The main purpose of this research was to examine differences in disease phenotypes and the genetic basis of resistance to P. medicaginis in M. truncatula. The resistant phenotype was characterized macroscopically by the presence of small hypersensitive response-like spots, whereas susceptible accessions showed typical stem and leaf lesions, within which pycnidia, or fruiting bodies, formed. In the resistant accession SA27063 pycnidia were not apparent, or where present they developed following a delay of 2–3 weeks, predominantly on senescing leaves. Microscopically, no differences were observed in penetration attempts between resistant and susceptible accessions, and no obvious changes uniquely associated with resistance were evident with the exception of fungal growth being restricted to one or a few plant cells. Production of H2O2 was observed in resistant SA27063-infected leaves but was also detectable in susceptible SA3054 and A17. H2O2 has been reported to play an important role in resistance involving a hypersensitive response [33, 34]. Build up of reactive oxygen species (ROS) is generally observed following penetration by fungi with a hemibiotrophic or biotrophic lifestyle [16, 35] and subsequently leads to resistance against such fungi, while necrotrophic pathogens may stimulate ROS production and subsequent cell death to facilitate subsequent infection . Examination of inoculated leaf tissue also showed accumulation of autofluorescent phenolic compounds around the infection site in both resistant and susceptible interactions. Phenolic compounds, such as flavonoids and isoflavonoids, are an important aspect in legume plant defence , and there is ample evidence for their accumulation in response to fungal pathogens in Medicago species [38–42]. Pilot transcriptional profiling at 12 hpi using the Mt16kOLI1plus microarray  showed significant changes in induced genes related to oxidative burst, cell wall strengthening, lipid metabolism, and the phenylpropanoid pathway leading to isoflavonoid production (L. Kamphuis, unpublished data). However, similar levels of induction occurred in both the resistant and susceptible interactions. Cytological similarities reported here are supported in oats, where cell death and induction of defence-related responses were observed in response to the HST victorin in sensitive genotypes (reviewed by Sweat et al., 2007 ), and in barley where a susceptible-specific second ROS burst was observed in the later stages of the infection . Detailed expression and metabolic profiling may help to identify differences in the HR response and the abundance and composition of defence-related compounds and their role in restricting P. medicaginis colonisation.
The lower proportion of polymorphic makers in the SA27063 × SA3054 cross suggested this is a narrower cross than SA27063 × A17. The total map length for SA27063 × SA3054 (488.3 cM) was smaller than previously reported genetic maps [29, 30, 32] and in the SA27063 × A17 map. The genetic distance between markers DK501R and CysPR1 on LG3 in particular was notably smaller than in the SA27063 × A17 map. LG3 corresponds with chromosome three, the largest chromosome of M. truncatula . One explanation for this phenomenon could be that accessions SA27063 and SA3054 are sympatric , and may share long stretches of homology in chromosome three, resulting in a lack of detectable recombination.
Quantitative trait loci (QTL) analysis is often used to identify and characterise loci conferring resistance. QTL mapping allows the roles of specific R-loci to be described, race-specificity of partial resistance genes can be assessed, and interactions between resistance genes, plant development and the environment to be analysed . QTL analysis for P. medicaginis resistance in the SA27063 × A17 mapping population revealed a QTL on chromosome four (rnpm1) and in the SA27063 × SA3054 population a QTL on chromosome eight (rnpm2), both recessive in nature. As explained in the results, although the rnpm2 QTL explained a relatively small amount of the total calculated variance for resistance (29.6%), SA27063-like or resistant individuals in the F2 mapping population correlated with individuals homozygous for SA27063 alleles at rnpm2 in approx. 80% of instances. Among the F3 families used in fine mapping and predicted to be resistant by their genotype, this proportion remained constant, allowing rnpm2 to be mapped to a 0.8 cM interval between markers h2_16a6a and h2_21h11d by recombination break point analysis. The proportion of resistant homozygous SA27063 rnpm2 individuals suggested rnpm2 is epistatic to one or several cis- or trans-acting genes or regulatory elements, possibly locus rnpm3 on LG1. As the detection and accuracy of minor QTLs in segregating populations largely depends on map quality and population size, increasing the initial mapping population size or genome-wide analysis of expressed QTLs  may reveal other minor QTLs and will enable thorough examination of the interaction between rnpm2 and rnpm3.
We did not proceed with fine mapping rnpm1 in the SA27063 × A17 cross, due to poor fertility and a large proportion of aberrant phenotypes among the F2 progeny (Table 1), which may cause biased representation of particular gamete genotypes. Furthermore, accession A17 bears a reciprocal translocation, involving chromosomes four and eight, and exhibits reduced pollen viability . All previously published M. truncatula genetic maps used accession A17 or a closely related line, J6, as one of the parents. The SA27063 × SA3054 genetic map is the first map produced in M. truncatula not involving A17 as a parental line and is therefore useful in identifying the correct position of ambiguously placed markers in A17-derived genetic maps. The reciprocal translocation involves markers distal to rnpm2 at approximately 30 cM or less and is therefore unlikely to affect the detection of this QTL among SA27063 × A17 progeny.
The simplest explanation for the existence of two separate QTLs among progeny in two crosses involving the same resistant parent is that genotype-specific susceptibility loci may interact with undetermined P. medicaginis OMT5 virulence factors or HSTs. The rnpm1 locus is tightly linked to marker AW256637 on BAC AC144658, which is in a contig containing a cluster of TIR-NBS-LRR and disease resistance protein-like genes , while no RGAs are apparent in the genomic sequence of the reference accession A17 at the rnpm2 locus. The recessive host genotype-specific nature of rnpm1 and rnpm2 bears similarities to the race-specific Pyrenophora tritici-repentis toxin insensitivity QTLs in wheat , where multiple, isolate-specific HSTs interact with a given host genotype to produce a susceptible phenotype. To date, no P. medicaginis phytotoxins have been characterised in vivo during infection. However, the presence of chlorotic symptoms in susceptible M. truncatula accessions inoculated with P. medicaginis in advance of the fungal hyphae, and the large number of toxins produced by other Phoma species suggests that susceptible plants with Rnpm1 and Rnpm2 are sensitive to a toxin and variation in these genes (rnpm1 and rnpm2) results in lack of sensitivity.
Future work will be directed at further fine mapping and isolation of rnpm2, in conjunction with gene expression studies to identify genes controlling resistance and candidates for rnpm2 in the region of interest. In addition, the possible involvement of fungal phytotoxins during infection will be characterised and their interaction with the resistance QTL evaluated.