MtQRRS1, an R-locus required for Medicago truncatula quantitative resistance to Ralstonia solanacearum


  • Cécile Ben,

    1. INP, UPS, Laboratoire d'Ecologie Fonctionnelle et Environnement (Ecolab), ENSAT, Université de Toulouse, Castanet Tolosan, France
    2. CNRS, Ecolab, Castanet Tolosan, France
    Search for more papers by this author
  • Frédéric Debellé,

    1. INRA, Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR441, Castanet Tolosan, France
    2. CNRS, Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR2594, Castanet Tolosan, France
    Search for more papers by this author
  • Hélène Berges,

    1. INRA, Centre National de Ressources Génomiques Végétales (CNRGV), Castanet Tolosan, France
    Search for more papers by this author
  • Arnaud Bellec,

    1. INRA, Centre National de Ressources Génomiques Végétales (CNRGV), Castanet Tolosan, France
    Search for more papers by this author
  • Marie-Françoise Jardinaud,

    1. INRA, Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR441, Castanet Tolosan, France
    2. CNRS, Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR2594, Castanet Tolosan, France
    3. INP, ENSAT, Université de Toulouse, Castanet Tolosan, France
    Search for more papers by this author
  • Philippe Anson,

    1. INP, ENSAT, Université de Toulouse, Castanet Tolosan, France
    Search for more papers by this author
  • Thierry Huguet,

    1. INP, ENSAT, Université de Toulouse, Castanet Tolosan, France
    Search for more papers by this author
  • Laurent Gentzbittel,

    1. INP, UPS, Laboratoire d'Ecologie Fonctionnelle et Environnement (Ecolab), ENSAT, Université de Toulouse, Castanet Tolosan, France
    2. CNRS, Ecolab, Castanet Tolosan, France
    Search for more papers by this author
  • Fabienne Vailleau

    Corresponding author
    1. INRA, Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR441, Castanet Tolosan, France
    2. CNRS, Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR2594, Castanet Tolosan, France
    3. INP, ENSAT, Université de Toulouse, Castanet Tolosan, France
    Search for more papers by this author


  • Ralstonia solanacearum is a major soilborne pathogen that attacks > 200 plant species, including major crops. To characterize MtQRRS1, a major quantitative trait locus (QTL) for resistance towards this bacterium in the model legume Medicago truncatula, genetic and functional approaches were combined.
  • QTL analyses together with disease scoring of heterogeneous inbred families were used to define the locus. The candidate region was studied by physical mapping using a bacterial artificial chromosome (BAC) library of the resistant line, and sequencing. In planta bacterial growth measurements, grafting experiments and gene expression analysis were performed to investigate the mechanisms by which this locus confers resistance to R. solanacearum.
  • The MtQRRS1 locus was localized to the same position in two recombinant inbred line populations and was narrowed down to a 64 kb region. Comparison of parental line sequences revealed 15 candidate genes with sequence polymorphisms, but no evidence of differential gene expression upon infection. A role for the hypocotyl in resistance establishment was shown.
  • These data indicate that the quantitative resistance to bacterial wilt conferred by MtQRRS1, which contains a cluster of seven R genes, is shared by different accessions and may act through intralocus interactions to promote resistance.


Ralstonia solanacearum is the causal agent of bacterial wilt, one of the most important plant bacterial diseases worldwide (Genin & Denny, 2012; Mansfield et al., 2012). This soilborne pathogen attacks > 200 plant species, including many agriculturally important crops such as tomato, potato, tobacco, banana/plantain, cashew, papaya and olive trees (Hayward, 1991). Bacterial wilt has also been reported on several legumes, including winged bean (Psophocarpus tetragonolobus), common bean (Phaseolus vulgaris), cowpea (Vigna sinensis), and groundnut (Arachis hypogaea) (Hayward, 1994).

Ralstonia solanacearum penetrates host plants through root tips, wounds or sites of lateral root emergence, depending on the host plant species. The bacterium then colonizes the vascular system, where it releases exopolysaccharides that block water uptake, leading to wilting and eventually death of the plants (Genin, 2010). Owing to its protected location inside the host plant, its ability to survive for many years in the soil, and the presence of reservoir plants (Alvarez et al., 2008), R. solanacearum is difficult to control. Examples of monogenic or polygenic resistances to R. solanacearum have been characterized on various plant species. Quantitative trait loci (QTLs) controlling resistance to bacterial wilt have been identified in several solanaceous crops, such as tomato (Danesh et al., 1994; Thoquet et al., 1996a,b; Mangin et al., 1999; Wang et al., 2000, 2012; Carmeille et al., 2006), tobacco (Qian et al., 2012) and eggplant (Lebeau et al., 2013), as well as in model plants such as Arabidopsis thaliana (Godiard et al., 2003) and Medicago truncatula (Vailleau et al., 2007). Up to now, only two resistance genes have been identified: the A. thaliana ERECTA gene involved in polygenic resistance (Godiard et al., 2003), and the A. thaliana RRS1 gene involved in monogenic resistance and tolerance (Deslandes et al., 2002; van der Linden et al., 2012).

Medicago truncatula became a key model plant for studying legume biology, mostly because of its relatively small genome (500 Mbp), autogamous simple diploid genetics, high synteny with legume crops and short seed-to-seed generation time (Barker et al., 1990; Young et al., 2011). Genomic tools available for M. truncatula include the annotated genome sequence of the A17 reference line (Young et al., 2011), mutant populations (Tadege et al., 2009; Pislariu et al., 2012) and large collections of natural accessions (Ellwood et al., 2006a; Ronfort et al., 2006; Lazrek et al., 2009; Branca et al., 2011). Owing to its interactions with various soilborne pathogens and symbionts, M. truncatula is a suitable model to study root–microbe interactions in legumes. With respect to pathogens, to date, M. truncatula has mainly been studied in terms of its interaction with pathogenic fungi (Torregrosa et al., 2004; Kemen et al., 2005; Ellwood et al., 2006b; Foster-Hartnett et al., 2007; Anderson et al., 2010; Ramírez-Suero et al., 2010; Anderson & Singh, 2011; Ben et al., 2013) or oomycetes (Djebali et al., 2009; Pilet-Nayel et al., 2009).

To study bacterial wilt in legumes, we previously set up a pathosystem based on R. solanacearum inoculation of M. truncatula (Vailleau et al., 2007). We showed that the A17 reference line was susceptible to the R. solanacearum GMI1000 strain, while the M. truncatula F83005.5 line was resistant. The two lines exhibited contrasting responses in terms of wilting symptoms. A reduced bacterial population was also observed in the leaves of F83005.5, R. solanacearum multiplication being restricted by a factor of 1 × 105 compared with that of A17. A quantitative genetic analysis on A17 × F83005.5 recombinant inbred lines (RILs) enabled us to identify a major QTL for resistance towards R. solanacearum, now called MtQRRS1 (Medicago truncatula quantitative resistance to Ralstonia solanacearum 1). MtQRRS1 was mapped to chromosome 5, within a region of 5 cM(Vailleau et al., 2007).

In the present work, we have improved the characterization of the pathosystem and genetically dissected the MtQRRS1 locus. To analyze a putative new genetic source for resistance to bacterial wilt, we used the A17 × DZA315.16 RIL population that involves a different resistant parent from F83005.5. A major QTL that colocalizes with MtQRRS1 was detected in this new cross. Using near-isogenic lines (NILs) derived from the A17 × F83005.5 RIL population, we defined the genetic position of the MtQRRS1 locus more accurately. The fine genomic structure within the locus was then characterized by comparing the genomic sequences of the F83005.5 resistant line with those of the A17 susceptible line. Within the MtQRRS1 region, 15 candidate genes were predicted, including a cluster of seven R genes. Organ-level control for bacterial wilt resistance was also investigated through grafting experiments. In planta bacterial quantification combined with candidate gene expressions, both in hypocotyls and foliar tissues, were performed in parallel.

Materials and Methods

Plant material and growth conditions

Seeds of Medicago truncatula Gaertn. genotypes Jemalong A17, F83005.5 and DZA315.16 were provided by the SGAP laboratory (INRA, Mauguio, France; The A17 × F83005.5 (LR5) and A17 × DZA315.16 (LR4) RIL populations were described previously by Vailleau et al. (2007) and Torregrosa et al. (2004), respectively. To help refine the MtQRRS1 position, 12 heterogeneous inbred families (HIFs) of NILs were selected from progenies of the LR5.F7.113 RIL, which is still heterozygous at the locus. Progenies that genetically recombined for the MtQRRS1 region were identified through a microsatellite-assisted screening (Supporting Information, Table S1). Twenty-nine NILs in the F11 generation derived from these recombinant HIFs, and containing either A17 or F83005.5 alleles at the locus, were then used for disease evaluation in response to R. solanacearum. Seeds were germinated as described by Vailleau et al. (2007) and plants were grown in Jiffy pots for 10 d under 16 : 8 h, light : dark, 23 : 20°C conditions, at 170 μmol m−2 s−1.

Plant inoculation procedure, disease symptom scoring and in planta bacterial growth measurements

The R. solanacearum GMI1000 strain was grown at 28°C in BGT medium (Boucher et al., 1985). Plants were root-inoculated by cutting c. 1 cm from the bottom of the Jiffy pot, and the exposed roots were immersed for 20 min in a suspension containing 108 bacteria ml−1. The plants were then transferred to a growth chamber at 28°C (12 h of light) and 100 μmol m−2 s−1. The disease symptom assessment was performed as previously described (Vailleau et al., 2007). In planta bacterial growth was measured in leaves and hypocotyls of A17- and F83005.5-inoculated plants. Entire plants were harvested at 3 and 5 d postinoculation (dpi), sterilized with 70% ethanol for 1 min, rinsed three times in sterile water and patted dry. Hypocotyls and first leaves (monofoliate leaves) were separated, weighed, crushed, and placed in water before determining bacterial concentrations by dilution plating on a modified semiselective medium from South Africa (SMSA) (Elphinstone et al., 1996). At each time point, bacteria quantifications correspond to three replicates of three plants. One representative experiment out of three independent biological repeats is presented. Statistical analyses were performed by fitting a generalized linear model (GLM) using a Poisson regression.

Experimental design for evaluation of bacterial wilt resistance in RIL populations and in HIFs, and statistical analysis of phenotypic data

The evaluation of the LR4 RIL population (A17 × DZA315.16) for bacterial wilt resistance was performed on 96 RILs in a randomized block design with four plants per block using three to four blocks per RIL. The LR5 RIL population (A17 × F83005.5) was previously evaluated by Vailleau et al. (2007). In order to narrow down the MtQRRS1 locus, additional phenotyping was performed on 16 LR5.F8 RILs with two to three independent replicates of 18 plants. Twenty-nine LR5.F11 NILs were also evaluated for bacterial wilt resistance in a randomized block design with four plants per block using two to four blocks per NIL.

Statistical analyses of the phenotypic data were performed with R software (R Development Core Team, 2012). For each independent experiment, statistical comparisons of the mean disease scores between lines were done at the time point of the disease kinetics when the susceptible parental line A17 had a score of 3.5–4. For ANOVA of QTL genotypic means, a model with ‘genotype at the QTL’ and ‘line within QTL genotype’ fixed effects was used, including a block effect when required. The Student–Newman–Keuls test for multiple mean comparisons was done using the R package ‘agricolae’.

QTL analysis

Quantitative trait loci for resistance to R. solanacearum were detected in the A17 × DZA315.16 (LR4) RIL population by multiple QTL mapping (MQM) (Jansen, 1993; Jansen & Stam, 1994) using the ‘qtl’ R package (Broman et al., 2003; Arends et al., 2010; R Development Core Team, 2012). Mean disease scores, adjusted for block effect and combining F7 and F8 replicates, were computed at 3, 5, 7, 10, 12, 14, 17, 19 and 21 dpi, in 96 RILs. These disease scores were then used as variables for QTL detection. Empirical threshold values for the Log-likelihood (LOD) scores were determined by computing 5000 permutations (Churchill & Doerge, 1994). Depending on the phenotype, the critical LOD score to indicate QTL significance ranges from 3.2 to 3.7. Heritabilities were computed using variance-component methods, by equating mean squares to their expectations and using the weighted least-squares (WLS) method for variance estimations (Kearsey & Pooni, 1996).

MtQRRS1 fine mapping: DNA extraction and molecular marker analysis

To fine-map the target region, single sequence repeat (SSR) and single nucleotide polymorphism (SNP) markers were developed based on the polymorphisms identified by sequencing the A17 and F83005.5 lines (Table S2). Total genomic DNA was extracted from leaf tissue, according to the protocol described by Gherardi et al. (1998). PCR for molecular marker analysis was performed as described by Lazrek et al. (2009). For SNP genotyping, sequences were obtained on an ABI Prism 3700 machine using BigDye Terminator v3.1 technology according to the manufacturer's instructions (Applied Biosystems, Life Technologies, Foster City, CA, USA). Polymorphisms between lines were revealed through multiple alignments of nucleotide sequences using MEGA 4 software.

The F83005.5 BAC library

The F83005.5 BAC library was constructed as described by Goubet et al. (2012). It consists of 55 296 clones with an average insert size of 65 kb (estimated genome coverage 6.8×). This library was screened by hybridization on high-density arrays using 32P-labeled probes corresponding to molecular markers identified in the region of interest.

BAC sequencing, assembly and annotation

Bacterial artificial chromosome clone culture and DNA extractions were conducted according to the manufacturer's specifications (Macherey Nagel Midi prep + kit). BAC pools were sequenced at a final coverage of 35–40X using the Sequencing Kit on a Roche/454 GS FLX sequencer according to the manufacturer's instructions (Roche Diagnostics). The 454 unpaired reads were assembled using the Newbler assembler (Roche- This resulted in three to eight sequence contigs per BAC. These contigs were ordered and oriented by comparison to the A17 reference sequence. PCR amplifications followed by Sanger sequencing allowed sequencing gaps to be filled. Final sequences were submitted to GenBank (accession JX457607). Annotation was carried out by comparison to the A17 genome (Young et al., 2011).

Grafting experiments

Grafting experiments were performed according to the protocol described by Journet et al. (2001), using 3-d-old seedlings. Grafted plants were directly transferred onto slanted agar in square Petri dishes with an interface of paper as previously described (Vailleau et al., 2007). Four different types of grafted seedlings (shoot/root) were made – test grafts A17/F83005.5 and F83005.5/A17, and control grafts A17/A17 and F83005.5/F83005.5 – and were grown for 14 d with 16 h light at 23°C and 100 μmol m−2 s−1. Roots of successfully grafted plants were sectioned 1 mm from the root tip and inoculated with 300 μl of a suspension of R. solanacearum GMI1000 at 108 bacteria ml−1. Petri dishes were maintained inclined at an angle of 45° and dishes were sealed with Parafilm, with several incisions allowing gas exchange. The grafts were then transferred to a growth chamber at 28°C (12 h of light) and 100 μmol m−2 s−1.

RNA extraction and quantitative real-time polymerase chain reaction (qRT-PCR) analysis

Leaves and hypocotyls of plants grown and inoculated in identical conditions were used for in planta bacterial growth measurement and relative gene expression analysis. Samples were collected at 0, 1, 2, 3, 4 and 5 dpi from five plants inoculated with R. solanacearum or with water (mock). Total RNA was extracted from two biological replicates using the RNAeasy Plant Mini Kit (Qiagen) following the manufacturer's procedure and analyzed with a Bioanalyser (Agilent Technologies, Life Technologies, Santa Clara, CA, USA). Reverse transcription was performed with 1 μg of total RNA using the superscript reverse transcriptase II (Invitrogen) and anchored oligo(dT). Quantitative PCR was conducted using a LightCycler 480 (Roche) with the manufacturer's recommended conditions and primers shown in Table S3. Cycling conditions were as follows: 95°C for 5 min, 45 cycles at 95°C for 15 s and 60°C for 1 min. For each cDNA sample, threshold cycle (CT) values for all selected genes were normalized to the CT value of four endogenous genes (∆CT), whose expression was constant on the 96 cDNA samples analyzed. The median of ∆CT was calculated for data analysis using the ∆∆CT method, using the mock-inoculated condition as the reference (Livak & Schmittgen, 2001). Specificity and efficiency of the amplifications were verified by analyses of melting curves and standard curves, respectively. No amplification was obtained for Medtr5g037640 and Medtr5g037660 genes (Table S3).


A major QTL on chromosome 5 colocalizes to the same locus in two RIL populations

With the aim of identifying new sources of resistance, we used the Algerian DZA315.16 line that exhibited a high degree of resistance towards R. solanacearum GMI1000, similar to that previously described for the resistant line F83005.5 (Vailleau et al., 2007) (Fig. 1a,b). The LR4 RIL population (derived from a cross between DZA315.16 and the susceptible A17 line) was thus used to map the loci responsible for bacterial wilt resistance in DZA315.16. The distribution of the LR4 RILs and of the previously analyzed LR5 RILs (derived from a cross between F83005.5 and A17) (Vailleau et al., 2007) according to their mean disease score showed continuous variations (Fig. 1c,d), indicating quantitative and polygenic control of the resistance response. Compared with the parental lines, some RILs of the LR4 population showed lower or higher mean disease score values, revealing transgressive phenotypes (Fig. 1d). We thus conducted a QTL analysis on the mean disease score of 96 RILs of the LR4 population. The genetic analysis was done for different time points along disease kinetics (from 3 to 21 dpi) using a dense genetic map comprising 370 SSR markers (Ben et al., 2013). Broad-sense heritability was estimated as 68% for disease symptoms at 21 dpi, indicating that a significant part of the total phenotypic variation is a result of the genetic variation. A QTL was detected on chromosome 5 as early as 3 dpi (Fig. S1). This QTL reached a maximum plateau LOD score value (LOD = 8.0) at 7 dpi. It is localized between SSR markers mtic349, located on BAC CR962134, and MTE33, located on BAC AC136286, within a 1 LOD support interval of 14 cM. This QTL accounts for 17.6% of the phenotypic variation at 21 dpi (Table 1). An additional QTL was detected on chromosome 7, but not reaching a significant LOD score before 10 dpi (Fig. S1). This second QTL is located between SSR markers mtic1092, located on BAC AC119411, and MTE46, located on BAC AC122729, within a 1 LOD support interval of 6 cM. At 21 dpi, the LOD score at peak was 6.1 and this QTL accounts for 16.7% of the phenotypic variation (Table 1). The model considering the two QTLs acting additively represents 37% of the phenotypic variance at 21 dpi. The two QTLs contribute to resistance from DZA315.16 alleles. We previously showed that three QTLs contribute to resistance to R. solanacearum in the F83005.5 line using an A17 × F83005.5 RIL population, with a major QTL detected on chromosome 5 (Vailleau et al., 2007). As shown in Fig. S2, QTLs on chromosomes 3, 5 and 7 were detected with the monitoring of the mean disease score at all the time points evaluated after R. solanacearum inoculation. The main QTL on chromosome 5 accounts for 38.5% of the phenotypic variation with a maximum LOD score of 14.0.

Table 1. Quantitative trait loci (QTLs) detected in the F7 and F8 recombinant inbred line (RIL) populations of Medicago truncatula derived from A17 × DZA315.16
CrossTraitTrait heritability h2(H2)aChromosomeNearest markerPeak LOD scorebPeak (cM)1 LOD support interval (cM)QTL effectcPercentage of phenotypic variance
  1. a

    h2 narrow heritability, H2 broad-sense heritability.

  2. b

    Log-likelihood (LOD).

  3. c

    A negative sign indicates that the increasing resistance of the QTL alleles were contributed by the resistant parent.

  4. dpi, d postinoculation.

LR4Mean disease score at 21 dpi0.52 (0.68)5mtic3498.42417–31−0.4617.6
(A17[S] x DZA315.16[R])  7mtic2406.1149–15−0.4416.7
Figure 1.

Medicago truncatula LR5 (A17 × F83005.5) (a, c) and LR4 (A17 × DZA315.16) (b, d) mean disease scores after Ralstonia solanacearum root inoculation at 108 bacteria ml−1. (a, b) Mean disease scores for parental lines after bacterial inoculation. Standard error bars shown for LR5 (a) and LR4 (b) analysis correspond to the inocula-tion of 90 and 108 plants, respectively, of each parental line. This was done following a three- to four-block design, with a main plot consisting of four plants. (c, d) Frequency distribution of resistance to bacterial wilt based on the mean disease score at 19 d postinoculation (dpi) for the LR5 recombinant inbred lines (RILs) (c) and 21 dpi for the LR4 RILs (d). F83005.5, open circles; A17, closed circles; DZA315.16, open squares.

The LR5 and LR4 genetic maps can be aligned using of a defined set of shared genetic markers that also allows consensus genetic maps to be constructed (see Dias et al., 2011 as an example). In this way we compared the genetic position of the QTLs detected on chromosome 5 in the LR4 and LR5 populations. In both populations, the 1 LOD support interval for the QTL is located between SSR markers MTE32 (= mtic511, located on BAC AC119414) and MTE33 (= mtic932, located on BAC AC136286) (Fig. 2). We called this major QTL ‘MtQRRS1’ (for M. truncatula quantitative resistance to R. solanacearum). We then decided to focus on the characterization of MtQRRS1 in the A17 × F83005.5 cross.

Figure 2.

A major quantitative trait locus (QTL) identified for resistance to bacterial wilt in LR5 (a) and LR4 (b) on chromosome 5 of Medicago truncatula colocalizes at the same locus on both recombinant inbred line (RIL) populations (c). Log-likelihood (LOD) score for mean disease score, as identified by multiple QTL mapping, are plotted against chromosome 5 of M. truncatula for LR5 (a) and LR4 (b) RIL populations. Significance of QTL is indicated by a LOD score above the threshold values, determined by permutation analysis at the genome-wise significance level of 0.05. For each QTL the 1 LOD support interval and QTL peak are indicated by black and red vertical lines, respectively. The use of shared genetic markers allows a comparison of the dense genetic maps of chromosome 5 in LR5 and LR4 (c). The 1 LOD support of the QTL described in Vailleau et al. (2007) is drawn in dark green in LR5; and the reduced QTL interval is drawn in red. The QTL detected in LR4 is drawn in dark green. The reduced QTL interval in LR5 overlaps with 1 LOD support of the QTL of LR4.

Genetic confirmation that MtQRRS1 is a major QTL of resistance to R. solanacearum

To confirm MtQRRS1 as a resistance factor against R. solanacearum GMI1000, we further analyzed A17 × F83005.5 RILs and newly derived HIFs that showed genetic recombination events in the QTL interval. First, the mean disease scores after R. solanacearum inoculation of 16 RILs from LR5.F8 were re-evaluated on two independent replicates of 18 plants each (Fig. 3a). These 16 RILs have a crossover between the markers MTE32_mtic511 and MTE33_mtic932. They were further genotyped with new SSR markers in the QTL interval to refine the localization of the recombination events (Fig. 3b). Comparison of the mean disease score with the RIL genotypes at the different markers by ANOVA of QTL genotypic means validated the MtQRRS1 locus with a high significance level. Using LR5.F7.76 and LR5.F7.129 RILs, the zone of interest was significantly reduced to 0.6 cM, between SSR markers mtic911 and mtic895. A physical map of the locus was established using overlapping BAC clones of the A17 line. Sequencing of four BACs covering this region (mth2-21j3, mte1-73d17, mte1-65b11 and mth2-54i22) was carried out as part of the M. truncatula genome sequencing program (Young et al., 2011). The RIL LR5.F7.113, which was still heterozygous at the MtQRRS1 locus and homozygous elsewhere in the genome, was retained as the starting point to create and screen HIFs with distinct recombination events in the QTL interval. Indeed, its residual heterozygosity at the locus allowed us to follow the segregation of resistance into its progeny. The overall methodology is shown in Fig. 4. Among 14 LR5.F8.113 offspring plants generated by selfing, four lines (LR5.F8.113B, G, H and L) were fixed for the A17 ‘A’ allele and four other lines (LR5.F8.113D, J, M, and N) were fixed for the F83005.5 ‘B’ allele (Fig. 5). With the exception of the LR5.F8.113B line, statistical analyses of the observed mean disease scores confirmed the QTL identified on chromosome 5 of M. truncatula for quantitative resistance to bacterial wilt (Fig. 5).

Figure 3.

Validation of MtQRRS1 as a major resistance factor to Ralstonia solanacearum and reduction of the confidence interval of the quantitative trait locus (QTL). (a) Mean disease score of 16 LR5.F8 recombinant inbred lines (RILs) and of the parental lines, 14 d after R. solanacearum root inoculation at 108 bacteria ml−1. For each line, two biological replicates of 18 plants per line are shown. Mean comparison of ‘A’ (3.29 ± 0.10; black bars) and ‘B’ (1.17 ± 0.13; gray bars) genotypes at the MtQRRS1 locus shows a significant difference following ANOVA (P = 3.45 × 10−9). Multiple mean disease score comparisons (Student–Newman–Keuls test) were performed and different letters above each bar indicate significant differences between groups (P-value < 0.05). Genotype ‘H’, dark grey bars. (b) Genotypes and disease phenotypes of 16 LR5.F7 RILs at the MtQRRS1 locus. Genotype ‘A’ corresponds to A17 parental line, genotype ‘B’ corresponds to F83005.5 parental line, and genotype ‘H’ corresponds to a heterozygous locus. Results from LR5.F7.76 and LR5.F7.129 RILs narrowed the locus between single sequence repeat (SSR) markers mtic911 and mtic895. The LR5.F7.113 RIL is heterozygous at this locus. R, resistant phenotype; S, susceptible phenotype; I, intermediate phenotype.

Figure 4.

Strategy for fine-mapping of the Medicago truncatula quantitative resistance to Ralstonia solanacearum 1 (MtQRRS1) locus and identification of candidate genes. HIF, heterogeneous inbred family; NIL, near-isogenic line; QTL, quantitative trait locus; RIL, recombinant inbred line.

Figure 5.

Genotypes and disease phenotypes of the LR5.F8.113 heterogeneous inbred families (HIFs) and of the parental lines. (a) Genotypes of recombinant inbred lines (RILs) at eight simple sequence repeat (SSR) markers. Genotype ‘A’ corresponds to the A17 parental line, genotype ‘B’ corresponds to the F83005.5 parental line, and genotype ‘H’ corresponds to a heterozygous locus. R, resistant phenotype; S, susceptible pheno-type. (b) Phenotypes of eight RILs and of the parental lines, 14 d after Ralstonia solanacearum root inoculation at 108 bacteria ml−1. For each recombinant line, 30 plants were tested and for the parental lines, 48 plants were tested. Mean disease score comparison of ‘A’ and ‘B’ genotypes at the MtQRRS1 locus shows a significant difference following an ANOVA (P < 2 × 10−16). Multiple mean comparisons (Student–Newman–Keuls test) were per-formed and different letters above each bar indicate significant differences between groups (P < 0.05).

Fine mapping of MtQRRS1

The genomic location of MtQRRS1 was refined by combining phenotyping data and dense genotyping data for RIL which recombine within the region of interest. The LR5.F8.113E, F, I and K lines, which are still heterozygous at the locus, were used to generate HIFs (Fig. 5). Genotyping of 1,023 offspring of these lines with three SSR markers revealed 12 inbreds (LR5.F9.113) with recombinant events close to the MtQRRS1 locus (Fig. S3). Segregation of the markers used for screening were found to have the expected 1 : 2 : 1 ratio using a χ2 test (Table S1). For each one of these 12 lines, dense genotyping data were obtained with the use of additional molecular markers evenly spaced in the region to accurately determine the position of the crossovers (Fig. S3, Table S2). These F9 lines, which still contained residual heterozygosity, were selfed to generate 130 inbred lines (LR5.F10.113). Among those, 56 NILs were identified as homozygous at the locus (27 with the A17 ‘A’ parental allele, and 29 with the F83005.5 ‘B’ parental allele) and were then selfed. Analysis of the phenotypes of 29 randomly chosen lines out of the 56 is shown in Fig. 6. All the NILs carrying the A17 allele (‘A’ genotype) between SNP markers 21j3.2 and 73d17.3 had scores that were similar to the susceptible parent. By contrast, all the NILs carrying the F83005.5 allele (‘B’ genotype) between these markers displayed a mean disease score statistically similar to the resistant parent (P-value < 2 × 10−16). These results confirmed the meaningful contribution of the MtQRRS1 locus to bacterial wilt resistance. Comparative analysis of the phenotypes and of the genotypes allowed the locus to be narrowed down to a 64 kb region (Fig. 7) according to the A17 M. truncatula genome assembly v3.5, released by the M. truncatula genome project (

Figure 6.

Combined analysis of disease phenotypes and genotypes delineates the MtQRRS1 locus between SNP markers 21j3.2 and 73d17.3. In the upper part, mean disease score of 29 LR5.F11.113 near-isogenic lines (NILs), 12 d after Ralstonia solanacearum root inoculation at 108 bacteria ml−1. For each NIL, eight to 16 plants were used following a two- to four-block design, with a main plot consisting of four plants. For the A17 and F83005.5 parental lines, 42 plants were phenotyped as controls. Mean disease score comparison of ‘A’ (2.87 ± 0.12; black bars) and ‘B’ (1.12 ± 0.11; gray bars) genotypes at the MtQRRS1 locus shows a significant difference following ANOVA (P < 2 × 10−16). In the bottom part, genotypes of the same NIL on the MtQRRS1 locus are shown. Genotype ‘A’ corresponds to the A17 parental line and genotype ‘B’ corresponds to the F83005.5 parental line.

Figure 7.

Genetic and physical maps of the MtQRRS1 locus pinpoint a zone with 15 candidate genes with a concentration of Toll-interleukin-1 receptor (TIR) domains. (a) Linkage map of chromosome 5 constructed using 111 F7 recombinant inbred lines (RILs) and 129 F8 RILs and focusing on 15 RILs showing crossovers in the zone of interest. The MtQRRS1 locus was mapped between simple sequence repeat (SSR) markers mtic911 and mtic895. The four A17 bacterial artificial chromosomes (BACs) covering this region were sequenced: 1, mth2-21j3; 2, mte1-73d17; 3, mte1-65b11; and 4, mth2-54i22. (b) Fine-mapping of the MtQRRS1 locus between single nucleotide polymorphism (SNP) markers 21j3.2 and 73d17.2. The number of recombination events between each pair of markers is indicated under the arrows. (c) High-resolution genetic mapping identifies two overlapping BACs, mth2-21j3 and mte1-73d17 (AC134049.24 and CR954189.3 in GenBank, respectively) and narrows the MtQRRS1 locus down to a 64 kb region. (d) Fifteen predicted genes at the MtQRRS1 locus according to the Medicago truncatula genome project ( Black arrows indicate R genes with NBS-ARC/LRR (nucleotide-binding adaptor shared by Apaf1, certain R genes and CED4/leucine rich repeat) domain (Medtr5g037590) or TIR domain (the others).

Fifteen candidate genes for the MtQRRS1 locus, including a cluster of seven R genes

The M. truncatula genome annotation v3.5 ( shows that 15 genes and two transposons are predicted in the 64 kb target region distributed on two BAC clones (mth2-21j3, GenBank ID AC134049.24 and mte1-73d17, GenBank ID CR954189.3) (Fig. 7, Table 2). Among these 15 genes, six are of unknown function (Medtr5g037640.1, Medtr5g037690.1, Medtr5g037740.1, Medtr5g037750.1, Medtr5g037760.1, and Medtr5g037770.1), one is predicted to have a DUF223 domain that can be found in nucleic acid-binding proteins with an OB-fold but whose function is unknown (Medtr5g037660.1), and one is an F-box family gene (Medtr5g037680.1). In addition, we observed a cluster of seven possible R genes (Medtr5g037590.1, Medtr5g037610.1, Medtr5g037630.1, Medtr5g037650.1, Medtr5g037700.1, Medtr5g037710.1, and Medtr5g037720.1), of which Medtr5g037630.1 and Medtr5g073650.1 were duplicated in tandem (Fig. 7). Six out of these seven genes contain only one Toll-interleukin-1 receptor (TIR) domain. Medtr5g037590.1 contains an NB-ARC (nucleotide-binding adaptor shared by Apaf1, certain R genes and CED4) domain and a leucine rich repeat (LRR) domain (Table 2). TIR, NB-ARC and LRRs are typical domains found in plant resistance genes (van der Biezen & Jones, 1998; Burch-Smith & Dinesh-Kumar, 2007; Takken & Goverse, 2012). In order to characterize sequence polymorphisms between the resistant and the susceptible lines, we created a BAC library in the F83005.5 resistant line. Using this BAC library, the corresponding 64 kb region was isolated and sequenced from that line, after physical mapping of four BACs. Sequence comparison with A17 revealed numerous sites of polymorphism, highlighting SNPs and short indels (insertions/deletions) in coding sequences (Table 2). Importantly, we did not detect any large insertion in F83005.5 in the 64 kb region in comparison to A17 that could have corresponded to a specific coding sequence in the resistant line. Among the numerous polymorphisms observed between the susceptible and the resistant genotype sequences, we can report numerous nonsynonymous SNPs; large deletions in F83005.5 exons in Medtr5g037680.1 and in Medtr5g037690.1; one putative gene in A17 that is not identified in F83005.5 (Medtr5g037660.1); and, interestingly, a two-nucleotide deletion in the F83005.5 sequence leading to a putative truncated protein for Medtr5g037590.1, devoid of its LRR domain because of the appearance of a stop codon. Only six expressed sequence tags (ESTs) were identified among the 15 candidate genes (Table 2).

Table 2. Annotated genes in the MtQRRS1 locus on chromosome 5 in A17, according to the Medicago truncatula genome project (
Gene IDPhysical location on chromosome 5PFAM domainsaSynonymous SNPbNon synonymous SNPbIndelbEST evidencec
  1. EST, expressed sequence tag; NB-ARC, nucleotide-binding adaptor shared by Apaf1, certain R genes and CED4; LRR, leucine rich repeat; TIR, Toll-interleukin-1 receptor.

  2. a

    NCBI BLAST search.

  3. b

    Single nucleotide polymorphism (SNP) and indel (insertions/deletions) located in exons. The SNP and indel information is from sequencing bacterial artificial chromosomes (BACs) derived from F83005.5.

  4. c

Medtr5g037590.11601992116023481NB-ARC domain, LRR117−2/
Medtr5g037610.11603379916034953TIR domain96+2 BF643381
Medtr5g037630.11603776516040308TIR domain72/ EY478706
Medtr5g037640.11604371516044354Unknown protein12//
Medtr5g037650.11604470616047605TIR domain35/ EY475925
Medtr5g037660.11605026416053354DUF223Absent in F83005.5 //
Medtr5g037680.11605915316061581F-box-associated FBZA124−21; deletion removing 520 bp at the end of the gene/
Medtr5g037690.11606537216066397Unknown protein53Deletion removing 335 bp at the end of the gene EX529168
Medtr5g037700.11606726516070592TIR domain10/ CF068018
Medtr5g037710.11607261616073984TIR domain10/ GE346813
Medtr5g037720.11607499316077849TIR domain01//
Medtr5g037740.11608180016082495Unknown protein22//
Medtr5g037750.11608403016084732Unknown protein00//
Medtr5g037760.11608543716087115Unknown protein01//
Medtr5g037770.11608730716087952Unknown protein01//

Hypocotyls: a putative site for establishment of R. solanacearum resistance

Previous data concerning in planta propagation of R. solanacearum showed lower bacterial populations in the leaves of the F83005.5 resistant line compared with those of the A17 susceptible line, whereas bacterial colonization in root systems was not significantly different (Vailleau et al., 2007). To further dissect the organ-level control of the resistance to bacterial wilt in M. truncatula, disease establishment was monitored (Fig. 8a) and, in parallel, in planta bacterial growth was measured in hypocotyls and leaves of both lines, at 3 and 5 dpi (Fig. 8b). Whatever the plant tissue under study, the multiplication of R. solanacearum was significantly lower in the F83005.5 line compared with the A17 susceptible line (P-value = 0.023 at 3 dpi and P-value = 0.002 at 5 dpi). Likewise, whatever the line under study, the multiplication of R. solanacearum was significantly lower in leaves compared with hypocotyls (P-value = 0.008 at 3 dpi and P-value = 0.003 at 5 dpi). Taken together, these results suggest that the hypocotyls and the lower stem tissues may contribute to bacterial wilt resistance in F83005.5 by reducing bacterial load in the tissues.

Figure 8.

The hypocotyl tissues contribute to the resistance conferred to Ralstonia solanacearum by the MtQRRS1 locus. (a) Mean disease scores of the A17 susceptible line and of the F83005.5 resistant line after bacterial inoculation. Standard error bars correspond to the inoculation of 18 plants. Two biological replicates were scored. F83005.5, open circles; A17, closed circles. (b) Measurement of in planta growth of R. solanacearum in A17 (closed bars) and in F83005.5 (open bars) lines, 3 d (d3) and 5 d (d5) after inoculation, in hypocotyls (H) and leaves (L), respectively. Standard deviations of three replicates of three plants each are shown from one representative experiment. Statistical analysis on three biological repeats found a significantly lower bacterial multiplication in the F83005.5 line compared with the A17 line in the hypocotyls and leaves (P = 0.023 and P = 0.002 at 3 and 5 d postinoculation (dpi), respectively). (c) Pictures of the typical phenotypes after R. solanacearum inoculation on Medicago truncatula grafted plants. Bacterial inoculation was performed 14 d after grafting and disease symptoms were scored 10–15 d later. In the grafting type, the plant before the slash (/) is the shoot and that after the slash is the root. n, number of plants observed.

In vitro reciprocal grafts between the A17 and F83005.5 lines, as well as the A17 and F83005.5 self-grafted plants, were generated (Fig. 8c). Successful grafts exhibiting green cotyledons and numerous secondary roots 1–2 wk after grafting were observed in 10% of the cases with F83005.5 roots and at a lower rate of 5% with A17 roots. Such composite plants were then inoculated with R. solanacearum strain GMI1000. Depending on the genotypes of the root and the shoot, two response profiles were observed at 14 d after root inoculation: a chlorosis of the aerial part followed by the death of the plant for the A17/A17 and F83005.5/A17 (shoot/root) grafts; and a development of the aerial part of the plant with the appearance of new trifoliate green leaves for the F83005.5/F83005.5 and A17/F83005.5 grafted plants (Fig. 8c). These grafting experiments demonstrate that resistance to wilting symptoms is mainly controlled by the root and/or the hypocotyl. Taken together, the results of in planta bacterial growth measurements in distinct plant organs and the disease phenotypes of chimeric grafted plants suggest that the resistance to bacterial wilt in M. truncatula may develop in F83005.5 hypocotyls and might be related to a strong decrease in bacterial populations in shoots.

To further characterize the 15 candidate genes predicted at the MtQRRS1 locus, a comparative analysis of their expression profiles was carried out in the hypocotyls and in the leaves in both the susceptible and the resistant lines, inoculated with R. solanacearum or mock-inoculated. qRT-PCR was performed on a time course study at 0, 1, 2, 3, 4, and 5 dpi. To characterize molecular response to the pathogen, the expression patterns of genes linked to the ethylene (ET) and salicylic acid (SA) pathways (Gao et al., 2008) were analyzed at the same time points. The expression of a candidate ET-responsive gene, HEL (hevein-like protein) and of a candidate SA-responsive gene, PR5 (pathogenesis-related protein 5), were strongly induced in the hypocotyl tissues compared to lower expressions in the leaves (Fig. 9). The transcript abundance of HEL increased in F83005.5 hypocotyls at 4 and 5 dpi, with a peak at 4 dpi, whereas in A17 a strong induction was not observed until 5 dpi (Fig. 9a,b). On the other hand, a huge increase of PR5 expression was observed in the A17 hypocotyls at 5 dpi, compared with lower expression levels detected in leaves (Fig. 9c,d). These data confirmed an induction of the plant defense responses in the hypocotyls after R. solanacearum root inoculation in both lines. In parallel, the analysis of the expression profiles of the 15 candidate genes in the MtQRRS1 locus was performed. For the 13 candidate genes whose expression was successfully analyzed, we could not detect any reproducible differences in expression patterns (data not shown). During the time-course of this experiment, the classical wilting responses upon inoculation were observed (Fig. 8a), with a significant restriction of bacterial multiplication in the leaves and in the hypocotyls of the resistant line (Fig. 8b), and a strong induction of putative marker genes in hypocotyls (Fig. 9).

Figure 9.

Expression profiles of genes representing ethylene (HEL) and salicylic acid (PR5) signaling pathways in hypocotyl and leaf tissues of A17 and F83008.5 Medicago truncatula plants, after Ralstonia solanacearum root inoculation. (a, b) HEL, hevein-like protein; (c, d) PR5, pathogenesis-related protein 5. Gene expression levels determined by quantitative real-time polymerase chain reaction (qRT-PCR) are relative expression units, normalized to the median of four endogenous genes (ubiquitine, EF-1α, GADPH, and helicase). Values represent the average fold difference of R. solanacearum-inoculated plants relative to mock-inoculated plants as determined by the math formula method of Livak & Schmittgen (2001). Bars indicate standard deviation (= 2). A17, closed bars; F83005.5, open bars; dpi, d postinoculation.


To date, plant genes underlying partial and quantitative resistance to pathogenic organisms remain largely unknown. Partial resistance genes against fungal pathogens were identified in wheat that involved a kinase-START gene (Fu et al., 2009) and a putative ABC transporter (Krattinger et al., 2009). A major QTL for resistance to rice blast disease was cloned in rice, encoding a mutated proline-rich protein (Fukuoka et al., 2009). Yet, more complex situations are described. For example, a quantitative fungal resistance in rice is partly controlled by a cluster of 12 germin-like protein genes (Manosalva et al., 2009).

In this study, we genetically confirmed MtQRRS1 as a locus of quantitative resistance to bacterial wilt caused by R. solanacearum in M. truncatula F83005.5. QTL analysis of a cross (LR4) involving the resistant line DZA315.16 suggests that the MtQRRS1 locus, originally identified from the cross with F83005.5, might be conserved in DZA315.16. However, from the resolution of our genetic analysis of the MtQRRS1 locus in LR4, we cannot definitively conclude whether the gene, or genes, underlying each QTL are the same in both populations. An allelism test between F83005.5 and DZA315.16 may help to determine whether resistance is conferred by these two genetic loci identified in these two accessions.

Heterogeneous inbred families developed from several RILs of the LR5 population allowed us to fine-map MtQRRS1 and to reduce the locus to 15 putative candidate genes. The annotation of this region identified seven putative R genes displaying either an NB-ARC, an LRR or a TIR domain, which are characteristic of genes associated with plant disease resistance. In soybean, the number of NBS-LRR genes is significantly correlated to the presence of disease resistance QTLs in the 2 Mb-surrounding regions. Also, several recently duplicated regions that contain NBS-LRR genes harbored disease resistance QTLs on both sides of the duplication (Kang et al., 2012). In M. truncatula, the major QTL conferring resistance to Aphanomyces euteiches, AER1, was identified on chromosome 3 and shows the presence of a cluster of NBS-LRR genes (Pilet-Nayel et al., 2009). AKR, an aphid resistance locus, maps to a region flanked by CC-NBS-LRR genes (Klingler et al., 2005). In our study, only one gene with the NB-ARC and LRR domains can be classified as a classical R gene (Medtr5g037690.1). The polymorphism in this gene detected between A17 and F83005.5 (absence of the LRR domain in F83005.5) defines it as a good candidate gene, potentially involved in R. solanacearum resistance. In addition, six candidate genes have a single TIR domain each. Burch-Smith et al. (2007) showed that the single TIR domain encoded by the N gene in tobacco is sufficient for its association with p50 (the helicase domain of the Tobacco mosaic virus replicase proteins), and that specificity of the resistance was dependent of the sequence of the TIR domain. However, in spite of the numerous TIR-domain-containing nonNBS-LRR proteins, no clear role of these domains has been demonstrated (Burch-Smith & Dinesh-Kumar, 2007). The MtQRRS1 locus contains a putative F-box family gene, which should also be considered as a potential candidate for resistance to bacterial wilt based on previous work by Djebali et al. (2009), which describes a cluster of nine F-box genes in a major QTL of resistance against A. euteiches. Among the seven clustered R genes found in our study, Medtr5g037630.1 and Medtr5g037650.1 are duplicated in tandem. Previous studies identified resistance genes acting in tandem. The rice blast resistance was shown to be conferred by a combination of two nonTIR NBS-LRR class genes, Pikm1-Ts and Pikm2-TS. (Ashikawa et al., 2008). In A. thaliana, two adjacent TIR-NBS-LRR genes have been identified acting cooperatively to specifically confer resistance to Peronospora parasitica isolate Cala2. These two genes, RPP2A and RPP2B, are part of an R-gene cluster containing two other TIR-NBS-LRR genes, and which may potentially recognize other pathogens or variants (Sinapidou et al., 2004). In A. thaliana, two adjacent TIR-NBS-LRR genes, RRS1 and RPS4, confer resistance, in tandem or individually, to three distinct pathogens (Narusaka et al., 2009).

We have to consider the hypothesis that the final responses explaining R. solanacearum resistance could be the effect of two or more genes of the locus. Such a situation was experimentally validated for a resistance QTL in rice. A cluster of 12 germin-like protein gene members acts against two different fungal pathogens (causal agents of rice blast and sheath blight diseases) and it was shown that, as more germin-like genes were suppressed, disease susceptibility of the plants increased (Manosalva et al., 2009). So, what could be the advantage of a cluster of R genes for the plant? The NB-ARC domains of APAF-1 in mammals, DARK1 in Drosophila and CED4 in C. elegans, have been shown to homo-oligomerize upon activation, to potentially provide a platform for nucleotide binding and to activate resistance mechanisms (van Ooijen et al., 2007). As several lines of evidence suggest that R proteins could act through multimeric complexes, we cannot exclude the possibility that several R proteins at the MtQRRS1 locus could act together in dynamic interactions. The M. truncatula genome has 764 NBS-LRR genes and almost 90% of them occur in clusters and superclusters (Young et al., 2011). Could the presence of R genes in superclusters participate in the emergence of new quantitative resistances and enhance the evolution of these genes, contributing to the plasticity required by the host to counteract ongoing pathogen evolution? Simultaneous transfer to A17 of several F83005.5 genes at the locus using Agrobacterium tumefaciens transformation might help test the hypothesis of a coordinate function of several genes under MtQRRS1. Genetic studies of mutants could also be used to analyze the potential implication of individual genes.

Relationships between disease symptoms and pathogen spread into the tissues define the tolerance and resistance concepts. For vascular wilt diseases, resistance is generally considered to be a state in which pathogen growth in the xylem is very limited and few, if any, symptoms occur (Robb et al., 2007). We showed that bacterial load in aerial tissues is reduced in the F83005.5 resistant line (by five orders of magnitude), in contrast to the situation described for root tissues (Vailleau et al., 2007). The hypocotyl is thus suggested as a critical zone for reducing bacterial growth. This situation is not encountered for bacterial wilt response in tomato, where the difference in bacterial populations between the resistant and susceptible lines is not greater than two orders of magnitude in the aerial tissues (Ishihara et al., 2012), with bacterial load close to that of the susceptible line of M. truncatula. Similar to our experimental results, analysis of systemic spread and vascular response to Verticillium in Brassica napus revealed that the resistance mechanisms prevent the aerial parts but not the roots from being invaded (Eynck et al., 2009). Whether or not a causal relationship exists between reduction in bacterial load and resistance to wilting symptoms remains to be established for resistance to bacterial wilt in M. truncatula. In any case, our data suggest that molecular events leading to resistance might occur in the hypocotyl. As an example of tissue-specific expression of disease resistance genes, it was shown that Crr1a, a TIR-NB-LRR gene that confers resistance to clubroot in Brassica rapa, is expressed in the stele and cortex of hypocotyls and roots, where secondary infection of the pathogen occurs (Hatakeyama et al., 2013).

Although contrasting wilt symptoms and defense-related gene expression were observed during experiments with A17 and F83005.5, we could not detect reproducible differential expression of the genes at the MtQRRS1 locus in hypocotyls or leaves in response to R. solanacearum inoculation. This suggests that, at the MtQRRS1 locus, differences in the gene expressions between A17 and F83005.5 may occur earlier than the sampled time points and/or in specific cells and/or that they might be subjected to strong environmental effects. Complex gene expression patterns were also observed for rice germin-like gene members underlying a QTL involved in response to blast disease, where not all the genes were expressed and not all at the same time points (Manosalva et al., 2009). For M. truncatula, Yang et al. (2008) described the identification of a monogenic TIR-NBS-LRR resistance gene against Colletotrichum trifolii, the RCT1 (resistance to C. trifolii) gene. The resistance/susceptible response of the plant is explained by alternative splicing and/or sequence-level polymorphism with 27 SNPs identified (Yang et al., 2008), and no significant differences in expression patterns were described between susceptible and resistant lines. Thus, we cannot exclude the implication of alternative splicing at the MtQRRS1 locus in the regulation of the observed resistance towards R. solanacearum. As the use of M. truncatula as a surrogate was a success in the cloning of RCT1, the characterization of the MtQRRS1 locus could also be useful in identifying orthologs in crops to set up durable strategies for the management of bacterial wilt in the field.


We thank Clare Gough and Sylvain Raffaele for critically reading the manuscript. We thank Jean Marie Prosperi for providing the A17 and F83005.5 M. truncatula seeds. We are also grateful to Michèle Ghérardi, Redouane Behima, Philippe Matter, Marianne Prevot and Amélie Estachy for designing and genotyping SSR or SNP markers. We thank Annie Perrault for technical assistance with genotyping LR5.F9.113 RIL, and Marie José Tavella and Christopher Leroux for help with inoculations. We thank Jerome Gouzy for help in F83005.5 sequence analysis. We also thank Etienne Pascal Journet for advice on grafting experiments and Ralph Pawlowiez for technical assistance with grafting. Sequencing and gene expression analysis were done with the GeT-PlaGe platform facilities ( We thank Nevin D. Young and Fabien Chardon for useful discussions. We acknowledge FR ‘Agrobiosciences, Interactions & Biodiversité’ ( for financial support. Part of the work was performed at the LIPM that is part of the Laboratoire d'Excellence (LABEX) entitled TULIP (ANR-10-LABX-41).