Introduction

Ascochyta blight, caused by Mycosphaerella pinodes (Berk. and Blox.), is the most important foliar disease of field pea (Pisum sativum L.) worldwide. Breeding pea varieties with resistance to M. pinodes is difficult due to the availability of only partial levels of resistance (Ali-Khan et al. 1973; Ali et al. 1978; Bretag 1989, 1991; Xue et al. 1996; Kraft et al. 1998; Xue and Warkentin 2001; Prioul et al. 2003) inherited as a complex polygenic trait (Zlamal 1984; Wroth 1999). In the last 3 years, QTL mapping studies have attempted to dissect the genetic basis of quantitative resistance to ascochyta blight in pea. Timmerman-Vaughan et al. (2002, 2004) identified numerous QTLs for resistance to field epidemics located on the seven pea linkage groups. Six QTLs common to two populations were identified on linkage groups II, III, IV, V and VII (Timmerman et al. 2004). However QTL specificity for pathogens of the ascochyta complex (Mycosphaerella pinodes, Ascochyta pisi and Phoma medicaginis) was not specified. Tar’an et al. (2003) reported three QTLs for resistance to Mycosphaerella blight under field conditions. In a previous study, we identified six and ten QTLs for resistance to M. pinodes at the seedling (controlled conditions) and at the adult plant stages (field conditions) respectively, four QTLs being common to both plant stages (Prioul et al. 2004). These QTL mapping studies have considerably increased our knowledge on the genetic architecture of pea partial resistance to M. pinodes, but the biological functions of the resistance factors underlying these QTLs still remain unknown. Such information on the biochemical mechanisms underlying resistance will be helpful to improve the efficiency of MAS based construction of resistant genotypes and provide a more efficient and durable resistance to M. pinodes in pea.

The candidate gene approach has been successfully used in plant genetics for QTL characterization, by testing associations among QTLs and genes potentially involved in the biochemical pathways leading to trait expression. Co-localizations between resistance QTLs and resistance genes (R genes), resistance-gene analogs (RGAs) or defense response genes (DR genes) were reported in a variety of plant species, leading to the hypothesis of a possible involvement of these genes in the effects of some resistance QTLs (for review, see Pflieger et al. 2001a).

In pea, R genes involved in resistance against fungi (Dirlewanger et al. 1994; Timmerman et al. 1994; Coyne et al. 2000), viruses (Gritton and Hagedorn 1980; Marx et al. 1985; Provvidenti and Hampton 1991; Timmerman et al. 1993; Dirlewanger et al. 1994), or bacteria (Hunter et al. 1998, 2001), as well as defense-related cloned sequences (Gilpin et al. 1997; Weeden et al. 1998, 1999) and NBS-LRR RGAs (Timmerman-Vaughan et al. 2000) were located on published pea maps. Up to now, only one major resistance gene has been cloned in pea, conferring resistance to a pathotype of PSbMV (Gao et al. 2004). Comparative mapping studies have also shown that three genomic regions containing RGAs also included QTLs for resistance to Ascochyta blight (Timmerman-Vaughan et al. 2002). Physiological and biochemical studies of the pea–M. pinodes interaction reported that M. pinodes elicitor (Shiraishi et al. 1978a, 1992; Matsubara and Kuroda 1987) induces many defense responses in pea, such as accumulation of pisatin, a major phytoalexin in pea (Shiraishi et al. 1978b), activation of the genes encoding phenylalanine ammonia-lyase (PAL) and chalcone synthase (for review, see Yamada et al. 1996), activation of PR proteins (β-1,3-glucanase, chitinases) (Yoshioka et al. 1992), generation of superoxide anion (Kiba et al. 1997), enhancement of ATPase activity (Kiba et al. 1995, 1996, 1997), activation of the polyphosphoinositide metabolism (Toyoda et al. 1992, 1993, 1998). However, the mechanisms by which M. pinodes elicitor recognition, activation of signal transduction pathways leading to defense responses and expression of the QTLs of partial resistance are connected are still not well understood.

The aim of the present study was to map candidate genes for resistance on the JI296 × DP genetic linkage map and compare their genomic localizations with QTLs for resistance to M. pinodes previously identified in the RIL population (Prioul et al. 2004). Candidate genes were (1) previously cloned pea DR genes chosen according to their potential role in the M. pinodes/pea interaction or in disease resistance mechanisms, (2) RGAs previously isolated in pea by Timmerman-Vaughan et al. (2000) and likely to map in similar genomic regions as QTLs identified in the JI296 × DP population, and (3) RGAs cloned in the present study. In this paper, we report the development and mapping of a set of PCR-based DR and RGA markers and discuss the genomic co-localizations between these candidate genes and QTLs for M. pinodes resistance in pea.

Materials and methods

Plant material

Two homozygous pea genotypes were used as sources of total genomic DNA for PCR amplification: DP and JI296 (i.e. cv. ‘Chemin Long’), partially resistant and susceptible to M. pinodes, respectively (Prioul et al. 2003, 2004). Genomic DNA was extracted from dried young leaves using the CTAB method of Doyle and Doyle (1990).

A 135 F2:6 recombinant inbred line (RIL) population derived from the cross JI296 × DP, previously described by Prioul et al. (2004), was used for mapping studies.

Isolation, cloning and sequencing of RGA sequences

Degenerate RGA-consensus primers were designed on the conserved motifs P-loop and GLPL (Meyers et al. 1999). Primer pairs were designated (1) RGAino, with forward sequence 5′-GGI GGI GTI GGI AAI ACI AC-3′ (Leister et al. 1996) and reverse sequence 5′-IAG IGC IAG IGG IA(A/G) ICC-3′ (Fourmann et al. 2001), and (2) RGAdeg with forward sequence 5′-GGT GGG GTT GGG AA(A/G) AC(A/T/C/G) AC-3′ (Fourmann et al. 2001) and reverse sequence 5′-CAA CGC TAG TGG (A/T/C/G)A(A/G) (A/T/C/G)CC-3′ (Fourmann et al. 2001).

PCR amplifications were performed in 50 μl containing 1.5 mM MgCl2, 0.2 mM of each dNTP, 1.5 μM of each primer, 200 ng of genomic DNA and 1.5U of Taq polymerase (Eurobio). PCR conditions were 5 min at 94°C; 1 min at 94°C, 1 min at the annealing temperature (Table 2) and 2 min at 72°C for 30 cycles; and 5 min at 72°C.

PCR products were purified and concentrated on centrifugal Filter Devices Microcon (Amicon process), cloned into pGEM-T vector system (Promega, Madison, Wis.) and used to electroporate DH10B cells (Gibco BRL). Clones containing inserts were screened by PCR using universal M13 primers. A total of 707 putative recombinant colonies were obtained, including 291 and 363 clones isolated from DP and JI296, respectively, using RGAino primers (clones individually named as IDx or IJx, where x is the number of the clone), and 5 and 48 clones obtained from DP and JI296, respectively, using RGAdeg primers (clones individually named as DDx or DJx).

To select non-redundant clones before sequencing, aliquots containing 15 μl of the PCR product from each clone were digested with the restriction enzymes HaeIII (Appligene) and either AluI (Appligene) or RsaI (Gibco BRL), following the manufacturer instructions. The restriction fragments were separated on a 5% agarose gel (1X TAE buffer, 150 V, 3 h30 migration). A total of 130 non redundant clones were retained for sequencing. All sequencing reactions were performed by Genome Express laboratory (Montreuil, France).

PCR amplification of RGAs and DR candidate genes

Three classes of candidate genes were tested for mapping: RGAs cloned in the present study, 4 RGAs previously cloned in pea by Timmerman-Vaughan et al. (2000), and 12 pea DR genes chosen in the GenBank database. The putative biological functions of cloned DR genes and RGAs are detailed in Table 1. Sequence-specific primers were designed using the OLIGO 6.0 software (http://www.oligo.net/). Primer specifications for DR genes and RGAs are indicated in Tables 2 and 3, respectively.

Table 1 Pea DR and RGA sequences previously published used for the candidate gene approach
Full size table
Table 2 Primer specifications, optimized PCR conditions and generated polymorphism between DP and JI296 for DR genes
Full size table
Table 3 Specific primer pairs defined on RGAs, PCR conditions and generated polymorphism between DP and JI296
Full size table

Amplifications were performed in 25 μl using 50 ng of genomic DNA of each parental line DP or JI296, 1.5–2.5 mM MgCl2, 0.5 mM of each dNTP, 1 μM of each primer, 1U Taq polymerase (Eurobio). PCR conditions were 5 min at 94°C; 1 min at 94°C, 1 min at the annealing temperature (Tables 2 and 3) and 2 min at 72°C for 30 cycles; and 5 min at 72°C. PCR products were separated either on a 1.4% agarose gel (1X TAE buffer, 110 V, 1 h30 migration) or on a 5% non-denaturing acrylamide gel (1X TBE buffer, 220 V, 15–20 h migration). DNA from bands of interest was removed from the gel and reamplified according to Brunel et al. (1999) with slightly different conditions: 5 min at 94°C; 1 min at 94°C, 45 sec at the annealing temperature, and 1 min at 72°C for 25 cycles; and 5 min at 72°C. These products were then directly sequenced to control for specific amplification of the expected sequence.

Sequence analysis

RGA sequences amplified from DP and JI296 were analyzed using the GCG software package (version 10.2, Genetics Computer Group, Madison). RGA clones presenting less than five different bases were considered as the same initial sequence, as reported by Fourmann et al. (2001). RGA multiple sequence alignments were performed using the CLUSTALW option of the GCG Wisconsin package with default parameters.

For DR genes and RGAs isolated by Timmerman-Vaughan et al. (2000), sequences amplified from DP and JI296 were validated for their homology with the chosen candidate gene sequence by performing similarity searches against sequences deposited in the non-redundant sequence databases (NCBI), using BlastN and BlastX algorithms (Altschul et al. 1997). To identify useful polymorphism for further marker development, pairwise comparisons of sequences and editing of restriction maps were performed using the PILEUP and MAP options of the GCG package, respectively.

Genetic mapping

Specific candidate gene markers showing polymorphism between JI296 and DP were genotyped on the RIL population. For each segregating marker, a chi-square analysis was used to test for deviations from the expected (1:1) segregation ratio. The MAPMAKER/EXP software (version 3.1) (Lincoln et al. 1992) was used to map the candidate genes on the JI296 × DP genetic map previously described by Prioul et al. (2004), using the “try” command to assign molecular markers to linkage group (minimum LOD score of 2.0; maximum recombinant fraction of 0.30), and the “ripple” command to test final orders (LOD threshold of 2.0). Three additionnal molecular markers were added to the JI296 × DP genetic map increasing the number of common markers with other published pea genetic maps: sL01 (Frew et al. 2002) was mapped on LG VI, sP2P5 (Timmerman-Vaughan et al. 2002) on LG II, and sY16 (Timmerman-Vaughan et al. 2002) on the distal part of LG Vb.

Statistical analyses

Statistical analyses for candidate gene-phenotype association were performed using the SAS version 6.12 package (SAS Institute, Cary, NC, USA) and the QTL CARTOGRAPHER Windows v1.30 software (Basten et al. 1994, 2001). Markers that co-localized with QTLs for ascochyta blight resistance were checked by one-way ANOVA, Kruskal–Wallis non-parametric test and composite interval mapping (CIM) analysis, as described in Prioul et al. (2004).

Results

RGA isolation

Out of the 130 sequenced clones, 64 showed high similarities with RGA sequences when compared with GenBank accessions (BlastX e-values varying from e −84 to e −46). The remaining 66 clones were not analyzed further since they corresponded to PCR products that (1) were generated by one of the primer alone, suggesting the existence of two opposite complementary sequences on DNA allowing PCR amplification, (2) showed similarities with retrotransposon sequences, myosins or integrases, or (3) did not show any similarities with any sequences in the GenBank database. Out of the 64 retained RGA sequences, 61 were obtained with RGAino primer pair and 3 with RGAdeg primer pair. Five sequences were not analyzed further due to the poor quality of their electrophoregrams. Considering a likely Taq polymerase (Eurobio) error frequency of 1 × 10−4 in base incorporation, the remaining 59 clone sequences were divided into 16 groups, sequences within each group being considered as identical. For each parental line, a reference clone was chosen as the most probable sequence for each group. The 16 reference sequences were conserved for further analysis, corresponding to insert sizes varying from 451 bp (incomplete sequence) to 516 bp.

Figure 1 presents a multiple alignment and classification of the 16 RGA putative amino acid sequences. The conserved domains P-loop (upper primer), RNBS-A-TIR, RNBS-A-nonTIR, kinase 2, RNBS-B, RNBS-C and GLPL (lower primer) previously defined by Meyers et al. (1999) were found in the sequences. Fifteen sequences had the RNBS-A-TIR domain and were divided into four classes based on the presence of specific motifs in the RNBS-A-TIR domain and along the sequence: class I consisted of three sequences isolated from both parental lines, class II contained nine fragments isolated from both parental lines, class III was composed of two fragments isolated from JI296 and class IV was composed of the unique JI296 sequence IJA3. DJ37 is the only sequence to possess the RNBS-A-nonTIR domain (class V). Except for class I, RGAino and RGAdeg primer pairs led to the identification of distinct RGA classes. No high similarities were found between pea RGA sequences and known R-genes, except with the tobacco N gene (genbank accession U15605) conferring resistance to the Tobacco Mosaic Virus or the tomato I2 gene (genbank accession AF118127), conferring resistance to Fusarium oxysporum. Amino acid identities of 45%, 38%, 34% and 36% were found between the tobacco N gene and reference sequences of classes I, II, III and IV, respectively. Class V DJ37 presented 44% amino acid identity with the tomato I2 gene. Highest homologies (up to 97% of amino acid identity) were found between pea RGA sequences identified in this study and published sequences putatively involved in plant disease resistance, isolated from Pisum sativum (RGA-G3A, RGA2.159) and closely related legume species such as Cicer arietinum (CP3), Cajanus cajun (PP10) or Medicago sativa (AF230841). High amino acid identity percentages were found within the same pea RGA class but varied from 23.9% to 70.4% between classes (Table 4). The relationships within the 16 pea RGA reference sequences identified in this study and Genbank accessions are illustrated in Fig. 1.

Fig. 1
figure 1

Alignment of the predicted amino acid sequences of the five classes of RGA sequences isolated from pea. Sequences of Nicotiana tabacum N (U15605), Lycopersicum esculentum I2 (AF118127), Cajanus cajun PP10 (AF186633), Cicer arietinum CP3 (AF186625), Medicago sativa (AF230841), Pisum sativum RGA-G3A (AF123703) and RGA2.159 (AF123702) are included in this analysis. The conserved motifs defined by Meyers et al. (1999) are indicated in bold

Full size image
Table 4 Amino acid identities (%) between the five classes of pea RGA sequences
Full size table

RGA mapping

Second-generation specific primer pairs designed for classes I, IV and V of isolated RGAs (primer pairs RGA1a, RGA1b, RGA4 and RGA5) amplified PCR products related to the corresponding groups, but no polymorphism useful for mapping could be found between DP and JI296. The primer pair RGA2 gave an electrophoresis profile with two bands on DP and JI296, both sharing high homology with the class II sequences. When comparing sequences obtained between the two parental lines, seven and two nucleic acid differences were found between the two lower bands and the two upper bands, respectively. A CAPS marker was developed (Table 3) and one of the RGA2 loci (corresponding to the lower band) was mapped on linkage group VII (Fig. 2). Two allele-specific primers, designed in regions of the class II IJB174 and IJB91 sequences, generated sequences showing polymorphism between the two parental genotypes (orphan bands in the JI296 parental line) and were mapped on linkage group VII, close to the RGA2 locus (Fig. 2). The primer pair RGA3 amplified a single band in DP and two bands in JI296. After having checked that all three PCR products were related to the class III, the polymorphic RGA3 locus was mapped on linkage group VII in the vicinity of the class II RGA IJB174 and IJB91 (Fig. 2). For each of the four RGA markers added to the genetic map on LGVII, a high segregation distortion (P < 0.0001) was observed favoring the DP allele (Fig. 2), as for most of the neighboring molecular markers in this genomic region.

Fig. 2
figure 2

Localization of 14 candidate gene markers on the JI296 × DP genetic map, including 8 DR gene markers and 6 RGAs (4 from this study and 2 previously isolated by Timmerman-Vaughan et al. 2000). Candidate genes are displayed in bold italic. The size of each linkage group is indicated in Kosambi centiMorgans (cM) below each group. Markers with segregation distortion at P ≤ 0.05, P ≤ 0.01 and P ≤ 0.001 are indicated by *, ** and ***, respectively. QTLs previously detected in the JI296 × DP RIL population (Prioul et al. 2004) are indicated by vertical bars to the left of each linkage group. Shaded bars represent QTLs identified at the seedling stage, under controlled conditions. Black bars represent QTLs identified at the adult stage, under field conditions. The names of the QTLs are reported in adjacent position. The percent number above each QTL is the range of percentage of phenotypic variation (R 2) individually explained by the QTL, depending on different resistance criteria used in Prioul et al. (2004). The length of the bar represents a 1-LOD support interval from the peak LOD of the QTL

Full size image

The specific primers designed on the four RGAs previously isolated in pea by Timmerman-Vaughan et al. (2000) successfully amplified the candidate sequences and generated polymorphism between the two parental lines. RGA-G3A and RGA2.97 were mapped on LG VII, in the vicinity of the four RGA markers developed in this study (Fig. 2). RGA1.1 was assigned to LG III but its accurate location still remained unclear. RGA2.65 was not mapped further due to difficulties in following the marker in the segregating population.

Genomic amplification and mapping of DR candidate genes

Out of the primer pairs defined for the 12 DR candidate genes, two (from Pschitin and DRR206-d sequences) successfully amplified target genomic sequences but did not reveal any polymorphism between parental lines (Table 2). These primer pairs were not used further. For the ten other DR genes, polymorphism between DP and JI296 was revealed and, except for PEAPAL1 and PsPRP4A, was generated by digestion with restriction enzymes. The CAPS markers are listed in Table 2. PCR amplifications from genomic DNA revealed the presence of introns for Pschitin, Hmm6 and DRR230-b sequences.

Out of the 12 candidate genes tested in the present study, 8 were mapped on the JI296 × DP genetic map (Fig. 2). PsPRP4A was located on LG II; Peachi21, PsMnSOD, DRR230-b and PsDof1 were mapped at different positions on LG III; peaβglu and DRR49a were mapped in a same genomic region on LG VI, and Hmm6 was located on LG VII in the vicinity of the RGA2.97 locus. The genes Hmm6, PsMnSOD and PsDof1 showed significant deviation from the expected 1:1 ratio (Fig. 2). PEAPAL1 and PEAPAL2 were not mapped due to difficulties in following these markers in the segregating population.

Genomic co-localizations between candidate genes and QTLs for resistance to M. pinodes

CIM analysis (data not shown), one-way ANOVA (P ≤ 5 × 10−3) and the Kruskall–Wallis non-parametric test (P ≤ 5 × 10−3), performed on different ascochyta blight resistance scoring criteria previously described in Prioul et al (2004), confirmed the co-segregations between PsDof1 and the QTL mpIII-1, DRR230-b and the QTL mpIII-4, and between IJB91, IJB174, RGA2 and RGA-G3A and the QTL mpVII-1 (Table 5, Fig. 2).

Table 5 Significant associations (P ≤ 5.10− 3) between candidate genes and the different disease scoring criteria used in the mapping QTL study for resistance to M. pinodes in the JI296 × DP RIL population, under both controlled and field conditions (Prioul et al. 2004). Probabilities were obtained using one-way ANOVA/Kruskal–Wallis non-parametric test
Full size table

Discussion

Using a PCR-based candidate gene approach, we isolated and mapped several RGAs and DR genes on the JI296 × DP genetic map, some of them occurring in genomic regions containing QTLs for resistance to M. pinodes on LGs III and VII.

Development and mapping of candidate RGA and DR gene markers

RGAs isolated in this work, using degenerate oligonucleotide primers based on NBS-LRR type of cloned R genes, fell into 5 distinct classes, with classes I, II and IV corresponding to new RGA types in comparison with the 9 pea RGAs previously identified by Timmerman-Vaughan et al (2000). Discrepancies between the two studies may be due either to the use of different pea genotypes or the possibility that some RGA sequences may be absent in some lines (Collins et al. 1998), or to the set of degenerate primers used and the fact that slight differences in primer sequences and/or primer combinations could lead to the amplification of different RGA fragments (Leister et al. 1996; Aarts et al. 1998). The close relatedeness of these RGAs with RGAs isolated from other closely related legumes corroborates the existence of legume specific families of NBS-LRR R genes (Meyers et al. 1999; Zhu et al. 2002). As previously reported in legumes (Yu et al. 1996; Zhu et al. 2002; Bertioli et al. 2003), an unbalanced ratio of TIR-NBS LRR and non-TIR NBS LRR sequences (15/1) was also observed in this study. Such a ratio may reflect either a bias during the PCR towards TIR sequences unrevealing differences in the P-loop and GLPL motifs between TIR and non-TIR NBS sequences, as hypothesized for the Arachis genome (Bertioli et al. 2003) and/or the presence of a smaller number of nonTIR-NBS than TIR sequences in the pea genome. Using a non-TIR-NBS specific primer should allow to test between both hypothesis. There are potentially much more RGAs to detect in pea. Indeed, 147 NBS-LRR RGAs have been identified in the model legume M. truncatula, mostly organized in clusters on the genome, several of these clusters being syntenic between M. truncatula and pea (Zhu et al. 2002). Although more RGAs should be mapped on the pea genome, our results and the ones of Timmerman-Vaughan et al. (2000) suggest the existence of a RGA cluster on LG VII including 5 RGAs distributed over 41 cM (Fig. 2). By comparative mapping with M. truncatula, this cluster appears to be syntenic to an extended cluster of at least 30 TIR NBS-LRR RGAs spanning the majority of the LG VI in M. truncatula (Zhu et al. 2002; NSF Plant Genome Project 2002). Syntenic blocks of R gene loci were also previously identified between LG I of pea and LG V of M. truncatula (Zhu et al. 2002). Consequently, NBS-LRR RGAs isolated and mapped in M. truncatula could be useful tools for identification of more RGAs in pea.

In the present study, we also designed PCR-based markers for 4 NBS-LRR sequences previously isolated in pea. The RGAs RGA-G3A and RGA2.97 were unambiguously mapped on the JI296 × DP genetic map, occuring at similar genomic locations as in the JI1794 x Slow population (Timmerman-Vaughan et al. 2000). Using the PCR-based approach on pea DR sequences available in public databases, we successfully mapped 8 DR genes on the JI296 × DP genetic map. Three of these genes were not mapped previously on published molecular linkage maps, namely PsDof1, Hmm6 and PsPRP4A. Our mapping results are in agreement with previously reported map locations found for DRR49a (corresponding to pi49), Peaβglu (corresponding to β-1,3-glucanase) and PsMnSOD (corresponding to Sodmt) (Gilpin et al. 1997; Weeden et al. 1998, 1999), but also clarify the locations of Peachi21 and DRR230. Indeed, we confirmed the linkage between Peachi21 and the P202 marker (a 5 cM interval; data not shown), as previously described in Gilpin et al. (1997), and located Peachi21 on LG III. We also unambiguously assigned DRR230-b (corresponding to pi39) to LG III, whereas it was previously located on LG III/IV by Gilpin et al. (1997). Because of the absence of polymorphism between the parental lines or difficulties to follow the markers in the segregating population, Pschitin, PEAPAL1 and PEAPAL2 were not mapped on the JI296 × DP map, whereas these genes were previously mapped on other linkage maps (Gilpin et al. 1997; Weeden et al. 1999).

Except for PsMnSOD, PEAPAL1 and PEAPAL2 for which PCR-based STS markers have been developped by Weeden et al. (1999), remaining DR genes and the 4 RGAs previously reported on linkage maps were located by other groups using RFLP probes. Based on PCR, our markers require a lower technical complexity for sample preparation and marker detection than RFLP, are less time-consuming, and can easily be transfered to any laboratory. The major problems we found with RGA and DR marker development were either the low rate of mutations in the coding sequence of the genes (i.e., PEAPAL1, PEAPAL2), thus limiting the usefulness of the CAPS strategy, or the inability to exploit mutations by digestion with restriction enzymes or allele-specific design with the 3′ end of the primer located on the mutation (Ugozzoli and Wallace 1991). The current development of SNP genotyping technologies should provide us new tools to take better advantage of these mutations and develop new types of markers.

Co-localizations between candidate genes and QTL for resistance to M. pinodes

A first co-localization was identified on LG III between the PsDof1 gene, encoding a putative transcription factor that may bind to DNA through a Dof (DNA-binding with one finger) domain (Seki et al. 2002), and the QTL mpIII-1, a major QTL explaining 18–42% of the total phenotypic variation at different growth-stages and in various environmental conditions (Prioul et al. 2004). As PsDof1 was initially isolated from a cDNA library constructed from M. pinodes elicitor-treated pea epicotyls (Seki et al. 2002), it has been suggested to be involved in the elicitor-induced activation of elicitor-responsive genes (Seki et al. 2003). Thus, PsDof1 represents a good candidate for the QTL mpIII-1.

A second genomic co-localization was observed on LG III between DRR230-b, a member of the pea defensin gene family, and the QTL mpIII-4 which explained 29% of the stem resistance in the field (Prioul et al. 2004). DRR230-b cDNA was first isolated from pea pods in response to infection by the fungal pathogen Fusarium solani (Chiang and Hadwiger 1991). More recently, Lai et al. (2002) reported the induction of two related defensin genes (DRR230-a and DRR230-c) in response to different fungal pathogens, including Ascochyta pinodes (teleomorph Mycosphaerella pinodes), and bacterial pathogens (compatible, incompatible and non-host strains). It has been hypothesized that pea defensins may play a role in the general plant defense responses, as suggested by their accumulation in response to wounding, ozone exposure and pathogen infection (Sävenstrand et al. 2000; Lai et al. 2002), and would act as general inducible determinants of disease resistance (Lai et al. 2002). Although further QTL analysis studies across environments should be performed to confirm particularly the involvement of the QTL mpIII-4 in pea partial resistance to M. pinodes (Prioul et al. 2004), DRR230-b can be considered as a putative candidate for this QTL.

The third genomic co-localization was found on linkage group VII between a cluster of RGA sequences (RGA2, RGA3, IJB174, IJB91, RGA-G3A) and mpVII-1, a minor-effect QTL identified from seedling and adult stage condition scorings (Prioul et al. 2004). In their previous study, Timmerman-Vaughan et al. (2002) also reported a co-localization on LG VII between RGA-G3A and a QTL for resistance to field epidemics of ascochyta blight. In our work, statistical analyses showed a specific association between the 4 RGAs belonging to class II (i.e., RGA-G3A, RGA2, IJB174 and IJB91) and the M. pinodes disease severity AUDPC scored at the seedling stage (Table 5), suggesting a stage-specific expression of these genetic factors. In our previous QTL mapping study, we used two distinct methodologies to assess partial resistance to M. pinodes: under controlled conditions, we focused on seedling resistance to infection only, scoring disease severity progress during 20 days after inoculation with M. pinodes; whereas under field conditions, we focused on the overall response of the adult plant to M. pinodes infection, including both resistance to infection and resistance to fungus progress upward on the plant. As the QTL mpVII-1 was detected in both conditions and given the global assessment methodology used in field trials, we can hypothesize an involvement of these genetic factors in the very early steps of adult-plant resistance. Although further experiments are still needed to establish a functional link between these RGAs and the expression of resistance, the four mapped RGAs could be considered as valuable candidate genes for the QTL mpVII-1.

When comparing our results with previous reports about the mapping of disease resistance genes in pea, we identified two chromosomal regions of the pea genome that might possess a hot spot of genes with a putative role in pea disease resistance. The distal part of linkage group III, where we located two QTLs for resistance to M. pinodes (Prioul et al. 2004) and the DR genes DRR230-b and PsDof1, was also previously reported to carry the Rmp4 gene, involved in stem resistance to M. pinodes in pea seedlings and linked to Np (Clulow et al. 1991), the Fw gene, conferring resistance to Fusarium oxysporum f. sp. pisi race 1 (Dirlewanger et al. 1994; Weeden et al. 1998), the QTL Asc3.1 for resistance to Ascochyta blight (Timmerman-Vaughan et al. 2002) and two RGAs, RGA1.1 and RGA2.65 (Timmerman-Vaughan et al. 2000). The chromosomal region of LG VII spanning from Q500 to DIMIN (covering ∼ 45 cM) carries QTLs for resistance to Ascochyta blight (Timmerman-Vaughan et al. 2002; Prioul et al. 2004), the Hmm6 gene (this study), encoding an enzyme involved in the terminal step for synthesis of pisatin and isolated in pea tissues after infection with the bacteria Nectria haematococca (Wu et al. 1997), RGAs (this work; Timmerman-Vaughan et al. 2002), the Ppi2 gene for resistance to Pseudomonas syringae pv. pisi (Hunter et al. 2001). Probably the further use in pea mapping of bridge markers such as microsatellites allowing to align genetic maps from different crosses (Loridon et al. 2005) will help in confirming map to map co-localisations between candidate genes and identified resistance genes or QTLs to various pathogens.

The DR genes assayed in the present study only represent a small subset of the genes potentially involved the M. pinodes/pea interaction. Moreover, only a small number of RGAs has been mapped on the JI296 × DP map, and we cannot exclude the hypothesis that R genes that do not encode NBS-LRR genes may also be involved in resistance to M. pinodes. We also know that the methodology used did not enable us to test and check all the alleles of a gene family. Further analyses based on the choice of additional candidate genes and the use of complementary methodologies, including methods aiming at evaluating gene transcription after inoculation with M. pinodes (i.e., SSH, micro-arrays...), are under investigation and should allow to identify new candidate genes for resistance QTLs.

Conclusion

RGA and DR markers developed in this study (1) successfully amplify potentially functional genes, (2) as far as RGAs are concerned, may be closely linked to disease R genes or QTL for resistance to various pathogens, and (3) are PCR-based, and can therefore be easily transferred to any laboratory and are of good value for a further use in marker-assisted selection (MAS). Moreover, all the candidate gene markers developed in the present study could be potentially useful in identifying the molecular factors underlying resistance to other pea diseases.

Using the same RIL population for both the candidate gene approach and the QTL mapping study, we reported three co-localizations between candidate genes and QTLs for resistance to M. pinodes, which suggest that quantitative resistance to M. pinodes could be explained by an association of genes acting at different levels in the M. pinodes/pea interaction. Our results corroborate previous findings in other plant species where the molecular basis for quantitative resistance were reported to be involved in recognition (Kanazin et al. 1996; Pflieger et al. 1999), transcriptional regulation (Trognitz et al. 2002) or plant defense (Faris et al. 1999; Geffroy et al. 2000; Pflieger et al. 2001b, Trognitz et al. 2002). Nevertheless, further experiments are required to (1) confirm the three co-localizations between the candidate genes and the resistance QTLs, (2) discard co-segregations that may have occured by chance only (Pflieger et al. 2001a), (3) validate the implication of the candidate gene(s) in the phenotypic variation.