In a plant-pathogen system that follow gene-for-gene interactions (Flor, 1971), genetically well-defined resistant host germplasm (differential lines) are an essential tool for identifying and characterizing avirulence (Avr) alleles present in pathogen isolates. Ideally, each plant differential line should represent a single resistance (R) gene in an otherwise susceptible host genotype, providing a common background for individual R genes. In addition to being a valuable tool for pathology, producing single R gene differential lines in a common genetic background would allow researchers to dissect and compare the robustness, efficiency and downstream molecular defense pathways for each R gene, free from the interference caused by variation in background genotypic effects.

Resistance against the hemibiotrophic fungal pathogen Leptosphaeria maculans, the causative agent of blackleg disease in of Brassica species (Howlett et al., 2001), is governed largely by race-specific R genes. Nineteen R genes, identified in several Brassica species, have been reported to convey race-specific resistance against L. maculans (reviewed in Raman et al., 2013). Twelve of these R genes have been positioned in the A genome of B. napus or B. rapa (Rlm1, Rlm2, Rlm3, Rlm4, Rlm7, Rlm9, LepR1, LepR2, LepR3, LepR4) or the B genome of B. juncea (LmJR1, LmJR2) via linkage mapping (Delourme et al., 2004; Yu et al., 2005, 2007, 2012; Christianson et al., 2006; Long et al., 2011; Raman et al., 2012a,b; Larkan et al., 2013, 2014, 2016), while two more have been assigned to B genome chromosomes of B. juncea and/or B. nigra (Rlm6, Rlm10) through marker-based assessment of recombinant or chromosome addition lines (Chèvre et al., 1996, 1997; Eber et al., 2011). RlmS, identified in the B. napus variety Surpass 400 (Van De Wouw et al., 2009) likely corresponds to the BlmR2 locus mapped to chromosome A10 (Long et al., 2011; Larkan et al., 2013). The remaining R genes (Rlm5, Rlm8, Rlm11, and RlmJ1) have yet to be genetically characterized in the host Brassica lines in which they reside. However, Rlm8 and Rlm11 are contained within the A genome of B. rapa varieties (Balesdent et al., 2002, 2013), while Rlm5 and RlmJ1 are found in B. juncea (genome AABB), though it has not yet been determined which genome they reside in or if these in fact represent the same gene (Balesdent et al., 2002; van de Wouw et al., 2014). Several other mapping efforts have localized resistance loci (LEM1, LmFr1, LmR1, ClmR1) to chromosome A07 of B. napus (Dion et al., 1995; Ferreira et al., 1995; Mayerhofer et al., 1997, 2005), though it has been assumed that these all correspond to Rlm4 based on map position and phenotypic response (Rouxel et al., 2003b; Rimmer, 2006).

Our understanding of Brassica—Leptosphaeria interactions has expanded tremendously over the last decade, aided greatly by the availability of both pathogen (Rouxel et al., 2011) and host (Wang et al., 2011; Chalhoub et al., 2014; Liu et al., 2014) genomes. To date, eight R gene-interacting effectors have been cloned from L. maculans; AvrLm1, AvrLm2, AvrLm3, AvrLm4-7 & AvrLm7, AvrLm6, AvrLm11, AvrLmJ1 (Gout et al., 2006; Fudal et al., 2007; Parlange et al., 2009; Balesdent et al., 2013; van de Wouw et al., 2014; Ghanbarnia et al., 2015; Plissonneau et al., 2016), along with two of the corresponding host R genes; LepR3 and Rlm2 (Larkan et al., 2013, 2015). Yet despite the rapid expansion of genetic information available for the pathosystem, most characterisations of fungal isolates and host germplasm still rely almost solely on pathological studies for generating Avr and/or R gene hypotheses. The differential host germplasm used in pathology-based studies often varies between labs, can contain multiple known R genes and is sometimes poorly characterized in terms of homozygosity, confirmation of the genetic interactions and the non-specific resistance effects observed during the interaction.

Here we describe the production of single-R gene B. napus introgression lines (ILs) for use in the pathological and molecular study of Brassica—Leptosphaeria interactions. Each IL contains one of seven blackleg resistance (R) genes (Rlm1, Rlm2, Rlm3, Rlm4, LepR1, LepR2 & LepR3) introgressed into a common susceptible B. napus line (Topas DH16516) allowing for the study of Avr-R gene interactions with a minimum of phenotypic ambiguity.

We present here the development and characterisation of single-blackleg R gene introgression lines for the study of the Brassica–Leptosphaeria pathosystem, allowing for the accurate assessment of Avr-R gene interactions in a highly homogenous genomic background free from ambiguous, phenotype-altering effects present in many other B. napus lines. Assessing the R gene content of B. napus material with varied genomic backgrounds poses several challenges. We noted several varieties which produced mild and inconsistent HR in a non-Avr dependent fashion, restricting the spread of infection in some tests. This phenomenon has been noted several times in the past (Rouxel et al., 2003b; Balesdent et al., 2005; Dilmaghani et al., 2009). Specifically, Dilmaghani et al. (2009) noted atypical reactions for tests performed using Jet Neuf which could not be explained by the presence of AvrLm4 in the affected isolates. Restriction of cotyledon lesions, resulting in less than full susceptibility, has also been observed and associated with adult-plant resistance and field-derived quantitative trait loci (QTL) in several studies (Delourme et al., 2004; Raman et al., 2012b; Huang et al., 2014; Larkan et al., 2016). While inoculation of variety lines affected by this minor restriction of infection still produced 14 dpi rating scores that were regarded as susceptible interactions (>6 on the 0–9 scale), using these lines as differentials under non-ideal conditions, or when using less vigorous inoculum, could lead to an incorrect assessment of either the virulence of the isolate being tested or the R gene content of the host line. Phenotyping of Brassica—Leptosphaeria interactions can also be affected by growth conditions, including temperature and moisture (Huang et al., 2006) and these appear to affect different R gene-mediated resistance responses unequally. It should be noted that we did not subject our seedlings to additional humidity to establish infection during the pathological assays, which is a common practice for L. maculans pathology. In our experience, L. maculans inoculum that fails to reach full infection on Topas DH16516 will also perform poorly on Westar and other susceptible varieties. Inoculum that is sub-optimal may then be subject to further inhibition by non-specific cotyledon resistance found in many lines, leading to a false classification of specific avirulence. The judicious use of susceptible controls is vital to evaluating the interaction of isolates with resistant material. Our results reflect as close to “ideal” conditions for pathotyping as we can achieve.

We also demonstrate here that a range of phenotypic responses can be observed for B. napus varieties carrying the same R gene. The most striking example of this phenomenon can be seen in the comparison between the IL Topas-LepR2 and 1135, the LepR2 donor line (Table 2). 1135 was produced from an initial B. napus x B. rapa ssp. “sylvestris” (BRS) F₁, were the B. napus parent was DH12075, a Crésor × Westar F₁-derived DH line (Mayerhofer et al., 2005). 1135 was produced from the DH12075 × BRS F₁ following successive backcrossing to the DH line Westar N-o-1 before selection at BC₄ (Yu et al., 2012), so the LepR2 gene is present in a largely Westar-derived genomic background in this line. In our testing, six of the eight non-transgenic isolates used in the characterisation of the ILs were able to produce mild to severe infection on 1135 cotyledons (averaged ratings of 7.3–9) while inoculation with the remaining two isolates (00-100s and WA30s) resulted in intermediate interactions (5.5. and 5.4, respectively). In contrast, seven of the eight isolates proved to be avirulent on the IL Topas-LepR2 (average scores 3.1 to 4.8) and only the reference isolate v23.1.3 proved to be truly virulent against LepR2 (8.2). Variation in resistance response was also clearly shown for Rlm3, in which we show a wide range of averaged lesion scores (2.0–6.0 for interactions with AvrLm3 isolate 00-100s) over 23 variety lines harboring the gene (Tables 2, 3). We speculate that this phenomenon can be attributed to variation in the genomic background of individual varieties, whereby the amplitude of HR is affected, either positively or negatively, by other host loci involved in recognition or signaling during pathogen invasion. The recognition of the L. maculans effectors AvrLm1 (Gout et al., 2006) and AvrLm2 (Ghanbarnia et al., 2015) by the R genes LepR3 (Larkan et al., 2013) and Rlm2 (Larkan et al., 2015), respectively, is known to involve at least two additional partner proteins in the membrane-bound signaling complex; SOBIR1 and BAK1 (Larkan et al., 2015; Ma and Borhan, 2015). Other host proteins may also be involved in the signaling complex, based on other plant interactions with apoplastic fungal pathogens (van der Hoorn and Kamoun, 2008; Stotz et al., 2014; Misas-Villamil et al., 2016). Subtle variation in either the expression or sequence of these partner proteins may account for the range of resistance responses observed. Indeed, in each case where we compared an R gene present in a largely Westar-derived genomic background; 1065 (LepR1), 1135 (LepR2), DH12075 (Rlm3) and the Rlm3 variety AC Excel (a selection of Karat x Westar Rakow, 1993), against the Topas DH16516-derived genomic background of the ILs (Topas-LepR1, Topas-LepR2 and Topas-Rlm3) we observed significantly stronger HR and much lower variance in IL cotyledon ratings across the test (Tables 2, 3, Figure 3). This would suggest that the Westar genomic background is somewhat deficient in its ability to fully express R gene-mediated cotyledon resistance. Another notable discrepancy was the comparatively weak reaction of Rlm1 in the donor parent line Quinta DH24288 compared to Topas-Rlm1 or Columbus when challenged with isolate v23.1.3 (AvrLm1) or the transgenic isolate 3R11:AvrLm1 (Table 2), again suggesting that substitution of the genomic background can affect the amplitude of the R gene response and hence the ability of the observer to accurately infer the presence or absence of corresponding Avr genes in the pathogen or the R gene present in the respective host plant.

Additionally, we have also observed what appears to be an effect of heterozygosity on the amplitude of HR. In particular, during the mapping of Rlm7, it was very difficult to determine the resistance status of individual BC₁F₁ seedlings, with most appearing intermediate to susceptible by 14 dpi. Only after assessment of the resulting BC₁F₂ progeny, where individuals harboring homozygous Rlm7 could be observed, was it possible to confidently assign resistance or susceptibility to each member of the mapping population (data not shown). This suggests a co-dominant phenotype for Rlm7. Additional testing for potential co-dominant phenotypic effects is required for blackleg R genes, particularly in regards to the accurate assessment of R gene content in modern hybrid (heterozygous) varieties.

Our results for the detection of Rlm3 and Rlm4 in B. napus cultivars conflict with several other reports in the literature. The blackleg R gene Rlm4 was first identified in the varieties Jet Neuf and Quinta V, a single-plant selection of cv. Quinta (Balesdent et al., 2001). However, in determining the Rlm4 resistance phenotype for Quinta V selection, the authors remarked on the heterogeneous nature of the Quinta variety seed, noting the isolate used to characterize the Rlm4 interaction showed virulence on the original seed lot obtained, which appeared to segregate for both Rlm1 and Rlm4 (Balesdent et al., 2001). As demonstrated in the present study and noted in previous reports (Kutcher et al., 2010; Larkan et al., 2013), the DH line Quinta DH24288, produced at AAFC Saskatoon from a Quinta seed lot that was originally obtained as breeder seed (Mengistu et al., 1991; P. H. Williams, pers. comm), is homozygous for Rlm1 and Rlm3. Genetic characterisation of the A07 blackleg R gene locus has a long history. A single dominant R gene was originally mapped from the French cultivar Major and termed LEM1 (Ferreira et al., 1995). LEM1 was later shown to induce resistance to the L. maculans PG2 isolate PHW1245 carrying the avirulence gene alm1 (Pongam et al., 1998). LEM1 from Major was later assumed to be Rlm4, as were mapped resistance loci LmFr1 from French cv. Crésor (Dion et al., 1995) and LmR1 from the Australian cultivars Shiralee and Maluka (Mayerhofer et al., 1997), based on map location and/or differential pathology tests (Rouxel et al., 2003b; Rimmer, 2006). The Crésor resistance, present in DH12075, a DH line produced from a Crésor × Westar F₁, was again mapped (as CLmR1) and shown to localize to the same chromosomal position as LmR1 from Shiralee (Mayerhofer et al., 2005). Our tests suggest that all of these lines in fact carry Rlm3, not Rlm4. This is supported in part by previous work in which cultivars Shiralee, Maluka, Quinta, and Major were all shown to be susceptible to an isolate carrying AvrLm4-7 (Raman et al., 2012b) and also by the pedigree of the Rlm3 varieties Quantum and Q2, which contain L. maculans resistance derived from Maluka and Shiralee, respectively (Stringam et al., 1995a,b; 1999). However, the B. napus variety lines tested here have been selected over several generations for homozygous phenotypes when challenged with L. maculans, with several of the selections coming from source seed that was clearly segregating for R gene content on first assessment (data not shown). The variety lines used in the testing therefore do not necessarily represent the cultivar as it was upon commercial release, particularly in regards to the homozygosity of the R genes carried by the variety, nor do they necessarily represent the variety lines of the same name held by other institutions. Disparities in either R gene classification of B. napus varieties or pathotyping of L. maculans isolates, due to either error or variation in germplasm, only highlights the need for unambiguous, standardized material for pathological studies.

Brassica napus is an allotetraploid species, consisting of two highly-homologous diploid genomes known to undergo inter-genomic homoeologous recombination which can result in complex genomic rearrangements (Nicolas et al., 2007; Cai et al., 2014; Mason et al., 2014). Three of the ILs appear to be missing the lower portion of chromosome C06, likely due to the parental lines containing non-compatible A07-C06 genomic arrangements (Osborn et al., 2003). The whole-genome SNP analysis indicated that these lines carry two copies of lower A07; one copy from the R gene donor parent and a second copy replacing lower C06, derived from Topas DH16516. This event affects C06 from approximately 12.6 Mb to the end of the lower chromosome arm, with the duplication occurring at approximately 16.85 Mb on A07, around 550 kb below the mapped Rlm3 and Rlm4 interval, leaving these genes unaffected. However, the Rlm1 region of A07, positioned approximately 4 Mb below this point (Figure 1) is represented twice in the Topas-Rlm1 IL. We infer that Topas-Rlm1 carries the Quinta DH24288 Rlm1 allele on A07, while the Topas DH16516 rlm1 allele is carried on C06. This duplication of lower A07 does not appear to affect the phenotypic expression of the R genes in question, with all three ILs producing consistent resistance responses equal to or better than other B. napus varieties harboring the same R genes. However, this genomic event is something that should be taken into consideration when analyzing these lines in complex, whole-genome molecular studies. Significant homoeologous translocation events were also detected in several of the other B. napus varieties surveyed, the most notable resulting in very large null regions detected for lower C01 (22.92 Mb) in Goéland and for upper C02 (33.68 Mb) in Scoop (Supplementary Table 5). In these cases no marker heterozygosity was detected for the corresponding homoeologous A genome regions, suggesting these varieties are affected by non-reciprocal translocations, where the varieties carry two highly-monomorphic copies of the corresponding A01 and A02 chromosomal regions, respectively, with total loss of the affected the C genome regions.

Our hope is that the ILs described here will be widely distributed amongst the Leptosphaeria research community and will provide a strong basis for comparison of Leptosphaeria—Brassica interactions between labs. By introgressing the R genes into a common genomic background we have produced highly standardized material for Leptosphaeria pathology and related studies of host gene expression, and largely eliminated both negative and positive background variation which could interfere with the unambiguous pathotyping of L. maculans isolates and determination of R gene- specific defense responses. Work to add to the IL set through the production of new ILs (Rlm7, Rlm9 etc.) is on-going.

NL, FY, DL, SR, and MB designed experiments. NL and FY carried out production of ILs and genetic mapping of R loci. NL performed all final pathological and genomic characterisations of lines. NL and MB drafted manuscript. All authors read and approved final manuscript with the exception of SR (deceased).

This work was funded by the Western Grains Development Fund, the AAFC-Industry Blackleg Consortium II (members include Agriculture Victoria Services Pty Ltd on behalf of the State of Victoria, Australia, Bayer, Crop Production Services (Canada), Dow AgroSciences, Lantmännen and Monsanto Canada) and the AAFC Canadian Agri-Science Cluster.