A functional genomics resource for Brassica napus: development of an EMS mutagenized population and discovery of FAE1 point mutations by TILLING


Author for correspondence:
Jinling Meng
Tel:+86 27 87282457
Fax:+86 27 87280016
Email: jmeng@mail.hzau.edu.cn


  • • Two ethylmethanesulfonate (EMS) mutant populations of the semi-winter rapeseed cv. Ningyou7 were constructed with high mutant load, to provide a TILLING platform for functional genomics in Brassica napus, and for introduction of novel allelic variation in rapeseed breeding.
  • • Forward genetic screening of mutants from the M2 populations resulted in identification of a large number of novel phenotypes. Reverse genetic screening focused on the potentially multi-paralogous gene FAE1 (fatty acid elongase1), which controls seed erucic acid synthesis in rapeseed. A B. napus BAC library was screened, and loci in a reference mapping population (TNDH) were mapped to conclude that there are two paralogous copies of FAE1, one on each of the B. napus A and C genomes.
  • • A new procedure is demonstrated to identify novel mutations in situations where two or more very similar paralogous gene copies exist in a genome. The procedure involves TILLING of single plants, using existing SNPs as a positive control, and is able to distinguish novel mutations based on primer pairs designed to amplify both FAE1 paralogues simultaneously.
  • • The procedure was applied to 1344 M2 plants, with 19 mutations identified, of which three were functionally compromised with reduced seed erucic acid content.


Rapeseed is a staple oilseed crop of enormous economic importance worldwide. The oil derived from crushing harvested seed is a major provider of calorific value to the human food chain, with variations in fatty acid profile including combinations of erucic acid, oleic and linolenic acid that are of industrial value as oils, lubricants, surfactants and high-value plastics (Metzger & Bornscheuer, 2006). There is an increasing need to focus rapeseed crop improvement on optimizing performance characteristics such as fatty acid composition, seed oil content and yield. More generally, crop genetic improvement needs to expand the available variation present within the existing gene pool in order to make significant advances in key traits. For many years, mutant populations generated with chemical mutagens such as ethylmethanesulfonate (EMS) and N-ethyl-N-nitrosourea (ENU) have been used to induce valuable variations in crop genomes (Konzak et al., 1976). However, whilst precursory construction of an EMS population for B. napus has recently been reported (Wu et al., 2007a), the resultant mutant populations have not been large enough to generate sufficient variation, nor has an efficient method for screening such variation been developed. There remains a requirement for development of large mutant populations, along with suitable high-efficiency screening methods suitable for rapeseed improvement.

Functional characterization of genes is an important component of plant genomic research and is increasingly informing crop improvement. Loss of function mutants are of particular interest since they provide valuable evidence for the role of specific genes in regulatory, developmental, biochemical and metabolic networks. Several general methods such as insertional mutagenesis and RNA interference have been used to obtain reduction-of-function or knockout mutations, and used successfully in model plants such as Arabidopsis (Arabidopsis thaliana) (Long & Coupland, 1998) and crops such as rice (Waterhouse et al., 1998). However, there are particular challenges in applying these approaches to rapeseed research and crop improvement, primarily because of either reliance on Agrobacterium T-DNA vectors for transmission or lack of availability of endogenous tagging systems. Transgenesis is inefficient in B. napus, and endogenous tagging is not accepted by regulatory bodies and consumers in most countries. TILLING (targeting-induced local lesions in genomes) is a more recent reverse genetic approach that has successfully been applied to both model plants such as Arabidopsis (McCallum et al., 2000) and crops such as rice (Till et al., 2007), maize (Till et al., 2004b) and wheat (Slade et al., 2005). TILLING is based on screening populations of plants that have been treated with traditional chemical mutagens that cause point mutations, followed by discovery in genes of interest using a very sensitive detection method. The main advantage is the ability to accumulate an allelic series of mutants with a range of modified functions, from wild-type to almost loss of function (Slade et al., 2005). The original TILLING method involves seeds that are mutagenized by treatment with EMS, which primarily introduces G/C to A/T transitions. The resulting M1 plants are self-fertilized, and M2 individuals are used to prepare DNA samples for mutational screening, whilst an inventory of their seeds is established for future and downstream research (Henikoff et al., 2004). The mutational screening relies on an enzyme which cleaves singleton DNA mismatches, as described by Oleykowski et al. (1998). Since TILLING relies on use of chemical mutagens that induce genome lesions randomly, it is relatively easy to construct a sufficiently large mutant population that has a high probability of containing mutations in all gene loci. Moreover, TILLING is applicable to targeting induction of mutations in genes of interest within the genome of almost any plant species (Till et al., 2006). Species with duplicated genomes are more tolerant of induced mutation as a result of the functional redundancy conferred by paralogous sets of genes (Slade et al., 2005). Since the rapeseed genome is particularly complex, it is likely to be quite tolerant of mutations induced by EMS. These considerations suggest that TILLING is better suited to rapeseed genomic research and trait improvement than other methods reported to date.

Although TILLING is potentially very powerful, there are some difficulties in applying the method to rapeseed. The most challenging issues arise from the need to develop gene locus-specific primer pairs within the complex rapeseed genome. B. napus (AACC, 2n = 4x = 38; c. 1150 Mbp) is an amphidiploid species originating from a spontaneous hybridization of Brassica rapa (AA, 2n = 2x = 20) and Brassica oleracea (CC, 2n = 2x = 18) (U, 1935), where the oginating diploids have been retained essentially intact (Li et al., 2005). Moreover, the Arabidopsis, B. rapa and B. oleracea genomes appear to have arisen from a common hexaploid ancestor c. 14.5–20.4 million yr ago (Lysak et al., 2005). Sequential duplication events within ancestral genomes have resulted in regions that are represented as single copies within Arabidopsis, being present as three segmental copies within the B. rapa and B. oleracea genomes. This compounds the complexity of the B. napus A and C genomes (Mayer et al., 2001; Park et al., 2005; Yang et al., 2006). From recent comparative genomic studies (Park et al., 2005; Town et al., 2006), it is possible to infer the presence of six paralogous regions in B. napus, with random gene loss resulting in, on average, four intact gene copies, with additional genomic rearrangements contributing further pseudogene fragments. Among the intact paralogous genes, some are able to be distinguished based on differences such as sequence lengths and intron composition. In such cases, it is relatively trivial to obtain gene-specific primer pairs based on sequence polymorphisms. Slade et al. (2005) have reported successful screening using a TILLING approach for this class of multi-paralogous gene, in both hexaploid and tetraploid wheat. For multi-paralogous genes that differ by only a few nucleotides among different paralogues, it is very difficult to design gene-specific primer pairs to screen for mutations, and also difficult to identify mutations reliably in distinct paralogues.

We were motivated to overcome the difficulties associated with application of TILLING in polyploid species, owing to the potential benefits for functional gene analysis in the closest crop relatives to the model Arabidopsis, and specifically for potential application in genetic crop improvement of B. napus. We therefore tested our system by focusing on screening using TILLING for mutations in the gene FAE1 (fatty acid elongase1), which had previously been shown by Southern blotting analysis to be present with at least two very similar paralogous copies within the B. napus genome (Barret et al., 1998). FAE1 is the key gene in seed erucic acid biosynthesis in rapeseed (Roscoe et al., 2001), and was originally cloned in Arabidopsis by directed transposon tagging with the maize element Activator (Ac) (James et al., 1995). The product of the gene is a condensing enzyme that extends the chain length of fatty acids from C18 to C20 and C22 (Lassner et al., 1996). More recently, the restriction enzyme AvrII has been used to discriminate between the two genes present in the A and C genomes of the two B. napus cultivars Zhongyou 821 and Zhongshuang No. 9, which possess high (HEA) and low seed erucic acid (LEA) content respectively. The B. napus FAE1 gene copies are intronless, with 98% sequence similarity between the paralogues (Wu et al., 2007b). A reference doubled haploid population (TNDH) developed by our laboratory from Ningyou7 (a Chinese semi-winter cultivar) and Tapidor (a European winter cultivar) was used to detect functional QTL for seed erucic acid content, with evidence from four independent field environments (Qiu et al., 2006). Two major seed erucic acid content QTL were detected, one located on linkage group/chromosome A8 (A genome) and the other on C3 (C genome), accounting for c. 71% of the genetic variation. Taken together, such evidence indicates there are likely to be two intact highly identical functional FAE1 gene paralogues in B. napus, one in each of the A and C genomes.

Here we report development of a TILLING platform for B. napus, involving generation of two structured EMS mutagenized populations derived from homozygous cv. Ningyou7 (one parent of the TNDH mapping population). In order to demonstrate the utility of these populations for gene discovery, a forward genetic screen was carried out, resulting in a large number of novel phenotypes. Reverse genetics was tested within a potentially multi-paralogous gene system. Comparative genomic sequence analysis involved screening of a B. napus cv. Tapidor (the other parent of the TNDH population) BAC library to obtain a more precise estimate of paralogous FAE1 gene copy. BAC-derived locus-specific sequences were then mapped to the A8 and C3 chromosomes within the TNDH mapping population (Qiu et al., 2006). Reverse genetic screening for FAE1 mutants by TILLING resulted in identification of 19 mutations from 1344 M2 plants, of which three had altered function as revealed by marked changes in the accumulation of erucic acid.

Materials and Methods

Construction of EMS mutant population

Self-pollinated seeds of the Brassica napus L. DH line cv. Ningyou7 were prepared in summer 2004. For the pilot dosage-response experiment (autumn 2004), dry high-quality mature seeds of Ningyou7 were soaked in water for 8 h at room temperature and then sets of 300 seeds were imbibed in different concentrations of EMS (Sigma-Aldrich Biotech, St Louis, MO, USA) aqueous solution ranging from 0.0% (w/v) to 1.2% at 0.2% intervals in a 50 ml flask at room temperature in the dark for 18h. Seed germination rate and germination time of M0-treated seed was recorded, as well as percentage self-pollinated seed set of M1 plants derived from the M0 seed (Table 1). In subsequent large-scale population generation experiments (autumn 2005), seeds of Ningyou7 (generated in 2004 and stored under optimum conditions) were treated in either 0.6% or 0.3% EMS. Large-scale DNA isolation, and storage of M1 and M2 seed followed the methods described for the Arabidopsis TILLING project (McCallum et al., 2000; Colbert et al., 2001; Greene et al., 2003).

Table 1.  Concentrations of aqueous ethylmethanesulfonate (EMS) solution used to treat seeds of Brassica napus Ningyou7
  • a

    Batches of 300 mature Ningyou7 seeds were imbibed at each of five different concentrations of EMS solution, with a control treatment (0%) of distilled water, for 18 h at 20°C.

  • b

    Germination time per treatment was recorded as the number of days from observation of the first germinated seed to the day when the last seed germinated.

  • c

    Seed set is expressed as the percentage of M1 plants from which viable seeds were set.

Concentration (w/v) 0.0%a 0.4% 0.6% 0.8% 1.0% 1.2%
Germination rate (%)1009582.381.976.775.9
Germination time (d)b 2 712121919
Seed set (%)c10045.325 8.3 6.7 4.7

Investigation of novel phenotypes amongst the two mutant populations was carried out on M2 plants grown in field conditions (Wuhan, China at 30°35′N 108°54′E). Morphological characters scored included plant architecture, leaf, floral and reproductive morpholology, with mutant phenotypes being compared to untreated wild type plants. Phenotypic trait categories are outlined in Table 2. Seed oil content of M3 seed was estimated using near infrared spectroscopy (NIR) with a standard method (Mika et al., 2003).

Table 2.  Summary of phenotypes observed in M2 Brassica napus plants from the two ethylmethanesulfonate (EMS)-treated (0.3 and 0.6%) populations
Morphological charactersNo. of plantsPercentage (%)
0.3% (2604 plants)0.6% (7110 plants)0.3%0.6%
  1. Mutants were categorized according to architectural, floral, leaf and reproductive traits. The numbers recorded represent characters scored on individual plants and thus may represent sibling aggregates. Furthermore, since the scores are nonexclusive, a single plant may contain more than one variation.

 Dwarf stature 3  90.120.13
 Increased stature231120.881.58
  Pure white 7 130.270.18
  Light-yellow/creamy25 980.961.38
 Abnormal petal211230.811.73
 Abnormal stamen362091.382.94
 Flowers absent 5 270.190.38
 others21 570.810.80
 Abnormal appearance352131.343.00
 Abnormal color371451.422.04

BAC library screening and BAC clones analysis

The JBnB BAC library was developed from DNA isolated from a B. napus doubled haploid line derived from cv. Tapidor (Rana et al., 2004) and screened by hybridization of a 33P-labeled 426 bp Tapidor FAE1 genomic DNA fragment (amplified by primer pair FAE426). Primer pairs FAE426 and Fae1primer1 to 5 (Table 3, Fig. 1) were designed from B. napus FAE1 cDNA sequence AF274750 (GenBank accession) and B. napus FAE1 promoter sequence AF275254. BAC DNA was isolated using a modified mini-alkaline lysis extraction procedure (Bendahmane, 1999). PCR walking on three representational BAC clones (BAC1, BAC7 and BAC11) followed standard procedures (Gibbons et al., 1991). PCR products were sequenced and six primer pairs (Fae1p1-1, Fae1p1-2 Fae1p1-3, Fae1E1, Fae1p2-1 and Fae1E2) (Table 3, Fig. 1) were designed based on these sequences. Gene-specific primer pairs, S-FAE1-A8 and S-FAE1-C3, were designed to provide specific amplification products that distinguished between the two groups of BAC clones.

Table 3.  Primer sequences used for PCR amplification
Primer nameForward primer sequence (5′→3′)Reverse primer sequence (5′→3′)
Figure 1.

Regions of the FAE1 BAC clones covered by primer pairs on the two groups (the names of the primer pairs are marked above their corresponding amplicons). Ten BAC clones in group 1 correspond to BnC3.FAE1 and three BAC clones in group 2 correspond to BnA8.FAE1. Lines represent 5′ and 3′ genomic sequence of FAE1 (c. 1 kb for 5′ and 3′ of BnA8.FAE1, c. 3.5 kb for 5′ of BnC3.FAE1 and c. 1 kb for 3′ sequence); rectangles represent coding region (1521 bp). Lines with triangles at the end indicate specific primers of BnA8.FAE1; lines with circles at the end indicate that primer pair can amplify both BnA8.FAE1 and BnC3.FAE1; lines with diamonds at the end indicate specific primers of BnC3.FAE1. Fae1primer1 and Faeprimer5 were used in the PCR assay in Fig. 4.

Based on the FAE1 sequence obtained from the B. napus cv. Ningyou7 and Tapidor DH parents of the TNDH mapping population, two polymorphic markers, FAE1-A8 and FAE1-C3 (using primer pairs M-Fae1-A8 and M-Fae1-C3, Table 3, Fig. 1), were developed. This enabled us to identify and map loci in the context of the TNDH linkage map constructed by Long et al. (2007) using JoinMap 3.0 software (http://www.kyazma.nl/index.php/mc.JoinMap). Seed erucic acid content data were obtained from field trials of the TNDH population conducted over four environments (Qiu et al., 2006). Seed erucic acid content QTLs were detected by the composite interval method (CIM) using Windows QTL cartographer 2.5 software (http://statgen.ncsu.edu/qtlcart/WQTLCart.htm).

FAE1 mutation screening

The primer pair FAE426 was used to screen the EMS mutant populations. The forward primer was labeled at the 5′ end with 700 nm IRD fluorescence and the reverse primer was labeled with 800 nm IRD fluorescence (MWG Biotech, Inc., Ebersberg, Germany). DNA samples isolated from M2 plants of the 0.6 and 0.3% EMS populations were screened individually, with the screening protocol following that used in rice and Arabidopsis TILLING projects (Till et al., 2004b; Till et al., 2006). CEL 1 was extracted from celery and stored in −80°C (Till et al., 2004a). FAE426 with unlabeled primer was used to amplify products from genomic DNA of M2 plants which had yielded novel cleaved bands at 700 and 800 nm in the TILLING gel. The PCR products were electrophoresed on agarose gel so as to recover the 426 bp bands and then these 426 bp PCR products were sequenced.

Mutation sequence analysis

Sequence derived from mutant and wild-type (WT) plants was analyzed to determine the likelihood of changes affecting protein function. The PARSESNP (Project Aligned Related Sequences and Evaluate SNPs; http://www.proweb.org/parsesnp/) and SIFT (Sorting Intolerant from Tolerant; http://blocks.fhcrc.org/sift/SIFT) software packages were used to predict the severity of each mutation (Ng & Henikoff, 2003; Taylor & Greene, 2003). SIFT values < 0.05 and PSSM scores > 10 indicated that the nucleotide change was likely to affect gene function.

The 426 bp sequences obtained following TILLING screening were used in BLAST comparison against the comprehensive nucleotide database of available Brassica sequences served at BrassicaDB (http://brassica.bbsrc.ac.uk/BrassicaDB/). All related B. napus DNA sequence accessions that were detected with an E-value < 10−15were batch-downloaded from the NCBI GenBank nucleotide database. The sequences were aligned using a local implementation of the NCBI standalone Blast software (packageblast-2.2.18-ia32-win32.exe, http://www.ncbi.nlm.nih.gov/BLAST/download.shtml). The 426 bp sequence was used as query and those downloaded sequences that appeared to contain a sequence error (a different base present in just one downloaded sequence) were deleted. Default parameters were used in Blast, except for the output format (its format was ‘m 4’ so that SNPs can be easily visualized). SNPs were assigned where different bases at a given nucleotide position appeared in two or more sequences.

Phenotypic and genotypic analysis

The erucic acid content of seeds harvested from WT Ningyou7 and M2 plants which contained FAE1 mutations was measured by gas chromatography (GC) using standard methods (Kaushik & Agnihotri, 2000) with multiple seeds tested from each plant. Wild-type Ningyou7 plants were grown in the same field and seeds were harvested at the same time as the mutant population. Single seed of the M3 generation of the two mutants L080-1 and N004-1 was used to establish the erucic acid content, using a half-seed GC analysis method (Conte et al., 1989).

To detect the location of mutations in fae1 paralogues in the A8 or C3 chromosomes, M2 DNA from mutants L080-1 and N004-1 was used to amplify products using the gene locus-specific primer pairs S-FAE1-A8 and S-FAE1-C3. PCR reactions were carried out with proof-reading pfu taq polymerase (Tiangen Biotech, Inc., Beijing, China), and the purified PCR products sequenced.

Genetic analysis of the two mutants L080-1 and N004-1 involved ANOVA and multi-comparision analysis of phenotypic and genotypic data, using SAS 8.2 software.


Mutant population development

We first established an optimal concentration of EMS for treatment of B. napus seeds required to generate a mutant population. Three hundred self-pollinated seeds from a Ningyou7 DH line were treated with EMS at each of seven different concentrations, from 0.0 to 1.2% (w/v), at 0.2% intervals. The M1 germination rate, germination time and self-pollinated percentage seed set were recorded (Table 1). At EMS concentrations of 0.8% or above, the germination rate was < 85%, and germination time was prolonged. Almost all the surviving M1 plants were unable to set seed through self-pollination, with M2 seed set < 10%. At 0.6% EMS, 25% of M1 plants had sufficient self-pollinated seeds, with a germination rate of c. 85%. An EMS concentration of 0.6% was therefore selected as an optimal treatment for Ningyou7 seeds in developing the mutant population. To determine whether a lower EMS concentration, generating a higher rate of M1 self-pollinating seed set, would be suitable for TILLING, a 0.3% EMS concentration was also selected to develop a secondary mutant population. Between one and four M2 plants were generated by random selection from each M1 plant. In total, 3926 lines were developed from the 0.6% EMS-treated population with 7110 M2 plants, and a further 933 lines from the 0.3% EMS-treated population with 2604 M2 plants (Table 4). DNA was extracted from all M2 plants and diluted to the same concentration suitable for PCR reactions. Self-pollinated seeds were also harvested from the individual M2 plants and stored in a seed bank.

Table 4.  Summary characteristics of the two Brassica napus cv. Ningyou7 ethylmethanesulfonate (EMS) populations
 0.3% EMS0.6% EMS
M1 generation
Number of plants~ 1080~ 13 000
M2 generation
Number of lines  933   3926
Number of plants 2604   7110
Mutations detected   4   15
from x M2 plants  614   730
Mutation load within plant 9174 28 941
Mutation density per plant  130.8 kb 41.5 kb

Forward genetic screening of mutants in the M2 population

To evaluate and compare the effect of mutagenisis in the two populations through forward genetics, all M2 plants were scrutinized for visible variant phenotypes. These included key developmental characters throughout the growing season, such as plant architecture, leaf, floral and reproductive morpholology (summarized in Table 2). As expected, the population treated with 0.6% EMS yielded a higher percentage of mutant phenotypes, with examples of visible phenotypic variants shown in Fig. 2. In order to assess variation in total seed oil content, 652 M3 seeds from the 0.6% EMS population and 516 from the 0.3% EMS population were compared with untreated seed (WT). ANOVA of these three groups indicated no significant difference in oil content between the two EMS populations, although a significant difference (P < 0.01) was observed between the 0.6% EMS population and WT (n = 34). A significant difference (P < 0.01) was also detected between the 0.3% EMS populations and WT. The frequency distribution of oil content amongst a total of 1168 M3 seeds (Fig. 3) shows a maximum value of 47.8%, considerably higher than WT (38.36 ± 1.10%).

Figure 2.

Examples of phenotypic mutants observed in the ethylmethanesulfonate (EMS)-treated M2 populations of Brassica napus. (a) Mutant plant with stocky (reduced stature) architecture; (b) curly leaf phenotype; (c–f) floral mutants: anatropous inflorescence (c); white flowers (d); absent and distorted stamens (e); aborted buds (f). Both (e) and (f) were also recorded as infertile mutants.

Figure 3.

Distribution of Brassica napus seed oil content within 1168 M3 plants (652 samples were from the 0.6% ethylmethanesulfonate (EMS)-treated population, and 516 were from the 0.3% EMS population). ANOVA did not detect any significant difference in the oil content between the two populations. However, the transgressive distribution appeared greater in the 0.6% EMS population, which contained both the maximum (30.62%) and minimum (47.93%) oil contents. The oil content of wild-type (WT) Ningyou7 was 38.36 ± 1.10% (n = 34). The maximum oil content of WT was 40.89% and the minimum was 36.36%. The values recorded were from M3 seed samples scored on an individual plant basis, and thus may represent sibling aggregates.

FAE1 homologous in B. napus cv. Tapidor and Ningyou7

To estimate the copy number of paralogous FAE1 genes in the B. napus AC genome, and to identify sequence differences amongst these copies, a 426 bp probe generated from Tapidor genomic DNA was used to screen the JBnB BAC library, which was developed from the doubled haploid B. napus cv. Tapidor (Rana et al., 2004). 13 BAC clones (JBnB060D12, JBnB015C15, JBnB060D12, JBnB154K14, JBnB170L17, JBnB160P09, JBnB079L09, JBnB009C03, JBnB065M23, JBnB149N05, JBnB032N18, JBnB178F05, and JBnB108J15), identified here as BAC1 to BAC13, were identified with positive hybridization signals. Five primer pairs were designed to amplify each of these BAC clones (Fae1primer1 to Fae1primer5, see Table 3, Fig. 1). Fae1primer5 allowed classification of the BAC clones into two groups, with BAC4, BAC7 and BAC10 in group 2 generating single amplified bands of the same size when visualized on agarose gels, and no products obtained from the remaining BACs of group 1 (Fig. 4b). PCR with Fae1primer1 resulted in products of predicted length from all BAC clones apart from BAC11 (Fig. 4a). PCR-walking was then used to obtain the 5′ and 3′ sequences of the FAE1 gene in each of three clones (BAC1, BAC4, BAC11) representing the different classes. The end of the BAC11 insert was located near the FAE1 promoter region, which explained why the Fae1primer1 was unable to amplify a product from this BAC, and suggested that this clone may in fact belong to the same locus-class as BAC1. We then obtained sequences flanking the FAE1 gene from BAC1, including c. 3.5 kb from 5′ and 1 kb from 3′ regions, together with c. 1 kb each of 5′ and 3′ sequence from BAC4. Using these sequences, six primer pairs were designed (Fae1p1-1, Fae1p1-2, Fae1p1-3, Fae1p2-1 amplifying the 5′ end region, Fae1E1, and Fae1E2 amplifying the 3′ end region) (Table 3, Fig. 1). In combination, these primers were able to be used in PCR reactions to generate products that enabled classification of the 13 BAC clones into two groups, corresponding to two distinct loci. The results were consistent with those obtained with Fae1primer5 (data not shown).

Figure 4.

PCR assay to analyze Brassica napus cv. Tapidor BAC clones. Thirteen BAC clones were amplified with Fae1primer1 (a) and Fae1primer5 (b). Lane numbers are indicated on the figures, with M representing the DNA ladder DL2000 (Promega. Inc). (a) Fae1primer1 was not able to amplify a product from BAC11. (b) Fae1primer5 was able to classify the 13 BAC clones into two groups. Group 2 included BAC4, BAC7 and BAC10, with the remainder in group 1. Subsequent investigation revealed BAC11 to be in group 1.

To further characterize the paralogues within the B. napus AC genomes, the FAE1 gene and its flanking sequence were obtained from each of the 13 BAC clones using the six specific primer pairs. The 10 BAC clones in group 1 were identical throughout c. 6kb of sequence (3.5 kb was 5′ of the FAE1 start codon, 1 kb was 3′ of the FAE1 stop codon, with a coding region of 1521 bp) (Fig. 1). The three BAC clones in group 2 also showed no sequence difference within the c. 3.5 kb region (with 1 kb 5′ end of the FAE1 start codon, and 1 kb 3′ end of the FAE1 stop codon) (Fig. 1).

Two markers, FAE1-A8 and FAE1-C3, corresponding to each group of BAC clones, were found to be polymorphic between the parents of the TNDH mapping population. These enabled us to locate the existing FAE1 BAC clones in the context of the TNDH genetic linkage map, on chromosomes A8 and C3. Both markers co-located with the peaks of the QTLs for seed erucic acid content (Fig. 5), indicating that these corresponded to the two paralogous copies of FAE1 in the B. napus genome. The two genes were designated as BnaA.FAE1.a and BnaC.FAE1.a, according to the standard nomenclature of Ostergaard & King (2008). However, to simplify subsequent reference within this paper, BnA8.FAE1 and BnC3.FAE1 are used to represent paralogues in the A8 and C3 chromosomes of B. napus, respectively. Two gene locus-specific primer pairs, S-FAE1-A8 and S-FAE1-C3, corresponding to BnA8.FAE1and BnC3.FAE1, respectively, were designed from the sequence flanking FAE1. These enabled us to obtain complete genomic sequences for both paralogues from the WT Ningyou7 DH parent of the EMS mutant populations that included 5′ and 3′ flanking regions. Alignment of these sequences (GenBank accessions: EU543282 and EU543283) with cDNA sequences present in the GenBank nucleotide database confirmed the lack of introns within this gene, as reported elsewhere (Wu et al., 2007b). The coding region of both paralogues was 1521 bp in Ningyou7, with a sequence similarity of 98.6%. Twenty-one SNPs were identified between the two paralogues. There was also very little (c. 6%) variation in the c. 400 bp upstream region of the start codon.

Figure 5.

Mapping of the two FAE1 paralogues in the context of the reference TNDH linkage groups, and detection of QTLs for seed erucic acid content. The two linkage groups were constructed based on data presented in Qiu et al. (2006) and Long et al. (2007). BnA8.FAE1 was located on Brassica napus chromosome A8, and BnC3.FAE1 was located on chromosome C3. Two hundred and two TNDH lines were grown in four environments, two in the south of China (Wuhan, at 30°35′N 108°54′E) and two in the north (Dali, at 34°52′N 109°56′E). Linkage map construction and QTL detection followed the methods outlined by Qiu et al. (2006), with the length of the vertical lines representing 95% confidence intervals of these QTLs. All QTLs shown were significant at P < 0.01.

Screening for FAE1 mutations in the EMS mutant populations

To screen for mutations in EMS-treated populations using TILLING, it is essential to utilize gene-specific primer pairs (Colbert et al., 2001). The maximum length of fragment that can be screened using the Li-Cor 4300 polyacryamide gel system is just over 1 kb. Although other authors have reported detection of mutants by TILLING in PCR products in the range 300 bp–1.6 kb (Slade et al., 2005), our experience was that we observed very diffuse bands with PCR fragments of > 1 kb on the Li-Cor 4300 gels. Moreover, although we were able to use gene-specific primer pairs for each of the paralogous copies of FAE1, the forward and reverse primers matched positions remote from the start and stop codons. As a result, it was not possible to obtain gene-specific primers that corresponded to the coding region suitable for TILLING of mutations within each of the two highly similar paralogues of FAE1. We therefore designed the primer pair FAE426, so that it simultaneously amplified 426 bp from both paralogues, for which five SNPs already existed to distinguish the WT alleles, and used this in TILLING for novel mutations in our EMS populations.

Co-amplification of two similar genes in a pooled sample can lead to an increased error (Cooper et al., 2008). We therefore carried out direct screening for novel mutations from DNA isolated from individual M2 plants, using the standard TILLING protocol (Till et al., 2006). TILLING gels were scanned using a Li-Cor 4300. Putative mutant lines and location of nucleotide changes were validated by sequencing of the PCR product generated from unlabeled FAE426 primers. Double peaks within the sequence trace not associated with the five known paralogous SNPs positions indicated an induced nucleotide change, which also corresponded with the cleaved band position in the Li-Cor TILLING gel. Fig. 6 shows one example of a mutation detected within M2 plant L209-1 (M2 plant field accession NO). From the Li-Cor gel system, 700 and 800 nm images are obtained (Fig. 6a,b). The cleaved bands corresponding to the induced mutation were located at c. 230 bp in the 700 nm image, and at c. 190 bp in the 800 nm image, thus totaling 426 bp. The sequence traces indicate the mutated region of L209-1 (Fig. 6c) and the corresponding region of nontreated WT Ningyou7 (Fig. 6d). In total, 19 mutations were discovered amongst 1344 M2 plants screened (one plant from each M2 line) (Table 5).

Figure 6.

Mutation within the FAE1 gene fragment from L209-1 identified by TILLING. 700 nm and 800 nm Li-Cor 4300 scanning gel photos are shown in (a) and (b). Two novel bands were found in the same lane, respectively, in both the 700 and 800 nm images, and these are magnified in (e) and (f). (c, d) Partial sequence trace corresponding region in L209-1 (c) and in wild type (d). The primer pair FAE426 was designed to amplify the two FAE1 paralogues simultaneously – this is demonstrated at position 480 where double peaks are observed in the sequence trace (which indicates an induced nucleotide change G to A in one of the two paralogous copies at this position), whilst a single peak is observed in the wild type at the same position (indicates both of the two paralogous copies FAE1 in this position are G).

Table 5. FAE1 mutations identified by TILLING
M2 plantNucleotide changeAmino acid changePSSM differencecSIFT scoredSeed erucic acid content (%)
  • a

    The code for nucleotide change refers to mutations at specific positions in the genomic sequence of the FAE1 gene, referenced from the from the first base (A) of the start codon. Thus, C266T indicates that the 266th nucleotide had a transition from C to T.

  • b

    The consequence of the nucleotide mutation in the FAE1 protein chain is referred to by a code whereby the initial and resulting amino acid is shown, referenced to the amino acid position of the start codon. Thus, P89L indicates that the 89th amino acid is predicted to be modified from P to L. =, indicates no change (synonymous mutation).

  • c,d

    SIFT scores < 0.05 and PSSM value > 10 are predicted to be deleterious to gene function.

  • e

    Plants indicated arose from the 0.3% EMS treated population, with the remainder from the 0.6% ethylmethanesulfonate (EMS)-treated population. Bold type indicates where SIFT scores or PSSM value were greater than the threshold for likelihood of causing a deleterious effect on protein function.

L324-1C266TaP89Lb  47.77
L457-1C347TT116M  50.27
N050-1G384AK128=  48.77
K026-1eC403TL135=  49.34
N128-1C420TH140=  52.53
L518-1C446TP149L  48.77
N142-1C449TP150L  49.34
N184-1C450TP150=  45.58
K024-1eG453AR151=  48.59
L421-1G478AE160K  50.08
L209-1eG480AE160=  51.26
L496-1G508AE170K  50.78
N377-1C627TL209=  48.56
L286-1G629AR210Q19.90.01no seed

Relationship between EMS concentration and mutation frequency

Of the 19 mutations detected, 15 arose from the 0.6% EMS mutangenized population, following screening of 730 M2 plants, and four from the 0.3% population following screening of 614 M2 plants (Table 5). Given that FAE426 was designed to amplify two FAE1 paralogous copies simultaneously, this corresponds to a single screen of 852 bp within the B. napus genome. Within the 0.6% EMS population, the mutation frequency was calculated at one mutation per 41.5 kb in each plant ((0.426 kb × 2 × 730 plants)/15 mutations = one mutation every 41.5 kb) and 28 941 mutations per plant (1 200 000 kb per plant/one mutation every 41.5 kb = 28 941 mutations per plant), and within the 0.3% EMS population at one mutation per 130.8 kb ((0.426 kb × 2 × 614 plants)/four mutations = one mutation every 130.8 kb) and 9174 mutations per plant (1 200 000 kb per plant/one mutation every 130.8 kb = 9174 mutations per plant) (Table 4).

Of the 19 FAE1 mutations, 18 represented G/C to A/T transitions and one represented an A to C transversion (Table 5), which is a close fit to previous reports that EMS mutagenesis primarily induces G/C to A/T transitions.

A comparative analysis of natural FAE1 SNPs in rapeseed cultivars and in the mutant population

Ecotilling is a new approach for detecting SNPs in natural populations (Comai et al., 2004). To determine whether the 19 induced mutations were present within existing lines representing selected varieties, breeding or experimental material, an in silico Ecotilling analysis was carried out. A total of 112 B. napus FAE1 sequences that corresponded to the 426 bp region screened in our TILLING experiments were aligned. This included 101 ESTs (note that this gene is intronless) and 11 genomic sequences. Once sequence errors were deleted (i.e. a different base present in just one downloaded sequence), nucleotide polymorphisms were detected at 18 positions, each of which was distinct from the 19 induced point mutations (Table 6).

Table 6. In silico Ecotilling with Brassica napus FAE1
  • a

    Position from the FAE1 start codon ATG, to the 264th nucleotide.

  • b

    Indicates that two alternative nucleotides were detected at this position, with G/A denoting that G was prevalent in some accessions, and A in others.

  • c

    Indicates the presence of an indel – with A/D denoting that A was present in some accession, whereas a deletion was present at the corresponding position in others.

  • d

    Indicates the presence of either two alternative nucleotides (G/T) or a deletion at this position. Comparison of the 18 SNPs to the 19 ethylmethanesulfonate (EMS)-induced mutations (Table 5) indicates that there was no overlap in the distribution of nucleotide changes.


Functional characterization of fae1 mutants with reduced seed erucic acid content

In total, seven (37%) synonymous mutations and 12 (63%) mis-sense mutations were identified from TILLING screening of 1344 M2 plants. Both SIFT score and PSSM values were used to predict whether the base change for individual mutations would be deleterious to FAE1 gene function (Ng & Henikoff, 2003; Taylor & Greene, 2003). Four of the 19 mutations were predicted to affect the FAE1 protein function since both the SIFT score and PSSM values for these events were at least at the threshold of severity (Table 5). Seed erucic acid content was determined by GC for M3 seeds of 18 available plants (multiple seeds tested from each plant). The M3 seed erucic acid content of plants N057-1, N004-1, L297-1 and L080-1 was lower than that of the wild-type. This corresponded well with the predicted probabilities of altered gene function derived from the four mutations (Table 5). Two individual M2 plants, L080-1 and N004-1, were selected to identify the mutated locus within the genome using gene-specific primer pairs S-FAE1-A8 and S-FAE1-C3 (Table 3), which amplified BnA8.FAE1 and BnC3.FAE1, respectively. The M2 mutant plant L080-1 was recessive homozygote bna8.fae1at position 599bp (assigned mutant a8fae1-599), while the M2 plant N004-1 was heterozygous BnC3.FAE1/ bnc3.fae1 at 400 bp (assigned mutant c3fae1-400) (Fig. 7). Alignment of these sequences to the corresponding paralogous copies in WT Ningyou7 showed that BnC3.FAE1 in L080-1 corresponded to WT BnC3.FAE1, and that BnA8.FAE1 in N004-1 corresponded to WT BnA8.FAE1. We therefore deduced that in each mutant line the paralogous copy contained no other induced mutations.

Figure 7.

Location of FAE1 mutations in lines L080-1 and N004-1. DNA from single M2 plants (L080-01 and N004-1) was amplified with the primer pair FAE426. (a, d) The amplified PCR products were sequenced and partial sequence traces are shown with double peaks at positions of nucleotide transition. Two gene-specific primer pairs, S-FAE1-A8 and S-FAE1-C3, amplified DNA from M2 plant of the two mutants L080-1 and N004-1, and the PCR products were sequenced. (b) The partial FAE1 sequence trace of L080-1 from chromosome A8. (c) Corresponding sequence from the C3 gene copy. (e) Partial FAE1 sequence trace from N004-1 in A8 chromosome. (f) Sequence from N004-1 corresponding to the chromosome C3 paralogue. From these four trace files, it was possible to deduce that a C to T transition mutation had occurred in the L080-1 FAE1 genomic sequence corresponding to the gene copy located on chromosome A8. Similarly, it was easy to deduce that a G to A transition mutation within the N004-1 FAE1 genomic sequence corresponded to the paralogue located on chromosome C3. Moreover, it was possible to predict that the FAE1 mutation within the L080-1 M2 plant was homozygous, whilst the FAE1 mutation within the N004-1 M2 plant was heterozygous.

Thirty-one M3 seeds were harvested from the M2 mutant plant N004-1, and the seed erucic acid content was determined using half of each seed. This involved dissection and analysis of one cotyledon. The remaining half seed included the viable shoot apical meristem and radicle, which was germinated and grown to yield M3 plants, allowing subsequent genetic analysis. The genotypic segregation in 31 M3 plants displayed Mendelian inheritance with 9 C3FAE1-400/C3FAE1-400 : 16 C3FAE1-400/c3fae1-400 : 6 c3fae1-400/c3fae1-400, corresponding to a 1 : 2 : 1 ratio supported by a chi-squared test (χ2 = 0.6130). The erucic acid content in the 31 seeds was significantly (P < 0.01) associated with their genotypes tested by LSD (least significant difference) all-pairwise comparison test multiple comparison (Table 7). We conclude that the point mutation in c3fae1-400 appears to be causal in reducing the seed erucic acid of N004-1. Moreover, no significant difference was observed between the seed erucic acid content (19.21 ± 0.53%) among 29 M3 seeds of L080-1 with ANOVA analysis, consistent with a recessive homozygote genotype at bna8.fae1.

Table 7.  Genetic analysis of the Brassica napus M3 generation from the heterozygous mutant c3fae1-400
 Plant numberSeed erucic acid content (%)LSDa all-pairwise comparisons test
  • a

    Least significant difference multiple comparisons.

  • b

    Significance at P < 0.01.

C3FAE1-400/C3FAE1-400 952.01 ± 1.14Ab
C3FAE1-400/c3fae1-4001645.71 ± 0.85B
c3fae1-400/c3fae1-400 631.83 ± 1.39C


We have successfully constructed two rapeseed mutant populations suitable for both forward and reverse genetic detection methods. In the forward genetic screening of mutants, a large number of novel mutated phenotypes were identified, and many of these represent a valuable genetic resource affecting crop traits such as male sterility, reduced stature and plants with increased oil content. In the reverse genetic TILLING of mutants, we screened 1344 M2 plants representing 1344 distinct lines (i.e. siblings were not screened), and detected a total of 19 FAE1 mutations, three of which had a change of function. In addition, in silico Ecotilling analysis demonstrated a clear distinction between sites of natural and induced SNPs. This is very encouraging for the application of TILLING within rapeseed crop genetic improvement as a highly efficient method for generating new genetic resources for breeding. We are also able to calculate that if a 1kb gene fragment were to be screened within the 0.6% EMS population, with a single plant selected from each line, then c. 100 mutations would be detected following a complete screening of the 0.6% EMS population (a total of 3926 lines). This suggests that the population carries a high mutation load, and is a valuable resource for breeding and research. However, for practical application of induced mutation within crop improvement, it is necessary to adopt a process that reduces the background caused by the mutations at unlinked loci. Such a schema should involve backcrossing to the WT or target elite cultivar, using molecular marker and phenotypic tracking for recurrent selection of the mutant locus. It is also important to eliminate epistatic effects and carry out controlled self-pollination of the candidate mutant plant before segregation analysis of the mutation identified and modified phenotype. Such information can be substantiated by analysis of transcript profile and in vitro protein expression for the target gene (Heckmann et al., 2006). We conclude that the populations generated are efficient for exploring novel sources of variation using both forward and reverse genetics approaches to screening mutants.

Although we were able to demonstrate development of an efficient new research platform through discovery of three novel induced functional low erucic acid mutant alleles, it is important to determine the optimal mutagen dosage for B. napus. The EMS concentration is a key factor in developing such populations, with higher concentrations leading to higher mutation frequencies. However, any higher mutation load achieved is offset by the effect of EMS on reducing seed set, viability and germination (Greene et al., 2003; Kim et al., 2006). For B. napus there have been no prior reports of an optimal mutagen dosage to guide development of a forward genetics or reverse genetics TILLING population. Here, we found that at EMS concentration of 0.6% yielded a calculated mutation density of one per 41.5 kb in each plant, compared with almost a third of that in the population treated at 0.3% EMS. These estimates may not accurately reflect the true mutation density of the two EMS populations, as it was not possible to screen the first 50 bp in a TILLING gel. However, the mutation frequencies within our two B. napus mutants were within the range of those reported elsewhere for monocot and dicot species, with one mutation per 24 kb reported in hexaploid wheat, one per 40 kb in tetraploid wheat (Slade et al., 2005), 300 kb in rice (Till et al., 2007), 100 kb in barley (Hordeum vulgare L.) (Caldwell et al., 2004), and 170 kb in Arabidopsis (Greene et al., 2003). In the forward genetic investigation, 23.25% M2 plants displayed observable mutant phenotypes in the 0.6% EMS population, compared with 12.98% in the 0.3% EMS population. These estimates may be low and not reflect the true mutant load, as some plants may carry more than one mutant phenotype. Moreover, when M2 plants were grown in field conditions, we found that the performance within the 0.3% EMS population was stronger than the 0.6% population. At concentrations of EMS above 0.6% we found that M1-treated Ningyou7 plants set almost no seed. Therefore, taking into account the mutation frequency and the performance of mutagenized plants in the field, we consider an EMS concentration between 0.3 and 0.6% to be optimal for construction of mutant populations in B. napus and probably also in other amphidploids of the Brassica triangle of U, for example B. juncea (2n = 36; AB genomes) and B. carinata (2n = 34; BC genomes).

Since the original report of TILLING in 2000, the technique has successfully been applied to many crops, notably in rice and wheat (Slade et al., 2005; Till et al., 2007). However, we are not aware of other reports to date of its application within any Brassica species. The main challenges arise from the complex genome of B. napus in which most of genes are multi-paralogous. This reduces the ease with which gene locus-specific primers can be designed, and also limits the ability to locate the mutations unequivocally to specific paralogous copies. Here we have successfully demonstrated that TILLING technology can also be applied in polyploid crops such as B. napus that comprise multiple closely related complex segmentally duplicated genomes. We have been able to allocate target genes within the B. napus genome into two classes. The first class comprises those genes which occur as multi-paralogous copies distinguished by clear sequence differences; and the second class includes multi-paralogous genes that possess only minor sequence differences between different copies. It is a relatively trivial exercise to design gene-specific primers for the first class. However, for the second class we have developed a novel strategy that is based on using a primer pair that is designed to amplify all paralogous copies of the gene simultaneously, albeit from individual rather than pooled DNA samples. The procedure is based on single plant TILLING, to identify mutations in situations where two or more very similar paralogous gene copies exist in a genome. We made use of existing SNPs as a positive control, and demonstrated the ability to distinguish novel mutations, based on use of primer pairs designed to amplify both FAE1 paralogues simultaneously. Slade et al. (2005) reported that using a primer pair that could amplify at least two gene copies would be less efficient. However, in our experiments we found that these types of primers provide a powerful approach to mutant detection. The precondition for successful screening of mutations using this approach is that the target fragment should be highly conserved amongst the different paralogues. In the case of FAE1, we found just five SNPs between the two paralogues in the 426 bp target region. These five SNPs were also detected as background signals in the Li-Cor gel images, but did not detract from our ability to detect novel induced mutations (Fig. 5). Within the TILLING gel, the cleaved bands that were present in all lanes with positions differing between 700 and 800 nm are predicted to arise from the SNPs that exist between the two FAE1 paralogues. However, the cleaved bands that were detected in only one lane within the 700 nm image, together with the corresponding cleaved band in the 800 nm image, indicated discovery of a novel mutation. Therefore, we can conclude that with a modified TILLING primer design we are able to screen for mutations in any gene in the EMS population of B. napus, even though this consists of very complex duplicated genomes. This modified method may improve and widen the application of TILLING technology for crop genomic research and improvement.

For rapeseed, the ability to achieve zero or low seed erucic acid (LEA) content has long been a major breeding objective, since erucic acid has been regarded as an antinutritional component in seed oil (Vles et al., 1978). In the 1960s, the first variant with LEA content was found in the animal feed rape cultivar ‘Liho’, with the first LEA rapeseed ‘Oro’ being derived following introduction of the LEA variation from ‘Liho’ (Downey & Craig, 1964). Subsequently, almost all LEA rapeseed cultivars have carried the LEA gene source from ‘Liho’ or ‘Oro’, and so this single genetic resource has probably contributed considerably to an inbreeding effect and associated genetic erosion through linkage drag in rapeseed breeding programmes (Sharpe & Lydiate, 2003). Although breeders have successfully developed many LEA rapeseed cultivars, new sources of LEA are still required to modulate the relevant pathways within the seed oil fatty acid synthesis and modification pathways (Barker et al., 2007). Wu et al. (2008) reported discovery of a four base deletion in the FAE1 gene that resulted in LEA content within six Chinese rapeseed cultivars. However, this new source of LEA may still contain ‘Liho’ or ‘Oro’ genetic background, as so many LEA rapeseed cultivars have derived from material including the parent ‘Liho’ or ‘Oro’.

In the work reported here, the WT genotype of our EMS mutant population was Ningyou7, which is a high erucic acid (HEA) content cultivar in China, distinct from many other modern rapeseed cultivars (Meng et al., 1996). This cultivar has little or no allelic genetic background in common with ‘Liho’ or ‘Oro’, and so the three novel LEA content rapeseed alleles induced by TILLING in this research represent a unique genetic resource. The gene locus-specific primer analysis confirmed that the mutant a8fae1-599 and c3fae1-400 were separately associated with bna8.fae1and bnC3.fae1, respectively. Seed erucic acid content in rapeseed has been shown to be primarily controlled by these two genes (Barret et al., 1998; Qiu et al., 2006; Wu, Y et al., 2007; Wu et al., 2008). Thus a cross between these two mutants would be expected to generate a new LEA cultivar within what is otherwise a HEA background. In conclusion, the three new LEA mutants, L080-1, N004-1 and L297-1, that were discovered by TILLING could be used as a new genetic resource for LEA rapeseed breeding without collateral inbreeding effect and genetic erosion.

In conclusion, we have reported the application of TILLING to the generation and identification of a novel LEA genetic resource for rapeseed improvement. Although TILLING was reported initially as a reverse genetic approach for Arabidopsis genomic research (McCallum et al., 2000), it will increasingly be very powerful to combine these two approaches to crop functional genomics and genetic improvement. Alongside this work focused on TILLING, QTL for rapeseed seed oil and erucic acid content have been detected in the TNDH genetic mapping population, followed by development of near-isogenic lines (NILs) to resolve these two regions (F. Tian & J. Meng, unpublished). Fine mapping of the two QTLs has resulted in FAE1 and several other genes being proposed as major candidates contributing to seed oil content (unpublished data). To validate gene function for these candidates, TILLING will now be the method of choice in B. napus, rather than conventional transgene knock-down approaches. Knockout of candidate gene function via TILLING overcomes complications associated with RNAi or anti-sense in a multi-paralogous system, and has the advantage of greatly reducing the time required to carry out such research. Taken together, we propose that the combination of QTL resolution making use of comparative genomic information and TILLING of candidates provides a powerful new approach to functional crop genomics, with the advantages of both forward and reverse genetics, and that this will accelerate genetic crop improvement.


The authors thank to Dr Andy Phillips, Dr Hai-Chun Jing and Mr Carlos Bayon (Rothamsted Research, Harpenden, Herts AL5 2JQ, UK) for their support in TILLING technique training to NW. The authors thank Dr Ian Bancroft (Department of Crop Genetics John Innes Centre, Norwich Research Park Colney Lane, Norwich NR4 7UH, UK) for his constructive suggestion to this work. The authors thank Mr Dianrong Li (Hybrid Rapeseed Centre of Shaanxi, Dali 715105, China) for his contribution to the field work and NW thanks graduate students within his laboratory for their contribution to the work of mutant population DNA extraction. Financial support for this work was provided by the National Basic Research and Development Program (2006CB101600), the Key Project of Chinese Ministry of Education (103175), and the National Nature Science Foundation of China (30600373).