, 1995). This suggests that R genes might be involved in non-host resistance to inappropriate formae speciales. A model was proposed in which both NB-LRR genes and pattern recognition receptors (PRRs) mediate non-host resistance and their relative contribution depends on the divergence time between the host and the non-host plant (Schulze-Lefert and Panstruga, 2011).

Resistance genes are generally assumed to be rapidly evolving (Michelmore and Meyers, 1998) and functional prediction of R gene homologues/orthologues in related species has been shown to be unreliable (Hulbert et al., 2001). Resistance function against different pathogens has been shown for the homologous pairs Tm-22/Rpi-vnt1.1 and Rpi-blb2/Mi-1 from Solanum species (van der Vossen et al., 2005; Foster et al., 2009) as well as for the three allelic Arabidopsis genes RPP8, HRT and RCY1 (Takahashi et al., 2002). Recently, the stem rust resistance gene Sr33 from Aegilops tauschii was found to be a homologue of the Mla resistance gene family in barley which mediates resistance to powdery mildew (Periyannan et al., 2013). In contrast, recognition of the same pathogen was shown for the homologous genes N′ from Nicotiana sylvestris and L from Capsicum against Tobamovirus spp. in a transient expression assay in Nicotiana benthamiana (Sekine et al., 2012) and three orthologous R genes at the Xa3/Xa26 locus in one cultivated and two wild rice cultivars against Xanthomonas oryzae (Li et al., 2012). An example of conserved fungal pathogen recognition from grass species comes from TmMla1 isolated from the wheat species Triticum monococcum mediating powdery mildew resistance like its barley homologue Mla (Jordan et al., 2011). These examples of R gene homologues gave the first insights into the evolution of R genes. However, the orthologous relationship remains unclear for most of the genes and little is known about the evolution of true orthologous R genes mediating resistance to the same pathogen.

In this study, we cloned the Pm8 gene by a homology-based cloning approach and found it to be the rye orthologue of the wheat Pm3 gene. Sequence analysis revealed that the PM8 and PM3B proteins share 81% sequence identity and that nucleotide diversity is located mainly in the solvent-exposed residues of the LRR domain. The data indicate that an orthologous gene in two grass species can co-evolve with related pathogens over a surprisingly long evolutionary time.


Homology-based cloning of a candidate gene for Pm8

Pm8 maps to a gene-dense region at the distal end of chromosome arm 1RS (Hulbert et al., 2001; Sandhu and Gill, 2002), whereas the wheat Pm3 powdery mildew resistance gene is located in the syntenic gene-dense region of chromosome arm 1AS of wheat, close to the genetic loci encoding the storage proteins Glu-3/Gli-1. Considering that gene order is highly conserved in grasses (Elliott et al., 2002; Mohler et al., 2002; Sandhu and Gill, 2002), we hypothesized that the two genes are orthologues.

To determine whether Pm3 and Pm8 show any similarity in their powdery mildew resistance spectra, we analysed the resistance spectra of Pm8 and different Pm3 alleles to 162 wheat powdery mildew isolates from Switzerland (Brunner et al., 2012), seven from USA and 24 from France (Methods S1). The isolates were tested on wheat differential lines for Pm8 and the Pm3ag alleles, except for 33 Swiss isolates for which Pm3e and Pm3g data were missing. The 35 isolates found to be avirulent on Pm8 were all also avirulent to at least one of the tested Pm3 alleles. The 13 isolates which were virulent on all tested Pm3 alleles were also virulent on Pm8. The same correlations were also detected in other virulence analyses on wheat powdery mildew isolates from Germany and England (Zeller et al., 2002; Huang and Röder, 2004; Lillemo et al., 2010; Schmolke et al., 2012). This functional similarity between Pm8 and the Pm3 alleles possibly reflects recognition of similar effector proteins and provided additional support for the hypothesized orthologous relationship between the two genes.

Southern blot analysis was carried out to examine the presence of Pm3 homologous genes on the rye chromosome arm 1RS from ‘Petkus’ carrying the powdery mildew resistance gene Pm8. A 184-bp fragment from the wheat cultivar Chul/8*Chancellor located 4 kb upstream of the Pm3b allele was used as a probe. This probe UP3 (Figure 1a) is a fragment of the restriction fragment length polymorphism (RFLP) probe TmRGL-1pro which has been shown to specifically detect Pm3 loci in hexaploid wheat (Yahiaoui et al., 2004; Srichumpa et al., 2005), T. monococcum and Triticum turgidum (Wicker et al., 2007b). It does not hybridize to closely related members of the large Pm3 resistance gene-like (RGL) family in these genomes. Remarkably, all tested 1BL.1RS wheat lines and the three ‘Petkus’ rye lines Petkuser Winter, Petkus II and Petkus 91 shared a distinct fragment of 4 kb (Figure 1b). This fragment is absent in Chul/8*Chancellor and the rye cultivar Imperial, two lines not carrying Pm8. In the wheat line Kavkaz, an additional fragment of 6.5 kb indicates the presence of the Pm3 allele Pm3-Kavkaz on wheat chromosome 1AS (Yahiaoui et al., 2006). Thus, Southern blot analysis indicated the presence of one Pm3 homologue on the chromosome arm 1 RS in lines carrying Pm8, and this sequence represented a good candidate for Pm8.

Figure 1.

Wheat lines carrying the 1BL.1RS chromosome and three ‘Petkus’ rye lines share a Pm3 homologous sequence. (a) The structure of the Pm8-candidate gene is shown. The black box indicates the intron and the other boxes the coding region with its three predicted domains [coiled coil (CC), nucleotide-binding (NB)-ARC, leucine-rich repeat (LRR)]. The 2.7-kb non-coding 5′ sequence including the promoter region and the 0.4 kb untranslated 3′ end are represented as lines. The primers used for cloning of the Pm8-candidate gene are indicated by arrows. The wheat probe UP3 was used for Southern blot analysis (its homologous sequence in the Pm8-candidate gene is indicated by a horizontal bar). (b) Wheat and rye genomic DNA was digested with HindIII and hybridized with the 184-bp UP3 probe. A fragment of 4 kb was detected in all tested wheat 1BL.1RS and ‘Petkus’ rye lines (indicated by the arrow head). Line Chul/8*Chancellor (Chul/8*CC) contains the Pm3b allele and was used as a positive control.

We then used available sequence information of the known Pm3 alleles for a homology-based approach to clone this candidate gene. First, the 5′ untranslated region and the 5′ end of the candidate gene were amplified using the primers UP3B/consLRR3B2 (Figure 1a). The 3′ end of the gene was amplified by 3′ rapid amplification of cDNA ends (RACE)-PCR (Methods S1 in Supporting Information). Based on the obtained 5′ and 3′ sequences, the full-length open reading frame was amplified from wheat line Kavkaz/4*Federation and rye line Petkus 91 genomic DNA using nested PCR similar to the amplification strategy for the Pm3 alleles (Srichumpa et al., 2005). The sequences amplified from the wheat translocation line and from rye were identical. The predicted gene has a total length of 4321 bp, with one 193-bp intron ending 84 bp upstream of the stop codon, as deduced from the 3′ RACE sequence information. Based on the successful amplification in the 3′ RACE experiment we conclude that the gene is expressed. It encodes a protein of 1375 amino acids and we refer to it as the Pm8-candidate gene.

The Pm8-candidate gene maps to the Pm8 locus

The sequence of the Pm8-candidate gene, including its 5′ and 3′ regions, was compared with all available sequences from wheat with homology to the Pm3 alleles (Yahiaoui et al., 2004; Wicker et al., 2007b; Bhullar et al., 2010). This allowed us to design a primer pair specific for the Pm8-candidate gene (SH43/SH46; Figure 1a), which amplified a 662-bp fragment in wheat and rye lines carrying Pm8 (Figure 2). Validation of this marker sfr43(Pm8) on a large set of wheat lines with and without the 1BL.1RS translocation as well as on rye lines proved it to be diagnostic for the presence of the Pm8 gene (Table S1).

Figure 2.

The Pm8-candidate gene marker sfr43(Pm8). The primer pair SH43/SH46 amplifies a 662-bp fragment in wheat 1BL.1RS lines (Kavkaz and Kavkaz/4*Federation) as well as in rye line Petkus 91. No fragment was amplified from wheat lines without the 1BL.1RS translocation [Federation, Amigo (1AL.1RS; Pm17) and Chul/8*Chancellor (Pm3b)] or from rye line Imperial known not to carry Pm8.

We used two independent populations to genetically and physically map the Pm8-candidate gene. In an earlier study of wheat 1BS/1RS homoeologous recombinants obtained in a ph1b mutant background, the Pm8 resistance phenotype was shown to be genetically located between the cluster of rust resistance genes Lr26, Sr31 and Yr9 and the Gli-1/Glu-3 loci (Lukaszewski, 2000; Sharma et al., 2009). To test if the Pm8-candidate gene maps to this same locus, we analysed six recombinants with physical breakpoints close to Pm8 with marker sfr43(Pm8), along with their parental lines Pavon (no Pm8) and Pavon 1RS.1BL (Pm8). The Pm8 resistance phenotype was confirmed in the recombinants using a Pm8 avirulent powdery mildew isolate, 07230, in a leaf segment infection test (Methods S1). The recombinants T9 and 1B+37 were resistant in the infection test and were positive for the marker sfr43(Pm8). The other four recombinants (T16, T18, T8 and 1B+14) were susceptible in the infection test and negative for the marker (Figure 3a). These results are consistent with previous infection tests and showed that the Pm8-candidate gene co-localized with the Pm8-mediated powdery mildew resistance and with the physical map position of the Pm8-locus.

Figure 3.

The Pm8-candidate gene marker sfr43(Pm8) maps to the Pm8 resistance locus. (a) Molecular analysis of wheat–rye recombinants. Black bars indicate the rye chromosome arm 1RS and white bars the wheat chromosome arm 1BS. On the left side, markers and gene locations are indicated (Sharma et al., 2009). Below the graph, results from the infection test with powdery mildew isolate 07230 (R, resistant; S, susceptible) and results of the Pm8-candidate gene marker sfr43(Pm8) analysis are given (+, 662-bp fragment present; –, no fragment). (b) The Pm8-candidate gene marker sfr43(Pm8) was integrated into the genetic map established by Mago et al. (2005) by testing 134 F2 wheat individuals. On the left, distances between markers are given in cM and the number of recombinants is indicated in brackets.

In a second approach, we used a high-resolution mapping population from a cross of two 1BL.1RS wheat lines earlier developed for the rust resistance genes Sr31, Lr26 and Yr9 on chromosome arm 1RS (Mago et al., 2005). This population was segregating for Pm8 but it was not phenotyped for this gene. No phenotypic analysis could be performed since no more seeds of the F2 population were available. In total, the genomic DNA of 134 F2 plants was screened with the marker sfr43(Pm8). It mapped 0.7 cM proximal to the marker Xiag95 and 1.7 cM distal to the rust resistance genes (Figure 3b). The location of Pm8 in this mapping population is in agreement with the location of Pm8 found in the 1BS/1RS homoeologous recombinants, where Pm8 is located between markers Gli-1/Glu-3 (and Iag95; Mago et al., 2002) and the rust resistance genes (Figure 3a).

The Pm8-candidate gene mediates race-specific resistance to powdery mildew

We validated the Pm8-candidate gene for resistance function by transiently expressing it in leaf epidermal cells of the wheat cultivars Federation and Chancellor, which are highly susceptible to powdery mildew. Transient expression of the Pm8-candidate gene resulted in a significant reduction of the haustorium index (HI) (HI 27% Federation and 26% Chancellor; Student's t-test, P < 0.001) compared with the empty vector control (HI 66% Federation and 67% Chancellor) when the leaves were inoculated with the Pm8 avirulent powdery mildew isolate 07230 (Figure 4a, Table S2). In contrast, there was no significant difference in the HI when a Pm8 virulent isolate 07250 was used in this assay (HI 62% Federation and 67% Chancellor; Student's t-test, P > 0.7) (Figure 4a, Table S2). This demonstrated that the Pm8-candidate gene mediated race-specific powdery mildew resistance to an isolate avirulent on Pm8.

Figure 4.

The Pm8-candidate gene mediates race-specific powdery mildew resistance against isolate Bgt 07230. (a) In the transient assay, the haustorium index (HI) was significantly lower when leaf segments of the susceptible wheat cultivars Federation or Chancellor were bombarded with the Pm8-candidate gene and infected with the Pm8 avirulent Bgt isolate 07230 compared with the empty vector control (Student's t-test, ***P < 0.001). No significant difference was found when using virulent Bgt isolate 07250 (Student's t-test, P > 0.7). The HIs represent the mean of three independent biological replicates and error bars are the standard deviations. (b) The wheat cultivar Bobwhite 26 (Bobwhite SH 98 26) was stably transformed with the Pm8-candidate gene. While Bobwhite 26 was highly susceptible to the isolates 07230 and 07250, T2 plants of three independent transformation events (Pm8#12, Pm8#34 and Pm8#59) were completely resistant to isolate 07230 but susceptible to isolate 07250. The control cultivars Bobwhite 56 (Bobwhite SH 98 56) and Fed*4/Kav (Federation*4/Kavkaz) carry an endogenous Pm8 gene and are resistant to isolate 07230 but susceptible to isolate 07250.

We then stably transformed the Pm8-candidate gene under the transcriptional control of the maize ubiquitin promoter (ubi) into cultivar Bobwhite SH 98 26, which is highly susceptible to powdery mildew and does not carry an endogenous Pm8 gene. Functionality of this construct was first confirmed in transient transformation assays. Three transformation events, which segregated for powdery mildew resistance in the T1 generation and carried one to three copies of the gene, as indicated by Southern blot analysis (Figure S1), were further analysed. In the T2 generation, the heterozygous families showed co-segregation of the transgene with Pm8 powdery mildew resistance, while plants of the homozygous families (T2-Pm8#12, T2-Pm8#34 and T2-Pm8#59) were resistant when inoculated with an isolate which is avirulent on Pm8 (07230). All plants were susceptible to Pm8-virulent isolate 07250 (Figure 4b).

Expression of the Pm8 transgene was verified in a reverse transcription, quantitative real-time polymerase chain reaction (RT-qPCR) assay along with the wheat–rye translocation lines Ambassador, Benno, Federation*4/Kavkaz and Veery#6 and the rye line Petkus 91, all carrying an endogenous Pm8 gene. The Pm8 expression levels in these control lines were not significantly different from each other. In contrast, transgenic line Pm8#12 showed an expression level approximately 430 times higher than the Federation*4/Kavkaz line, while lines Pm8#34 and Pm8#59 had expression levels about 255 and 161 times higher, respectively (Figure 5). The complete resistance of the transgenic lines to the Pm8-avirulent isolate compared with the incomplete resistance of the endogenous Pm8 line Federation*4/Kavkaz (Figure 4b) could be due to the high expression level of Pm8 under the ubiquitin promoter. The genetic background may also play a role, since a related Bobwhite line carrying an endogenous Pm8 gene (Bobwhite SH 98 56) was more resistant than Federation*4/Kavkaz (Figure 4b). In summary, the Pm8-candidate gene-mediated race-specific powdery mildew resistance is identical to the resistance mediated by the endogenous Pm8 gene both in a transient assay and in stably transformed lines. The functionality of the Pm8-candidate gene in three distinct wheat cultivars in the absence of rye chromosome 1RS indicates that Pm8 function is cultivar independent and does not rely on additional genes present on the rye 1RS translocation. We conclude that the isolated gene is Pm8.

Figure 5.

The transgenic Pm8 lines showed a higher expression level of Pm8 than the wheat-rye translocation lines and rye. The relative expression levels of Pm8 were determined by reverse transcription, quantitative real-time polymerase chain reaction (RT-qPCR) in the first leaf of 12-day-old plants. They are plotted relative to line Fed*4/Kav (Federation*4/Kavkaz) and represent the means (untransformed) of three biological replicates (two replicates for Petkus 91). The 95% confidence intervals (back-transformed) are plotted. On top of the bars, different letters denote a significant difference in expression level (Tukey's honestly significant difference test, α = 0.050).

Pm3-specific haplotype sequence at the Pm8 locus reveals orthology

A BLAST similarity search revealed that Pm8 is most similar to the Pm3 alleles showing 87% identity at the DNA level and 80% at the protein level to the Pm3 allele Pm3CS. Pm3CS is a susceptible allele present in several wheat lines and represents the consensus sequence of the known wheat Pm3 resistance alleles (Yahiaoui et al., 2006). The highest similarity was found to the functional allele Pm3b with 81% identity at the protein level. The Pm8 gene codes for a protein of 1375 amino acids, whereas Pm3b codes for 1415 amino acids. The intron size is highly similar with 193 bp for Pm8 versus 200 bp for Pm3b and is conserved in its position at the C-terminal end of the protein. All of the characteristic domains of coiled-coil NB-LRR resistance proteins as well as all the motifs in the NB-ARC domain (Takken et al., 2006) are conserved in length and position between the Pm3 alleles and Pm8 (Figure S2).

The UP3 Southern blot probe was shown in earlier studies to specifically detect Pm3 among a large number of sequence-related genes in wheat (Yahiaoui et al., 2004). A BLASTN search including the whole wheat survey sequences (http://www.wheatgenome.org/) revealed that the probe had no other homology than to the 5′ region of the Pm3 alleles, a bacterial artificial chromosome clone from T. monococcum which was used for mapping of the Pm3 locus and a Pm3 homologue on chromosome 1BS (Pm3-1B). In the barley genome (Mayer et al., 2012) the sequence is not conserved, while in rye three copies are present as indicated from Southern blot analysis (Figure 1b). Sequence comparison of Pm8 and Pm3 showed that UP3 is part of a 294-bp region that is highly conserved between the two genes (94% identity; Figure S3). This is even more remarkable considering that the complete 5′ regions of the two genes are in general highly diverse with 54% identity (Figure S3). The fact that the Pm3 haplotype-specific sequence is conserved in Pm8 strongly supports the conclusion from physical and genetic mapping experiments that Pm8 and Pm3 are orthologues.

Very high sequence conservation in the CC domain between Pm8 and Pm3

Among the CC, NB-ARC and LRR domains, the CC domain shows the highest sequence conservation, with only 10 out of 158 amino acids being different between PM8 and PM3B in this domain (Figure S2). This is reflected by a very low nucleotide diversity value of 0.051 (π(total); Table 1), while the nucleotide diversity value of Pm8 compared with Pm3b for the entire gene is 0.097. Actually, in the predicted CC structure only one amino acid change is found. Interestingly, no polymorphisms were found in the CC domain among the 31 Pm3 alleles known from hexaploid wheat, and only three nucleotide polymorphisms were found in the 23 alleles from tetraploid wheat (Yahiaoui et al., 2009; Bhullar et al., 2010). The CC domain of NB-LRR R proteins has been shown to specifically interact with other proteins or to form oligomers (Shen et al., 2007; Rairdan et al., 2008; Jordan et al., 2011; Maekawa et al., 2011; Chang et al., 2013). We speculate that the CC domains of PM3 and PM8 may interact with the same proteins and are involved in the same signalling pathway based on their high sequence conservation.

Table 1. Nucleotide diversity between Pm8 and Pm3b
StructureNumber of sitesPolymorphic sitesSynonymous changesNon-synonymous changesπ (total)aπ synaπ nsa
  1. CC, coiled coil; NB, nucleotide-binding; LRR, leucine-rich repeat.

  2. a

    π is the nucleotide diversity which gives the average number of nucleotide differences per site between sequences. syn, synonymous; ns, non-synonymous.

Entire gene41224001462530.0970.140.084
CC domain4742414100.0510.1240.029
NB domain5197831460.150.2450.121
ARC1 domain210306240.1430.0820.161
ARC2 domain3275410.0150.0530.006
LRR domain2397231801510.0960.1290.087
Solvent-exposed residues40216460.0960.1430.157

Sequence comparison of the LRR domains

It was shown previously that LRR6 and LRR7 in the LRR domain encoded by the Pm3 alleles are nearly identical in sequence and therefore result from a duplication event (Yahiaoui et al., 2004; Wicker et al., 2007a). This LRR duplication was also found to be present in a Pm3-like gene isolated from the D genome progenitor of wheat, Ae. tauschii (Figures 6 and S2). However, it did not occur in several other Pm3-like genes from T. aestivum, T. monococcum, T. turgidum, barley and rice (Wicker et al., 2007a), nor in Pm8 (Figure S2). Therefore, this duplication event must have occurred after the speciation of wheat and rye about 7.5 million years ago and most likely before the three wheat diploid ancestor genomes separated about 2.5 million years ago (Huang et al., 2002a; Chalupska et al., 2008).

Figure 6.

PM8 groups outside the wheat PM3 proteins. Neighbour-joining phylogenetic tree of full-length proteins of PM3CS, PM3A-PM3G, PM8, PM3-1B, AetPM3 and HvPM3. The complete deletion option was chosen to eliminate gaps. The percentages of 1000 bootstrap replicates are depicted and the bar indicates 1% dissimilarity. On the right, the domain structure of the proteins is given and highly conserved or identical sequence blocks are shaded. The two arrows indicate the presence of the leucine-rich-repeat(LRR)6/LRR7 duplication.

Amino acid changes in the LRR domain encoded by the 17 functional Pm3 alleles are mainly found in the predicted solvent-exposed residues close to the C-terminal end of the LRR domain and to a lesser extent at the N-terminal end and in polymorphic blocks in the middle of the domain (Yahiaoui et al., 2006; Bhullar et al., 2010; Brunner et al., 2010). Furthermore, Brunner et al. (2010) showed that polymorphic amino acids at both ends of the LRR domain are necessary for pathogen recognition specificity. Comparison of PM8 with PM3B revealed that the number of non-synonymous changes (46) in the solvent-exposed residues clearly exceeds the number of synonymous changes (16) (Table 1) and that there is an accumulation of polymorphisms found in the solvent-exposed residues of LRRs 20–28, while in LRRs 1–19 polymorphisms in solvent-exposed and non-solvent-exposed residues are present in a similar ratio (Figures 7 and S4). Thus, amino acid changes in solvent-exposed residues are concentrated in the most C-terminal region of the LRR domain if the functional PM3 proteins are compared with each other as well as in a comparison of PM3 and PM8 proteins. This strongly suggests that evolutionary forces acted in both proteins on this particular part of the LRR domain.

Figure 7.

Amino acid changes occur prevalently in solvent-exposed residues in the leucine-rich repeats (LRRs) 20–28. Green columns represent the number of amino acid polymorphisms among the five solvent-exposed residues of the LxxLxLxx motif for each LRR. The percentage of amino acid differences between PM8 and PM3B in the remaining positions is indicated in orange for each LRR.

Pm8 and Pm3-like genes from different grass species are a complex mosaic of common ancient haplotypes

Among the total of 54 Pm3 alleles (functional or non-functional in resistance) known so far in wheat, Pm3a and Pm3b share a unique ARC2 subdomain which is very different from the ARC2 consensus sequence (Yahiaoui et al., 2006; Bhullar et al., 2010). Interestingly, Pm8 has exactly the same ARC2 sequence (Figures S2 and S5a). We searched for this sequence block in Pm3-like genes from barley, Brachypodium, rice, wheat and Ae. tauschii (Wicker et al., 2007b; Vogel et al., 2010; Mayer et al., 2012; Jia et al., 2013). We found one PM3-like protein from Ae. tauschii (AetPM3) carrying this sequence block in the ARC2 subdomain (Figure S5a). Furthermore, one PM3-like protein from H. vulgare (HvPM3) and one protein cloned from the 1BS chromosome of hexaploid wheat (PM3-1B) had a very similar haplotype in the first half of the ARC2 subdomain (Figure S5a). The haplotype could not be found in any of the five full-length Pm3 resistance gene analogues (RGAs) found in T. monococcum, T. aestivum and T. turgidum, nor in rice (Wicker et al., 2007b). We conclude that the specific ARC2 domain is of ancient origin and was already present in the gene pool of the Triticeae ancestor.

While the CC domains of PM8 and PM3 are highly conserved, none of the PM3-like proteins PM3-1B, AetPM3 and HvPM3 showed high sequence conservation in this domain compared to PM3 and PM8 (Figures 6 and S2). In contrast, the Ae. tauschii sequence (AetPM3) shares an identical 67 amino acid long region with the PM3 alleles but not PM8 at the end of the NB domain (Figures 6, S2 and S5b). Therefore, PM8 shares with AetPM3 only the unique ARC2 haplotype, but not a conserved CC nor the NB domain and the duplication of LRR6/7. In the LRR region, no conserved sequence blocks between PM8 and any of the other above analysed sequences, or any of the 54 known Pm3 alleles, were found. These data support a model of PM8 evolution where a reshuffling of genes present at the Pm3/Pm8 orthologous loci occurred before wheat and rye diverged (Figure 6). We conclude that a substantial part or most of the sequence variability in these two active resistance genes in modern wheat and rye genotypes was already present as sequence segments in the ancestors of Triticeae species and have been reused over an evolutionarily long time period.


It is known that the activity of a number of resistance genes is actually based on two independent genes which must act together to confer resistance (Eitas and Dangl, 2010). As chromosomal translocations carry hundreds of genes, it is not known if the resistances conferred by them are based on a single gene or on two or even more genes. Based on gene isolation and subsequent analysis of transient or stable transformation experiments we could show in three different genetic backgrounds that a single gene from the 1RS translocation is sufficient to confer Pm8 race-specific resistance.

Unique haplotype sequence reveals orthology

Our study shows that Pm3 and Pm8 represent a pair of orthologous genes with a conserved resistance gene function against powdery mildew and demonstrate a high evolutionary conservation which must involve both host and pathogen components. It is generally difficult to demonstrate orthologous relationships of genes if they are members of gene families, most particularly those in gene clusters. Gene loss can result in deep paralogues which are then misinterpreted as orthologues (Wicker et al., 2007b). In addition, gene families in different species can evolve very differently based on unequal recombination and gene conversion, making the definition of orthologous genes impossible. These problems have mostly prevented the identification of clear resistance gene orthologues between different species in earlier studies (Takahashi et al., 2002; Li et al., 2012). Orthology between Pm3 and Pm8 was strongly suggested by genetic and physical mapping, but could ultimately be established by the unique haplotype conservation of the UP3 upstream sequence which is a single copy sequence in the wheat A and B genomes. It is likely that other known or yet to be identified Pm3 orthologues also function as powdery mildew resistance genes, but such a function remains to be demonstrated.

Sequence conservation suggests recognition of the same effector class by Pm8 and Pm3

The highest sequence diversity in PM8 compared with PM3 was found in the solvent-exposed residues of the LRR domain at the C-terminal end. This is similar to earlier findings in other plant resistance genes: For several NB-LRR resistance genes it was shown that the solvent-exposed residues of the LRR domain are under diversifying selection (Parniske et al., 1997; Meyers et al., 1998; Ellis et al., 1999; Rose et al., 2004; Yahiaoui et al., 2006; Seeholzer et al., 2010) and the LRR domain has been shown in several cases to interact with pathogen effectors (Ellis et al., 2000; Bergelson et al., 2001). Solvent-exposed residues were predicted to directly interact with pathogen avirulence proteins (Jones and Jones, 1997; Takken and Goverse, 2012). Recently, it was shown that positively selected sites are clustered at the N- and C-terminal ends of the LRR domain of the flax L resistance alleles and that these sites are required for interaction with AvrL567 (Ravensdale et al., 2012). Furthermore, race specificity was found to be determined in polymorphisms throughout the LRR domain in the flax resistance alleles L6 and L11 (Ellis et al., 2007). For PM3, we have previously shown that polymorphic amino acids reside mainly in the solvent-exposed residues of the C-terminal LRR domain and that they play a crucial role in pathogen recognition specificity (Brunner et al., 2010). For PM8, an accumulation of polymorphisms was found in the solvent-exposed residues mainly at C-terminal LRRs in comparison with PM3 proteins. Therefore, the LRR domain of PM8 and PM3 must have been subjected to a very similar evolutionary pressure in both rye and wheat. Given that the overall sequence is very well conserved between PM3 and PM8, it is likely that the C-terminal part of the LRR domain is very important for pathogen recognition specificity. Possibly, the same protein is recognized by PM3 and PM8. This would also be in agreement with the data showing that there is a certain overlap in race specificity between Pm8 and Pm3 alleles. However, the resistance activity of Pm8 in wheat demonstrates that an effector from wheat powdery mildew is recognized, be it directly or indirectly. Thus, although effector complements have been found to be highly specific and unique for a given pathogen species (Baxter et al., 2010), very similar effectors may have been conserved in mildews over more than 7 million years. This might indicate that effectors detected by PM8 and PM3 are important for mildews and are not easily lost. So far, no effector or avirulence genes have been cloned for Bgt or Bgs. It will be interesting to study and compare the molecular interactions and downstream signalling pathways of Pm3 and Pm8.

Evolution of Pm8 by the reshuffling of sequence blocks

There are many examples of R gene evolution that suggest a complex evolutionary history involving the reshuffling of sequence blocks among homologous genes (Meyers et al., 2005). It is known from a genome-wide study of NBS-LRR genes in Arabidopsis thaliana that the diversity of this gene family was generated by extensive duplication and ectopic rearrangement events (Meyers et al., 2003). The CC domain of wheat TmMLA1 shows sequence homology to two barley HvMLA subfamilies while the LRR domain is distinct from both families (Jordan et al., 2011). The R gene N′ from Nicotiana sylvestris is closely related in DNA sequence to I2 and R3a in the NB domain but closer to its Capsicum orthologue L in the C-terminal half of the LRR domain, suggesting a complex evolution by recombination and gene conversion (Sekine et al., 2012). Comparison of the Pm3 loci from diploid, tetraploid and hexaploid wheat revealed low sequence conservation and extensive rearrangements of the Pm3-like genes and their up- and downstream sequences (Wicker et al., 2007b).

The ARC2 sequence unique for Pm3a and Pm3b among the Pm3 alleles is present in Pm8 and furthermore in a Pm3-like gene from Ae. tauschii, AetPm3. This sequence block shows a high number of synonymous mutations when compared with the wheat reference allele Pm3CS (Yahiaoui et al., 2006), indicating an ancient evolutionary origin. This sequence block must have evolved before wheat and rye separated. Interestingly, the PM3A/PM3B specific ARC2 domain has been shown to enhance resistance gene function and might be retained for this reason during this long time period (Brunner et al., 2010). The duplication of LRR6/7 which is present in the Pm3 alleles was also found in AetPM3, but is absent in Pm8. Considering that both Pm8 and the Pm3 alleles are functional R genes, this duplication does not seem to affect gene function. The observation of highly characteristic segments of Pm8 and Pm3 genes in several Triticeae species indicates that there is possibly a quite limited, natural haplotype diversity leading to functional R genes. It remains to be seen if approaches based on artificial resistance genes can successfully broaden useful sequence diversity for breeding (Farnham and Baulcombe, 2006; Brunner et al., 2010).

Functional conservation of Pm8 and Pm3 gene function after host-species divergence

The ancestral gene of Pm8 and Pm3 must have been already present approximately 7.5 million years ago (Huang et al., 2002a,b) before wheat and rye diverged. Given the common function of Pm8 and Pm3, the ancestral gene was possibly already an active resistance gene against powdery mildew at that time. Consequently, the resistance function has been conserved for a very long time period. In wheat, the Pm3 gene diverged to an allelic series after its domestication (Yahiaoui et al., 2006). In contrast, in rye only one gene, Pm17, has been suggested to be a Pm8 allele based on progeny testing of F2 and F3 plants (Hsam and Zeller, 1997). While the race specificity of the Pm3 alleles to wheat powdery mildew was thoroughly studied in Pm3 differential lines, less is known about the function of Pm8 in rye. In agricultural ecosystems, major resistance genes introduced into an elite cultivar used by farmers are frequently only effective for a few years, as was also experienced for Pm8 (Bennett, 1984). This rapid breakdown of the Pm8 resistance gene in the agricultural environment most likely reflects directional selection of virulence in the pathogen population due to a high selection pressure generated by the high abundance of the R gene. In contrast, the long life of the Pm8 gene in natural ecosystems indicates balancing selection, where resistant and susceptible alleles are maintained over long periods of time (Stahl et al., 1999; Tiffin and Moeller, 2006; Brown and Tellier, 2011). Further studies which will identify the effector and avirulence genes from powdery mildew are needed to clarify the molecular nature of selection.

Experimental procedures

Southern blot analysis

Isolation of genomic DNA from leaf material and Southern hybridization were performed as described (Stein et al., 2001; Travella et al., 2006). For analysis with the probe UP3, genomic DNA was digested with the restriction enzyme HindIII. Probe UP3 was amplified by PCR with the primers UP3B (5′-TGGTTGCACAGACAATCC-3′) and UP3A (5′-GACAAATGTGGCGTTTGC-3′).

Amplification of Pm8 by nested PCR and sequencing

The coding region of Pm8 was amplified in a two-step nested PCR. A first PCR was carried out on the wheat line Kavkaz/4*Federation and rye line Petkus 91 using the primers SH32 (5′-TGCCGACCAGGCTTTGAATC-3′) and N3'SP3R (5′-ACAATCAGGGATCAGGCC-3′). On the obtained PCR products, a nested PCR was performed with primer pair SH33 (5′-TTAATTGGATCCCAATGGCAGAGCTGGTGGTC-3′) and Sl_1 (5′-TATATAGTCGACGCTTCAGCTCCGGCAGGCCTG-3′) adding a BamHI and a SalI restriction site, respectively. For all PCR reactions, the Herculase-II fusion high-fidelity DNA polymerase (600675; Agilent Technologies, http://www.chem.agilent.com/) was used.

Mapping of Pm8

For the specific detection of Pm8, the primer pair SH43 (5′-TGGCTTCCAACAGCCCTAGC-3′) and SH46 (5′-AGGCTTTTGCACCTTCTCTC-3′) was used and designated as marker sfr43(Pm8). Polymerase chain reaction was performed in a total volume of 25 μl with 0.05 units/μl Taq DNA polymerase (D1806; Sigma-Aldrich, http://www.sigmaaldrich.com/) and an annealing temperature of 65°C for 30 s.

The results of the marker sfr43(Pm8) on the F2 mapping population ‘Federation*4/Kavkaz’ x ‘King II-derivative’ were integrated into an existing 1RS genetic map using mapmaker Version 2.0 and the Kosambi mapping function for converting recombination frequencies into centimorgans.

Single-cell transient expression assay

The coding region of Pm8 was amplified with the primer pair SH33/Sl_1 and ligated into the multiple cloning site of the PGY1 vector (Schweizer et al., 1999) under the control of the constitutively expressing 35S CaMV promoter. Seedlings of Chancellor and Federation were bombarded with 3-mg gold particles coated with 1.3 μg of the plasmid construct pGY1-Pm8 or the empty vector control plasmid pGY1 per shot together with 1.3 μg of the reporter plasmid pUbi-GUS. The significance of the differences between the HIs was analysed by Student's t-test. For a more detailed description see Methods S1.

Wheat transformation and analysis of transgenic Pm8 plants

The entire 4.4-kb coding region of Pm8 was amplified from the plasmid construct pGY1-Pm8 using primers SH033 and TJ064 (5′-CATCATGGATCCTCACAAATCTTCTTCAGAAATCAACTTTTGTTCGCTCCGGCAGGCCTGCCTCCGC-3′) thereby adding a myc-epitop tag at the 3′ end of the gene and cloned into the pAHC17 vector (Christensen and Quail, 1996). As a selectable marker, the phosphomannose isomerase gene was used (Reed et al., 2001). Transgenic plants were produced by particle bombardment of immature wheat embryos of the cultivar Bobwhite SH 98 26 essentially as described in (Brunner et al., 2011).

Quantitative real-time PCR analysis for detection of Pm8 expression

Expression of Pm8 was quantified in a reverse transcription, quantitative real-time polymerase chain reaction (RT-qPCR) assay using a CFX96 Real-Time System C1000TM Thermal cycler (Bio-Rad, http://www.bio-rad.com/). Per wheat or rye line, technical triplicates of three biological replicates each (two for rye) were analysed. Each biological replicate consisted of three pooled first leaves of 12-day-old plants. Three reference genes (ADP, RLIL, TA.6863) were included which match both the rye and wheat sequences (Table S3). Data analysis was performed using the statistical package jmp version 9.0 (SAS Institute, http://www.sas.com/). For a more detailed description see Methods S1.

Sequence analysis

The DNA sequences were analysed with Clone Manager Professional Suite version 8 (Sci-Ed Software, http://www.scied.com/). Sequence alignments were performed with mega5 (Tamura et al., 2011) and corrected manually. Nucleotide diversity was calculated with DnaSP version 5.0 (Librado and Rozas, 2009) and phylogenetic analysis was done with mega5.


We thank Roi Ben-David for statistical help, A. J. Lukaszewski for providing seeds of the wheat–rye recombinant lines and Christina Cowger for supplying us with US mildew isolates. This project was financially supported by the Swiss National Science Foundation (310030B-144081), an Advanced Investigator Grant of the European Research Council (ERC-2009-AdG 249996, Durableresistance) and the Indo-Swiss Collaboration in Biotechnology.