Medicago truncatula SARDI core collection accessions were previously screened for their response to three P. medicaginis isolates. Accession SA27063 was resistant to P. medicaginis OMT5, whereas A17 and SA3054 were susceptible 18. When leaves of three-week-old SA27063 plants were spot inoculated, the fungus was limited to the inoculation site and no chlorosis or necrosis was observed 7–10 days post infection (dpi, Figure 1a). In susceptible interactions the fungus successfully penetrated host cells and spread beyond the inoculation site, accompanied by a halo of chlorosis ahead of the infection zone. By 7–10 dpi infected leaves were entirely chlorotic or necrotic and pycnidia were apparent (Figure 1b, e). Similar symptoms were observed following spray inoculation: small, microscopic hypersensitive response-like lesions and no chlorosis of leaves occurred in resistant SA27063 (10 dpi; Figure 1c); in the susceptible interactions disease symptoms were visible to the naked eye at 7 dpi. P. medicaginis colonised the surrounding plant tissue, resulting in macroscopic necrotic spots on both leaves and stems surrounded by spreading chlorosis (10 dpi; Figure 1d). As early as 10 dpi P. medicaginis produced pycnidia on susceptible leaves.

Histological staining using trypan blue or DiOC6 was performed to examine differences in the infection process between resistant and susceptible accessions at the cellular level (Figures 2a–e). Following inoculation, P. medicaginis spores germinated on the leaf surface and successful penetration of host cells was observed as early as 6 hours post inoculation (hpi). In both interactions, penetration attempts occurred by three routes; directly through stomata followed by penetration of the underlying mesophyll cells (Figure 2a), directly through the epidermal cells (Figure 2c), or between epidermal cells (Figure 2d). Following penetration, fungal colonisation proceeded rapidly in the susceptible accessions SA3054 and A17, whereas in the resistant accession SA27063 hyphal development was limited to individual or a few epidermal cells. Reddish-brown diaminobenzidine-tetrahydrochloride (DAB) staining to detect endogenous H₂O₂ production in cells at or around the site of penetration was observed in accession SA27063 (Figure 2b, c). However, H₂O₂ was inconsistently associated with penetration attempts and was also evident in susceptible accessions. Autofluorescence was also observed around the site of infection in both the susceptible and resistant interaction, indicating the production of phenolics (Figure 2d, e).

Accession SA27063 was crossed with susceptible accessions A17 and SA3054 to create two F₂ mapping populations. F₁ individuals of both crosses showed P. medicaginis OMT5 disease symptoms equivalent to the susceptible parents, indicating recessive resistance. F₂ individuals were screened for their disease phenotype in both the SA27063 × A17 (n = 92) and SA27063 × SA3054 (n = 94) populations. Both populations differed significantly from a normal distribution (P_W < 0.0001; based on the Shapiro and Wilk test 22, which tests the null-hypothesis that a given sample derives from a normally distributed population).

The SA27063 × A17 population was not chosen for further genetic studies due to poor fertility, the presence of a reciprocal translocation between chromosomes four and eight in accession A17 23, and a large proportion of aberrant F₂ individuals (Table 1). In the SA27063 × SA3054 cross, F₃ families (number of individuals per family ≥ 16) were phenotyped to confirm individual F₂ phenotypes. When the same F₃ families were inoculated between different experiments, mean disease scores did not alter significantly showing the phenotyping method was reliable and reproducible. The proportions of SA27063 × SA3054 F₃ families distributed against the mean disease score (Figure 3a) was significantly different from a normal distribution (P_W < 0.0001). The relationship between the families' means and the variance within each family is depicted in Figure 3b. This relationship was evaluated using the Fain's test which predicts a pattern of maximum variability in intermediate families whenever at least one locus with major effect is involved 2425. Analysis of the F₃ family disease scores using Fain's test revealed that the quadratic term was highly significant (P_/t/ = 0.0027; Figure 3b), indicating one or a few loci with major effects were involved in resistance to P. medicaginis.

Genetic maps for the SA27063 × SA3054 and SA27063 × A17 mapping populations were created using both gene-based and microsatellite markers. Markers were initially selected to be evenly distributed over each linkage group and were obtained from several published sources 2627282930. A total of 115 markers were characterised for the SA27063 × SA3054 (n = 94) and 78 for SA27063 × A17 (n = 92) populations.

Forty-six and 58 markers for SA27063 × SA3054 and SA27063 × A17 respectively were single nucleotide polymorphisms (SNPs). The majority of these SNPs were converted to cleaved amplified polymorphic sequences [CAPS, 31]. SNPs that could not be assayed as CAPS were genotyped using SNapShot (AB) single fluorescent nucleotide extension. A total 53% of markers were polymorphic between SA27063 and A17 and 35% between SA27063 and SA3054. Marker polymorphism data, primer sequences, and references for SA27063 × SA3054 and for SA27063 × A17 markers have been provided [see Additional files 1 and 2].

The genetic maps for SA27063 × A17 and SA27063 × SA3054 are shown in Figures 4 and 5, respectively. As expected, the SA27063 × SA3054 map was composed of eight linkage groups, in accordance with the basic number of chromosomes in M. truncatula (x = 8). The map spans a total of 488.3 centimorgans (cM) with an average distance between markers of 4.2 cM. Very few published markers were polymorphic between SA27063 and SA3054 on the lower half of linkage group 2.

In contrast to the SA27063 × SA3054 map, the SA27063 × A17 map was organized into seven linkage groups, with LG4 and LG8 forming a single linkage group (Figure 4). These linkage groups have previously been reported as single linkage groups 293032. The establishment of new linkage relationships between LG4 and LG8 combined with observations of semisterility in hybrids involving A17 provides strong evidence that A17 bears a reciprocal translocation, which differentiates it from other M. truncatula accessions. This reciprocal translocation involves the lower arm of chromosome 4 and the lower arm of chromosome 8 23. All publicly available markers in the middle of LG7 were monomorphic and LG7 was therefore split into two small linkage groups named LG7a and LG7b. The SA27063 × A17 genetic map spans a total of 497.8 cM with an average distance between markers of 6.4 cM. The linkage groups of the genetic maps presented here are similar in length to previously published maps 2930 with the exception of LG3. In SA27063 × SA3054 the distance between DK501R and CysPR1 was 50.9 cM, whereas in SA27063 × A17 this distance was 79.6 cM.

Two major quantitative trait loci (QTLs) for resistance to P. medicaginis OMT5 were identified, one in each mapping population. In SA27063 × A17, the effect of the QTL (d = 1.4) was significantly different from zero (P < 0.00001; LOD = 7.37). The locus was named

r

n

r

n

To initiate fine mapping of rnpm2 on the long arm of chromosome eight, an additional 434 F₂ individuals (524 in total) were screened for recombination within a 8.2 cM interval between markers MtB294, MtB174 and MtB139 (Figure 5). F₃ families from the selected recombinant F₂ individuals were phenotypically characterised. As mentioned above, locus rnpm2 explains 29.6% of the total variance in the population's response to the fungal infection when measured in a 1–5 scale. However, recombinant families that showed the SA27063-like (resistant) phenotype correlated with homozygous SA27063 alleles at rnpm2 in 78.5% of instances. The resistant parent consistently showed little or no visible necrotic symptoms, while disease progression in the susceptible parent was variable and ranged from 2.5 – 4.5. The variance in disease scores was therefore partly a function of the continuum of mean quantitative susceptible disease scores and the sum of disease scores in heterozygous F₃ families. Recombination break point analysis was therefore used to fine map rnpm2, with phenotyping data from thirty-eight informative F₃ families in conjunction with further polymorphic markers to resolve recombination break points. The location of rnpm2 was mapped to a 0.8 cM interval between markers h2_16a6a and h2_21h11d (Figure 6).

To ensure even germination, seeds were scarified with sand using a mortar and pestle, transferred to a Petri dish lined with blotting paper, and irrigated with sterile water. The seeds were kept at 4°C for 48 h, and germinated at room temperature overnight. Seedlings were sown in vermiculite (Richgro Garden Products, Jandakot, Western Australia 6164), fertilized twice a week using Optigrow fertilizer (Growth Technology, O'Connor, Western Australia 6163) and grown under natural light in a temperature controlled greenhouse (22°C day, 18°C night).

Parental M. truncatula accessions were identified from the disease phenotype screens described by as being either resistant (SA27063) or susceptible (SA3054 and A17) to P. medicaginis OMT5 18. Crosses between SA27063 and either A17 or SA3054, as pollen donors, were obtained by a manual crossing procedure 32. F₁ seeds were collected and heterozygosity verified with genetic markers polymorphic between the parental accessions.

P. medicaginis OMT5 inoculum was prepared by growing the isolate on wheat meal agar plates (WMA, 12 g ground wheat meal, 12 g agar, and 1 litre of distilled water). The plates were incubated at 22°C for 28–42 days under U/V. Conidia were harvested by incubating plates with 10 ml milliQ water for 20 min. and the suspension filtered through a glasswool syringe. Inoculations of were performed at the fourth trifoliate stage. Two methods of inoculation were used: spraying with an artists airbrush to runoff (2 × 10⁶ spores/ml, 0.05% Tween 20), using a rotating platform to ensure an even distribution over plant surfaces; and spot inoculation with 10 μl droplets of spore suspension (1 × 10⁶ spores/ml, 0.05% Tween 20) on five leaves per plant. These were the unifoliate, two leaflets of the first trifoliate, and two leaflets of the second trifoliate. The third leaflets of the second and third trifoliates were mock inoculated. To ensure high humidity to stimulate conidia germination, plants were placed in sealed propagator trays for 48 hours.

F₂ individuals were spot inoculated and macroscopically evaluated at 7 days post inoculation (dpi), then rescored at 10 dpi to confirm more resistant disease reactions as on a 1 (resistant) – 5 (susceptible) scale described by Ellwood et al. 18. F₃ families were spray inoculated using a minimum of 16 F₃ individuals per family and a randomised plot design with two representative F₃ individuals per family per plot. Disease symptoms were evaluated at 7 and 10 dpi according to the 1–5 scale devised by Salter and Leath 50. All inoculations incorporated parental control accessions.

Three different staining methods were used to study the M. truncatula – P. medicaginis interaction microscopically:

1) Superficial fungal growth was observed by staining with fluorescent dye 3'3-dihexyloxacarbocyanine iodide (DiOC6₍₃₎, Sigma-Aldrich). Whole or part-leaf samples were immersed in fresh aqueous solutions of the fluorescent dye 3 DiOC6₍₃₎ at 50 mg mL^-1, prepared from DiOC6₍₃₎ stock solution in ethanol (0.5 mg mL^-1, stored at -20°C as described by Duckett and Read 51. Between 2–3 minutes of exposure normally gave an adequate level of stain absorption. The samples were then placed on slides and the excess stain solution drawn off gently from the edges with tissue. The samples were examined under UV with a light fluorescent Olympus BH-2 microscope fitted with an epifluorescence filter B2A (450–490 nm excitation filter and a 520 nm barrier filter). Hyphae stained bright yellow, conidiospores stained bright green to yellow with increasing age, plant cells exhibiting a hypersensitive response were dark brown or dark green, and healthy cells appeared a deep red colour due to the autofluorescence of intact chloroplasts. Yellow autofluorescence of phenolic compounds could also be observed under UV-light.

2) Superficial and intracellular fungal growth was observed using trypan blue staining. Leaves were fixed in Farmers Fluid until completely cleared. The cleared leaves were immersed in a 0.03% trypan blue stain concentrate mixed with equal volume of 100% ethanol for 30 minutes, then destained in 2.5 g/mL chloral hydrate and examined under a Olympus BH-2 light-microscope.

3) To visualize production of H₂O₂ in response to P. medicaginis penetration, a DAB (Diaminobenzidine-tetrahydrochloride) staining solution was used 34. The petioles were placed in the DAB staining solution (1 mg DAB/mL H₂O pH 3.8, Sigma) for 48 hours. The leaves were fixed and cleared in a solution of ethanol: chloroform (v/v 4:1) and 0.15% TCA for 2–3 hours. After DAB staining leaves were stained with Trypan Blue staining solution for 5–10 minutes. The stained leaves were examined under an Olympus BH-2 light-microscope.

Departure from normality for the distribution of disease scores was tested using the W test of Shapiro and Wilk 22. Fain's test was used to evaluate the relationship between the means of the F3 families and the variance within the families 25. This test assumes that if resistance is determined by one or a few genes with a large effect, that families possessing the most extreme phenotypes are likely to be homozygous, exhibiting low variance within the family. Families with intermediate phenotypes are more likely to be heterozygous, exhibiting large variances (for resistance) within each family. The relationship between families can be described using a quadratic equation, where a significant value (P_/t/ < 0.05) indicates the presence of one (or a few) major genes. All the statistical analysis of the data described above was carried out using JMP-IN 5.1 software (SAS Institute, Cary, NC).

Different types of PCR markers were used to identify polymorphic markers; EST-based intron targeted markers 30; microsatellites located in introns 28; microsatellites from sequenced BACs 52 and other published sources 2629.

Temperature gradient PCR was used to identify the optimum annealing temperature for each primer pair using the standard reference accession M. truncatula A17 as a positive control. The following basic PCR protocol was used with minor variations: 50 ng of genomic DNA template, 1 × PCR reaction buffer, 2 mM MgCl₂, 0.25 mM of each dNTP, 10 pmol of each primer and 1 unit of Taq DNA polymerase. Thermocycling conditions (with minor variations) were: 2 minutes initial denaturation at 94°C followed by 36 cycles of 94°C for 30 sec, marker-specific annealing temperatures for 30 sec, 72°C for 60 sec, then a final extension step of 5 min at 72°C. Where no length or published restriction enzyme polymorphisms were available, PCR products with a single amplicon were direct sequenced using BigDye 3.1 Terminator Cycle Sequencing Ready Reaction Mix (Applied Biosystems [AB], Foster City, California) and the products run on an AB Prism 3730 DNA sequencer. Polymorphisms in the DNA sequences were identified by aligning the sequences in Vector NTI Suite 9.0 (Invitrogen, Carlsbad, California). Where restriction enzymes recognising differences in DNA sequence were available, markers were run as co-dominant CAPS 31. Single Nucleotide Polymorphisms (SNPs) for which no restriction enzyme was available were detected using SNaPshot (AB), and analysed using GeneMapper v. 3.7 (AB). Large fragment size polymorphisms were resolved by agarose gel electrophoresis and visualised using ethidium bromide staining. Small fragment size polymorphisms were resolved by native polyacrylamide gel electrophoresis or by AB Prism 3730 capillary sequencer using fluorescently labelled primers and a GeneScan™ 500 LIZ^® Size Standard (AB).

Genomic DNA from the parents and F₂ individuals were extracted as previously described 21 and polymorphic markers genotyped as described above. A Pearson Chi-square analysis was applied to test the observed segregation ratio of parental alleles against the expected Mendelian segregation ratio for co-dominant inheritance in a F₂ population, 1:2:1, to remove highly distorted markers. Marker order and map distances were calculated with Multipoint v1.2 software 53 using a maximum recombination fraction (rf) of 0.210 and the Kosambi mapping function. To test the stability of the order of markers for each linkage group (LG), a Jackknife re-sampling approach was used with 5000 iterations. Markers that violated monotonic increase of rfs (i.e. deviation from the expected increase of rf between a marker and its subsequent neighbours) were detected and removed using the control of monotony function. Marker orders were accepted if the probability was greater than 0.90. Removed markers were re-attached to the genetic map in the interval of most probable fit (e.g. minimum increase in the number of recombination events).

QTL analysis was performed with MultiQTL v2.5 53 by applying a general interval mapping and marker restoration method as described by Lichtenzveig et al. 54. The hypotheses that a single locus or two linked loci have an effect on resistance to P. medicaginis were evaluated. Firstly, 5000 permutation tests were performed on the hypothesis that one locus on a chromosome has an effect on the disease resistance (H₁) versus the null hypothesis (H₀) that the locus has no effect on the disease resistance. Secondly, 3000 permutation tests were run on the hypothesis that a single locus has an effect on disease resistance versus two linked loci. The model with the highest LOD score was fitted to the QTL and when the models did not differ significantly the simpler model was chosen ('one locus-one trait'). Five thousand bootstrap samples were run to assess the estimates and the standard deviation (SD) of the main parameters: locus effect, its chromosomal position, its LOD score and the proportion of explained variability (PEV).