Independent breakdown events of the Brassica napus Rlm7 resistance gene including via the off-target impact of a dual-specificity avirulence interaction.

Protection of many crops is achieved through the use of genetic resistance. Leptosphaeria maculans, the causal agent of blackleg disease of Brassica napus, has emerged as a model for understanding gene-for-gene interactions that occur between plants and pathogens. Whilst many of the characterized avirulence effector genes interact with a single resistance gene in the host, the AvrLm4-7 avirulence gene is recognized by two resistance genes, Rlm4 and Rlm7. Here, we report the "breakdown" of the Rlm7 resistance gene in Australia, under two different field conditions. The first, and more typical, breakdown probably resulted from widescale use of Rlm7-containing cultivars whereby selection has led to an increase of individuals in the L. maculans population that have undergone repeat-induced point (RIP) mutations at the AvrLm4-7 locus. This has rendered the AvrLm4-7 gene ineffective and therefore these isolates have become virulent towards both Rlm4 and Rlm7. The second, more atypical, situation was the widescale use of Rlm4 cultivars. Whilst a single-nucleotide polymorphism is the more common mechanism of virulence towards Rlm4, in this field situation, RIP mutations have been selected leading to the breakdown of resistance for both Rlm4 and Rlm7. This is an example of a resistance gene being rendered ineffective without having grown cultivars with the corresponding resistance gene due to the dual specificity of the avirulence gene. These findings highlight the value of pathogen surveillance in the context of expanded knowledge about potential complexities for Avr-R interactions for the deployment of appropriate resistance gene strategies.

INTRODUCTION

Brassica napus (canola, oilseed rape) is the second most important oilseed crop worldwide, with 72 million tonnes of seed harvested in 2019 (Zheng et al., 2020). One of the major production constraints of canola is blackleg disease, caused by the fungus Leptosphaeria maculans. This disease is found worldwide, with the exception of China, and causes annual yield losses of 5%–20% in Europe, Canada, and Australia, with localized epidemics resulting in up to 90% yield loss (Fitt et al., 2006; Hwang et al., 2011; Sprague et al., 2006a; Zheng et al., 2020). L. maculans is a sexually reproducing pathogen that releases ascospores from the colonized crop debris (stubble) with each rainfall event during the growing season. These ascospores land on the leaves, flowers, upper stems or branches and can colonize the tissue and grow within the vascular tissue, stopping nutrient flow throughout the plant (Hammond et al., 1985; Sprague et al., 2018).

The impact of blackleg disease can be minimized through the use of cultural, chemical, and genetic approaches. Cultural practices such as rotation of crops and isolation from the previous year's stubble can significantly reduce disease (Marcroft et al., 2003, 2004); however, in regions such as Australia, these approaches are becoming difficult to achieve due to increases in canola production (Van de Wouw et al., 2021). Fungicides can minimize blackleg disease with applications possible at seeding, four‐ to six‐leaf, and 30% bloom growth stages (Peng et al., 2020; Van de Wouw et al., 2016, 2021). Fungicides are available worldwide for control of blackleg disease, with growers in Australia using at least one fungicide on approximately 90% of crops (Van de Wouw et al., 2021).

Whilst the type and dependency of cultural and chemical practices vary across growing regions, the use of genetic resistance is a universal approach to minimizing blackleg disease. There are two types of resistance involved in the L. maculans–B. napus interaction: quantitative (minor) and qualitative (major) (Delourme et al., 2006). Quantitative or minor gene resistance, whilst poorly understood, is thought to involve the small contribution of a number of minor genes to give an overall resistant phenotype (Stuthmann et al., 2007). Quantitative resistance is generally thought to be expressed in the adult plants, be extremely durable, and provide equal protection against all isolates (Stuthmann et al., 2007). However, in the L. maculans–B. napus interaction quantitative resistance appears to be isolate‐specific and expressed throughout the life of the plant (Jiquel et al., 2021; Marcroft et al., 2012a; Schnippenkoetter et al., 2021).

Qualitative or major gene resistance is well understood in the L. maculans–B. napus interaction and occurs via a gene‐for‐gene interaction whereby for each Brassica resistance gene there is a corresponding avirulence (Avr) effector gene in L. maculans (Balesdent et al., 2005; Flor, 1955). These resistance genes provide protection throughout the life of the plant and have been shown to be effective in all plant parts in the B. napus–L. maculans interaction (Balesdent et al., 2005; Elliott et al., 2016; Marcroft et al., 2012a; Sprague et al., 2018). To date, 19 resistance genes or alleles of genes have been identified from Brassica species (Rlm1–Rlm11, Rlm13, Rlm14, RlmS, and LepR1–LepR4) and five of these genes, LepR3, Rlm2, Rlm4, Rlm7 and Rlm9, have been cloned (Degrave et al., 2021; Delourme et al., 2006; Haddadi et al., 2021; Larkan et al., 2013, 2015, 2020; Long et al., 2011; Raman et al., 2021; Van de Wouw et al., 2009; Yu et al., 2005, 2008, 2013). In comparison, 11 of the corresponding Avr effector genes have been cloned from L. maculans (Balesdent et al., 2013; Degrave et al., 2021; Fudal et al., 2007; Ghanbarnia et al., 2015, 2018; Gout et al., 2006; Parlange et al., 2009; Petit‐Houdent et al., 2019; Plissonneau et al., 2016, 2018; Van de Wouw et al., 2014b).

Following the identification of many of the Avr effector genes from L. maculans, the gene‐for‐gene interaction was found not to be as simple as first thought. To date, three of the L. maculans Avr genes have been shown to have dual specificity (AvrLm1‐R3, AvrLm4‐7 and AvrLm5‐9), whereby a single avirulence gene is recognized by two separate resistance genes (for review see Van de Wouw & Howlett, 2020). The reverse has also been shown whereby two avirulence genes, AvrLm10A and AvrLm10B, are recognized by a single resistance gene (Rlm10) (Petit‐Houdent et al., 2019). To complicate the interaction even further, AvrLm4‐7 is epistatic over AvrLm3 and AvrLm9 whereby when an isolate is avirulent towards Rlm7, the AvrLm3–Rlm3 and AvrLm9–Rlm9 interactions are masked, and the isolates appear virulent towards both Rlm3 and Rlm9 regardless of their genotype (Ghanbarnia et al., 2018; Plissonneau et al., 2016).

The mechanisms of virulence also vary dramatically depending on the avirulence gene (for review see Van de Wouw & Howlett, 2020). Virulence towards Rlm1 and Rlm6, for example, is usually conferred through deletion of the entire corresponding avirulence genes (Fudal et al., 2009; Gout et al., 2007). Virulence towards Rlm2 and Rlm4 is usually conferred through single‐nucleotide polymorphisms (SNPs) that result in a single amino acid substitution (Ghanbarnia et al., 2015; Parlange et al., 2009). Virulence towards Rlm7 is conferred by either deletion of the gene or through repeat‐induced point (RIP) mutations (Mitrousia et al., 2018; Parlange et al., 2009), which is a genome defence mechanism that occurs during sexual crossing and generates G to A and C to T transitions within repetitive sequences (Selker, 1990).

Whilst genetic resistance provides protection to blackleg, if the pathogen population changes then resistance can be overcome and rendered ineffective. When cultivars harbouring major resistance genes are grown continuously, this creates strong selection pressure towards any isolates in the population that are virulent towards that corresponding resistance gene. These isolates then increase in number, eventually reaching a frequency whereby they render the resistance gene ineffective. This scenario has occurred many times for the B. napus–L. maculans interaction, such as the breakdown of LepR3 and LepR1 in Australia, Rlm3 in Canada, Rlm1 in France, Rlm7 in the UK, and Rlm6 in trial experiments in France (Brun et al., 2010; Mitrousia et al., 2018; Rouxel et al., 2003; Sprague et al., 2006a; Van de Wouw et al., 2014c; Zhang et al., 2016). This boom‐and‐bust nature of major gene resistance has led to the development of resistance gene management strategies, initially in Australia and then in Canada, whereby resistance genes are rotated in space and time to minimize the risk of resistance being overcome (Marcroft et al., 2012b; Van de Wouw et al., 2016; Zhang & Fernando, 2018). All commercial cultivars are released to industry with associated resistance groups, which are letters given to represent different resistance genes. Growers can then rotate cultivars with different resistance genes to minimize the risk of resistance being overcome. In Australia, this management strategy is then supported through the monitoring of disease severity in these resistance groups in field sites across canola‐growing regions of Australia, so that warnings can be provided to industry when the effectiveness of specific resistance genes is in jeopardy (Van de Wouw et al., 2016). This monitoring has previously led to issuing warnings on the Eyre Peninsula, South Australia, which saved growers over A$13 million in potential yield losses (Van de Wouw et al., 2014c). In the current study, we report on the breakdown of Rlm7 resistance in two separate locations in Australia, whereby the different selection regimes resulted in different allele diversity patterns.

RESULTS

Blackleg disease is monitored in canola cultivars representing different resistance groups at field sites across Australia every year (Table S1). This monitoring allows detection of potential resistance breakdown at regional levels and has previously been used to provide warnings to growers that significant yield losses are imminent due to resistance being overcome (Van de Wouw et al., 2014c). In the current study, this regional monitoring identifies breakdown of Rlm7 resistance.

Breakdown of Rlm7 resistance due to extensive use of Rlm7‐containing cultivars

Disease data (leaf lesion severity and percentage internal infection at the crown) were collected from the Hamilton, Victoria site in 2020 (Figure 1). Leaf lesion data collected at the four‐leaf growth stage showed that the disease severity for the Rlm7 (Group H) cultivar was not significantly different from the Rlm1 (Group A), Rlm4 (Group B), Rlm3 (Group C), and Rlm4, Rlm6 (Group BF) cultivars, which harbour resistance genes that are already reported as overcome in Australia (Van de Wouw et al., 2018). In comparison, the LepR1, Rlm1 (Group AD), Rlm1, Rlm4, LepR1 (Group ABD), and Rlm1, Rlm4, LepR1, Rlm6 (Group ABDF) cultivars had significantly lower levels of disease than the Rlm7 cultivar (Figure 1c). Lastly, the Rlm7 cultivar had five times more disease at the Hamilton site compared to the national average (Figure 1c).

FIGURE 1

Disease severity in the Group H (Rlm7) cultivar at the Hamilton, Victoria, monitoring site. (a, b) Leaf lesions were detected on the Rlm7 (Group H) cultivar at the Hamilton (a) site but not at other sites such as Kaniva, Victoria (b). (c) At the Hamilton site, the average numbers of leaf lesions per plant were recorded for cultivars with different resistance genes (indicated by resistance group on the x axis), including the Group H representative cultivar at the four‐leaf stage. The level of leaf lesion infection was much higher in the Group H cultivar at the Hamilton site compared to the national average (last bar on graph, determined from 14 sites). (d) The percentage of crown canker severity was also determined for all cultivars at the Hamilton site at the end of the growing season. Again, the level of crown canker infection was much higher in the Group H cultivar at the Hamilton site compared to the national average. Data are the average of three replicate plots. Bars with the same letter indicate no significant difference (p < 0.01)

Similarly, at plant maturity at the end of the growing season, the Rlm7 cultivar displayed similar levels of crown canker disease to the Rlm1 (Group A), Rlm4 (Group B), and Rlm3 (Group C) cultivars (Figure 1d). Significantly lower levels of disease were observed in the LepR1, Rlm1 (Group AD), Rlm1, Rlm4, LepR1 (Group ABD), Rlm4, Rlm6 (Group BF), and Rlm1, Rlm4, LepR1, Rlm6 (Group ABDF) cultivars compared to the Rlm7. Lastly, the Rlm7 cultivar had almost three times higher disease at the Hamilton site compared to the national average (Figure 1d). These combined results of high levels of disease severity early and late in the growing season and at different stages of plant development suggest that the Rlm7 resistance may be overcome in the Hamilton, Victoria region.

Four isolates were cultured from lesions following surface sterilization of infected leaves collected at the Hamilton site in 2020. The AvrLm4‐7 gene was PCR amplified and sequenced from these four isolates to determine the genotype at this locus. Each isolate harboured an allele of the AvrLm4‐7 gene that had undergone RIP mutation, resulting in 38 or 40 C‐T or G‐A transitions throughout the coding region. Three of these four isolates were screened for virulence towards both Rlm4 and Rlm7 using in planta phenotyping assays to confirm that the RIP mutations inactivate AvrLm4‐7 and therefore result in virulence. All three isolates (20BL001–20BL003) were virulent towards both Rlm4 and Rlm7 (Figure 2 and Table S2). One of these isolates, 20BL001, was transformed with the wild‐type copy of the AvrLm4‐7 gene, which restored the avirulence phenotype towards both Rlm4 and Rlm7 (Figure S1). Stubble (crop debris) from the Rlm7 cultivar was collected from the Hamilton site at the end of the 2020 growing season and a further 19 isolates were obtained from the mature stubble in 2021. All 19 isolates (21BL001–21BL019; Table S2) harboured avrLm4‐7 alleles that had undergone RIP mutation, resulting in 35 to 41 C‐T or G‐A transitions throughout the gene.

FIGURE 2

Repeat‐induced point (RIP) mutations within the AvrLm4‐7 gene of Leptosphaeria maculans confers virulence towards Rlm7. (a) Eleven isolates (two controls, D3 and D22, and nine representative isolates from 2020, harbouring RIP alleles at AvrLm4‐7), were screened for virulence towards Westar (no R genes, black bars), Rlm4 (white bars), and Rlm7 (grey bars) lines. Lesions developed on Westar (b) following inoculation with isolate D22, with a functioning copy of AvrLm4‐7, but not Topas‐Rlm7 (c), as expected. Lesions on Westar (d) and Topas‐Rlm7 (e) formed following inoculation with isolate 20BL193, which harbours RIP mutations throughout the AvrLm4‐7 gene. Bars represent the average pathogenicity score from eight plants. Error bars represent standard errors

Information on farming practices was collected from the grower on whose property the disease monitoring site was located. This grower had been using Grain ‘n’ Graze farming practices for the past three of four years, whereby cultivars with vernalization requirements are sown in March and then grazed with sheep for a number of months before being allowed to grow through to maturity and harvested for yield (Kirkegaard et al., 2008). The grower had predominately been using the Grain ‘n’ Graze cultivar Hyola 970CL during this time (G. Kreeck, Southern Farming Systems, personal communication), which harbours resistance gene Rlm7 (https://grdc.com.au/GRDC‐FS‐BlacklegManagementGuide). This use of an Rlm7 cultivar over multiple and successive years has presumably led to selection of the avrLm7 isolates within the L. maculans population, resulting in the Rlm7 resistance gene being overcome.

Breakdown of Rlm7 resistance due to dual specificity of AvrLm4‐7

Disease severity data were collected at crop maturity from the Yeelanna, South Australia, site on the Eyre Peninsula in 2019 (Figure 3). The Rlm7 cultivar had similar levels of disease as the Rlm1 (Group A) and Rlm4 (Group B) cultivars at this site, although these were significantly lower than the Rlm3 (Group C) and Rlm4, Rlm6 (Group BF) cultivars (Figure 3). The Rlm1, LepR1 (Group AD), Rlm1, Rlm4, LepR1 (Group ABD), and Rlm1, Rlm4, LepR1, Rlm6 (Group ABDF) cultivars had the lowest level of disease. The Rlm7 cultivar at the Yeelanna site had five times more disease than the national average.

FIGURE 3

Disease severity in the Group H (Rlm7) cultivar at the Yeelanna, South Australia, monitoring site. The percentage of crown canker severity was determined for all cultivars at the end of the growing season. The level of disease observed in the Group H cultivar was much higher at the Yeelanna site compared to the national average (last bar on graph). Data are the average of three replicate plots. Bars with the same letter indicate no significant difference (p < 0.01)

Stubble was collected from the Rlm7 cultivar at the Yeelanna site at the end of 2019 and 17 isolates were obtained from the mature stubble in 2020 (Table S2). The AvrLm4‐7 allele was PCR amplified and sequenced from each of these, with eight isolates harbouring the AvrLm47_02 allele, which results in virulence towards Rlm4 but not Rlm7 (Table 1). The remaining nine isolates all harboured either the AvrLm47‐09 or AvrLm47‐10 alleles, which both contained RIP mutations throughout the coding region. A subset of six of the isolates harbouring these RIP alleles (20BL183, 20BL184, 20BL186, 20BL187, 20BL188, 20BL193; Table S2) was screened for virulence towards Rlm7. As expected, all six isolates were virulent towards both Rlm4 and Rlm7 (Figure 2). Five of these isolates (20BL183, 20BL184, 20BL186, 20BL187, 20BL188) were complemented with a wild‐type copy of the AvrLm4‐7 gene and shown to restore the avirulence phenotype towards both Rlm4 and Rlm7 (Figure S1).

TABLE 1

Alleles of the AvrLm4‐7 gene of Leptosphaeria maculans

Allele a	Frequency of allele (number)	Populations identified in b	Mutations c		Translational consequences of mutations			Phenotypic consequences d
			Non‐RIP	RIP	Syn. subs.	Non‐syn. subs.	Stop codons	Rlm4	Rlm7
AvrLm47‐0*	1% (1)	2	0	0	0	0	0	Avirulent	Avirulent
AvrLm47‐1*	7% (7)	2, 4	2	0	0	2	0	Virulent	Avirulent
AvrLm47‐2*	40% (40)	2–5	2	0	0	2	0	Virulent	Avirulent
AvrLm47‐4*	3% (3)	5, 6	3	0	0	3	0	Virulent	Avirulent
AvrLm47‐5*	1% (1)	7	0	39	10	19	9	Virulent	Virulent
AvrLm47‐6*	1% (1)	7	0	42	10	22	9	Virulent	Virulent
AvrLm47‐7*	1% (1)	7	0	37	8	19	9	Virulent	Virulent
AvrLm47‐8*	1% (1)	7	0	39	9	20	9	Virulent	Virulent
AvrLm47‐9	15% (15)	1, 3, 5, 6	0	37	8	19	9	Virulent	Virulent
AvrLm47‐10	1% (1)	3	0	36	7	19	9	NT	NT
AvrLm47‐11	6% (6)	1, 5, 6	0	38	9	20	8	Virulent	Virulent
AvrLm47‐12	1% (1)	1	0	40	11	19	9	Virulent	Virulent
AvrLm47‐13	1% (1)	1	0	38	9	19	9	NT	NT
AvrLm47‐14	1% (1)	5	0	39	9	20	9	NT	NT
AvrLm47‐15	2% (2)	5, 6	0	39	10	19	9	NT	NT
AvrLm47‐16	2% (2)	6	0	38	8	20	9	NT	NT
AvrLm47‐17	1% (1)	6	0	42	11	21	9	NT	NT
AvrLm47‐18	1% (1)	1	0	38	10	18	9	NT	NT
AvrLm47‐19	1% (1)	1	0	38	9	19	9	NT	NT
AvrLm47‐20	1% (1)	1	0	39	8	21	8	NT	NT
AvrLm47‐21	3% (3)	1	0	41	8	20	10	NT	NT
AvrLm47‐22	6% (6)	1	0	39	9	21	8	NT	NT
AvrLm47‐23	1% (1)	1	0	35	6	18	8	NT	NT
AvrLm47‐24	2% (2)	1	0	40	8	21	8	NT	NT

Allele nomenclature based on Van de Wouw and Howlett (2012). *AvrLm47‐0 through AvrLm47‐8 previously published.

Population details: 1, Hamilton, Victoria, Rlm7; 2, Hamilton, Victoria, Rlm3; 3, Yeelanna, South Australia, Rlm7; 4, Yeelanna, South Australia, Rlm1, LepR1; 5, Canberra, Australian Capital Territory, Rlm7; 6, Springfield, New South Wales, Rlm7; 7, historic isolates, unknown.

Repeat‐induced point (RIP) mutations were identified as CpA to TpA and TpG to TpA mutations as previously reported (Selker & Garrett, 1988). Non‐RIP mutations were identified as other classes of transitions and transversions.

Phenotypes are based on either previously published findings (Van Parlange et al., 2009; Van de Wouw & Howlett, 2012) or this study (see Figure 2). NT refers to not tested.

Unlike the situation in Hamilton, Victoria, Rlm7 cultivars have never been grown on the Eyre Peninsula, as there are no cultivars harbouring Rlm7 with suitable phenology for the region. However, there has been a high dependency on Rlm4 cultivars in this region for the past 3 years (Figure 4). Since 2018, over 80% of seed sold on the Eyre Peninsula has been cultivars containing the Rlm4 resistance gene, with 2019 and 2020 seeing 95% of total seed sold harbouring Rlm4 (Nutrien Cummins Ag Supplies, personal communication). This suggests that the breakdown of Rlm7 on the Eyre Peninsula is associated with the high use of Rlm4 cultivars and the dual specificity of the corresponding avirulence gene, AvrLm4‐7.

FIGURE 4

Percentage of Brassica napus cultivars sown on the Eyre Peninsula, South Australia, between 2016 and 2020 that harbour the Rlm4 resistance gene

Diversity of RIP alleles differs under contrasting selection regimes

The allelic variation for the RIP isolates varied across different selection regimes (Table S2). Of the nine avrLm7 isolates collected from Yeelanna, only two different RIP alleles were detected, AvrLm47_09 and AvrLm47_10. These two alleles contained 37 (AvrLm47_09) and 36 (AvrLm47_10) C‐T or G‐A transitions, each generating 19 amino acid substitutions and nine stop codons (Table 1). Of the eight AvrLm7 isolates, only a single allele, AvrLm47_02, was detected. Conversely, of the 23 avrLm7 isolates collected from Hamilton, Victoria, 11 different RIP alleles were identified (AvrLm47_9, AvrLm47_11, AvrLm47_12, AvrLm47_13, AvrLm47_18, AvrLm47_19, AvrLm47_20, AvrLm47_21, AvrLm47_22, AvrLm47_23, AvrLm47_24), suggesting that the diversity of RIP alleles was much higher in the isolates collected from Rlm7 cultivars at Hamilton, Victoria, compared to Yeelanna, South Australia.

As a comparison, the AvrLm4‐7 alleles were sequenced from additional isolate populations from Hamilton and Yeelanna that were collected from alternative resistance sources, not Rlm7. No RIP alleles were identified in any of these isolates. For the isolates collected from the Rlm1, LepR1 (Group AD) cultivar at the Yeelanna site, two non‐RIP alleles were detected, AvrLm47_01 and AvrLm47_02, both phenotypically avrLm4 (Table 1). For the isolates from the Rlm3 (Group C) cultivar at the Hamilton site, three non‐RIP alleles were detected: AvrLm47_00, AvrLm47_01, and AvrLm47_02. The AvrLm47_00 isolate is avirulent towards both Rlm4 and Rlm7 (Table 1).

Lastly, isolates were also collected from Rlm7 cultivars from experimental plots at Springfield (New South Wales) and Canberra (Australian Capital Territory) whereby recurrent Rlm7 selection had also occurred (Table S2). From eight Springfield isolates, six isolates had five different RIP alleles: AvrLm47_09, AvrLm47_11, AvrLm47_15, AvrLm47_16, and AvrLm47_17. These five RIP alleles harboured up to 42 RIP mutations resulting in 19–21 amino acid substitutions and eight or nine stop codons (Table 1). The remaining two isolates both contained the AvrLm47_04 allele, which harbours three non‐RIP‐like mutations resulting in virulence towards Rlm4 but not Rlm7 (Table 1). From 10 Canberra isolates, seven had four different RIP alleles: AvrLm47_09, AvrLm47_11, AvrLm47_14, and AvrLm47_15 (Table 1). The remaining three isolates harboured two different alleles, AvrLm47_02 and AvrLm47_04.

Molecular variation was analysed for the AvrLm4‐7 locus in the isolates using analysis of molecular variance (AMOVA), taking into consideration geographical location (region) and the cultivar as the source of resistance from where the isolates originated (population). For all 100 isolates significant differences in genetic diversity among regions and populations were detected as well as within populations (Table 2). The largest source of variation was within populations (69%), followed by among populations (22%). The genetic diversity of each population was calculated using Shannon indices, whereby the Hamilton (Vic) population collected from a Rlm7 cultivar showed the highest level of diversity (Shannon index, I = 2.08; genetic diversity, h = 0.85) and the Yeelanna (South Australia) population collected from a LepR1, Rlm1 cultivar showed the lowest (I = 0.349, h = 0.198) (Figure 5a). The Yeelanna population collected from a Rlm7 cultivar showed similar levels of diversity (I = 1.053, h = 0.602) as the Hamilton population from a Rlm3 cultivar (I = 1.129, h = 0.560) but much lower than each of the populations collected from Rlm7 cultivars (I = 1.696–2.080, h = 0.813–0.851). The lower diversity of alleles for the Yeelana‐Rlm7 population was further exemplified when the genetic diversity indices were calculated only on the 49 isolates harbouring RIP alleles (Figure 5b), whereby this population obtained I index of 0.349 and h index of 0.198. This compares to a range of I indices of 1.386–2.163 and h indices of 0.694–0.858 for each of the other RIP populations. These data suggest that the diversity in RIP alleles is much lower for the Yeelanna Rlm7 populations compared to all other Rlm7 populations.

TABLE 2

Analysis of molecular variance results for the AvrLm4‐7 alleles identified across 100 isolates

Source of variation a	df	SS	MS	Est. var.	%	p value
Among regions	4	7.851	1.963	0.036	8%	0.001
Among populations	2	4.332	2.166	0.096	22%	0.001
Within populations	93	27.866	0.300	0.300	69%	0.001

Note: Analysis of molecular variance was performed with 999 random permutations. SS, sum of squares; MS, mean of square value.

Regions refer to different geographical locations. Populations refer to different cultivars, and therefore sources of resistance, that the isolates were cultured from.

FIGURE 5

Differences in genetic diversity were calculated using Shannon indices for the AvrLm4‐7 allele of Leptosphaeria maculans populations. (a) When all seven populations representing 100 isolates were analysed, population 4 showed the lowest diversity and population 1 showed the highest. (b) When the 49 isolates harbouring repeat‐induced point mutation alleles were analysed, large differences in diversity were observed for population 3 compared to all other Rlm7 populations. Population details: 1, Hamilton, Victoria, Rlm7; 2, Hamilton, Victoria, Rlm3; 3, Yeelanna, South Australia, Rlm7; 4, Yeelanna, South Australia, Rlm1, LepR1; 5, Canberra, Australian Capital Territory, Rlm7; 6, Springfield, New South Wales, Rlm7; 7, historic isolates, unknown. Error bars represent the standard errors for the Shannon index (I) and genetic diversity (h) as calculated using GenAlEx v. 6.503

Across the 20 different RIP alleles identified, 33 mutations are common between all alleles (Figures 6a and S2), with a further three to nine mutations responsible for the variation between the different RIP alleles. When compared to a single published RIP allele from isolate Nz‐T4 (GenBank: KT804641.1), the same 33 RIP mutations are also in common with this allele (data not shown).

FIGURE 6

Haplotype network (a) and phylogenetic tree (b) of the 24 different alleles identified for the AvrLm4‐7 locus of Leptosphaeria maculans. AvrLm47_00 confers an avirulent phenotype towards both Rlm4 and Rlm7; AvrLm47_01, AvrLm47_02, and AvrLm47_04 confer a virulent phenotype towards Rlm4 and an avirulent phenotype towards Rlm7 (represented in red text). Alleles AvrLm47_05 to AvrLm47_24 confer virulence towards both Rlm4 and Rlm7, and were generated through repeat‐induced point (RIP) mutations (represented in black text). Population details: 1, Hamilton, Victoria, Rlm7; 2, Hamilton, Victoria, Rlm3; 3, Yeelanna, South Australia, Rlm7; 4, Yeelanna, South Australia, Rlm1, LepR1; 5, Canberra, Australian Capital Territory, Rlm7; 6, Springfield, New South Wales, Rlm7; 7, historic isolates, unknown. The size of the circle represents the number of samples detected for each allele (see key for details). The phylogenetic tree features bootstrap values (from 1000 replications) greater than 65%

The avrLm7 RIP alleles (AvrLm47_05 to AvrLm47_24) are phylogenetically distinct from the AvrLm7 alleles (AvrLm47_00 to AvrLm47_4), forming a separate clade on the phylogenetic tree, which is supported with bootstrap values (Figure 6b). However, no clear pattern of divergence can be seen within the RIP alleles with limited bootstrap support (Figure 6).

DISCUSSION

The use of major gene resistance is an essential tool in the fight to minimize the impact of plant diseases like blackleg of canola. However, as seen on numerous occasions in Europe, Canada, and Australia, L. maculans populations have the ability to evolve rapidly and overcome these resistance genes (Mitrousia et al., 2018; Sprague et al., 2006a; Van de Wouw et al., 2014c; Zhang et al., 2016). Previous resistance breakdowns have involved either the widescale use of the resistance gene on farms or recurring selection in trial settings (e.g., Delourme et al., 2014; Van de Wouw et al., 2014c). The selection pressure from the continual use of the resistance gene has selected for isolates in the population that are virulent, which then increase in frequency, resulting in the corresponding resistance gene being overcome. In the current study, we have identified the breakdown of Rlm7 associated with two different selection regimes.

The first involves continual use of Rlm7 cultivars on a specific farm, resulting in the selection of avrLm7 isolates and therefore the breakdown of Rlm7 resistance. This is similar to situations in Europe whereby breakdown of Rlm7 resistance occurred in response to widespread sowing of cultivars carrying this resistance gene (Daverdin et al., 2012; Mitrousia et al., 2018). In the scenario in Australia, RIP mutation in the avrLm4‐7 gene was identified as the only mechanism conferring virulence towards Rlm7, whilst in the UK and France gene deletion was identified as the major mechanisms of virulence (Daverdin et al., 2012; Mitrousia et al., 2018). However, in French surveys it was found that in the first year, RIP mutations were the most frequent event giving rise to virulence and these decreased in frequency in the second year (Daverdin et al., 2012). Continual monitoring in Australia might identify a similar situation in which RIP mutations decrease in frequency and deletion alleles increase. Because this was the first time monitoring had been done at this location, it is unknown when the frequency of virulent avrLm7 isolates began to increase; however, it is likely that it has occurred within a 3‐ or 4‐year time frame that would coincide with the use of the Rlm7 cultivar Hyola 970 on the farm. In comparison, the breakdown of Rlm7 resistance was reported over a 4‐year window in specific trials in France and was much slower under UK conditions (Daverdin et al., 2012; Mitrousia et al., 2018). It is unclear why the breakdown of resistance occurred much faster in Australia, but this finding is consistent with other resistance genes such as Rlm1, which was overcome in 3 years in Australia compared to 10+ years in France, and may be a reflection of the fungal life cycle in Australia, whereby the sexual cycle during which RIP mutation occurs is essential to produce the ascospores to initiate disease (Rouxel et al., 2003; Sprague et al., 2006b). Previously, Rlm7 has only been available in winter cultivars, and therefore predominantly only used in Grain ‘n’ Graze farming systems. In 2020 the Rlm7 cultivar Hyola 970CL set an Australian harvest record of 7.16 t/ha (Thyer, 2021), and in 2021 spring‐type Rlm7 cultivars were released for the first time, which will expand the potential growing area for this resistance gene. Therefore, close monitoring of disease levels will be required to prevent yield losses from resistance breakdown in Rlm7 cultivars.

The second scenario presented in this study was the breakdown of Rlm7 on the Eyre Peninsula, South Australia, where no Rlm7 cultivars have ever been sown. Instead, this region has predominately been sown with Rlm4 cultivars, suggesting that the breakdown of the Rlm7 resistance is due to the dual specificity of AvrLm4‐7, a single avirulence gene recognized by two different resistance genes, Rlm4 and Rlm7. A SNP is the most common mechanism conferring virulence towards Rlm4 (Parlange et al., 2009) and has been reported in populations in Europe, Canada, and Australia (Daverdin et al., 2012; Fernando et al., 2018; Van de Wouw et al., 2018). Conversely, the most common mechanism for virulence towards Rlm7 is either deletion or RIP mutation of the AvrLm4‐7 gene that results in virulence towards both Rlm7 and Rlm4 (Daverdin et al., 2012). Previously, the identification of RIP alleles has only been reported from populations that have undergone Rlm7 selection (either field or trials) or historic populations where the selection regime is unknown (Daverdin et al., 2012; Van de Wouw & Howlett, 2012). In the current study, RIP alleles were detected following multiple years of Rlm4 cultivation, suggesting that strong selection pressure due to Rlm4 can lead to selection of both SNPs and RIP alleles conferring virulence towards Rlm4 and, consequently, virulence towards Rlm7. However, under Rlm4 selection the diversity of RIP alleles was significantly lower than in populations that had been selected through Rlm7 cultivation (e.g., Figure 7). These data suggest that RIP mutation is potentially triggered at a lower frequency under Rlm4 selection compared to Rlm7 selection. Therefore, only a small number of RIP alleles are present in the population during Rlm4 selection, and then when Rlm7 cultivars are grown the small number of RIP alleles present are selected for and increase in frequency with low genetic diversity (Figure 7a). However, under Rlm7 selection, RIP mutation is being triggered more readily and therefore there is a larger pool of avrLm4‐7 alleles being generated and selected (Figure 7b).

FIGURE 7

Representation of different selection regimes and the impact on genetic diversity within the AvrLm4‐7 locus of Leptosphaeria maculans. Different colours within a shape represent different alleles for each phenotype class

The mechanisms of RIP mutation are still poorly understood in non‐model organisms although the process has been proposed to exist in many fungal species based on genome analyses, especially within the ascomycetes (van Wyk et al., 2021). With the exception of L. maculans and some model fungal species, there is limited experimental evidence of RIP mutation existing in other species. Furthermore, there is no report of RIP mutation conferring virulence in other plant pathogen species. Of the 20 RIP alleles identified in this study, 33 mutations were common between them. Do these findings suggest that there is a precursor allele that is predominant in the population that is selected for under Rlm4/Rlm7 selection? In line with this possibility, the 33 mutations were in common in four historical isolates that were collected in the 1980s. However, these 33 mutations were also shared with an allele sequenced from isolate Nz‐T4 from New Zealand. An alternative possibility is that multiple rounds of RIP mutation occur during meiosis, which results in a first‐round of mutations that target specific bases, and therefore common in all RIP alleles, and this is then followed by a second round of mutations that randomly targets remaining potential sites. This is consistent with recent findings in L. maculans whereby at least two rounds of RIP mutation occur during meiosis (Van de Wouw et al., 2019). In experiments characterizing RIP mutation using specific constructs carrying DNA duplication, multiple independent alleles are created after crossing. However, the trigger for mutations at the AvrLm4‐7 locus is probably the large regions of repetitive DNA sequence on either side, which may impose a different intensity of the mutation process and hence those 33 sites may always be targeted. Counter to this hypothesis is the range of different mutations found in RIP‐derived alleles of AvrLm6, which is also found embedded in repetitive DNA (Van de Wouw et al., 2010).

Hence, it is unclear whether the RIP alleles arise independently each time there is selection pressure being imposed by Rlm4 or Rlm7 cultivation or if these alleles are always being maintained at low frequency in the population. Fitness penalties in strains that are virulent towards Rlm4 have been reported from field‐ and laboratory‐based experiments in the UK (Huang et al., 2006, 2010). No such experiments have been done for strains associated with selection with Rlm7; however, it may be reasonable to assume there would be considering the differences associated between avirulent and virulent alleles.

The dual specificity of Avr genes has implications for management strategies that are being implemented for growers. In both Australia and Canada, growers are advised to rotate resistance genes in the field to minimize the risk of resistance being overcome (Marcroft et al., 2012b; Van de Wouw & Howlett, 2020; Zhang & Fernando, 2018). When a single avirulence protein is recognized by two difference resistance proteins then these resistance genes potentially need to be treated as a single R gene when deploying management strategies, depending on the mechanisms conferring virulence. Resistance genes LepR3 and Rlm1 are another blackleg disease example whereby two separate R genes in B. napus recognize a single Avr gene in L. maculans, AvrLm1‐R3. Similar to the situation described in the current study, the widescale cultivation of sylvestris cultivars harbouring LepR3 resistance genes led to the selection of virulence at the AvrLm1‐R3 locus. In this situation, gene deletion is the most common mechanism for loss of avirulence and therefore will always result in virulence towards both genes. In Australia, LepR3 and Rlm1 are treated as a single resistance group due to the fact that virulence towards one of the genes results in virulence towards both. Due to the previous findings that the emergence of virulence under Rlm4 selection was due to a single SNP, and therefore only virulence towards Rlm4 and not Rlm7, these two resistance genes have been treated as separate resistance groups in Australia. Rotation to Rlm7 following Rlm4 cultivation has therefore been recommended as isolates would only be virulent towards Rlm4. However, rotation to Rlm4 following Rlm7 selection was not recommended as virulence towards Rlm7 inactivates the gene and always also leads to virulence towards Rlm4. The findings presented in this current study suggest that Rlm4 and Rlm7 should not be rotated in either manner because Rlm4 cultivation can lead to generation of RIP alleles, albeit at a lower frequency.

As more R–Avr interactions are discovered and characterized in plant–microbe interactions, and that information moved into crop protection, establishing situations in which multigene interactions occur can be useful for breeding prioritization strategies and the deployment of new resistance genes.

EXPERIMENTAL PROCEDURES

Disease monitoring sites

Blackleg severity was monitored in 14 field sites across Australia in 2019 and 2020 (Figure S3). These disease monitoring sites were located directly adjacent to National Variety Trial yield evaluation sites where cultivars/breeding lines are sown. Two of these sites, Hamilton in Victoria and Yeelanna in South Australia, displayed unusual levels of disease for the Rlm7 (Group H) cultivar in 2020 and 2019, respectively. For the full list of disease monitoring sites, see Table S1.

At each of these disease monitoring sites, eight cultivars representing different resistance groups were sown. The cultivars grown at each site were ATR‐BonitoPBR (Rlm1; Group A), SFR650917TT (Rlm4; Group B), ATR‐StingrayPBR (Rlm3; Group C), HyTTec® Trophy (Rlm1, LepR1; Group AD), Hyola® 559TT (Rlm1, Rlm4, LepR1; Group ABD), SF Turbine TT (Rlm4, Rlm6; Group BF), Hyola® 350TT (Rlm1, Rlm4, LepR1, Rlm6; Group ABDF), and DG1901TT (Rlm7; Group H). Cultivars were sown in three replicate plots (10 × 1 m) with complete randomization, designed using DiGGer software (www.austatgen.org/files/software).

In 2019, disease severity was assessed at maturity as described previously (Van de Wouw et al., 2014c). In brief, for each replicate plot 20 plants were chosen at random, uprooted, and then cut at the crown. The cross‐section was visually assessed for the percentage of internal infection (discolouration). The average crown canker severity was determined from the 60 plants (20 per replicate) for each cultivar at each site. The national average was determined by averaging the crown canker severity for the Rlm7 cultivar from all 14 disease monitoring sites.

In 2020, in addition to the maturity assessments described above, the severity of leaf lesions was also determined. At the four‐leaf growth stage, the total number of lesions was counted for each of 20 consecutive plants within each plot. For consistency, the 20 plants were always selected from the second row, starting 1 m in from the front of the plot. The average number of lesions per plant was determined for each replicate plot for each cultivar at each site. The national average was determined for the leaf lesions per plant for the Rlm7 cultivar from all 14 disease monitoring sites.

Differences in disease severity were determined through analysis of variance of the log‐transformed mean for each plot. Data were analysed using GenStat v. 16.1.0.

Farm practice information was collected. For the Hamilton (Victoria) site, direct information regarding cultivar use was obtained from the farmer where the site was located by G. Kreeck, the trial manager for Southern Farming Systems who ran the trial. For the Yeelanna (South Australia) site, cultivar use in the Eyre Peninsula region was collected from Nutrien Ag Supplies, Cummins. This company is the largest distributor of seed for the Eyre Peninsula region and therefore a relevant representative of farming practices.

Isolate culturing and pathogenicity screening

All isolates used in this study are listed in Table S2 and the location of their collection sites represented in Figure S3. Isolates were collected following one of two protocols. For the isolates collected from Hamilton, Victoria, in 2020, leaf lesions were collected from the Rlm7 cultivar from the disease monitoring site. Lesions were excised and isolates cultured using the surface sterilization technique as previously described (Van de Wouw et al., 2017). For the remaining isolates used in this study, stubble (crop debris) was collected from disease monitoring sites, matured, and then individual ascospores discharged as previously described (Van de Wouw et al., 2018). All isolates were maintained on 10% Campbell's V8 juice agar. Four historic isolates were included in which alleles harbouring RIP mutations had been previously described (Van de Wouw & Howlett, 2012).

To complement the AvrLm4‐7 mutations, a wild‐type copy of the gene was transformed into six representative isolates by Agrobacterium tumefaciens‐mediated delivery, as described previously (Van de Wouw et al., 2014a). Transformants were selected on medium containing hygromycin (50 µg/ml) and cefotaxime (100 µg/ml), then isolates obtained from colonies derived from single spores.

The virulence phenotype of each of the isolates was determined by inoculating pycnidiospores onto cotyledons of B. napus seedlings of lines Westar (harbouring no major gene resistance), Topas‐Rlm4 (harbouring Rlm4 resistance gene), and Topas‐Rlm7 (harbouring Rlm7 resistance gene) (Larkan et al., 2016; Van de Wouw et al., 2018). For each isolate × host interaction, eight plants were inoculated and symptoms determined 14 days postinoculation on a 0 (no darkening around wounds) to 9 (large lesions with prolific sporulation) scale (Koch et al., 1991). Average pathogenicity scores of >5.0 were considered virulent and average pathogenicity scores of <3.0 were considered avirulent (Marcroft et al., 2012a). Isolates D2 and D22 were used as controls, with isolate D2 being virulent towards both Rlm4 and Rlm7, whilst isolate D22 is virulent towards Rlm4 and avirulent towards Rlm7.

AvrLm4‐7 allele diversity analysis

Alleles of AvrLm4‐7 were identified by PCR amplification and Sanger sequencing. Genomic DNA was extracted from mycelia of each of the isolates as previously reported (Van de Wouw et al., 2010). The AvrLm4‐7 gene was amplified using primers (5′‐AGAAGGGTAAGGGGCAAGTC‐3′ and 5′‐GAAGAACCCTGCTAGATAGGTAAGC‐3′) and PCR conditions previously reported (Van de Wouw & Howlett, 2012). Amplicons were purified using a PCR purification kit as per manufacturer's specifications (Qiagen) and then sequenced using BigDye terminator cycling at the Australian Genome Research Facility (Melbourne, Australia). Sequences were analysed using Geneious R9.1.8. All AvrLm4‐7 sequences were aligned to the avirulent AvrLm4‐7 reference sequence of L. maculans isolate v23.1.3 (GenBank accession no. AM998638; Parlange et al., 2009) as well as the RIP allele of isolate Nz‐T4 (GenBank accession no. KT804641). Allele nomenclature was based on alleles previously published in Van de Wouw and Howlett (2012). RIP mutations were identified as CpA to TpA and TpG to TpA mutations as previously reported (Selker & Garrett, 1988). All allele sequences were uploaded to GenBank (accession numbers OM470482–OM470505).

Differences in genetic diversity and genetic indices were analysed using AMOVA, Nei's genetic distance calculations, Shannon's information index, and diversity index in GenAlEx v. 6.503 (Peakall & Smouse, 2012). AMOVAs were performed with 999 random permutations. Haplotype maps of the various AvrLm4‐7 alleles were generated using PopART (http://popart.otago.ac.nz) using median‐joining and minimum‐spanning networks (Bandelt et al., 1999). Maximum‐likelihood phylogenetic analyses were conducted using the bootstrap method (1000 replications) using the software MEGA v. 10.2.6 (Kumar et al., 2018).

FIGURE S1 Sequence alignment of all alleles identified in this study. The red nucleotides are the repeat‐induced point (RIP) mutations that are in common across all RIP alleles

Click here for additional data file.

FIGURE S2 Complementation of avrLm7 isolates with a functional copy of the AvrLm4‐7 gene restores avirulence towards both Rlm4 and Rlm7. (a) Average pathogenicity scores of isolates with or without a functional copy of AvrLm4‐7 inoculated onto Westar (no R gene), Topas‐Rlm4 (Rlm4), and Topas‐Rlm7 (Rlm7) lines. (b) Representative photos of cotyledons of Westar (no R gene), Topas‐Rlm4 (Rlm4) and Topas‐Rlm7 (Rlm7) inoculated with either isolate 20BL183 or 20BL183 complemented with AvrLm4‐7 (20BL183+AvrLm47)

Click here for additional data file.

FIGURE S3 Location of both field sites and isolate collection sites used in this study. Numbers of sites refer to the locations listed in Table S1

Click here for additional data file.

TABLE S1 Locations of blackleg disease monitoring sites across Australia in 2019 and 2020

Click here for additional data file.

TABLE S2 Leptosphaeria maculans isolates used in this study

Click here for additional data file.