The dispensable chromosome of Leptosphaeria maculans shelters an effector gene conferring avirulence towards Brassica rapa


Author for correspondence:

Marie-Hélène Balesdent

Tel: +33 1 30 81 45 73



  • Phytopathogenic fungi frequently contain dispensable chromosomes, some of which contribute to host range or pathogenicity. In Leptosphaeria maculans, the stem canker agent of oilseed rape (Brassica napus), the minichromosome was previously suggested to be dispensable, without evidence for any role in pathogenicity.
  • Using genetic and genomic approaches, we investigated the inheritance and molecular determinant of an L. maculans–Brassica rapa incompatible interaction.
  • Single gene control of the resistance was found, while all markers located on the L. maculans minichromosome, absent in the virulent parental isolate, co-segregated with the avirulent phenotype. Only one candidate avirulence gene was identified on the minichromosome, validated by complementation experiments and termed AvrLm11. The minichromosome was frequently lost following meiosis, but the frequency of isolates lacking it remained stable in field populations sampled at a 10-yr time interval, despite a yearly sexual stage in the L. maculans life cycle.
  • This work led to the cloning of a new ‘lost in the middle of nowhere’ avirulence gene of L. maculans, interacting with a B. rapa resistance gene termed Rlm11 and introgressed into B. napus. It demonstrated the dispensability of the L. maculans minichromosome and suggested that its loss generates a fitness deficit.


In addition to the normal complement of essential chromosomes, many animal, plant, fungal, and oomycete species contain B chromosomes (also known as supernumerary or dispensable chromosomes) which are inherited in a non-Mendelian manner (Jones, 1995; Covert, 1998). These chromosomes are not essential for life and are lacking in some individuals. Most plant B chromosomes are enriched in transposable elements. As a consequence, they are mainly or entirely heterochromatic, and largely noncoding. Persistence in a species is thus dependent on either higher transmission rates (Muñoz-Pajares et al., 2011) or the presence of the few coding sequences that present a selective advantage, at least under some growth conditions. According to Covert (1998), ‘supernumerary chromosomes that confer an adaptive advantage in certain habitats, such as the ability to cause disease on a specific host, may be referred to as ‘conditionally dispensable’ chromosomes (CDC) in order to reflect their importance in some, but not all, growth conditions.’

In fungi, B chromosomes or CDCs have been described in species belonging to the classes Sordariomycetes and Dothideomycetes, including the plant pathogens Magnaporthe oryzae (Chuma et al., 2003, 2011), Fusarium oxysporum (Ma et al., 2010), Nectria haematococca (Coleman et al., 2009), Alternaria alternata (Hatta et al., 2002), Alternaria arborescens (Hu et al., 2012), Cochliobolus heterostrophus (Tzeng et al., 1992) and Mycosphaerella graminicola (Wittenberg et al., 2009) (for additional examples: Covert, 1998). CDCs are not required for saprophytic growth, but their contribution to other stages of the fungal life cycle seems to be variable from one species to another. The most commonly reported selective advantage lies in host range delineation: the presence in the CDC of genes encoding host-selective toxins (HSTs) or effectors allows the pathogen to infect new plant species (Mehrabi et al., 2011) while, in some cases, lack of the CDC results in isolates with a saprophytic behaviour (Johnson et al., 2001). For instance, in A. alternata pathotypes, CDCs carry clusters of genes encoding secondary metabolites acting as HSTs responsible for pathogenicity on distinct host plants, some of them being duplicated in the essential genome (Mehrabi et al., 2011). In N. haematococca, at least three B chromosomes have been identified, one of which is dispensable for in vitro growth and carries genes involved in pathogenicity to pea (Pisum sativum), such as genes involved in phytoalexin detoxification (Mehrabi et al., 2011). An F. oxysporum CDC was shown to harbour the SIX (Secreted In Xylem) genes, encoding small secreted proteins (SSPs) acting as effectors and virulence/avirulence factors. In this species the genes harboured by the CDC were postulated to be the main determinants for adaptation to tomato (Solanum lycopersicum) (Ma et al., 2010). While their loss has been associated with reduced fertility, M. oryzae CDCs are believed to be nonessential for growth and pathogenicity, probably because of the occurrence of duplicated copies of the avirulence genes in the core genome (Chuma et al., 2003, 2011). In the fungal species with the greatest number of dispensable chromosomes reported to date, M. graminicola, dispensable chromosomes harbour, among others, genes with inactivated paralogues in the essential genome, while all pathogenicity genes currently described are harboured by the core genome (Wittenberg et al., 2009; Goodwin et al., 2011).

In Leptosphaeria maculans, the causal agent of stem canker of oilseed rape (Brassica napus), the smallest chromosome, termed the minichromosome (MC), was shown to be inherited in a non-Mendelian manner following in vitro crosses and was lacking in the electrokaryotypes of some isolates. Dispensability of its coding sequences, rather than translocation to larger chromosomes, could, however, not be demonstrated, preventing it from being definitively classified as a CDC (Leclair et al., 1996). Whole-genome sequencing of L. maculans showed that the MC is extremely enriched in transposable elements (TEs) and contains only a few predicted genes (Rouxel et al., 2011).

In L. maculans, genes encoding effectors involved in gene-for-gene interactions (avirulence genes) are characterized by their genome location in large, TE-rich chromosomal landscapes of the genome termed AT-isochores (Gout et al., 2006b; Fudal et al., 2007; Parlange et al., 2009). In addition, most of the putative effector genes (i.e. encoding SSPs) showing increased expression in the first stage of plant infection are located in such genomic landscapes (Rouxel et al., 2011). These features suggested an improved cloning strategy to rapidly identify new candidate avirulence (Avr) genes following phenotyping, by combining genetic mapping of the gene and selection of candidate effector genes based on their genome location and induced expression in planta.

In this study, we identified a new resistance (R) source in Brassica rapa, a plant species closely related to B. napus, and investigated the genetic control of the interaction in the plant and the pathogen. The new resistance was shown to be under monogenic, dominant genetic control, while an apparently more complex genetic control was found in the fungus. Cloning of the avirulence determinants showed that the avirulent phenotype was nevertheless caused by a single gene, AvrLm11, resident on the L. maculans MC. The dispensability and frequent loss of the MC in vitro were then demonstrated, while its maintenance in natural populations suggested its involvement in fungal fitness. The L. maculans MC thus can be considered as a CDC that may, like those of other fungal phytopathogens, host genes involved in host range definition.

Materials and Methods

Leptosphaeria maculans isolates

Thirty-one isolates from the IBCN and IMASCORE collections of isolates (Eckert et al., 2005), previously phenotyped for the nine Avr alleles (AvrLm1–AvrLm9; Balesdent et al., 2005) and originating from Europe, Canada, Australia and New Zealand, along with isolate v11.1.1 (Rouxel et al., 2003), were used in the preliminary screening experiments leading to the identification of a new source of resistance in Brassica rapa L. Among these, isolate IBCN14, belonging to race Av5-6 (i.e. combining the avirulence alleles at the AvrLm5 and AvrLm6 loci only; Balesdent et al., 2005), was found to be virulent on the resistant B. rapa line. AvrLm11 segregation analysis and mapping were carried out following an in vitro cross (cross 38) between two isolates of opposite mating-types (Mat), the sequenced isolate v23.1.3 (Av1-4-5-6-7-8, Mat1-2) and IBCN14 (Av5-6, Mat1-1), from which 84 progeny isolates were recovered and termed v38.x.y, with x referring to the pseudothecia from which the isolate originates and y to the isolate number.

Progeny of three available crosses were used to analyse the segregation and loss of the MC: crosses 11, 28 and 37 (Table 1). One isolate per pair of twin genotypes from tetrad Z (Z1, MC+; Z2, MC; Z4, MC, Z6, MC+; Leclair et al., 1996) were also used. Three hundred and ten single-conidia isolates from a wide-scale sampling performed on naturally infected oilseed rape leaves of a single cultivar in 15 locations in France in 2000–2001 (Balesdent et al., 2006) were randomly selected within each site and screened for the occurrence of the MC (Table 2). Ten years later (2010–2011), a comparable wide-scale sampling was performed in seven of these sites using the same protocol and trap cultivar, giving rise to a collection of 497 isolates (Table 2). Procedures for isolate culture and maintenance, sporulation and in vitro crosses were as previously established (Gall et al., 1994; Ansan-Melayah et al., 1995). Conidia suspensions were recovered from 12-d-old V8-juice agar culture in sterile water. Undiluted spore suspensions (for DNA extraction) or suspensions adjusted to 107 conidia ml−1 (for inoculation tests) were stored at −20°C until used.

Table 1. List and characteristics of Leptosphaeria maculans in vitro crosses analysed in this study
Cross no.Parental isolate 1 (Mat 1-1)Parental isolate 2 (Mat 1-2)No. of progeny analysedMinichromosome (MC) segregation
  1. a

    Cross 11 is as described by Ansan-Melayah et al. (1995); crosses 28 and 37 are BC3 and BC5 described by Huang et al. (2006); cross 38, this study.

  2. b

    MC+, evidence (obtained using pulsed field gel electrophoresis (PFGE) or amplification of Super-Contig 22 specific markers) for the presence of the minichromosome; MC, no PFGE MC band or no amplification of any markers specific for Super-Contig 22.

11aa.2 (MC+)bH5 (MC+)8582 (MC+):3 (MC) (3.5%)
28v27.10.1 (MC+)v23.1.3 (MC+)2827 (MC+):1 (MC) (3.6%)
37v36.2.13 (MC+)v23.1.3 (MC+)179167 (MC+):12 (MC) (6.7%)
38IBCN14 (MC)v23.1.3 (MC+)8470 (MC+):14 (MC) (16.7%)
Table 2. List and characteristics of French Leptosphaeria maculans populations characterized for the occurrence of AvrLm11 and minichromosome-specific markers
LocationRegiona2000–2001 samplingsb2010–2011 samplingsc
No. of isolates analysedNo. (%) of isolates without AvrLm11 and MC markersNo. of isolates analysedNo. (%) of isolates without AvrLm11 and MC markers
  1. a

    C, central region; E, eastern region; NC, north central region; W, western region; S, southern region; for more information on the sites, see Balesdent et al. (2006).

  2. b

    Isolates were randomly subsampled from the Balesdent et al. (2006) samplings.

  3. c

    In 2010–2011, L. maculans populations were sampled in seven selected sites, on the same host plant genotype Drakkar and with the same procedure as for the 2000–2001 sampling.

  4. d

    nd, not done.

ChartrainvilliersC20 1ndd 
ChateaurouxC18 0nd 
LutzC20 1nd 
OucquesC20 080 4
TouryC16 180 1
MartincourtE20 0nd 
Mons en PevèleN20 1nd 
PrémesquesN19 180 2
GrignonNC35 280 3
VerneuilNC19 180 1
VersaillesNC24 2nd 
La BénateW20 1nd 
Le RheuW22 332 0
Les AlleudsW19 0nd 
MondonvilleS20 065 1
Total 31214 (4.5)49712 (2.4)

Inoculation tests

Isolates were phenotyped for their virulence towards hosts carrying specific resistance genes following a cotyledon-inoculation test (Balesdent et al., 2001; Supporting Information Methods S1). Spore suspensions were inoculated on 10–12 plants of the B. napus lines Westar (no R gene), 01-23-2-1 (Rlm7), Jet Neuf or Pixel (Rlm4), Columbus (Rlm1, Rlm3), 00-156-2-1 (Rlm8) (Balesdent et al., 2002) and 02-159-4-1 (B. rapa resistant line; this work). IBCN14 was also inoculated on Surpass400 (Rlm1, RlmS; van de Wouw et al., 2008) and a B. napus line carrying Rlm10 (Eber et al., 2011). Symptoms were scored 12–21 d after inoculation using a 1 (avirulent) to 6 (virulent) rating scale (Methods S1). Isolates were classified as avirulent or virulent for a given locus whenever the mean disease rating on the corresponding differential line was ≤ 3 or > 3, respectively (Balesdent et al., 2001).

Plant multiplication and crossing

Brassica napus is a natural amphidiploid species (AC genome, 2n = 38) derived from hybridization between B. rapa (A genome, 2n = 20) and Brassica oleracea (C genome, 2n = 18; U, 1935). One plant (plant 2323) from the B. rapa (AA) resistant accession 02-159-4-1 was crossed with a susceptible B. rapa doubled haploid line, Z1 (kindly provided by Agriculture and Agri-Food Canada, Saskatoon, Canada), and with two B. napus (AACC) winter oilseed rape varieties, Darmor and Eurol (provided by the Genetic Resource Center, BrACySol, UMR IGEPP, Ploudaniel, France). More than 130 seeds were produced from AA F1 resistant hybrids crossed to Z1 (backcross BC1). Simultaneously, in the progeny of AAC resistant F1 hybrids crossed to the recurrent B. napus varieties (BC1), > 110 plants were analysed and resistant plants with 2n = 38 were selected for two additional backcrosses (BC2 and BC3). Selfing progenies of two BC3 plants per origin were screened to produce homozygous plants.

Plant cytogenetic analyses

Flow cytometry was used at the seedling stage to assess the plant chromosome number of each F1 hybrid as described by Leflon et al. (2006). For the establishment of meiotic behaviour, metaphase I (MI) was analysed for 20 to 30 pollen mother cells of young floral buds. They were fixed in Carnoy's solution (alcohol : chloroform : acetic acid, 6 : 3 : 1) for 24 h at room temperature and stored in 50% ethanol at 4°C. Anthers were squashed and stained in a drop of 1% acetocarmine solution.

DNA and RNA extractions

Fungal DNA was extracted from 1 ml of undiluted conidial suspension, centrifuged at 6000 g for 10 min. Spores were ground with carbide beads using a Retsch MM380 grinder and DNA was extracted using the BioRobot3000 and the DNeasy 96 Plant Kit (Qiagen S.A., Courtaboeuf, France) according to the manufacturer's recommendations. RNA from B. napus cotyledons inoculated with L. maculans was extracted 7 d post inoculation as described by Fudal et al. (2007).

Genetic mapping

The FONZIE pipeline (Bally et al., 2010) was used to identify single-copy minisatellite (MS) markers and to design primers, departing from the whole-genome sequence (WGS) of isolate v23.1.3. Polymorphic MS markers were amplified with standard procedures (Gout et al., 2006a) and separated in 2% agarose gels. The size of MS bands in progeny isolates was compared with those of the parental isolates used as controls. Linkage analyses among MS or Avr loci was performed using the Mapmaker/EXP 3.0 software (available online; Whitehead/MIT Center for Genome Research, Cambridge, MA, USA) with an LOD (logarithm of odds) score of 4.0 and a maximum recombination frequency of 20 cM.

Gene annotation

Annotation of untranscribed regions (UTRs), transcriptional start and stop sites and intron positions was performed following PCR amplification and sequencing of the 3′ and 5′ ends of cDNA using the Creator SMART cDNA Library Construction Kit (Clontech, Palo Alto, CA, USA) according to the manufacturer's recommendations and using AvrLm11-5UTRU, AvrLm11-5UTRnestU, AvrLm11-3UTRU and AvrLm11-3UTRnestU as specific primers (Table S1).

Cloning and transformation

The binary vector pPZPNat1, which carries the nourseothricin acetyltransferase 1 (nat1) gene (conferring resistance to the antibiotic nourseothricin) under control of the Aspergillus nidulans indole-3-glycerolphosphate synthase (trpC) promoter as a selectable marker, was used to introduce the candidate AvrLm11 gene into L. maculans isolate IBCN14 using Agrobacterium tumefaciens-mediated transformation (ATMT). AvrLm11 was amplified from genomic DNA of v23.1.3 using primers AvrLm11U and AvrLm11-XhoL (which introduces a XhoI restriction site; Table S1). The 1485-bp fragment obtained was digested by NheI and XhoI and ligated into pPZPNat1 previously digested by SpeI and XhoI. The construct, termed pPZPNat1-AvrLm11, was cloned into Escherichia coli, re-extracted and sequenced. Finally, the construct was introduced into A. tumefaciens strain C58 by electroporation at 1.5 kV, 200 ohms and 25 μF and used for transformation of the virulent isolate IBCN14. Transformation of L. maculans was performed as described by Gout et al. (2006b). Fungal transformants were selected on 50 μg ml−1 nourseothricin (Werner BioAgents, Jena, Germany).

Gene expression

Expression of genes located in super-contig (SC) 22 was analysed using whole-genome oligoarray data (Rouxel et al., 2011) corresponding to hybridization using four different conditions (mycelium grown for 7 d in Fries liquid medium, and oilseed rape infected cotyledons 3, 7 and 14 d post inoculation). Background correction, normalization and calculation of average expression levels were performed as described by Rouxel et al. (2011). In summary, gene models with an expression higher than three times the median of random probe intensities in at least two of three biological replicates were considered as transcribed. Transcripts showing a 1.5-fold change in transcript level between in vitro and in planta conditions with a P-value < 0.05 were considered to be significantly differentially expressed during infection compared with mycelial growth.

Multiplex PCR assay

A multiplex PCR assay was used to detect the presence of AvrLm11 in all field or progeny isolates. This assay aimed to amplify the mating type gene (Cozijnsen & Howlett, 2003) and AvrLm11. Using this assay, AvrLm11 was amplified along with one of the two mating-type alleles, used as internal controls of PCR amplification for isolates for which no AvrLm11 amplification could be obtained. PCR amplifications were performed in a total volume of 15 μl containing 0.2 μM of each deoxynucleotide triphosphate, 0.67 μM of each of the five primers MATU, MATL1, MATL2, AvrLm11_U2 and AvrLm11_L (Table S1), 0.6 U of Taq DNA polymerase (Qbiogen, Illkirch, France), 1.5 μl of the MgCl2-containing 10X reaction buffer supplied with the enzyme, and 1 μl of genomic DNA (between 10 and 30 ng of DNA). PCR amplifications were performed in an Eppendorf Mastercycler EP Gradient thermocycler (Eppendorf, Le Pecq, France), with 30 cycles of 94°C for 30 s, 59°C for 30 s, and 72°C for 60 s, with a final extension at 72°C for 5 min.

Data analyses

Segregations of phenotypes or genotypes were compared with expected segregation ratios using χ2 tests. Nonparametric tests were used to compare frequencies of genes between different populations of L. maculans (Wilcoxon tests for paired observations or Mann–Whitney tests for unpaired observations) or to compare genomic features between different genomic compartments (Kruskal–Wallis test and Steel–Dwass–Critchlow–Fligner multiple comparisons). Exact binomial tests were performed to test the enrichment of the MC in SSPs. All statistical analyses were performed using xlstat v2010.5.01 (Addinsoft, Paris, France) or R scripts.


Selection of a resistant B. rapa line

In the course of screening a germplasm collection for resistance to European L. maculans isolates (Rouxel et al., 2003), one plant out of 10 from the B. rapa oleifera accession IPK cr 1564/96 was found to be resistant to four isolates representative of the main European L. maculans race, race Av5-6-7-8 (v11.1.1, UK1, PT1 and Raw4; Balesdent et al., 2005; Stachowiak et al., 2006). The plant was selfed and the resulting selfing (99-159-1-1) was tested for resistance to v11.1.1, UK1 and 27 isolates from the IBCN collection. The line was fully resistant to 23 isolates, fully susceptible to two isolates including IBCN14, and displayed a heterogeneous resistance (i.e. plants being either resistant or susceptible) to four isolates. Isolates showing either virulent or heterogeneous interaction phenotypes were all characterized by a virulence allele at the AvrLm7 locus, suggesting that Rlm7 could have been heterozygous in the selected B. rapa plant, as previously described in another B. rapa accession (Leflon et al., 2007), in addition to a new R source. Two additional selfings were performed starting from plants selected in line 99-159-1-1 for their susceptibility to both IBCN14 and v38.2.5 (an AvrLm7 progeny isolate from IBCN14), to ensure the production of a B. rapa line (02-159-4-1) with a new fixed resistance but lacking Rlm7.

Genetic control of resistance and introgression into B. napus

One plant (2323) from line 02-159-4-1 was both selfed and crossed with the susceptible B. rapa line Z1. In the selfing (11 plants) and the 2323 × Z1 F1 (23 plants), all plants were susceptible to IBCN14 (Av5-6) and v38.2.5 (Av5-6-7), and resistant to the reference isolate v23.1.2 (Av5-6-7-8), consistent with the hypothesis of a dominant and homozygous resistance gene in plant 2323. One F1 plant was back-crossed to the susceptible B. rapa line. In the resulting BC1, all plants were susceptible to IBCN14, while a 66 : 69 resistant : susceptible (R : S) segregation to v23.1.2 was observed, consistent with the expected 50 : 50 R : S segregation under the hypothesis of a single dominant, homozygous resistance gene in plant 2323.

The same plant (2323) was crossed with two B. napus varieties, Darmor and Eurol. More than 100 plants of the AAC F1 hybrid progeny crossed with their recurrent parent were analysed for resistance, giving an R : S segregation ratio close to 50 : 50, confirming the presence of a major resistance gene (Tables S2, S3). The chromosome number of 47 and 34 resistant plants from Darmor and Eurol BC1, respectively, was assessed by flow cytometry (Fig. S1) and revealed a segregation of C chromosomes as already described by Leflon et al. (2006). Three plants per cross with 2n = 38 were selected. Their meiotic behaviour was highly unstable, with only 0–25% of cells showing 19 bivalents as expected for B. napus (Tables S2, S3). BC1, BC2 and BC3 progeny frequently showed an R : S distortion of segregation which can be explained by the meiotic instability, illustrated by plants with up to 2n = 39 (Table S2). These data were confirmed by the analysis of the selfing progeny of the four selected BC3 plants (Table S2); the three most stable plants gave rise to 94R : 34S plants, fitting the expected 75R : 25S segregation for one major gene (= 0.683, χ2 < 3.84), whereas the selfing progeny of the BC3 plant showing only 50% of pollen mother cells with 19 bivalents revealed a distortion (15R : 30S).

Genetic control of avirulence

The virulent isolate IBCN14 (Av5-6, Mat1-1) was crossed with the avirulent isolate v23.1.3 (Av1-4-5-6-7-8, Mat1-2). Apart from avirulence on line 02-159-4-1, three Avr genes were polymorphic in this cross, allowing us to test allelism between the avirulent phenotype on line 02-159-4-1 and the three independent Avr genes AvrLm1, AvrLm4-7, and AvrLm8. Considering these three genes, the eight possible phenotypic classes were recovered in the progeny, with proportions consistent with those expected for three independent loci (P[χ2] = 0.333; Table 3). The 14 : 70 segregation ratio of virulence to avirulence towards line 02-159-4-1 was more consistent with that of three independent Avr genes (= 0.248) than two (P = 0.06) or one gene (< 0.001). Notably, virulent and avirulent isolates were found in almost all previously defined phenotypic classes (Table 3), suggesting that, whatever the number of genes involved, they all are distinct from AvrLm4-7, AvrLm1 or AvrLm8. In addition, as AvrLm2, AvrLm3, AvrLm5, AvrLm6 and AvrLm9 are not segregating in cross 38, and IBCN14 was not virulent on Surpass400, harbouring RlmS, or on B. napus lines harbouring Rlm10 (data not shown), the interaction between v23.1.3 and line 02-159-4-1 involved at least one new avirulence gene.

Table 3. Segregation of interaction phenotypes in Leptosphaeria maculans cross 38 (IBCN14 × v23.1.3)
Avr LocusParental isolatesPhenotypic classes in the progenya
  1. a

    The eight phenotypic classes as defined by the combination of the three independent avirulence (Avr) genes that are polymorphic in cross 38, AvrLm4-7, AvrLm1 and AvrLm8.

  2. b

    V, the isolate is virulent, or A, the isolate is avirulent, following inoculation on lines with the corresponding resistance gene (Brassica napus Jet Neuf or Pixel (Rlm4), Columbus (Rlm1, Rlm3), B. rapa 00-156-2-1 (Rlm8) (Balesdent et al., 2002), 02-159-4-1 (B. rapa resistant line; this work)).

Number of isolates in each class1311111611949
Virulence on line 02-159-4-1VAVAVAVAVAVAVAVAVA
Number of isolates in each class11201147313110360427

The avirulence locus is located on the minichromosome

Among 468 MS markers showing size polymorphism in a range of L. maculans isolates (Bally et al., 2010), 235 were polymorphic between v23.1.3 and IBCN14. Of these, 227 showed size polymorphism between the alleles amplified in v23.1.3 and IBCN14, whereas seven showed a presence/absence polymorphism, with the locus amplified in v23.1.3 but not in IBCN14. One of these had been generated from the sequence of SC01 and the six others from that of SC22, previously shown to correspond to the L. maculans MC (Rouxel et al., 2011) (‘Min’ markers; Fig. 1d). In the progeny of cross 38, the presence/absence polymorphic marker from SC01 segregated independently from avirulence to line 02-159-4-1. By contrast, all polymorphic markers from SC22 co-segregated with the interaction phenotype on line 02-159-4-1; the 14 virulent isolates did not amplify the markers whereas those amplifying the markers were all avirulent. Consistent with the avirulence segregation ratio, all SC22 MS markers segregated with a very strong segregation distortion. From these data we hypothesized that the Avr locus (loci) matching the line 02-159-4-1 R gene was (were) located on SC22 and that the whole SC22 was missing in the virulent parental isolate IBCN14. In this context, it was not possible to further reduce the genetic interval around the avirulence locus.

Figure 1.

Genome structure of Leptosphaeria maculans Super-Contig 22, gene predictions and their transcriptomic profile. (a) Number of predicted genes along SC22; numbers in brackets indicate the number of genes with at least one transcriptomic support; (b) G+C content variation along SC22; (c) schematic representation of the nine AT-rich (red boxes) and the eight GC-equilibrated (blue boxes) isochores along SC22; (d) location of molecular markers generated along SC22; Minx, minisatellite markers, Px, markers amplifying portions of predicted genes. (e) In planta relative expression of predicted genes compared with in vitro conditions. Only those genes showing significant over- (closed symbols) or under- (open symbols) expression during oilseed rape infection are shown. Values are the logarithm of the ratio of mean normalized expression values at (green circles) 3 d post infection (dpi), (red squares) 7 dpi or (blue diamonds) 14 dpi over mean normalized expression under in vitro conditions, calculated from means of two technical and three biological repeats.

Segregation distortion of MC markers could be explained either by the presence of a duplicated copy of the MC in isolate v23.1.3, although the 14 : 70 segregation of the markers is only poorly supported by statistics (= 0.06), or by lower viability or germination rate of MC ascospores. Notably, in vitro cross 37 (MC+ × MC+ isolates) and cross 38 (MC+ × MC isolates; Table 1) were established at the same time under the same conditions but gave rise to progeny with clear differences in ascospore germination rate. The germination rate of ascospores discharged from all pseudothecia from cross 37 was > 90%, whereas only 67.3 ± 5% of ascospores ejected from cross 38 pseudothecia germinated.

Analysis of SC22 identified a unique avirulence gene candidate

SC22 (731 443 bp) encompasses nine AT-rich isochores (AT1–AT9; Fig. 1b,c) covering 92.5% of its length (Rouxel et al., 2011) and eight GC-equilibrated isochores representing 54 791 bp only. SC22 contains 36 predicted genes, one of them being located within an AT-rich isochore (Fig. 1). Compared with gene models of the whole L. maculans genome, a low percentage (six out of 36) of translated SC22 gene models had BLAST hits to nonredundant protein databases, but most of them (30 genes, including all genes with BLAST hits) were validated by transcriptomic and/or expressed sequence tag (EST) data (Tables 4, S4, S5). Six genes were predicted to encode secreted proteins, among which four were small (< 300 AA) proteins (SSPs). Only one of them, Lema_uP119060.1, combined all the characteristics of previously described L. maculans avirulence genes (Gout et al., 2006b; Fudal et al., 2007; Parlange et al., 2009) and therefore was selected as an avirulence candidate gene; it is located within an AT-rich isochore, is over-expressed 7 d post inoculation (Fig. 1e) and encodes a 95 AA, cystein-rich SSP (Fig. 2) with no homology in the databases. Lema_uP119060.1 was further annotated by RACE PCR. One intron of 71 nt, a 5′-UTR of 52 bp and a 3′-UTR of 210 bp were identified (Fig. 2).

Table 4. Features of the dispensable chromosome (Super-Contig 22) compared with the whole genome or with AT-rich isochores of Leptosphaeria maculans
 Super-Contig 22Whole v23.1.3 genomeAT-rich isochoresa
  1. a

    Excluding the small transition zones (mean 859 bp) between AT-rich and GC-equilibrated regions (Rouxel et al., 2011).

  2. b

    NA, not applicable.

  3. c

    Percentage of predicted proteins with a BLAST hit to proteins in the ‘Non Redundant’ (NR) database, September 2012. Cut-off value = e-10.

Size (Mb)0.7345.1215.42
Number of predicted genes3612469148
Gene density (nb of genes per kb)0.0490.270.0096
Core genome size (corresponding to GC-isochores; Mb)0.05529.7NAb
Gene density in the core genome (nb of genes per kb)0.650.42NA
Predicted proteins with BLAST hits (%)c16.684.824.3
Genes with EST, transcriptomic or proteomic support (%)82.8684.856.75
Average predicted gene size (bp)112813231084
Average predicted protein size (aa)328418.4156
Mean TpA/ApT index of predicted genes1.081.041.45
Genes with a TpA/ApT index > 1.5 (%)13.86.937
Number (%) of genes encoding predicted Small Secreted Proteins (SSP)4 (11.1)651 (5.2)57 (38.5)
Figure 2.

Nucleotidic sequence of region surrounding AvrLm11 (accession number CBX90019.1), used for complementation, and amino acid sequence of the corresponding predicted protein. 5′ and 3′ untranslated regions of the gene are underlined. The intron is indicated as italicized lowercase (77 nt). The predicted signal peptide (15 aa) is indicated by dark grey shading and the eight cysteine residues are indicated by grey shading. The asterisk indicates the stop codon.

AvrLm11, a ‘classical’ L. maculans avirulence gene

To test the ability of Lema_uP119060.1 to confer avirulence towards the 02-159-4-1 B. rapa resistant line, a complementation assay was carried out. A fragment of 1453 bp corresponding to the complete open reading frame of Lema_uP119060.1, a promoter region of 418 bp and a terminator region of 358 bp (Fig. 2) was cloned into the binary vector pPZPNat1 and introduced into the virulent isolate IBCN14 by ATMT. As a control, the SC22 gene Lema_P119130.1 (Fig. 1), also encoding an SSP, was similarly introduced into IBCN14. Overall, 18 and 14 independent transformants were isolated for Lema_uP119060.1 and Lema_P119130.1, respectively. Only transformation of pPZPNat1-Lema_uP119060.1 into IBCN14 resulted in an avirulent phenotype following inoculation on line 02-159-4-1 for all 18 transformants, whereas their virulence towards Westar (no R gene) or 02-23-3-1 (Rlm7) was unchanged (Figs 3, S2). Therefore, Lema_uP119060.1 is the gene responsible for the avirulent phenotype on line 02-159-4-1. Based on (1) the single, major gene control of the resistance in the B. rapa 02-159-4-1, (2) the control of the corresponding avirulent phenotype by a single avirulence gene and (3) the independence of this interaction with the previously defined AvrLm1-AvrLm10 and AvrLmS Avr genes, Lema_uP119060 was termed AvrLm11 and the corresponding resistance, Rlm11. Further inoculation of a B. napus differential set with IBCN14 complemented with AvrLm11 confirmed that Rlm11 is distinct from known R genes to L. maculans (Table S6).

Figure 3.

Differential interaction phenotypes of Leptosphaeria maculans isolates on two resistance sources, Rlm7 and Rlm11. Cotyledons of Westar (no R gene), 02-23-3-1 (Rlm7) and 02-159-4-1 (Rlm11) were inoculated with the wild-type isolates v23.1.2 (Av5-6-7-8-11), IBCN14 (Av5-6) and v38.2.5 (Av4-5-6-7) and with isolate IBCN14 complemented with the AvrLm11 candidate gene. Photographs were taken 15 d post inoculation and show that the introduction of AvrLm11 in IBCN14 is sufficient to trigger the Rlm11-mediated resistance, whereas the virulent phenotype on the Rlm7 line is unaltered.

Occurrence of AvrLm11 and of the MC in progeny of crosses and in field populations

In cross 11, for which a detailed genetic map has been built (Kuhn et al., 2006; Bally et al., 2010), AvrLm11 was present in both parental isolates and three SC22 markers (Min22.0, Min22.4 and Min17884; Fig. 1d) showed size polymorphism between parental isolates. Three out of 85 (3.5%) cross 11 progeny isolates lacked these three markers, along with AvrLm11. Two isolates displayed the two alleles for the three SC22 markers but only one allele (i.e. one or the other parental allele) for the 237 MS markers located on other SCs and mapped in cross 11. These data indicated that mating has led either to the loss or to the duplication of the whole MC with a very high frequency (5.9%). In two additional in vitro crosses, the multiplex PCR assay (Fig. 4) revealed the loss of AvrLm11 in 3.5–6.7% of their progeny (Table 1). Taking into account these three crosses, AvrLm11, along with all markers from SC22, were lost with a mean frequency of 4.8% (Table 1; data not shown). Field isolates sampled in France in 2000/2001 and 2010/2011 were multiplex-PCR assayed for AvrLm11 presence/absence. Depending on the year and the site, AvrLm11 was absent in 0–13.6% of each population. Overall, 3.2% of French isolates lacked AvrLm11 (Table 2). No statistically significant evolution of AvrLm11 frequency could be found in this 10-yr time interval (Wilcoxon test for paired observations (for the seven sites sampled in both decades), = 0.248; Mann–Whitney test (if considering all sites sampled), = 0.206). To confirm the complete dispensability of the MC, additional PCR markers that amplified predicted genes were generated for each of the eight GC-isochores of SC22 (‘Px’ markers; Fig. 1d, Table S1). Similarly to IBCN14, all field isolates lacking AvrLm11 in both collections also lacked all SC22 PCR markers. Finally, the presence of all SC22 markers, including AvrLm11, was analysed in the single-ascospore isolates of the complete tetrad ‘Z’, in which the loss of the MC band was first identified by pulsed field gel electrophoresis (PFGE) (Leclair et al., 1996). AvrLm11 and SC22 markers were amplified in the two parental isolates and in the four isolates of the tetrad showing the MC band, whereas they were not amplified in the four isolates lacking the MC band (data not shown).

Figure 4.

Multiplex PCR identification of the presence of AvrLm11 in Leptosphaeria maculans isolates. (a) Multiplex assay for the reference isolates v23.1.2 (Mat1-1, AvrLm11), v23.1.3 (Mat1-2, AvrLm11), IBCN14 (Mat1-1, avrLm11) and the water control. (b) An example of the use of the multiplex assay for wide-scale screening of the presence of AvrLm11 in field isolates. The arrow indicates the PCR product for AvrLm11, the double arrows the PCR products for mating-type alleles, and the asterisks field isolates missing AvrLm11.

RIP signatures in SC22 genes

Repeat-induced point mutation (RIP) is a pre-meiotic mechanism of inactivation of duplicated sequences in some ascomycetes (Galagan & Selker, 2004). RIP has been shown to be active in L. maculans and to contribute to rapid evolution of effector genes, leading to either their diversification or their inactivation (Idnurm & Howlett, 2003; Fudal et al., 2009; Rouxel et al., 2011; Daverdin et al., 2012). As duplication of the MC has been observed to occur in some progeny isolates (2.3% in cross 11; see also Leclair et al., 1996), it was questioned whether this duplication could lead to higher RIP signatures on MC genes compared with other chromosomes. The RIP index TpA/ApT was calculated for all L. maculans genes located in GC-isochores, in AT-isochores or in SC22. The RIP index of SC22 genes was intermediate between, but not significantly different from, those of GC-isochores and AT-isochores (= 0.837 and = 0.218, respectively; nonparametric Steel–Dwass–Critchlow–Fligner multiple comparisons). However, five (13.8%) SC22 predicted genes, including AvrLm11, had a TpA/ApT > 1.5, which represents twice the proportion observed for the whole L. maculans genome (6.9%) but half that for genes within AT-isochores (37%; Table 4).


In the 1990s, PFGE of L. maculans karyoptypes demonstrated both major chromosome-size polymorphism, generated through meiosis, and the apparent loss of the smallest chromosomal band (Plummer & Howlett, 1993, 1995; Leclair et al., 1996). Leclair et al. (1996) failed to identify single-copy sequences specific for the MC, probably because of the large proportion of repeats it harbours, therefore preventing a formal Southern blot demonstration that the absence of the smallest PFGE band really corresponded to loss of genetic material rather than translocation to larger chromosomes. Dispensability of the MC therefore remained questioned until this study. WGS of L. maculans confirmed that the MC indeed was mainly made up of repeated elements, in a much higher proportion than in the rest of the genome, with a low GC content (35.2% GC for SC22 compared with 44.1% for the WGS; Rouxel et al., 2011). The search for MS markers targeted to GC-isochores enabled us to generate here single-copy markers along the entire MC. The analysis of their occurrence in isolates from a tetrad previously described as lacking the MC PFGE band (Leclair et al., 1996) demonstrated the complete loss of this chromosome, including all GC-isochores and genes. Therefore, the smallest chromosome of L. maculans has to be considered as a B chromosome or a CDC.

The L. maculans CDC shares some characteristics with previously described fungal CDCs. Its size, ranging between 650 and 950 kb depending on the isolate (Leclair et al., 1996), is close to those of the supernumerary chromosomes 15 (0.75 Mp) and 17 (0.53 Mp) of N. haematococca (Coleman et al., 2009), the 1.05 Mb A. alternata CDC (Hatta et al., 2002), the 0.41–0.77 Mb size range of the eight M. graminicola CDCs (Goodwin et al., 2011) and the 1 Mb CDC of A. arborescens (Hu et al., 2012). Similarly to CDCs of N. haematococca, F. oxysporum, or M. graminicola, a higher repeat content along with a lower GC content was also observed in the L. maculans CDC. However, a few striking features of the L. maculans CDC are noteworthy. First, its predicted gene content is far lower than those reported in other species, with a gene density of < 0.05 genes per kb, compared with c. 0.09–0.18 genes per kb for the three N. haematococca CDCs (Coleman et al., 2009), 0.14 genes per kb in M. graminicola (Goodwin et al., 2011) and 0.21 genes per kb for the A. arborescens CDC (Hu et al., 2012). Secondly, a predicted function could be attributed to none of the CDC predicted proteins, and a very low percentage of them had BLAST hits. For example, 160 out of 209 genes (76.5%) of the A. arborescens CDCs had BLAST hits compared with 16.6% for L. maculans CDC genes. In this respect, the L. maculans CDC more closely resembles those of M. graminicola, for which 10% of the genes in the dispensome had BLAST hits compared with 59% for the core chromosome genes. However, the M. graminicola CDCs contained a lower proportion of secreted proteins or other pathogenicity genes than the core chromosomes, while in L. maculans the MC contained a proportion of SSP-encoding genes twice as high as, but not significantly different (binomial exact test, = 0.115) from, that of the whole genome (Table 4).

The genome of L. maculans is structured into alternate regions with homogenous GC content abruptly changing from AT-rich regions (or isochores) to GC-equilibrated regions, and this organization is observed for all chromosomes (Rouxel et al., 2011). Many characteristics of the L. maculans CDC are intermediate between those of AT-rich and GC-equilibrated isochores, such as gene density, gene size, RIP indices and effector gene content (Table 4). However, BLAST hits of predicted CDC proteins were much lower than predicted for the rest of the genome, including AT-rich regions, while the percentage of CDC genes with transcriptomic support was similar to that found for the entire L. maculans genome. These data highlight the specificity of the gene content of the CDC compared with other chromosomes, with still unknown functions or functions that may be specific for its interaction with its host plants or for its life cycle.

Our work shows that meiosis could generate either the loss or the duplication of the L. maculans CDC at an appreciable frequency (c. 5%). In L. maculans, inactivation of duplicated sequences by RIP is frequent and RIP signatures can be found even in single-copy genes embedded within repeated regions (Rouxel et al., 2011). Duplication of the CDC in some isolates at a nonnegligible frequency, as found here or in previous reports (Leclair et al., 1996), could lead to RIP mutations in the genes it harbours. Activity of RIP to lower the GC content of the dispensome has been described in M. graminicola (Goodwin et al., 2011) and RIP following duplication may have led to a diversification of the genes harboured by the MC in L. maculans. However, genes located in GC-isochores of SC22 display only slightly higher RIP indexes than the genes located in GC-isochores from other chromosomes, maybe because higher rates of RIP mutations would have led to inactivation of important proteins and have been counter-selected.

The foreign origin and horizontal transfer of CDCs has been suggested or demonstrated in a few fungal species, leading to a change in host range or in pathogenic vs saprophytic status (Mehrabi et al., 2011). The low BLAST hits and lack of predicted function of proteins encoded by the L. maculans CDC could be an argument in favour of such a hypothesis; however, the current very low number of BLAST hits of CDC proteins to prokaryotic or eukaryotic sequences prevents any attempt to test this hypothesis and trace back the phylogenetic origin of the L. maculans CDC. In addition, because the CDC structure resembles the unique isochore structure of the core chromosomes of L. maculans, it is more plausible that its CDC evolved from the core chromosomes.

A few recent reports describe the occurrence of avirulence genes in fungal CDCs. Here we show that the L. maculans CDC also shelters a high proportion of genes encoding SSPs, one of which, AvrLm11, encodes an avirulence effector. AvrLm11 is the first cloned L. maculans Avr gene corresponding to a resistance source identified in B. rapa. AvrLm11 has all the features of L. maculans Avr genes: it is a ‘middle of nowhere’ gene (Gout et al., 2006b) located within a large (321-kb) AT-rich region, it is expressed during infection, it is over-expressed 7 d after inoculation and it encodes a predicted cystein-rich SSP. Its location on a CDC makes the study of its evolution pattern particularly interesting. We have shown how frequently the loss of the CDC, including AvrLm11, may happen during sexual reproduction. Sexual mating occurs each year during the L. maculans infectious cycle, and ascospores produced following meiosis are in most countries regarded as the primary source of inoculum. The first leaf lesions observed in autumn in Europe correspond to ascospore infection sites, with each leaf lesion corresponding to a distinct haplotype generated by sexual recombination (Gout et al., 2006a). The frequency of the loss of the CDC following meiosis in controlled conditions (4.8%) was comparable to that observed in samples collected in autumn from leaf lesions (3.2%). Paradoxically, we did not observe any increase in the frequency of isolates having lost the CDC in a 10-yr time interval. These data suggest that maintenance of the L. maculans CDC confers a selective advantage for its life cycle completion because, without any fitness cost, annual random matings combined with a ‘mutation’ rate (loss of the MC) of 0.048 would have led, even in the absence of the Rlm11 selection pressure in the field, to a shift from 4.5% (2000/2001 sampling) to c. 35% of isolates lacking the CDC 10 yr later (2010/2011) (Methods S1, Fig. S3). Leclair et al. (1996) tested the host range and virulence of a few L. maculans isolates having lost or not the MC and did not find any fitness cost linked to its loss. However, only the early stages of the interaction were investigated, using cotyledon inoculations and qualitative rating scales (Leclair et al., 1996). Genes located on the CDC could be important in the life cycle of the fungus at a later stage, less accessible to controlled condition tests but eventually revealed by field population analyses. Whether this fitness cost is linked to the loss of AvrLm11 itself, as an effector gene, or of other CDC genes with currently unknown function, is still unresolved. In the present study, cross 38 involving isolate IBCN14, lacking the CDC, gave rise to abundant ascospore discharge, but only 70% of them germinated. In addition, a strong segregation distortion was found in its progeny for the transmission of the CDC, compared with the 1 : 1 expected segregation ratio. It can thus also be hypothesized that the loss of the CDC reduces viability of the progeny and counterbalances the high frequency of MC loss.

All these data therefore question the potential durability of the AvrLm11-matching resistance gene, Rlm11, and whether the ease by which the CDC can be lost will be counterbalanced by the apparent strong fitness cost resulting from its loss. Molecular events responsible for virulence have been investigated for three L. maculans Avr genes located in the core chromosomes (Gout et al., 2007; Fudal et al., 2009; Daverdin et al., 2012). In most cases the resistance was overcome by either gene inactivation or complete deletion of the gene and its AT-rich environment. In contrast to the findings of the present study, complete deletion only encompassed one coding sequence regardless of the size of the deletion, while in the case of AvrLm11 the whole CDC including all its GC-isochores and genes was lost. The present work has led to the introduction of Rlm11 from B. rapa to oilseed rape genotypes in a context where avirulent isolates represent > 95% of the French L. maculans population. This material is now available not only for plant breeding but also to assess in field conditions the durability of an R gene corresponding to an Avr gene located on a CDC, in comparison with Avr genes present in the core chromosomes. The survey of virulent isolates towards Rlm11 in experimental fields will be facilitated by the multiplex PCR assay to detect AvrLm11 and the CDC single-copy molecular markers designed here.


This work was funded by the French agency Agence Nationale de la Recherche (ANR) contract ANR-07-GPLA-015 (‘AVirLep’) under the framework of the Génoplante 2010 programme. The 2010–2011 large-scale sampling was funded by the CTPS project ‘Evolep’ coordinated by Xavier Pinochet (CETIOM). The authors wish to thank Laurent Coudard, Sabrina Frouillou, Julien Carpezat, Juliette Linglin, Bertrand Auclair and Martin Willigsecker for technical assistance.