A gene-for-gene interaction involving a 'late' effector contributes to quantitative resistance to the stem canker disease in Brassica napus.

The control of stem canker disease of Brassica napus (rapeseed), caused by the fungus Leptosphaeria maculans is based largely on plant genetic resistance: single-gene specific resistance (Rlm genes) or quantitative, polygenic, adult-stage resistance. Our working hypothesis was that quantitative resistance partly obeys the gene-for-gene model, with resistance genes 'recognizing' fungal effectors expressed during late systemic colonization. Five LmSTEE (stem-expressed effector) genes were selected and placed under the control of the AvrLm4-7 promoter, an effector gene highly expressed at the cotyledon stage of infection, for miniaturized cotyledon inoculation test screening of a gene pool of 204 rapeseed genotypes. We identified a rapeseed genotype, 'Yudal', expressing hypersensitive response to LmSTEE98. The LmSTEE98-RlmSTEE98 interaction was further validated by inactivation of the LmSTEE98 gene with a CRISPR-Cas9 approach. Isolates with mutated versions of LmSTEE98 induced more severe stem symptoms than the wild-type isolate in 'Yudal'. This single-gene resistance was mapped in a 0.6 cM interval of the 'Darmor_bzh' × 'Yudal' genetic map. One typical gene-for-gene interaction contributes partly to quantitative resistance when L. maculans colonizes the stems of rapeseed. With numerous other effectors specific to stem colonization, our study provides a new route for resistance gene discovery, elucidation of quantitative resistance mechanisms and selection for durable resistance.

Introduction

Genetic resistance to plant diseases is a highly desirable trait for plant breeding, to ensure global food security and avoid the adverse effects of pesticides. Furthermore, genetic resistance is the only means of disease control available in some crop species lacking other efficient or sustainable control methods. Two types of resistance are deployed in the field for efficient plant disease control: qualitative and quantitative resistance. Qualitative resistance usually is controlled by a single resistance gene and confers complete resistance to pathogen populations harboring the corresponding avirulence gene. It is often associated with a hypersensitive response (HR) and localized cell death, preventing the pathogen from colonizing its host (Greenberg & Yao, 2004). By contrast, quantitative resistance does not prevent the pathogen from infecting the plant but instead limits symptom severity, thereby reducing the impact on yield. It is based on the combined effects of several genes, associated with genomic regions or QTL (quantitative trait loci), each making a quantitative contribution to plant defense (Delourme et␣al., 2006; St Clair, 2010; Niks et␣al., 2015). Many disease resistance QTL have been identified in plants over the last two decades. For example, Wilfert & Schmid‐Hempel (2008) analyzed 194 publications describing 445 QTL involved in resistance to various bioaggressors in diverse plants, such as wheat, barley, maize and melon. Despite the importance of quantitative resistance to the control of plant diseases, the mechanisms underlying such resistance remain largely unknown (Poland et␣al., 2009) and are probably highly diverse (Kushalappa et␣al., 2016; Corwin & Kliebenstein, 2017). Several studies have suggested that some quantitative resistance may be isolate‐specific (Arru et␣al., 2003; Calenge et␣al., 2004; Rocherieux et␣al., 2004; Zenbayashi‐Sawata et␣al., 2005; Niks et␣al., 2015), and that partial resistance may be at least partly due to gene‐for‐gene interactions, as hypothesized by Parlevliet & Zadoks (1977).

The Dothideomycete Leptosphaeria maculans is responsible for one of the most damaging diseases of rapeseed (Brassica napus), phoma stem canker disease, also known as blackleg. This disease, which has been known to cause losses exceeding US$900 million per year, is controlled principally by genetic resistance in most parts of the world (Fitt et␣al., 2008). The ascospores of L. maculans produced on stem residues land on aerial organs of the plant, where they germinate, and the hyphae penetrate the leaves and cotyledons through stomata. After entering the plant, the fungus goes through a short biotrophic stage (5–15 d), during which it colonizes the apoplast, subsequently switching to a necrotrophic lifestyle, in which it causes leaf spots (Fitt et␣al., 2006; Rouxel & Balesdent, 2005). After this first stage of colonization, the fungus migrates toward the stem tissues, from the leaves to the crown, during an asymptomatic phase that may last several months, before switching to a necrotrophic lifestyle in which necrosis is induced at the stem base (Hammond et␣al., 1985; West et␣al., 2001; Huang et␣al., 2014; Rouxel & Balesdent, 2005).

Qualitative and quantitative resistances to L. maculans have been described in B. napus (Delourme et␣al., 2006). Qualitative resistance typically is conferred by race‐specific resistance genes, the Rlmx or LepRx genes, conferring complete resistance to avirulent isolates and preventing such isolates from colonizing cotyledons or leaves (Balesdent et␣al., 2001; Huang et␣al., 2006; Larkan et␣al., 2013). Qualitative resistance can be assessed easily in controlled conditions using cotyledon inoculation assays, which are reproducible and used widely, and provide a reliable method for the high‐throughput screening of large collections of B. napus or populations of L. maculans (Balesdent et␣al., 2006; Rouxel et␣al., 2003; Ghanbarnia et␣al., 2012). Quantitative resistance, which operates during colonization of the petiole and stem, is a partial resistance under polygenic control mediated by a large number of QTL; its efficacy for limiting necrosis can be highly dependent on environmental conditions (Pilet et␣al., 2001; Fopa Fomeju et␣al., 2015; Jestin et␣al., 2015; Huang et␣al., 2016; Raman et␣al., 2016; Kumar et␣al., 2018). Quantitative resistance limits the onset and severity of stem necrosis, but its mechanistic determinants are unknown. However, effects on the growth rate of the pathogen in tissues, the ability of the pathogen to move from the petiole to the stem and stem necrosis development have been described (Huang et␣al., 2009, 2014). By contrast to qualitative resistance, reproducible methods for assessing quantitative resistance in controlled conditions have yet to be developed. This type of resistance currently is evaluated in field assays, by scoring disease severity on mature plants at the end of the growing season, which can last from six to 10 months, depending on geographical location. Furthermore, owing to the influence of the environment on quantitative resistance, the identification of stable resistance QTL requires replicated field‐plot experiments at different sites and in different years (Jestin et␣al., 2015; Huang et␣al., 2016; Kumar et␣al., 2018).

As already reported for other pathogens, the infection and colonization of rapeseed by L. maculans is dependent on a cocktail of effectors, molecules mainly corresponding to small secreted proteins which modulate plant immunity and facilitate infection (Sánchez‐Vallet et␣al., 2018). Two waves of L. maculans effector gene expression were identified initially, the molecules involved in the first wave being described by Gervais et␣al. (2017) as ‘early effectors’ and those of the second wave, ‘late effectors’ (aka ‘LmSTEE genes’ for ‘L. maculans stem expressed effector’). More extensive studies have since discriminated eight waves of genes specifically expressed during the interaction with the plant, all being enriched in genes encoding effectors (Gay et␣al., 2021). ‘Early effector’ genes, including all known avirulence genes (AvrLm) ‘recognized’ by the plant and leading to qualitative resistance, are upregulated during cotyledon/leaf colonization. Genetic manipulations or crosses of the fungus to generate isolates harboring the smallest possible number of AvrLm genes and isogenic isolates differing by only a single AvrLm gene, are now routinely used as tools for identifying the corresponding Rlm/LepR genes in Brassica genotypes, for screening genetic resources, or for use in plant breeding (Balesdent et␣al., 2002; Rouxel et␣al., 2003; Van de Wouw et␣al., 2014; Larkan et␣al., 2015). An easy‐to‐use, medium‐ to high‐throughput cotyledon inoculation test is used widely worldwide for such purposes (Larkan et␣al., 2016b). Quantitative resistance operates during the systemic colonization of rapeseed, and, during this phase, L. maculans expresses several genes encoding effectors thought to interfere with plant defenses (Gervais et␣al., 2017). These LmSTEE genes are expressed only very weakly, if at all, during axenic growth and cotyledon/leaf colonization, but are strongly expressed during stem colonization, many months after the initial leaf infection. Our working hypothesis is that at least some LmSTEE genes encode products recognized by specific plant resistance gene products during systemic colonization, in a manner similar to that described for AvrLm–Rlm interactions, accounting for at least part of the quantitative resistance. However, it is not possible to detect gene‐for‐gene interactions between LmSTEE genes and the resistance genes of B. napus in the usual cotyledon inoculation assay, because the corresponding avirulence proteins are not produced at this stage. In addition, stem inoculation assays (Gervais et␣al., 2017; Gay et␣al., 2021) for the identification of such resistance are not amenable to medium‐/high‐throughput screening, due to the length of the process, low reproducibility, and the time and space required to screen a large number of genotypes of fully grown plants. We therefore designed a new strategy, allowing the expression of ‘late’ effectors during cotyledon infection, for the identification of putative new resistance genes in rapeseed interacting with the late effectors from L. maculans in the cotyledon inoculation test.

With this approach we identified a B. napus genotype expressing hypersensitive resistance to one late effector: LmSTEE98. The LmSTEE98–RlmSTEE98 interaction was further validated by inactivation of the LmSTEE98 gene with a CRISPR‐Cas9 approach. The genetic control of this response was investigated, and a double‐haploid (DH) population was used to map the associated region on the B. napus genome. These results provide proof‐of‐principle for this approach to the identification of new, but difficult‐to‐detect, sources of resistance for the effective control of L. maculans and show for the first time in this model that gene‐for‐gene interactions contribute to quantitative resistance in grown‐up plants.

Materials and Methods

Fungal and plant materials

The sequenced isolate JN3 (v.23.1.3; Rouxel et␣al., 2011) and the closely related isolate JN2 (v.23.1.2) correspond to two sister progenies from an in␣vitro cross between European field isolates (Balesdent et␣al., 2001). They were used as controls in inoculation tests. A representative selection of isolates from worldwide collections and reference isolates from Australia, Canada, the USA, Chile, France, Mexico and New Zealand, also was analyzed (Supporting Information Table S1; Mendes‐Pereira et␣al., 2003; Dilmaghani et␣al., 2009, 2012; this study). INV13.269 is a single pycnidiospore isolate recovered from a French field in 2013.

Fungal cultures were maintained on V8 juice agar medium, and sporulating cultures were obtained on V8 medium, as described previously (Ansan‐Melayah et␣al., 1995).

A collection of 204 genotypes of Brassica napus (Table S2) was screened for gene‐for‐gene interactions with LmSTEE genes. This panel mostly comprised varieties previously used for association studies of quantitative resistance to Leptosphaeria maculans (Fopa Fomeju et␣al., 2015). The segregating doubled‐haploid population BnaDYDH was derived from a ‘Darmor‐bzh’ × ‘Yudal’ cross (Foisset et␣al., 1996). The ‘Darmor‐bzh’ parent is a French winter rapeseed cultivar and the ‘Yudal’ parent is a Korean spring rapeseed cultivar.

Plant inoculations

The cotyledons of 10‐d‐old seedlings were inoculated, by puncture, with 10 µl of inoculum (107 pycnidiospores ml−1), as described by Balesdent et␣al. (2001). Four different isolates, one per half‐cotyledon, were used to inoculate each plant. At least six plants were inoculated with each isolate. Plants were incubated for two days at room temperature in the dark, and then in a growth chamber at 19°C (night) and 24°C (day) with a 16 h : 8 h, light : dark photoperiod and a relative humidity of 90%. Symptoms were scored 10, 14 and 18 d post inoculation (dpi), with the IMASCORE rating scale (Balesdent et␣al., 2001). Mean scores for symptoms and the percentage of virulent and avirulent phenotypes induced by the LmSTEE ‘over‐expressed in cotyledons’ (OEC) transformants were determined for each cultivar and compared with the symptoms induced by the wild‐type (WT) isolate on the same cultivar, on the same date (Kruskal–Wallis test).

Adult plants were inoculated as described by Gervais et␣al., 2017. For the analysis of stem colonization by L. maculans, we cut the petiole of the second leaf horizontally, 1 cm from the insertion point of the leaf. Inoculum (10 µl, containing 107 pycnidiospores ml−1) was applied to the wounds. Inoculated plants were incubated as described previously. The necrosis of infected stems was assessed by cutting the whole stem section between the insertion point of the inoculated petiole on the stem and the ground into successive 3‐mm slices. The sections were scanned and the percentage of the stem section showing internal necrosis was measured with imagej.

Vector construction and fungal transformation

The promoter of the AvrLm4‐7 gene was amplified with the primers indicated in Table S3. It was then digested with NheI and BamHI, and ligated into a SpeI‐BamHI‐digested pPZPNat1 vector. The LmSTEE genes (LmSTEE1, LmSTEE30, LmSTEE35, LmSTEE98 and LmSTEE78) were amplified from their Start codon to their terminator regions and the amplicons were digested with EcoRI and XhoI or SalI and XhoI (Table S3). The resulting fragments were ligated into the pPZPNat1 vector containing the AvrLm4‐7 promoter, digested with the same enzymes.

For CRISPR‐Cas9 gene inactivation, the pLAU2 (hygromycin resistance), containing the Cas9 gene, and pLAU53 (geneticin resistance), in which the guide RNA was inserted, were used, as described by Idnurm et␣al. (2017). The CRISPOR prediction tool was used, with L. maculans as the reference genome (Dutreux et␣al., 2018), to design the guide RNA (gRNA) targeting the gene for inactivation (http://crispor.tefor.net/; Table S3). The guide RNA was selected as the sequence with the fewest off‐target predictions. The corresponding DNA fragment then was amplified with the universal primers MAI0309 and MAI0310 (Table S3). Gibson assembly (Silayeva & Barnes, 2018) was used to insert the guide RNA into the XhoI site of pLAU53, generating pLAU53‐gRNA plasmids.

The four mutated versions of LmSTEE98 (see below) were amplified from the Start codon to the terminator region, as for the WT version of LmSTEE98 (Table S3). The ΔlmSTEE98_06 and ΔlmSTEE98_10 variants were cloned by Gibson assembly (Silayeva & Barnes, 2018), with the primers Gib_pA4‐7:Lm98_F and Gib_pA4‐7:Lm98_R. The ΔlmSTEE98_20 and ΔlmSTEE98_30 variants were amplified with the LmSTEE98_SalI_Up and LmSTEE98_XhoI_Low primers and the resulting amplicons were digested with SalI and XhoI (Table S3). For both cloning techniques, the pPZPNat1 vector containing the AvrLm4‐7 promoter was digested with the same enzymes and the fragments were inserted, by Gibson assembly or enzyme‐mediated ligation. Both cloning methods produced constructs identical to those used for genotype screening (pA4‐7::LmSTEE98), except for the mutated sites.

The various plasmids were introduced into Agrobacterium tumefaciens C58 by electroporation at 2.5 kV, 200 Ω and 25 µF. Leptosphaeria maculans was transformed with the resulting A. tumefaciens strains as described in a previous study (Gout et␣al., 2006). For CRISPR‐Cas9‐mediated gene inactivation, pLAU2‐Cas9 was introduced into the JN2 isolate. Transformants then were transformed with pLAU53‐gRNA plasmids. Fungal transformants were selected on nourseothricin (50 µg ml−1) for pPZPNat1, hygromycin (50 µg ml−1) for pLAU2‐Cas9 and geneticin (50 µg ml−1) for pLAU53‐gRNA.

We checked that the fungal transformants with LmSTEE98 inactivation displayed no growth defects in␣vitro. A plug of mycelium was deposited in the center of a Petri dish containing V8 agar media, and mycelial growth was measured every 2 d for 6 d (nine replicates).

DNA manipulation and high‐resolution melting (HRM) experiments

Genomic DNA was extracted from suspensions of conidia with the DNeasy 96 or DNeasy Plant Mini Kit (Qiagen), in accordance with the manufacturer’s recommendations. LmSTEE genes were amplified by PCR with the primers in Table S3.

We performed HRM as described by Plissonneau et␣al. (2016), to analyze SNPs in the natural population of L. maculans for the various LmSTEE genes considered, with the primers described in Table S3. If variant HRM profiles were detected, the LmSTEE genes were sequenced by Eurofins Genomics (Eurofins, Ebersberg, Germany).

RNA manipulation and quantitative reverse transcription (qRT)‐PCR

Total RNA was extracted from the inoculated cotyledons 7 dpi, the time point corresponding to the peak of AvrLm4‐7 expression (Parlange et␣al., 2009), or from inoculated stems 30 dpi. The content of all RNA samples was adjusted to 1 µg of RNA and single‐strand cDNA was generated by oligo‐dT‐primed reverse transcription with the PowerScript reverse transcriptase (Clontech, Palo Alto, CA, USA), according to the manufacturer’s protocol. Plants inoculated with water were used as negative controls. For each condition, two to three technical replicates were performed on one (stem samples) or two (cotyledon samples) biological replicates. The qRT‐PCR experiments were performed as described by Fudal et␣al. (2007), with the primers indicated in Table S3. Ct values were analyzed as described elsewhere (Muller et␣al., 2002). Actin was used as a constitutively expressed reference gene and levels of tubulin expression relative to actin expression were used as a control.

Genetic mapping

Genotyping data for the BnDYDH population included an analysis of SNPs from the Brassica 60K Illumina Infinium SNP array (Clarke et␣al., 2016), the 20K Illumina Infinium SNP array (Chalhoub et␣al., 2014) and the 8K Illumina Infinium SNP array (Delourme et␣al., 2013). The genetic map contained 28 000 loci that represented 3592 unique loci, and covered 2128.2 cM (Laperche et␣al., 2017). The phenotypic data were mapped as quantitative traits with the R/qtl package (Broman et␣al., 2003) or as a qualitative traits where scores of 1 to 3 were considered to indicate plant resistance and scores of 4 to 6 were considered to indicate susceptibility.

Bioinformatics and statistical analyses

Blast analyses were performed by aligning the five LmSTEE protein sequences against the non‐redundant protein sequences from the GenPept, Swissprot, PIR, PDF, PDB and NCBI RefSeq databases with default parameters. Hits with an e‐value lower than 1.10–5 were selected. The PSI‐Blast method with default parameters was used to align the LmSTEE protein sequences against all proteins of Leptosphaeria biglobosa (Dutreux et␣al., 2018). Hits with an e‐value lower than 5 × 10–2 were selected.

Statistical analyses were performed with R (R Core Team, 2015). The effect of the mutants used for inoculation on the area affected by stem necrosis was assessed by applying a generalized linear model (GLM), assuming a quasipoisson distribution of the data.

Results

Selection of five late effector candidates

We selected five LmSTEE genes from the 307 effector candidates upregulated in the stem (Gervais et␣al., 2017). These genes were selected to cover the range of characteristics of the effector gene repertoire (Table 1) and were thus representative of the diversity present in this gene pool. The five LmSTEE genes encoded for small predicted secreted proteins with size ranging from 55 to 291 amino acids (Table 1). Some of them were particularly enriched in cysteins, such as LmSTEE1 with 10 cysteins for 80 amino acids (12.5%) compared to only two for LmSTEE78 (0.7%). The five genes belonged to three of the eight clusters of gene expression identified by Gay et␣al. (2021); LmSTEE35 and LmSTEE98 belonged to cluster 4 (‘biotrophy to necrotrophy transition’), LmSTEE1 and LmSTEE30 to cluster 5 (‘stem biotrophy’) and LmSTEE78 to cluster 6 (‘stem necrotrophy’; Table 1). All genes were much less expressed in cotyledons than ‘early’ effector genes and were all highly expressed in stems after inoculation in controlled conditions (Fig. S1). LmSTEE1 and LmSTEE30 were particularly highly expressed during stem colonization. Under field conditions, each gene had a specific expression pattern, with distinct time of peak expression (March for LmSTEE35, April for LmSTEE1, LmSTEE78 and LmSTEE98, and May for LmSTEE30; Fig. S1). Following its peak of expression in April, LmStee98 expression level remained intermediate at all stages except at the last time point, corresponding to stem necrosis (Fig. S1; Gay et␣al., 2021).

Table 1

Characteristics of the ‘late’ effector candidates of Leptosphaeria maculans studied here.

Name	Accession Nb	Size (amino acids)	Cysteines	Expression ranka	Expression waveb	Number of Blast hitsc	Closest Blast hit	E‐value (% Identities)	Blast hit coverage	Pfam domain (E‐value)	CAZymes hit (E‐value)	Number of hits on L. biglobosa 'brassicae' proteinsd
LmSTEE1	Lmb_jn3_03177	80	10 (12.5%)	2	Stem biotrophy (cluster 5)	24	Aspergillus bombycis hypothetical protein	3.00E‐19 (56%)	99%	None	None	1
LmSTEE30	Lmb_jn3_04778	107	10 (9.3%)	12	Stem biotrophy (cluster 5)	230	Epicoccum nigrum hypothetical protein	8E‐35 (68%)	98%	CFEM (2.7E‐10)	None	4
LmSTEE35	Lmb_jn3_08094	123	7 (5.7%)	93	Biotrophy‐to‐necrotrophy transition (cluster 4)	26	Colletotrichum salicis hypothetical protein	3E‐17 (39%)	94%	None	None	0
LmSTEE78	Lmb_jn3_00617	291	2 (0.7%)	114	Stem necrotrophy (cluster 6)	458	Stagonospora sp. SRC1lsM3a hypothetical protein	4.00E‐156 (74%)	99%	None	GH131 (7.8E‐64)	3
LmSTEE98	Lmb_jn3_11364	55	6 (10.9%)	101	Biotrophy‐to‐necrotrophy transition (cluster 4)	0	None	None	None	None	None	2

Fold‐change in rank between stems and cotyledons of L. maculans effectors from RNA‐seq data (Gervais et␣al., 2017).

Names of expression waves and corresponding genes cluster, as defined by Gay et␣al. (2021).

Number of hits with an E‐value < 1E‐5.

Results from PSI‐Blast against all L. biglobosa proteins with E‐values < 5E‐2 (Supporting Information Table S5).

With the exception of LmSTEE98, we were able to find homologs for all the LmSTEE proteins in databases, but all the closest hits corresponded to hypothetical proteins (Table 1). Three proteins paralogous to LmSTEE35 were also found in L. maculans (54% identity for the best hit; Table S4), but not in the related species from the same species complex, L. biglobosa, another pathogen of oilseed rape (Dilmaghani et␣al, 2009). For each of the other LmSTEE proteins, we identified one to four homologs in L. biglobosa (Table S5).

Conservation of late effector candidates in field populations of L. maculans

We used a collection of 186 isolates from around the world (Dilmaghani et␣al., 2009, 2012), to study the conservation of the LmSTEE genes in populations of L. maculans (Table S1). All isolates were obtained from B. napus, with the exception of the Mexican isolates, which were obtained from broccoli (Brassica oleracea).

The five LmSTEE genes were amplified by PCR in almost all isolates (Table 2). Only two French isolates, from two different regions, lacked LmSTEE35, and another two unrelated French isolates yielded no amplicon for LmSTEE98. Sequence polymorphism also was rare, and the sequences of LmSTEE1, LmSTEE35 and LmSTEE78 obtained were invariant. One silent mutation was detected in LmSTEE30, in 18% of the isolates tested, mostly from Western or Central Canada. Only LmSTEE98 displayed sequence polymorphism at the protein level, with one allele containing two nonsynonymous mutations leading to two substitutions in the protein sequence. This allele was found at only one site in Mexico, where it was present in 70% of the local population (17% of the Mexican isolates analyzed here; Tables 2, S1).

Table 2

PCR amplification and high‐resolution melting (HRM) analysis of LmSTEE genes in Leptosphaeria maculans isolates.

Origin of isolatesa	LmSTEE1			LmSTEE30			LmSTEE35			LmSTEE78			LmSTEE98
	Presence	Absenceb	Mutationc (no. isolates)	Presence	Absenceb	Mutationc (no. isolates)	Presence	Absenceb	Mutationc (no. isolates)	Presence	Absenceb	Mutationc (no. isolates)	Presence	Absenceb	Mutationc (no. isolates)
France	93	0	nd	93	0	nd	91	2	nd	93	0	nd	91	2	nd
Oceania	20	0	0/18	20	0	1/20d	20	0	0/20	20	0	0/20	20	0	0/18
Canada	17	0	0/14	17	0	11/17d	17	0	0/17	17	0	0/16	17	0	0/17
USA	10	0	0/9	10	0	1/10d	10	0	0/9	10	0	0/10	10	0	0/10
Mexico	41	0	0/37	41	0	0/41	41	0	0/40	41	0	0/41	41	0	7/41e
Chile	5	0	0/5	5	0	3/5d	5	0	0/5	5	0	0/5	5	0	0/5
Total	186	0	0/83	186	0	16/88d	184	2	0/91	186	0	0/88	184	2	7/90e

nd, not done.

Isolates are described in Supporting Information Table S3.

Lack of PCR amplification.

Number of isolates with single nucleotide polymorphisms (SNPs) compared to the reference sequence JN3, based on HRM results coupled with sequencing.

All mutations for LmSTEE30 corresponded to one single silent mutation.

Observed mutations for LmSTEE98 corresponded to two SNPs modifying the protein sequence: A > I(17)and I > P(27).

The use of an avirulence gene promoter makes it possible to overexpress LmSTEE genes during cotyledon colonization

We induced the expression of LmSTEE genes during cotyledon colonization, by creating mutant strains expressing the genes under the control of the promoter of the avirulence gene AvrLm4‐7 (pA4‐7), which is known to be upregulated by 7 dpi in cotyledons (Parlange et␣al., 2009; Fig. S1). Constructs were introduced into the INV13.269 isolate, which is virulent against all Rlm genes potentially present in current B. napus resources (Rlm1, Rlm2, Rlm3, Rlm4, Rlm7 and Rlm9), for the subsequent screening of a large panel of genotypes without interference with known AvrLm–Rlm interactions. The resulting OEC transformants (pA4‐7::LmSTEEi) strongly expressed LmSTEE genes during cotyledon infection (Fig. 1). We obtained such transformants for four genes, but not for LmSTEE30, which we were unable to overexpress with this strategy. Expression in the cotyledons varied between transformants and genes, but was similar to that of the native AvrLm4‐7 gene. The gene most highly expressed in OEC transformants was LmSTEE1, for which the level of expression was upregulated by a factor of 1000 in the transformants at 7 dpi; by contrast, transformants obtained with a construct containing the native promoter had expression levels similar to that of the WT isolate (Fig. 1). These findings demonstrate the usefulness of the AvrLm4‐7 promoter for modifying the expression profile of ‘late’ effector genes. The two transformants per construct displaying the strongest overexpression (Fig. 1) were selected for the screening of rapeseed genotypes in the cotyledon inoculation assay.

Fig. 1

Expression of LmSTEE genes in ‘over‐expressed in cotyledons’ (OEC) transformants of Leptosphaeria maculans during early stages of cotyledon colonization. We constructed OEC transformants specifically overexpressing the LmSTEE1, LmSTEE35, LmSTEE78 and LmSTEE98 genes under the control of the promoter of the AvrLm4‐7 gene during the early colonization of cotyledons. The expression of LmSTEE genes in the various OEC transformants was assessed by quantitative reverse transcription (qRT)‐PCR in infected cotyledons of the cultivar ‘Eurol’, 7 d post inoculation. Gray bars, expression data from three to four different transformants; black bars, expression in the wild‐type isolate (INV13.269); white bars, expression in control transformants with the endogenous promoter (LmSTEE1 only); red arrows, OEC transformants selected for the screening of plant genotypes. Mean expression is normalized against actin, with tubulin used as a control (Fudal et␣al., 2007). Error bars represent the SE for two biological and three technical replicates.

Two LmSTEE genes expressed in cotyledons induce resistance response in two different cultivars

The OEC transformants were used to inoculate a panel of 204 B. napus genotypes (Table S2). This panel comprised mostly European winter‐type rapeseed genotypes, and included only one spring genotype of Asian origin. With this approach, we identified two cultivars displaying specific resistance to two late effectors: ‘Dariot’, which was resistant to LmSTEE1 and ‘Yudal’, which was resistant to LmSTEE98 (Figs 2, S2). These genotypes displayed resistance responses with the two independent OEC transformants of each construct, but not with the other OEC transformants or the WT isolate. Heterogeneous resistance to other effectors was observed in ‘Dariot’ (Fig. 2), justifying the focusing of our analysis on the specific HR‐type resistance to LmSTEE98 observed in ‘Yudal’. We hypothesized that there must be a resistance gene, RlmSTEE98, interacting with LmSTEE98, at least at the cotyledon stage.

Fig. 2

Resistance response of the rapeseed cultivars ‘Yudal’ and ‘Dariot’ to two ‘late’ effector genes. Leptosphaeria maculans transformants expressing LmSTEE35, LmSTEE1, LmSTEE98 or LmSTEE78 under the control of the promoter of the avirulence gene AvrLm4‐7 were used to inoculate the ‘Yudal’, ‘Dariot’ and susceptible ‘Eurol’ cultivars. Two independent transformants were used per construct. The bars represent the percentage of plants displaying susceptible (in red) or resistant (in green) phenotypes, based on the IMASCORE rating scale (Balesdent et␣al., 2001), 14 d post inoculation. INV13.269 is the wild‐type isolate and the JN2 and JN3 isolates were used as additional controls. The resistance of ‘Dariot’ to JN3 and JN2 is linked to the presence of Rlm7, which recognizes AvrLm7 in these isolates.

LmSTEE98 inactivation suppresses its recognition in ‘Yudal’ at the cotyledon stage

For validation of the interaction between LmSTEE98 and RlmSTEE98, we used the CRISPR‐Cas9 technology (Idnurm et␣al., 2017) to inactivate LmSTEE98 in L. maculans. Using the transformant strain JN2 expressing the Cas9 protein (JN2_Cas9), we generated L. maculans LmSTEE98 mutants. Four independent mutants with four different mutations resulting in different truncations or modifications to the protein were selected: ΔLmSTEE98_06 (+1 bp), ΔLmSTEE98_10 (−2 bp), ΔLmSTEE98_20 (−9 bp) and ΔLmSTEE98_30 (−17 bp; Fig. 3a).

Fig. 3

Generation of CRISPR‐Cas9‐mutated versions of LmSTEE98 and effect of the mutations on their recognition by RlmSTEE98 in cultivar ‘Yudal’. Four CRISPR‐Cas9‐mutated versions of LmSTE98 were generated and their deduced protein sequences were aligned with that of the wild‐type (WT) protein in (a). The mutated alleles were placed under the control of the promoter of the avirulence gene AvrLm4‐7, and transformants of Leptosphaeria maculans were used to inoculate the susceptible cultivar ‘Darmor’ (b) and the resistant cultivar ‘Yudal’, containing the RlmSTEE98 gene (c). Three independent transformants were used per construct. Scoring was based on the IMASCORE rating scale, in which scores of 1–3 indicate resistance and scores of 4–6 indicate susceptibility (Balesdent et␣al., 2001), 14 d post inoculation. INV13.269 is the WT isolate. Error bars represent the SE for ≥10 biological replicates.

In order to test whether these altered versions of the protein were still able to elicit the RlmSTEE98‐mediated resistance, these four mutated versions of LmSTEE98 were placed under the control of the AvrLm4‐7 promoter and introduced into INV13.269, to generate OEC mutants. All transformants displayed high levels of expression of the mutated versions of LmSTEE98 on cotyledons at 7 dpi (Fig. S3). All of the OEC mutants were virulent on the susceptible genotype ‘Darmor’, as were the control isolates with the WT version of LmSTEE98 (Fig. 3b). Three OEC mutants with major defects of LmSTEE98 – ΔLmSTEE98_06, ΔLmSTEE98_10 and ΔLmSTEE98_30 – were virulent on ‘Yudal’, suggesting that the corresponding altered versions of the protein were not recognized by the plant (Fig. 3c). By contrast, the pA4‐7::ΔLmSTEE98_20 OEC mutant, lacking only three amino acids, was avirulent (Fig. 3c), as were the control isolates pA4‐7::LmSTEE98_4 and pA4‐7::LmSTEE98_9 (Fig. 2), suggesting that this mutant version of the protein was still recognized by the plant. These results indicate that the recognition of LmSTEE98 at the cotyledon stage is impaired by major mutations or truncations of LmSTEE98.

LmSTEE98 inactivation leads to stronger symptoms during stem colonization

We investigated the role of LmSTEE98 in stem colonization, using the four mutants inactivated with CRISPR‐Cas9 (Fig. 3a). These mutants displayed no in␣vitro growth or sporulation defects and were as virulent as the control isolate JN2_Cas9 when used to inoculate the cotyledons of a susceptible cultivar (Fig. S4). Moreover, they displayed no defect of LmSTEE98 gene expression during stem colonization (Fig. S5). We then tested the ability of these four mutants to cause stem necrosis on ‘Yudal’, and on ‘Darmor’ as a control.

The three mutants with LmSTEE98 gene impairments – ΔLmSTEE98_06, ΔLmSTEE98_10 and ΔLmSTEE98_30 – generated significantly higher percentage areas of stem necrosis on ‘Yudal’ than the control isolate JN2_Cas9 (Fig. 4a). By contrast, the effects of the ΔLmSTEE98_20 mutant, which lacked only three amino acids, did not differ significantly from those of JN2_Cas9. On ‘Darmor’; no significant effect of LmSTEE98 gene inactivation was observed (Fig. 4b).

Fig. 4

The inactivation of LmSTEE98 by CRISPR‐Cas9 method increases stem necrosis in ‘Yudal’ rapeseed stems. LmSTEE98 mutants of Leptosphaeria maculans (Fig. 3a) were used to inoculate the petiole of the second true leaf of Brassica napus ‘Darmor’ and ‘Yudal’, cut 1 cm from the point of insertion into the stem. Symptoms were evaluated on stem sections, 30 d post inoculation. (a) Percentage of the stem section displaying necrosis for cultivar ‘Yudal’ (n ≥ 18). (b) Percentage of the stem section displaying necrosis for cultivar ‘Darmor’ (n ≥ 18). The asterisks indicate a significant difference in the area under the curve between the control isolate JN2_Cas9 and the LmSTEE98 mutants, for the two biological replicates of the experiment (GLM: **, P < 0.05; ***, P < 0.01).

Genetic control and mapping of resistance to LmSTEE98

A doubled‐haploid population derived from the ‘Darmor‐bzh’ × ‘Yudal’ cross (BnDYDH; Foisset et␣al., 1996) was used to investigate the segregation of the phenotype of resistance to LmSTEE98 in ‘Yudal’. We found that 87 of the 258 offspring tested were resistant to the LmSTEE98 OEC transformants, whereas 171 were susceptible (Table S6). This segregation does not correspond to the expected 50 : 50 (resistant : susceptible) ratio for monogenic resistance in a DH population. However, genetic mapping of the resistant phenotype indicated that the resistance was controlled by a single locus mapping to a 0.6 cM interval on oilseed rape chromosome A09, flanked on one side by the marker Bn‐A09‐p33777987 and on the other side by Bn‐A09‐p33971771 (Table S6; Fig. 5). This genetic interval corresponded to a 266.33 kb sequence of chromosome A09 and an unanchored 84 kb sequence (potentially located on chromosome A09). In the reference genome of ‘Darmor‐bzh’, this interval was predicted to contain 70 genes (Chalhoub et␣al., 2014), but a comparison with this genetic interval in ‘Yudal’ was not possible, as a consequence of a lack of sequence data for ‘Yudal’. The control of the interaction by single genes in both the plant and the fungus is consistent with a typical gene‐for‐gene interaction, and LmSTEE98 was therefore renamed AvrLmSTEE98.

Fig. 5

Genetic mapping of the resistance gene RlmSTEE98 in the ‘Darmor‐bzh’ × ‘Yudal’ cross. RlmSTEE98 was mapped to chromosome A09 of Brassica napus. The genetic distances (in cM), are shown on the left, with marker order shown on the right.

Discussion

The use of resistance genes is the most efficient strategy for controlling stem canker disease in rapeseed crops. However, the resistance to avirulent populations of Leptosphaeria maculans conferred by Rlm/LepR genes is rapidly broken down in the field after the release of resistant varieties (Rouxel & Balesdent, 2017). New sources of resistance to the pathogen are, therefore, urgently required. High‐throughput screening for resistance at the cotyledon stage, with genetically improved isolates, is now well developed. However, screening for adult‐stage resistance is much more difficult, time‐consuming, and less reproducible, with high levels of variation between experiments and between plants. We hypothesized that adult‐stage resistance was at least partly dependent on gene‐for‐gene interactions involving fungal effectors expressed only during stem colonization. We tested this hypothesis by selecting a small number of effectors belonging to stem‐specific waves of expression, and designing an innovative strategy in which the genes were placed under the control of an early effector gene promoter, making it possible to screen a collection of Brassica napus genotypes for resistance genes operating in the stem in a miniaturized cotyledon test. Not only did this efficient strategy identify and map a new resistance source, but, following CRISPR‐Cas9 inactivation of the late effector gene identified, we were able to demonstrate its involvement in stem canker severity and recognition by the cognate resistance gene.

We selected five late effector candidates from three different waves of expression that had previously been shown to be specifically upregulated in rapeseed stems during systemic colonization (Gay et␣al., 2021). We recently showed that all ‘early’ effectors (including all known avirulence effectors) are actually specifically upregulated during all of the biotrophic stages of plant colonization (leaf/cotyledon, petiole and stems; Gay et␣al., 2021). The decrease in expression of early effectors is relayed, at least during stem colonization, by a stem‐specific biotrophic wave of gene expression including LmSTEE1 and LmSTEE30, followed by another wave involved in the biotrophy‐to‐necrotrophy transition, which includes LmSTEE78. LmSTEE35 and LmSTEE98 belong to a less well‐defined wave, illustrating a series of biotrophy‐to‐necrotrophy transitions. In field samples, the expression of LmSTEE genes other than LmSTEE98 and LmSTEE78, decreases with the development of stem necrosis (Gay et␣al., 2021). Some AvrLm effectors, produced during the initial stages of cotyledon colonization, may decrease the magnitude of symptom expression on B. napus cotyledons (Petit‐Houdenot et␣al., 2019). Likewise, the LmSTEE genes associated with the long systemic colonization of the tissues and involved in biotrophic stem colonization have been suggested to manipulate plant innate immunity, to enable the fungus to colonize its host efficiently, ultimately leading to the development of stem necrosis (Gay et␣al., 2021).

Both ‘early’ and ‘late’ candidate effectors are structurally similar, as they are small secreted proteins (SSPs) generally enriched in cysteine residues, and most have no predicted function or recognizable domain. However, they differ by their location within the genome. ‘Early’ effector genes are located in dispensable regions of the genome enriched in transposable elements acting as drivers of evolution by favoring deletions and translocations, and diversifying or inactivating point mutations (Rouxel & Balesdent, 2017; Gay et␣al., 2021; Soyer et␣al., 2020). By contrast, ‘late’ effector genes are located in GC‐equilibrated regions of the genome, an unusual feature for most effector genes in fungi, and a genome environment much less conducive in generating deletions or inactivation by Repeat‐Induced Point mutation, as frequently found for AvrLm genes (Rouxel et␣al., 2011; Rouxel & Balesdent, 2017). As a consequence, LmSTEE genes also differ from ‘early’ effector genes by their lack of polymorphism or presence/absence variation in field populations of L. maculans, with only one nonsynonymous mutation identified in LmSTEE98. This suggests that, unlike effector genes expressed at the onset of plant colonization, LmSTEE genes, which are not expressed in early stages of infection, are not readily subjected to selection pressures exerted by the plant, such as those generated by Rlm genes. As the corresponding RlmSTEE genes also may have only a partial effect, contributing to adult‐stage resistance, they may contribute to more stable and durable resistance in B. napus.

We designed a new approach based on the widely used cotyledon inoculation test, for identifying new sources of resistance to L. maculans in B. napus that could contribute to quantitative resistance. This type of resistance has not, to date, been amenable to miniaturized, medium‐throughput screening in growth chambers. We modified the expression of the LmSTEE genes such that they are expressed, under the control of the AvrLm4‐7 promoter, during cotyledon infection. This approach was particularly efficient, with all strains transformed with the constructs expressing the target genes to at least the same level as avirulence genes on cotyledons at 7 dpi. We screened a large collection of rapeseed genotypes to identify easy‐to‐record plant resistance responses to late effectors. We identified at least one such interaction, between LmSTEE98 and a resistance gene in the ‘Yudal’ cultivar. This interaction would not have been identified with the usual screening approaches.

Following inoculation with LmSTEE98 ‘over‐expressed in cotyledons’ (OEC) transformants, we observed a clear hypersensitive response (HR) phenotype, which segregated in the BnDYDH population. The segregation pattern in this population was not consistent with the 1 : 1 ratio expected for monogenic control of the resistance. However, markers closely linked to RlmSTEE98 on A09 also displayed distorted segregation patterns. Such distortions often have been observed in segregating doubled‐haploid (DH) populations in B. napus, even in other regions of the genome, and are thought to be due to in␣vitro androgenesis (Foisset et␣al., 1996; Delourme et␣al., 2013). To confirm that our approach revealed a genuine gene‐for‐gene interaction contributing to quantitative resistance, we used the CRISPR‐Cas9 method to introduce a range of mutations into AvrLmSTEE98. Heavily altered versions of the protein did not induce a HR on cotyledons, suggesting a lack of recognition by RlmSTEE98. In addition, the mutant isolates induced larger areas of stem necrosis in ‘Yudal’, also suggesting that the RlmSTEE98‐AvrLmSTEE98 interaction contributes to stem resistance to wild‐type (WT) isolates.

Our results thus highlighted a role for a single genomic region in ‘Yudal’, and functional analyses strongly suggest that the interaction was controlled by a single resistance gene. The participation of single major genes in adult‐stage resistance through gene‐for‐gene interactions already has been reported in other models, such as the Brassica oleracea/Plasmodiophora brassicae (Rocherieux et␣al., 2004), apple tree/Venturia inaequalis (Calenge et␣al., 2004) and barley/Pyrenophora graminea (Arru et␣al., 2003) pathosystems, but has never before been reported for the B. napus–L. maculans interaction. Gene‐for‐gene interactions underlying partial resistance also have been identified in a few models, such as the wheat/Zymoseptoria tritici system (Meile et␣al., 2018), or the rice/Magnaporthe oryzae pathosystem, in which an avirulence gene (AvrPi34) corresponding to the single dominant resistance gene Pi34 was shown to be responsible for the partial resistance phenotype (Zenbayashi‐Sawata et␣al., 2005).

The BnDYDH population has been used extensively to search for quantitative trait loci (QTL) for resistance to L. maculans following the scoring of stem necrosis severity in the field (Pilet et␣al., 1998, 2001; Jestin et␣al., 2011; Huang et␣al., 2016). Using the same BnDYDH population we mapped RlmSTEE98 to chromosome A09, but, surprisingly, this locus originated from ‘Yudal’, the parent of the cross considered susceptible under field conditions. QTL for resistance may originate from the susceptible parent, and resistance QTL have been identified in the vicinity of the genomic region containing RlmSTEE98, but no such QTL for resistance was identified in ‘Yudal’ in previous studies, the closest QTL associated with stem canker resistance located on A09 originating instead from ‘Darmor‐bzh’ (Jestin et␣al., 2011). Stem canker resistance QTL or associated markers on A09 from other cultivars also were detected by Jestin et␣al. (2015), Larkan et␣al. (2016a) and Raman et␣al. (2016). However, all these regions mapped to areas upstream from the physical position of RlmSTEE98.

There are four possible reasons for the identification of a resistance QTL originating from the susceptible genotype and not identified previously at this genomic location: (1) RlmSTEE98 is differentially expressed at different stages of plant growth, being expressed in the cotyledons and leaves, for example, but not in stems, preventing the recognition of AvrLmSTEE98 during stem colonization, when this gene is expressed. Stage‐specific expression patterns have been observed for other resistance genes, such as the wheat Lr34 gene, which is weakly expressed at the cotyledon stage but strongly induced in adult plants, conferring quantitative resistance to several pathogens (Krattinger et␣al., 2009). This hypothesis also is supported by the identification of differential expression profiles between cotyledons and stems for the genes present in the RlmSTEE98 region in ‘Darmor‐bzh’ (data not shown). However, our results showing that AvrLmSTEE98 inactivation increases the size of stem lesions in ‘Yudal’, but not in ‘Darmor‐bzh’ in controlled conditions strongly suggests that recognition takes place in the stem, rendering this hypothesis highly unlikely. Genomic and transcriptomic data for different developmental stages are now required for ‘Yudal’, to resolve this issue. (2) Alternatively, AvrLmSTEE98 may be expressed too late or at too low a level during systemic colonization, leading to late or insufficient recognition by the plant, and an inability to prevent the development of stem canker at this stage. Indeed, transcriptomic analyses in ‘Darmor‐bzh’ have shown that AvrLmSTEE98 is not expressed in field conditions from November to March, and that its expression peaks in April, but at a lower level than for two of the other LmSTEE genes analyzed here, LmSTEE1 and LmSTEE30 (Gay et␣al., 2021; Fig. S1). According to this hypothesis, the choice of LmSTEE genes for resistance screening in the approach developed here, should be based on a precise characterization of their expression profiles under field conditions, with particular attention paid to effectors expressed at the earliest stages of stem colonization. (3) Another possible explanation for the absence of resistance QTL associated with this region is that the RlmSTEE98 region is effectively associated with a resistance QTL conferring partial resistance to the fungus but not identified as such under field conditions. Several studies have shown that the detection of resistance QTL can be strongly influenced by environmental conditions, particularly for the B. napus–L. maculans system under European cropping conditions (Kaur et␣al., 2009; Raman et␣al., 2012). The effect of the RlmSTEE98 gene may be subject to such influences, preventing its detection. (4) Finally, the partial resistance conferred by RlmSTEE98 may have remained undetected owing to a combination of insufficient sensitivity and accuracy of the disease severity scoring in the field and a relatively low contribution to quantitative resistance relative to other QTL in this cross.

This study provides a new approach and tools for enriching the Rlm gene pool for a system in which the scarcity of Rlm genes has proved a critical limitation to crop sustainability. Our study also highlights the importance of the diversity of the plant material screened. Our screening was of limited diversity, with only European winter rapeseed varieties plus ‘Yudal’. More distantly related genotypes also might display new specific resistances to LmSTEE effectors. Also, choice of LmSTEE genes expressed early during stem infection could allow the identification of new RlmSTEE resistance in winter‐type European cultivars known to display a high level of adult‐stage resistance. Many other late effector candidates have been described, with 40 effector candidates belonging to the stem biotrophy wave of expression (Gay et␣al., 2021). These genes will be targets of choice for this strategy, now that proof‐of‐principle has been obtained for its validity.

With time, the B. napus–L. maculans system has become a model of choice to conceptualize mechanisms of plant–fungal pathogen coevolution, with a range of adaptive mechanisms set up by the fungus to break‐down qualitative resistance while maintaining effector function (Rouxel & Balesdent, 2017). The finding that LmSTEE genes may behave as avirulence effectors operating very late in the disease cycle and not prone to accelerated mutation rate make the picture more complex and suggests the plant surveillance machinery has been able to adapt to all tissue‐specific waves of effector gene expression to recognize the fungal presence.

The data obtained thus provide new tools and strategies for knowledge‐driven breeding for quantitative resistance in B. napus. They also cast new light on the mechanisms of quantitative resistance and question on the durable quantitative resistance paradigm in this model.

Author contributions

M‐HB, TR, RD designed the research; AJ, JG, AGK, BO, IF, M‐HB and TR performed the research; AJ, EJG, RD and M‐HB analyzed, collected and interpreted the data; M‐HB, TR and SF supervised the work; and AJ, JG, TR, M‐HB, RD and SF wrote the manuscript. AJ and JG contributed equally to this work.

Fig.␣S1 Expression profiles of LmSTEE genes during rapeseed colonization.

Fig. S2 Mean disease ratings of three rapeseed cultivars post inoculation with L. maculans transformants expressing ‘late’ effector genes in cotyledons.

Fig. S3 Expression of the mutated LmSTEE98 gene in OEC transformants during early stages of cotyledon colonization.

Fig. S4In vitro growth and pathogenicity phenotypes of L. maculans LmSTEE98 mutants.

Fig. S5 Expression of LmSTEE98 in mutants during stem infection.

Click here for additional data file.

Table␣S1 Isolates used for polymorphism studies in L. maculans populations.

Table S2 List of B. napus genotypes screened for resistance.

Table S3 List of primers used in this study.

Table S4 The top 20 Blast results for each ‘late’ effector.

Table S5 PSI‐Blast results for each late effector against L. biglobosa proteins.

Table S6 Cosegregation of resistance to LmSTEE98 and two molecular markers in the ‘Darmor‐bzh’ × ‘Yudal’ doubled haploid population.

Please note: Wiley Blackwell are not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

Click here for additional data file.