Logo of geneticsGeneticsCurrent IssueInformation for AuthorsEditorial BoardSubscribeSubmit a Manuscript
Genetics. Oct 2007; 177(2): 1207–1216.
PMCID: PMC2034624

Contrasting Rates of Evolution in Pm3 Loci From Three Wheat Species and Rice


The Pm3 gene from wheat confers resistance against powdery mildew and recent studies have shown that it is a member of a multigene family in the wheat genome. We compared genomic sequences ranging from 178 to 332 kb containing six Pm3-like genes and five gene fragments from orthologous loci in the A genome of wheat at three different ploidy levels. We found that the wheat Pm3 loci display an extremely dynamic evolution where sequence conservation is minimal between species and basically limited to very short sequences containing the genetic markers that define the orthology. The Pm3-like genes and their up- and downstream regions were reshuffled by multiple rearrangements, resulting in a complex mosaic of conserved and unique sequences. Comparison with rice showed that the known wheat Pm3-like genes represent only one branch of a large superfamily with several clusters in rice and suggests the presence of additional similar genes in the wheat genome. Estimates of divergence times and transposable-element insertions indicate that the Pm3 locus in wheat has undergone more drastic changes in its recent evolution than its counterpart in rice. This indicates that loci containing homologous resistance gene analogs can evolve at highly variable speeds in different species.

AS a defense system against pathogens, plant genomes contain resistance genes and resistance gene analogs (RGAs). Resistance gene products directly or indirectly detect specific pathogen proteins and are able to trigger a defense reaction (reviewed by Jones and Dangl 2006). RGAs are characterized by their sequence similarity to known resistance genes but have no known function and many of them appear to be pseudogenes. Most RGAs described are of the nucelotide binding site–leucine-rich repeats (NBS–LRR) type and encode proteins with a NBS domain, followed by a series of LRRs. The NBS domain is probably involved in signal transduction, whereas the LRR domain is thought to be largely responsible for specificity of pathogen recognition (Jones and Jones 1997).

Genomewide surveys in Arabidopsis and rice showed that RGAs are often found in clusters where genes from the same family are arranged in the same orientation with no other genes between them (Meyers et al. 2003; Zhou et al. 2004). RGA loci are known to evolve rapidly and several mechanisms were discovered that help explain their dynamics. The “birth-and-death” model (Michelmore and Meyers 1998) describes how new types of resistance genes arise and disappear. In this model, recombination or gene conversion between orthologs and paralogs as well as mutations in solvent-exposed residues of the LRRs are sources of genetic variation.

Sequence exchange through unequal crossing over or gene conversion leading to the creation of hybrid genes and thus to genetic variability was shown in a wide range of species such as Arabidopsis (Noël et al. 1999), tomato (Van Der Hoorn et al. 2001), lettuce (Kuang et al. 2004), and maize (Smith and Hulbert 2005). Each unequal crossing over between paralogs changes the number of genes and causes differences in RGA numbers even between closely related species (Kuang et al. 2004). Frequent sequence exchange between paralogs at the Arabidopsis RPP5 locus resulted in a mosaic of shared sequences among paralogs (Noël et al. 1999). The described mechanisms of unequal crossing over, gene conversion, and mutation portray RGA clusters as “reservoirs of variation” (Michelmore and Meyers 1998), from which new combinations are permanently created by a background recombination and mutation activity.

The tribe of the Triticeae diverged from rice ~50 MYA (Paterson et al. 2004) and contains important crops such as wheat and barley. Commercially used wheat species are polyploids, which originate from hybridizations of diploid ancestor (A, B, and D) genomes. The first hybridization event in wheat brought together the two diploid wheat species Triticum urartu (A genome) and a probably extinct close relative of Aegilops speltoides (B genome) to form the tetraploid wheat T. turgidum ssp. dicoccoides (genome formula AABB). Hexaploid bread wheat T. aestivum (genome formula AABBDD) formed ~10,000 years ago through hybridization with the D genome diploid grass Ae. tauschii (Feldman et al. 1995).

The cultivated diploid wheat T. monococcum, a close relative of T. urartu (the donor of the A genome), was used in previous comparative genomics studies. Comparison of low-molecular-weight (LMW) and high-molecular-weight (HMW) glutenin loci as well as the Lr10 locus with their orthologs from tetraploid wheat (T. turgidum) and hexaploid wheat (T. aestivum) showed that genes are largely conserved whereas intergenic regions have undergone various changes (Wicker et al. 2003; Isidore et al. 2005; Gu et al. 2006). Gu et al. (2006) showed a close relationship between T. aestivum and T. turgidum (ssp. durum) with extensive conservation of intergenic regions and estimated their divergence time to ~800,000 years. This estimate is based on the divergence of intergenic DNA (mostly transposable elements), which is believed to be largely free from selection pressure (Petrov 2001). Thus, it is expected to accumulate mutations randomly at a basic rate estimated to be 6E-9 substitutions per site per year in grasses (Gaut et al. 1996). Similarly, the divergence of T. monococcum from the T. turgidum/T. aestivum lineage was estimated to have occurred ~2.6–3 MYA (Wicker et al. 2003; Isidore et al. 2005). However, a recent study recommended an approximately twofold higher basic substitution rate of 1.3E-8 (Ma and Bennetzen 2004), which shifts the previous estimates to more recent times of 0.4 and 1.3–1.5 million years, respectively.

Due to the large genomes and high content of repetitive DNA of wheat species, map-based cloning of resistance genes is very labor-intensive and time-consuming (reviewed by Keller et al. 2005). The recent cloning of the Pm3 gene from hexaploid wheat (Yahiaoui et al. 2004, 2006) provided a wealth of genetic and physical mapping data on the population of Pm3-like genes in wheat. Active Pm3 alleles conferring resistance against powdery mildew are only found in a few wheat cultivars and landraces. However, all wheat lines analyzed by Southern hybridization contain at least 10–15 Pm3-like sequences each (Yahiaoui et al. 2004), which could be pseudogenes or active genes.

Here we present a comparative analysis of three orthologous loci containing a total of six Pm3-like genes and five gene fragments in the wheat A genome at three different ploidy levels. We found that the Pm3 loci went through an extremely dynamic evolution, resulting in minimal sequence conservation, which is basically limited to a few short sequences between wheat species. Comparison with rice showed that the known wheat Pm3-like genes represent only one branch of a large superfamily with several clusters in rice and that the Pm3 locus in wheat has evolved more rapidly than its homolog in rice.


Shotgun sequencing:

BAC DNA was isolated with the QIAGEN (Valencia, CA) large construct kit, mechanically sheared into fragments of 3–5 kb in a Hydroshear (GeneMachines), and ligated into the Topo-Blunt vector (Invitrogen, Carlsbad, CA). For transformation, electrocompetent DH-10B E. coli were used. Plasmid DNA was isolated in a 96-well format on a QIAROBOT (QIAGEN) and sequenced on an ABI3730 automated sequencer (Applied Biosystems, Foster City, CA). Base calling and quality trimming of the sequences were done using PHRED (Ewing et al. 1998) and the initial assembly of BAC sequences was done with the PHRAP assembly engine (version 0.990319, provided by P. Green and available at http://www.phrap.org). Gaps in the BAC sequences were closed by resequencing shotgun clones spanning the gaps, with 1 m betaine added to the sequencing reaction. Alternatively, gaps were closed by primer walking on shotgun clones or by PCR amplication of fragments from BAC DNA. BAC clones from hexaploid wheat cultivar Chinese Spring (Allouis et al. 2003) were provided by the joint consortium of the John Innes Centre, the Biotechnology and Biological Research Council, and Unité de Recherches en Génomique Végétale (INRA).

Sequence analysis:

For sequence analysis, programs from the EMBOSS package (http://emboss.sourceforge.net/), CLUSTAL W (Thompson et al. 1994), and DOTTER (Sonnhammer and Durbin 1995) were used. In a first step, all known repetitive elements were identified through BLAST (Altschul et al. 1997) against the database for Triticeae repetitive elements (TREP) (Wicker et al. 2002, http://wheat.pw.usda.gov/ITMI/Repeats) and annotated. The remaining sequence, not annotated, was screened for the presence of putative genes by BLASTX against all rice and Arabidopsis proteins and BLASTN against all Triticeae ESTs. RGAs were identified by similarity to published Pm3 gene sequences. The structure of the coding sequence was determined using DOTTER. Frameshifts in pseudogenes were also identified with the DOTTER. Identified repetitive elements were submitted to the TREP database. For the processing of large data sets Perl programs were written.

The complete sequence of the rice genome (ssp. japonica) was obtained from 12 pseudomolecules (version 3) from TIGR (http://www.tigr.org). Sequences for O. sativa ssp. indica were obtained from (ftp://ftp.genomics.org.cn/pub/ricedb/SynVs9311/9311/Sequence/Chromosome/). Homologs of Triticeae RGAs in rice were identified by BLASTN. Only loci producing alignments >70 bp were examined further. A program for sliding-window analysis was written in Perl. The alignment used as input was produced with the program WATER (EMBOSS package) using a gap creation penalty of 30.0 and a gap extension penalty of 0.1 for all alignments. Only aligned base pairs or amino acids were used to calculate sequence identity whereas gaps were ignored. Visual representation was done using the Perl Tk module.

Phylogenetic analysis of Pm3-like genes from rice and wheat was done with the PHYLIP package (http://evolution.genetics.washington.edu/phylip/). A multiple alignment was produced with CLUSTAL W using a gap creation penalty of 4.0 and a gap extension penalty of 0.01. The multiple alignment was used for bootstrapping with SEQBOOT, doing 100 bootstrap repetitions. A series of possible trees (jumbling the order of sequences three times for each tree) was produced with PROTPARS. The consensus tree was generated with CONSENSE.

Defining orthology at Pm3 loci of three wheat species:

To estimate what degree of DNA sequence identity can be expected from orthologous regions between wheat species, we reanalyzed previously published sequences from orthologous wheat loci. The sequences from HMW glutenin loci (Gu et al. 2006) provided a large sample of repetitive elements conserved between T. aestivum and T. turgidum. Comparison of four retrotransposons with a cumulative size of 35.8 kb showed an overall sequence identity of 98.9% between T. aestivum and T. turgidum. Comparison of the H1 haplotype Lr10 loci from T. monococcum and T. turgidum (Isidore et al. 2005) provided 18.3 kb of orthologous repetitive sequences and revealed an overall sequence identity of 96.6%. Similarly, at the Pm3 locus, the ~9-kb region of intergenic sequences conserved between T. monococcum and T. turgidum showed 96.9% sequence identity (region ε, Figure 1, Wicker et al. 2003). Therefore, we defined as orthologous all noncoding sequences that are 98–99% identical between T. turgidum and T. aestivum and 96–97% identical between T. monococcum and either T. turgidum or T. aestivum.

Figure 1.
Physical maps of orthologous Pm3 loci from T. monococcum, T. aestivum, and T. turgidum. Genes are labeled below the maps with arrows indicating transcriptional orientation. RFLP markers used in genetic mapping are indicated above the maps. Orthologous ...


Pm3 loci in wheat species from three different ploidy levels:

To investigate the genomic context and evolution of Pm3-like genes in the wheat gene pool, we sequenced BAC clones from the Pm3 locus of tetraploid and hexaploid wheat that were isolated from the respective genomic libraries using the RFLP markers TmRGL-1pro, SFR159, and 294D11-31. These markers are highly specific for wheat chromosome 1A and are closely linked to Pm3 in a hexaploid wheat mapping population (Yahiaoui et al. 2004). Two overlapping BACs from T. aestivum (cv. Chinese Spring) with a combined length of 178 kb were sequenced, and a previously described 142-kb sequence of T. turgidum BAC 107G22 (Wicker et al. 2003) was extended to 332 kb by sequencing two additional overlapping BAC clones. The two sequences were compared with each other and with a previously described 284-kb sequence from T. monococcum (Wicker et al. 2003).

The three genomic regions have similar contents of repetitive DNA (70–80%), constituting a representative sample of all major types of transposable elements. All three sequences have a relatively high content of CACTA transposons as they contain four to nine copies each (Figure 1). Genes were found interspersed between large blocks of repetitive DNA (Figure 1). Besides the previously described orthologous regions surrounding the LMW glutenin genes (region ε, Figure 1, Wicker et al. 2003), only very few additional conserved stretches could be identified. These are basically limited to small regions containing the RFLP markers TmRGL-1pro, SFR159, and 294D11-31 used earlier for the BAC screening plus a few other short stretches (Figure 1). These markers (TmRGL-1pro, SFR159, and 294D11-31) have previously been shown by genetic mapping to specifically identify Pm3 orthologous loci on chromosome 1A in the three wheat species. TmRGL-1pro and 294D11-31 are single-copy markers (Yahiaoui et al. 2004) whereas SFR159 is single copy in T. turgidum and T. aestivum but has three copies in T. monococcum. This is due to a local multiplication that leads to three tandemly repeated paralogous loci in T. monococcum (Wicker et al. 2003). Two of them are covered by the sequenced BACs (Figure 1).

Nevertheless, the fact that so few stretches are conserved compelled us to find additional evidence that they are indeed orthologs and not simply conserved motifs that are found in this region by chance. Therefore, we also required putative orthologous regions to have the same degree of DNA sequence conservation as orthologous loci described in previous studies (for details, see materials and methods). Thus, for this study, we defined as orthologous all sequences that are 98–99% identical between T. turgidum and T. aestivum and 96–97% identical between T. monococcum and either T. turgidum or T. aestivum.

Using these criteria, two regions totaling 3.9 kb were found to be orthologous between T. monococcum and T. aestivum (regions α and β, Figure 1), a 10.9-kb sequence is orthologous between T. monococcum and T. turgidum (region ε, Figure 1), and two stretches totaling 5.6 kb are orthologous between T. turgidum and T. aestivum (regions γ and δ, Figure 1). It should be noted here that the assumption was made that the putative orthologous regions are free from selection pressure and thus diverge at the basic substitution rate of 1.3E-8 substitutions per site per year (Ma and Bennetzen 2004). Indeed, most sequences we found to be orthologous are noncoding or consist of known repetitive elements. Only regions δ and ε (Figure 1) contain some coding sequences. However, no difference in sequence conservation was observed between noncoding and coding sequences in these two regions.

Sequence organization of Pm3-like genes:

The three wheat species from which Pm3 loci were analyzed do not have functional Pm3 genes conferring resistance against powdery mildew. The TaPm3CS gene from T. aestivum (cv. Chinese Spring) is allelic to active Pm3 genes in other wheat lines and differs from some of them by only a few nucleotides (>99% sequence identity, Yahiaoui et al. 2006), whereas all the other Pm3-like genes from the regions analyzed are at most 92.8% identical with active Pm3 genes (see below). Thus, no Pm3-like genes found in the three species are orthologous according to the requirement defined above.

The genomic regions sequenced from the three species contain six Pm3-like RGAs and five RGA fragments (Figure 2). For the characterization of the identified RGAs, we use the term “full-size gene” for genes encoding both NBS and LRR domains. The partial RGAs are 3′ fragments covering only parts of the LRR domain (Figure 2). Three full-size RGAs (TdRGL-1, TdRGL-2, and TdRGL-3) and one partial RGA (TdLRR-1) gene were identified on the T. turgidum sequence. The T. aestivum sequence contains two full-size (TaPm3CS and TaRGL-9) and three partial RGAs (TaLRR-1, TaLRR-2, and TaLRR-3), whereas the T. monococcum sequence contains only one full-size RGA (TmRGL-1).

Figure 2.
Sequence organization of Pm3-like genes from wheat. Positions of protein domains along the coding sequence are indicated at the top. Coding sequences are depicted as shaded boxes and the intron close to the 3′ end is indicated with a solid horizontal ...

The full-size genes encode an NBS domain of ~600 amino acids, which is followed by 26–29 LRR units. They are composed of two exons with one intron of 85–200 bp at a conserved position close to the 3′ end. The stop codon is located 86–215 bp downstream of the intron. The coding region of the second exon encodes the highly variable C terminus of the protein. The six full-size genes encode hypothetical proteins ranging in size from 1286 amino acids (TmRGL-1) to 1549 amino acids (TdRGL-3). The differences in size are due to duplications in the LRR region and an ~220-bp deletion in TmRGL-1 (Figure 2).

Three of the six full-length RGAs are probably pseudogenes since they contain frameshifts or in-frame stop codons that interrupt the open reading frame (Figure 2). Most of those mutations appear to be recent since open reading frames of the RGA pseudogenes are still largely intact and do not show typical signs of a long period of degeneration (e.g., large deletions or insertions of transposable elements). Only the TmRGL-1 gene from T. monococcum appears to be strongly degenerated as it contains one major deletion, three frameshifts, and four in-frame stop codons (Figure 2). For subsequent analysis, frameshifts were removed from the coding sequences of the pseudogenes to obtain hypothetical protein sequences for all RGAs and partial genes.

At the DNA level, the Pm3-like genes are 78–92.8% identical to each other and the hypothetical proteins are 74–92% similar to each other. TaPm3CS and its next neighbor, the partial gene TaLRR-1, are most closely related, whereas TmRGL-1 and TaRGL-9 are most distantly related. Pairwise comparison of all RGAs showed that sequence identity in the intron is usually 5–10% lower than in coding regions. No evidence was found for sequence homogenization of the intron as described by Kuang et al. (2004). Indeed, some introns have diverged to a degree that no reliable alignments with introns from other RGAs were possible.

The Pm3-like genes are embedded in a mosaic of conserved and unique upstream and downstream sequences:

Comparison of the genomic sequences from the three species showed that, besides the orthologous regions, some other motifs are conserved in the up- and downstream regions of the RGAs. However, most of these conserved motifs showed a lower degree of DNA sequence identity (75–95%) than the orthologous regions, indicating that these sequences diverged at different times during evolution. Most of these conserved motifs were found within 10 kb up- or downstream of the respective genes. Thus, we isolated (in silico) enough flanking sequences for all genes that, after removal of transposable elements unique to the respective genes, at least 10 kb of flanking sequence were left to compare (Figure 3). TmRGL-1 was the only gene for which conserved regions were identified more than 10 kb upstream of the gene, which is why we used 20 kb of upstream region for our analysis (Figure 3). Additionally, for TmRGL-1, only ~4 kb of downstream sequence are covered by the BAC clone sequenced. The comparison revealed that the up- and downstream regions of Pm3-like genes are composed of a mosaic of unique sequences mixed in with motifs conserved between different genes (Figure 3).

Figure 3.
Comparison of Pm3-like genes from wheat and their upstream and downstream regions. The coding regions of the genes and gene fragments are indicated by shaded boxes and their transcriptional orientation is indicated by arrows. Colored boxes indicate regions ...

The TaPm3CS gene contains the largest amount of conserved sequences in its flanking regions. Much of its upstream region is orthologous (i.e., 96% identical) to the T. monococcum TmRGL1 locus whereas a large region downstream is orthologous (i.e., 98.8% identical) to TdRGL2 from T. turgidum. Interestingly, the TaPm3CS gene itself shows an overall sequence identity of only ~85% to each of the other two genes. The most divergent locus is the one that contains the partial LRR sequence TdLRR1 in T. turgidum, as it shares no flanking sequence motifs with any of the other Pm3-like genes.

The Pm3 loci from T. aestivum and T. monococcum evolved in very different ways:

Comparison of the region upstream of TaPm3CS from T. aestivum and TmRGL1 from T. monococcum showed that up- and downstream regions of both genes have undergone a multitude of changes since the loci diverged. The most striking finding was that an ~10-kb region containing fragments of a pseudogene (TmPIK1) upstream of TmRGL1 appeared to be completely removed and replaced by a block of nested transposable elements in T. aestivum (Figure 4A ). Upon closer inspection, it became clear that the apparent replacement of genic with repetitive sequences must be the product of an interelement crossing over (Figure 4B).

Figure 4.
Evolution of orthologous regions upstream of Pm3-like genes from T. aestivum and T. monococcum. (A) Comparison of T. aestivum (top) and T. monococcum (bottom). Orthologous regions are indicated by turquoise areas that correspond to regions α and ...

A model for the sequence of events following the divergence of the two loci is depicted in Figure 4B. After divergence, two gypsy retrotransposons of the Jeli type inserted independently to the left and the right of the region, containing parts of the PIK gene in T. aestivum. Subsequently, an unequal crossing over occurred between the two Jeli elements, resulting in a hybrid element and eliminating the region between them. Evidence for that scenario is that the Jeli element, although complete, is not flanked by a typical 5-bp target-site duplication. Additionally, the Karin-1 element into which one Jeli inserted is fragmented on one side of the Jeli element but does not continue on the other side as would be expected from a simple insertion of Jeli into Karin-1. This introduces an irregularity in the nesting pattern, since one Jeli element inserted directly in the genomic DNA (nesting level 0) whereas the second Jeli element has the nesting level 1, as it is inserted into the already present Karin-1 element. Thus, the recombined Jeli element has two different nesting levels at its two ends (Figure 4B). In T. monococcum, two deletions eliminated a fragment upstream of TmRGL-1 and parts of a HORGY-1 retrotransposon (Figure 4B).

We estimated the time of divergence of the two loci as well as insertion times of the three complete LTR retrotransposons upstream of TaPm3CS (Angela-3, Laura-2, and WIS-1) using the method of Sanmiguel et al. (1998) with a basic substitution rate of 1.3E-8 (Ma and Bennetzen 2004). Estimates for insertion times as well as for the divergence of conserved parts of the promoter region were consistent with the proposed sequence of events; younger retrotransposons are inserted into older ones and all three inserted after the divergence of the two loci ~1.3 MYA. For the recombined Jeli-1 element, no insertion time estimate was possible since its LTRs originate from two different copies (Figure 4B).

Pm3-like genes in rice:

Using the nucleotide sequence of TaPm3CS in a BLASTN search against the rice genome (O. sativa ssp. japonica), we identified 12 Pm3 homologs in rice. The identified Pm3 homologs in rice were the only sequences in the rice genome that showed homology at the DNA level with Pm3-like genes from wheat. They are found in three clusters on chromosomes 1, 3, and 10, respectively, as well as in a single-gene locus on chromosome 2 (Table 1). The cluster on chromosome 1 is the largest, consisting of five RGAs whereas those on chromosomes 3 and 10 contain three RGAs each. Within clusters, all identified Pm3-like RGAs are in the same transcriptional orientation and share 64–91% DNA sequence identity. RGAs of the chromosome 3 cluster are the most closely related to each other (88–91%) whereas those on chromosome 1 are the most divergent (64–84% identical). Genes from different clusters are 64–67% identical. The rice RGAs are 63–70% identical with TaPm3CS from T. aestivum. Sequence conservation between rice and wheat RGAs is limited to the coding region whereas up- and downstream regions are totally divergent.

Approximate positions of clusters of Pm3-like genes in rice (O. sativa ssp. japonica, TIGR version 3)

Coding sequences of identified rice homologs were annotated manually on the basis of alignments with wheat Pm3-like genes. Hypothetical protein sequences were deduced for all Pm3 homologs. In pseudogenes, frameshifts and in-frame stop codons were removed to obtain the hypothetical amino acid sequence.

The predicted amino acid sequences of all rice and wheat Pm3-like genes were used in a phylogenetic analysis. As an outgroup, we used the closest homolog from Arabidopsis (At3g14460), which was identified by BLASTX. At3g14460 is only ~37% identical at the protein level to Pm3-like sequences from wheat and rice. At the DNA level, At3g14460 could not be aligned with the rice or wheat sequences. The phylogenetic tree showed that Pm3-like sequences can be classified into two major groups (Figure 5A). Group 1 contains all wheat sequences, the three genes from rice chromosome 3, as well as two from chromosome 1 (OsPm3_1-4 and OsPm3_1-5) while group 2 contains genes from chromosomes 1, 2, and 10. These data indicate that divergence of main lineages of Pm3 homologs started already in the common ancestor of rice and wheat. Interestingly, the cluster on chromosome 1 contains RGAs from three different lineages, OsPm3_1-4 and -5 (group 1), OsPm3_1-1 and -2 (group 2), as well as the most divergent rice gene OsPm3_1-3 (Figure 5B). This suggests the cluster on chromosome 1 to be the ancestor cluster that gave rise to the other loci possibly through gene movement of single genes from chromosome 1 to other chromosomes where the moved genes were duplicated and gave rise to new clusters.

Figure 5.
Pm3-like genes in rice. (A) Phylogenetic analysis of Pm3-like genes from rice and wheat. Rice sequences have the prefix “OsPm3” followed by an underscore and a number indicating the chromosome on which the gene was found. The individual ...

Gene loss through unequal crossing over in rice:

To study the recent evolution of Pm3 genes in rice, we compared the clusters of Pm3-like genes from the two rice subspecies indica and japonica (Figure 6). Due to the unfinished nature of the indica rice genome sequence, only the cluster on chromosome 3 had sufficiently high sequence quality to allow a comparison.

Figure 6.
Comparison of orthologous Pm3-like RGA loci on chromosome 3 of O. sativa ssp. japonica and O. sativa ssp. indica. (A) Comparison of the region containing the RGA cluster. Conserved regions are indicated by shaded areas (RGA1, -2, and -3 from ssp. japonica ...

We compared the sequences containing the Pm3-like RGAs plus several kilobases of the flanking region from both subspecies (O. sativa ssp. japonica chromosome 3, positions 35,590,000–35,700,000, and ssp. indica chromosome 3, positions 40,020,000–40,085,000). Sequence alignment of the two loci showed that over a large distance, the region containing several genes is highly conserved in both species with 99.3–99.4% sequence identity. The only exception is a copia retrotransposon insertion in japonica that is not present in indica (Figure 6A).

The major difference, however, is that indica contains only one RGA on chromosome 3 whereas japonica contains three. The single gene in indica appears to originate from two unequal crossing-over events that combined the sequences of three different genes into one. Breakpoints of putative unequal crossing-over events were identified by a sudden decrease in sequence conservation when the aligned sequences were compared in sliding windows of 200 bp (Figure 6B). The indica gene combines the 5′ end plus an extended upstream region of the first japonica RGA with the 3′ end plus downstream regions of the third japonica gene. Interestingly, the central region of the gene is only ~90% identical to any of the japonica genes (Figure 6B). This suggests that the particular RGA that provided this region was eliminated from the japonica genome. Additionally, these data show that 90% sequence identity is sufficient for an unequal crossing-over event. Unequal crossing over between RGAs was described previously, however, between more closely related sequences (Noël et al. 1999).


Our analysis of Pm3-like genes from wheat and rice revealed a multitude of highly informative details about the evolution of this gene family. We could demonstrate a complete reshuffling of the Pm3 locus in wheat within a relatively short evolutionary period. The observed complex mosaic pattern and an almost complete disruption of orthology made it virtually impossible to assess precise evolutionary relationships between genes or other genomic regions. Thus, conclusions based solely on genetic mapping of RGA clusters even in closely related species must be treated with caution since terms such as “ortholog” and “paralog” may become very problematic if not utterly meaningless. On the other hand, we found that the Pm3 family of RGA is highly conserved in plants at least since before the divergence of rice and wheat 50 MYA (Paterson et al. 2004). This finding is intriguing as it demonstrates a possible diversification of functions of RGAs. Although the function of the Pm3-like genes from rice is unknown, it must differ from that in wheat because there is no powdery mildew pathogen in rice.

The Pm3 loci in wheat are gene poor and rich in CACTA transposons:

The sequences studied have a content of repetitive DNA that is comparable to the one found in most other large genomic regions from Triticeae. The repeat content ranges from 70% in T. turgidum and T. aestivum to slightly over 80% in T. monococcum. However, compared to most previously described regions, the Pm3 loci studied have very low gene contents. Considering only those genes that appear to be intact (i.e., do not contain frameshifts or stop codons), the Pm3 loci of T. aestivum and T. monococcum have gene densities of one gene every 89 and 95 kb, respectively, and the Pm3 locus of T. turgidum contains only a single intact gene in the entire 332-kb region. Even if all identified pseudogenes are included, the three loci have relatively low gene densities of one gene every 35, 57, and 66 kb, respectively.

Most previous studies in Triticeae reported gene densities of roughly one gene every 20–45 kb (Dubcovsky et al. 2001; Feuillet et al. 2001; Wicker et al. 2001; Sanmiguel et al. 2002; Isidore et al. 2005). In some cases, gene densities similar to that in Arabidopsis were reported (Wei et al. 2002; Chantret et al. 2004). Such relatively high gene densities were expected since all of the above studies were done on distal regions of the chromosomes, which are known to be gene-rich (Qi et al. 2004). Less frequently, very low gene densities such as one gene per 175 kb (Dvorak et al. 2006) or 220 kb (Wicker et al. 2005) were reported, and the recent isolation of the barley vrs1 gene yielded only one gene in a 518-kb region (Komatsuda et al. 2007). Therefore, together with the three above studies, the three Pm3 loci demonstrate that low gene densities have to be expected frequently even within distal chromosomal regions where gene density is supposed to be high.

A notable characteristic of the repetitive fraction at all three Pm3 loci is the presence of numerous CACTA transposons. From earlier studies, it appears that most Triticeae genomic regions are rich in retrotransposons but contain only a few CACTA elements or none at all (Dubcovsky et al. 2001; Wicker et al. 2001; Sanmiguel et al. 2002; Wei et al. 2002; Chantret et al. 2004; Isidore et al. 2005; Gu et al. 2006), whereas in a few others, CACTA elements contribute substantially to the repetitive fraction (Wicker et al. 2005; Dvorak et al. 2006; Komatsuda et al. 2007). If one includes the data from Pm3 loci presented here, high content of CACTA elements seems to coincide with reports of low gene density. This suggests a preference of CACTA transposons for certain chromosomal regions and a local amplification of these elements.

Pm3-like genes are an ancient family:

The finding that a distinct family of resistance genes was identified in rice solely on the basis of DNA sequence homology is intriguing and contrary to the general perception that RGAs are fast evolving genes. Thus, the basic sequence organization of these genes is ancient and conserved despite obvious differences in function (e.g., there is no powdery mildew in rice). The fact that all wheat Pm3-like genes clearly cluster together with one subgroup of the rice homologs indicates that the Pm3 gene family has evolved multiple subfamilies well before the divergence of rice and wheat. Additionally, the multiple subfamilies of Pm3-like genes in rice suggest that the wheat genome might also contain additional clusters with RGAs homologous to those.

The observed strong conservation of the coding region of the Pm3-like genes between wheat and rice is in sharp contrast to the intense reshuffling of the sequences surrounding these genes. In some cases, we could show that unequal crossing over is a likely cause for genomic rearrangements and probably was an important mechanism that led to the observed mosaic structure of the up- and downstream regions of Pm3-like genes in wheat. It is, for example, a good explanation for the variation in gene numbers in the cluster of Pm3-like genes of the rice indica and japonica subspecies. Additionally, we could demonstrate how unequal recombination between two nearby repetitive elements can lead to the loss of genic sequences and, thus, interrupt sequence colinearity between species. Such events have long been postulated but, to our knowledge, were not actually reported.

The finding that none of the four loci in the rice genome containing Pm3 homologs was found in the region syntenic to wheat chromosome 1 came as no surprise, as RGAs are usually not found in collinear positions between species (reviewed by Leister 2004). It is assumed that the tendency of RGAs (and/or active resistance genes) to change their genomic location reflects selection pressure to remove successful resistance genes from clusters, thus protecting them from sequence homogenization through unequal crossing over or gene conversion (Leister 2004).

Rapid evolution of the Pm3 locus in wheat:

The method of estimating divergence time on the basis of conserved repetitive (i.e., selectively neutral DNA) has been applied several times in the past and has increased our understanding of the recent evolution of the wheat genomes (Wicker et al. 2003; Isidore et al. 2005; Gu et al. 2006). We made several comparisons between different wheat species using ours as well as previously published sequences. When the obtained sequence alignments are used for divergence time estimates using the rate of 1.3E-8 (Ma and Bennetzen 2004), T. monococcum is consistently placed at ~1.3 million years from the T. aestivum/T. turgidum lineage both with sequences presented in this study and with those from the Lr10 locus (Isidore et al. 2005). Similarly, the divergence of T. aestivum/T. turgidum is estimated to have occurred ~0.4 MYA on the basis of sequences from the Pm3 locus and previously published sequences (Gu et al. 2006).

The consistency of levels of DNA sequence conservation and resulting estimated divergence times from independent loci also makes it unlikely that the many differences between orthologous Pm3 loci are simply due to different ancient haplotypes being compared. A previous study showed that two different haplotypes can be found at the wheat Lr10 locus. Their comparison showed them to be highly divergent and indicated that they were present already before the divergence of the wheat A genomes (Isidore et al. 2005). Thus, for our study, it was crucial to use levels of DNA sequence identity as an indicator for true orthology in addition to genetic mapping data.

The relatively recent divergence times are surprising given the observed drastic changes in the regions compared. Indeed, especially between T. aestivum and T. turgidum, sequence conservation was minimal, reduced almost exclusively to the regions that contained the probes used for the BAC screening. In contrast, a recent study showed extensive sequence conservation even in highly repetitive regions between T. aestivum and T. turgidum (Gu et al. 2006). Similarly, recent comparison of T. turgidum and T. monococcum showed >25 kb of conserved transposable elements (Isidore et al. 2005). These data from wheat species suggest that the Pm3 locus evolved more rapidly than the other loci analyzed.

Although many studies show that resistance gene loci usually evolve faster than their surrounding genomic regions, our data indicate that the Pm3 locus in wheat evolved at an unusual speed even for RGA loci. For example, the Lr10 locus, which contains a resistance gene against wheat leaf rust, shows much more extensive sequence conservation. Additionally, the homologous locus of Pm3 in rice also showed less drastic rearrangements. The two rice subspecies diverged ~0.44 MYA (Ma and Bennetzen 2004) and the observed rearrangements can be clearly explained by a few insertions and deletions as well as two unequal crossing-over events. In contrast, the rearrangements that occurred in the wheat Pm3 locus are so manifold that only very few events could be traced back, and most of the sequences appear to reflect an extremely dynamic evolution in which these genomic loci have been reshaped vigorously.

The presented data show that Pm3-like RGAs are representatives of an ancient gene family that is highly conserved in wheat and rice. Nevertheless, it is surprising that such closely related genes can display such differences in evolutionary dynamics as well as in their apparent function (i.e., there is no powdery mildew in rice). Future studies should determine whether such variability can also be observed in other families of RGAs.


BAC clones of Triticum aestivum (Chinese Spring) were kindly provided by the joint consortium of the John Innes Centre, the Biotechnology and Biological Research Council, and the Institut National de la Recherche Agronomique (Unité de Recherches en Génomique Végétale). This work was supported by the Swiss National Science Foundation grant 3100A0-105620.


Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under accession nos. AY146587, AY146588, and DQ251490.


  • Allouis, S., G. Moore and A. Bellec, 2003. Construction and characterisation of a hexaploid wheat (Triticum aestivum L.) BAC library from the reference germplasm ‘Chinese Spring’. Cereal Res. Commun. 31: 331–338.
  • Altschul, S. F., T. L. Madden, A. A. Schaffer, J. Zhang, Z. Zhang et al., 1997. Gapped BLAST and PSI-BLAST, a new generation of protein database search programs. Nucleic Acids Res. 25: 3389–3402. [PMC free article] [PubMed]
  • Chantret, N., A. Cenci, F. Sabot, O. Anderson and J. Dubcovsky, 2004. Sequencing of the Triticum monococcum hardness locus reveals good microcolinearity with rice. Mol. Genet. Genomics 271: 377–386. [PubMed]
  • Dubcovsky, J., W. Ramakrishna, P. J. Sanmiguel, C. S. Busso, L. Yan et al., 2001. Comparative sequence analysis of colinear barley and rice bacterial artificial chromosomes. Plant Physiol. 125: 1342–1353. [PMC free article] [PubMed]
  • Dvorak, J., E. D. Akhunov, A. R. Akhunov, K. R. Deal and M. C. Luo, 2006. Molecular characterization of a diagnostic DNA marker for domesticated tetraploid wheat provides evidence for gene flow from wild tetraploid wheat to hexaploid wheat. Mol. Biol. Evol. 23: 1386–1396. [PubMed]
  • Ewing, B., L. Hillier, M. C. Wendl and P. Green, 1998. Base-calling of automated sequencer traces using phred. I. Accuracy assessment. Genome Res. 8: 175–185. [PubMed]
  • Feldman, M., F. G. H. Lupton and T. E. Miller, 1995. Wheats, pp. 184–192 in Evolution of Crops, Ed. 2, edited by J. Smartt and N. W. Simmonds. Longman Scientific, London.
  • Feuillet, C., A. Penger, K. Gellner, A. Mast and B. Keller, 2001. Molecular evolution of receptor-like kinase genes in hexaploid wheat. Independent evolution of orthologs after polyploidization and mechanisms of local rearrangements at paralogous loci. Plant Physiol. 125: 1304–1313. [PMC free article] [PubMed]
  • Gaut, B. S., B. R. Morton, B. C. Mccaig and M. T. Clegg, 1996. Substitution rate comparisons between grasses and palms: synonymous rate differences at the nuclear gene Adh parallel rate differences at the plastid gene rbcL. Proc. Natl. Acad. Sci. USA 93: 10274–10279. [PMC free article] [PubMed]
  • Gu, Y. Q., J. Salse, D. Coleman-Derr, A. Dupin, C. Crossman et al., 2006. Types and rates of sequence evolution at the high-molecular-weight glutenin locus in hexaploid wheat and its ancestral genomes. Genetics 174: 1493–1504. [PMC free article] [PubMed]
  • Isidore, E., B. Scherrer, B. Chalhoub, C. Feuillet and B. Keller, 2005. Ancient haplotypes resulting from extensive molecular rearrangements in the wheat A genome have been maintained in species of three different ploidy levels. Genome Res. 15: 526–536. [PMC free article] [PubMed]
  • Jones, J. D., and J. L. Dangl, 2006. The plant immune system. Nature 444: 323–329. [PubMed]
  • Jones, D., and J. D. Jones, 1997. The role of leucine-rich repeat proteins in plant defenses. Adv. Bot. Res. Adv. Plant Pathol. 24: 89–167.
  • Keller, B., C. Feuillet and N. Yahiaoui, 2005. Map-based isolation of disease resistance genes from bread wheat: cloning in a supersize genome. Genet. Res. 85: 93–100. [PubMed]
  • Komatsuda, T., M. Pourkheirandish, C. He, P. Azhaguvel, H. D. Kanamori et al., 2007. Six-rowed barley originated from a mutation in a homeodomain-leucine zipper I-class homeobox gene. Proc. Natl. Acad. Sci. USA 23: 1424–1429. [PMC free article] [PubMed]
  • Kuang, H., S. S. Woo, B. C. Meyers, E. Nevo and R. W. Michelmore, 2004. Multiple genetic processes result in heterogeneous rates of evolution within the major cluster disease resistance genes in lettuce. Plant Cell 16: 2870–2894. [PMC free article] [PubMed]
  • Leister, D., 2004. Tandem and segmental gene duplication and recombination in the evolution of plant disease resistance gene. Trends Genet. 20: 116–122. [PubMed]
  • Ma, J., and J. L. Bennetzen, 2004. Rapid recent growth and divergence of rice nuclear genomes. Proc. Natl. Acad. Sci. USA 101: 12404–12410. [PMC free article] [PubMed]
  • Meyers, B. C., A. Kozik, A. Griego, H. Kuang and R. W. Michelmore, 2003. Genome-wide analysis of NBS-LRR-encoding genes in Arabidopsis. Plant Cell 15: 809–834 (erratum: Plant Cell 15: 1634). [PMC free article] [PubMed]
  • Michelmore, R. W., and B. C. Meyers, 1998. Clusters of resistance genes in plants evolve by divergent selection and a birth-and-death process. Genome Res. 8: 1113–1130. [PubMed]
  • Noël, L., T. L. Moores, E. A. Van Der Biezen, M. Parniske, M. J. Daniels et al., 1999. Pronounced intraspecific haplotype divergence at the RPP5 complex disease resistance locus of Arabidopsis. Plant Cell 11: 2099–2112. [PMC free article] [PubMed]
  • Paterson, A. H., J. E. Bowers and B. A. Chapman, 2004. Ancient polyploidization predating divergence of the cereals and its consequences for comparative genomics. Proc. Natl. Acad. Sci. USA 101: 9903–9908. [PMC free article] [PubMed]
  • Petrov, D. A., 2001. Evolution of genome size: new approaches to an old problem. Trends Genet. 17: 23–28. [PubMed]
  • Qi, L. L., B. Echalier, S. Chao, G. R. Lazo, G. E. Butler et al., 2004. A chromosome bin map of 16,000 expressed sequence tag loci and distribution of genes among the three genomes of polyploid wheat. Genetics 168: 701–712. [PMC free article] [PubMed]
  • Sanmiguel, P., B. S. Gaut, A. Tikhonov, Y. Nakajima and B. L. Bennetzen, 1998. The paleontology of intergene retrotransposons of maize. Nat. Genet. 20: 43–45. [PubMed]
  • Sanmiguel, P., W. Ramakrishna, J. L. Bennetzen, C. Busso and J. Dubcovsky, 2002. Transposable elements, genes and recombination in a 215-kb contig from wheat chromosome 5A. Funct. Integr. Genomics 2: 70–80. [PubMed]
  • Smith, S. M., and S. H. Hulbert, 2005. Recombination events generating a novel Rp1 race specificity. Mol. Plant-Microbe Interact. 18: 220–228. [PubMed]
  • Sonnhammer, E. L., and R. Durbin, 1995. A dot-matrix program with dynamic threshold control suited for genomic DNA and protein sequence analysis. Gene 167: 1–10. [PubMed]
  • Thompson, J. D., D. G. Higgins and T. J. Gibson, 1994. CLUSTAL W, improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22: 4673–4680. [PMC free article] [PubMed]
  • Van Der Hoorn, R. A., M. Kruijt, R. Roth, B. F. Brandwagt, M. H. Joosten et al., 2001. Intragenic recombination generated two distinct Cf genes that mediate AVR9 recognition in the natural population of Lycopersicon pimpinellifolium. Proc. Natl. Acad. Sci. USA 98: 10493–10498. [PMC free article] [PubMed]
  • Wei, F., R. A. Wing and R. P. Wise, 2002. Genome dynamics and evolution of the Mla (powdery mildew) resistance locus in barley. Plant Cell 14: 1903–1917. [PMC free article] [PubMed]
  • Wicker, T., N. Stein, L. Albar, C. Feuillet, E. Schlagenhauf et al., 2001. Analysis of a contiguous 211 kb sequence in diploid wheat (Triticum monococcum L.) reveals multiple mechanism of genome evolution. Plant J. 26: 307–316. [PubMed]
  • Wicker, T., D. E. Matthews and B. Keller, 2002. TREP, a database for Triticeae repetitive elements. Trends Plant Sci. 7: 561–562.
  • Wicker, T., N. Yahiaoui, R. Guyot, E. Schlagenhauf, Z-D. Liu et al., 2003. Rapid genome divergence at orthologous low molecular weight glutenin loci of the A and Am genomes of wheat. Plant Cell 15: 1186–1197. [PMC free article] [PubMed]
  • Wicker, T., W. Zimmermann, D. Perovic, A. H. Paterson, M. Ganal et al., 2005. A detailed look at 7 million years of genome evolution in a 439 kb contiguous sequence at the barley Hv-eIF4E locus: recombination, rearrangements and repeats. Plant J. 41: 184–194. [PubMed]
  • Yahiaoui, N., P. Srichumpa, R. Dudler and B. Keller, 2004. Genome analysis at different ploidy levels allows cloning of the powdery mildew resistance gene Pm3b from hexaploid wheat. Plant J. 37: 528–538. [PubMed]
  • Yahiaoui, N., S. Brunner and B. Keller, 2006. Rapid generation of new powdery mildew resistance genes after wheat domestication. Plant J. 47: 85–98. [PubMed]
  • Zhou, T., Y. Wang, J. Q. Chen, H. Araki, Z. Jing et al., 2004. Genome-wide identification of NBS genes in japonica rice reveals significant expansion of divergent non-TIR NBS-LRR genes. Mol. Genet. Genomics 271: 402–415. [PubMed]

Articles from Genetics are provided here courtesy of Genetics Society of America
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...