High Resolution Melt (HRM) analysis is an efficient tool to genotype EMS mutants in complex crop genomes

Background Targeted Induced Loci Lesions IN Genomes (TILLING) is increasingly being used to generate and identify mutations in target genes of crop genomes. TILLING populations of several thousand lines have been generated in a number of crop species including Brassica rapa. Genetic analysis of mutants identified by TILLING requires an efficient, high-throughput and cost effective genotyping method to track the mutations through numerous generations. High resolution melt (HRM) analysis has been used in a number of systems to identify single nucleotide polymorphisms (SNPs) and insertion/deletions (IN/DELs) enabling the genotyping of different types of samples. HRM is ideally suited to high-throughput genotyping of multiple TILLING mutants in complex crop genomes. To date it has been used to identify mutants and genotype single mutations. The aim of this study was to determine if HRM can facilitate downstream analysis of multiple mutant lines identified by TILLING in order to characterise allelic series of EMS induced mutations in target genes across a number of generations in complex crop genomes. Results We demonstrate that HRM can be used to genotype allelic series of mutations in two genes, BraA.CAX1a and BraA.MET1.a in Brassica rapa. We analysed 12 mutations in BraA.CAX1.a and five in BraA.MET1.a over two generations including a back-cross to the wild-type. Using a commercially available HRM kit and the Lightscanner™ system we were able to detect mutations in heterozygous and homozygous states for both genes. Conclusions Using HRM genotyping on TILLING derived mutants, it is possible to generate an allelic series of mutations within multiple target genes rapidly. Lines suitable for phenotypic analysis can be isolated approximately 8-9 months (3 generations) from receiving M3 seed of Brassica rapa from the RevGenUK TILLING service.


Background
The identification and analysis of gene mutations in plants is fundamental to the investigation of gene function. One approach, Targeted Induced Loci Lesions IN Genomes (TILLING) was originally developed in Arabidopsis [1] and has subsequently been successful in a range of crop plants [2,3]. In this reverse genetic approach, an ethyl methane sulfonate (EMS) mutagenised population is screened for SNPs within target genes [4]. EMS mutagenesis generates multiple alleles within each gene, including nonsense, missense, splicing and cis-regulatory mutants, in comparison to T-DNA and transposon mutagenesis that generate only knockout mutants [5,6]. Analysis of an allelic series can provide information on important domains or amino acids within the protein of interest. A number of TILLING populations have been developed for a variety of crops of different genome size, including rice [2], wheat [7,8], Brassica rapa [3], Brassica napus [9], Lotus japonicus [10], Medicago trunculata [11], Arachis hypogaea [12], and Solanum lycopersicum [13]. These populations have subsequently been used to isolate mutations in a variety of genes, including those involved in starch metabolism in Lotus japonicas [14], peanut allergens [12] and a gene encoding a fatty acid elongase [9].
High resolution melt (HRM) analysis is a technique that measures the disassociation of double-stranded DNA at high temperature resolution, and permits the analysis of genetic variations (SNPs, mutations, methylation) in PCR amplicons [15][16][17]. This technique allows genotyping and mutation scanning without the need for costly labeled probes, as it uses high fidelity heteroduplex-detecting double-stranded DNA binding dyes, such as EvaGreen which exhibit equal binding affinities for GC-rich and AT-rich regions and no sequence preference [18,19]. The heteroduplex products are detected by the presence of a second low-temperature melting transition [20]. This enables the reaction to be performed in a single tube, making it cost-effective and suitable for high-throughput screening [17]. HRM has been used extensively in the genotyping of human tissue samples for the identification of genes associated with diseases [21,22] as well as identification of clinically important fungal species [23]. HRM has been used for quantitative detection of adulteration with related species of pants used for medicinal purposes [24]. HRM has been used previously to identify mutations in TILLING populations of tomato [13], wheat [8] and Medaka [25]. In the tomato study, HRM was used to identify mutations using DNA pools from an EMS mutagenised population of plants. The wheat studies demonstrated that HRM can be used to isolate novel mutations in starch branching enzyme IIa (SBEIIa) genes from an EMS population [8] and in mutation scans in mixed PCR amplicons containing three homeologous gene fragments [19]. It was also demonstrated that homozygous and heterozygous individuals of a single mutation could be genotyped using HRM [19].
The genus Brassica includes the closest crop relatives of Arabidopsis thaliana, such as B. rapa (A-genome, 2n = 2x = 20,~550 Mbp), which includes vegetable crops (e.g. turnip, Chinese cabbage) and oil-seed crops, B. oleracea (C-genome, 2n = 2x = 18, > 600 Mbp) which includes vegetable crops (cauliflower, broccoli, cabbage) and the amphidiploid B. napus (AC-genome, 2n = 4x = 38,~1100 Mbp), which includes oil-seed crops (canola, oilseed rape) and swede. As with many crop plants Brassica genomes are complex arising from a series of duplication events that has resulted in most genes being present in multiple paralogous and homeologous copies [26,27]. A TILLING population has recently been generated in B. rapa to enable gene function to be analysed [3] and has previously been used to identify mutations in the B.rapa orthologue of the INDEHISCENT gene [28]. The aim of this study is to determine if HRM analysis can be used to genotype TILLING mutants in the crop species B. rapa and thereby to improve the efficiency of functional genomics approaches in this species.

Gene selection and TILLING
In order to demonstrate that HRM is suitable for efficient genotyping of mutant lines identified using TIL-LING, two target genes were chosen, BraA.CAX1.a and BraA.MET1.a. In A. thaliana, CAX1 has been shown to encode a calcium-proton antiporter [29] and MET1 has been shown to encode a DNA methyltransferase [30]. The CAX1 sequence from A. thaliana [Genbank: NM_129373.3] was used to identify the homologous sequence from B. rapa and from this sequence, 1.5 kb including the transcriptional start was used as the target for TILLING. To isolate BraA.MET1.a from B. rapa Ro-18, primers were designed based on the B. rapa Chiifu-401 sequence [Genbank:AB251937] [31]. These produced a 2110 bp DNA fragment covering 136 bp upstream of the translational start site together with the first intron and the first 407 bp of the second exon. Both sequences were submitted to the RevGenUK TIL-LING service, which resulted in 20 and 17 mutations being identified in BraA.CAX1.a and BraA.MET1.a respectively (Table 1). These mutations included silent, missense, nonsense and non-coding mutations.

Genotyping of TILLING lines using HRM
From the lines identified using TILLING, twelve (BraA.cax1.a-1, -4, -6, -7, -8, -10, -11, -12, -14, -16, -18, -20) and five (BraA.met1.a-3, -6, -7, -10 and -14) lines carrying mutations in BraA.CAX1 and BraA. MET1.a respectively were chosen to be characterised based on the nature and position of the mutation. In order to genotype the lines, a HRM assay was developed for each gene. To increase the specificity of the HRM reaction a nested PCR approach was used for the BraA.cax1.a lines, first using the TILLING primers to amplify genomic DNA, prior to using the HRM primers with this product. In total, three primer pairs were designed to genotype all 12 mutations within the BraA.CAX1.a gene (amplifying fragments of 200, 203 and 260 bp; Table 2) and one primer pair was designed to genotype all the mutations within the BraA.MET1.a gene (amplifying a 399 bp fragment; Table 2). Individual M 3 plants of the TILLING lines and wild-type Ro-18 plants were then genotyped using the HRM assay using a single primer pair per plate and three technical replicates per plant, allowing multiple alleles to be genotyped in a single scanning run. Analysis of the difference plots for the fluorescent signals detected individual plants that were wild-type, homozygous or heterozygous for the mutation in alleles of both genes ( Figure 1). Three distinct groups can readily be observed in the difference plots. To confirm these results and assign zygosity groups, the PCR products from selected lines were sequenced. Sequencing the HRM PCR products identified by HRM as heterozygous from BraA.cax1.a-11 and BraA.met1.a-6, showed a double peak at the position of the mutation ( Figure  2), confirming that these were indeed heterozygous for the mutations. In addition, sequencing of PCR products from BraA.cax1.a-1 and BraA.met1.a-6 identified as homozygous by HRM, confirmed that they were homozygous for the mutation (Figure 2). These results demonstrate that HRM can be used both for rapid genotyping of TILLING mutants and for distinguishing wild-type, heterozygous and homozygous plants. The results also demonstrate that multiple alleles can be genotyped using a single primer pair on one PCR plate, allowing multiple lines and individual plants to be genotyped simultaneously.

Backcross analysis
To reduce the mutation load, reciprocal crosses of the TILLING lines and wild-type R-o-18 plants were performed with both homozygous and heterozygous mutant  (Table  3). For example, the F 1 plants of a cross using a homozygous BraA.cax1.a-1 plant were all heterozygous for the mutation (n = 5). In contrast, the F 1 progeny of heterozygous plants showed segregation for the mutation (Table 3). These results demonstrate that HRM can be used to follow mutations and zygosity through multiple generations and crossing events, making it an ideal method for genotyping of TILLING mutants.

Conclusions
To analyse an allelic series of mutants generated using the TILLING approach efficiently, a cost effective, highthroughput genotyping technique is required. Techniques that have been used previously include PCR-RFLP, but this relies on the mutation causing a change in a restriction site [32] and TaqMan assays, which can require expensive probes and assay optimisation [18,33]. Alternatively, PCR products covering the mutations could be sequenced. However, this is time consuming and potentially more expensive; HRM analysis of samples in triplicate costs~30% of Sanger sequencing costs.
The results presented here demonstrate that HRM can be used to genotype mutations identified by TILLING from EMS populations in B. rapa. The results also demonstrate that a single PCR primer pair could be used to identify mutations at multiple positions with the PCR fragment. This reduces the number of primer pairs needed to genotype an allelic series and hence the cost. Due to the nature of the HRM process, the DNA extraction, PCRs and melt curve analysis can all be performed in 96-well plates. This enables high-throughput genotyping of multiple alleles of a number of genes to be performed quickly and efficiently. This is particularly important in complex crop species such as B rapa, where paralogous genes may be present and accumulation of mutations may be required to generate a phenotype.

Plant material
Seed of the highly inbred, homozygous, self-compatible, rapid cycling B. rapa ssp. trilocularis line R-o-18 was used as wild-type control. Seeds for lines (M 3 generation) containing mutations in target genes were obtained from the R-o-18 TILLING population through the RevGenUK service [3,34]. After in silico comparative sequence analyses, Brassica sequences were assigned to the member of the A. thaliana gene family with which they shared the highest sequence identity, thus removing false positives. All selected sequences which were orthologous to AtCAX1 were formatted into FASTA and entered into a contig assembly program (ContigExpress, VectorNTI 11, Invitrogen, Paisley, UK) using default settings to form longer contiguous Brassica sequences and identify possible locus specific paralogues. These nascent Brassica gene sequences were aligned to the A. thaliana gene of interest, flanked at either end with 1 Mbp of sequence, using AlignX software at default settings. The genomic structure of these contigs was elucidated and annotated in the Vector map software (VectorNTI 11). Primers ( Table 2) were designed to amplify~1.2 kbp fragments of BraA.CAX1.a beginning upstream from the deduced transcriptional start site to a region within the gene using Primer 3 (Version 0.4.0) [40] with the parameters "Max Self Complementarity" and "Max 3' Self Complementarity" adjusted to 2, to avoid hairpin loops and potential dimerisation. Primers to amplify the BraA. MET1.a gene (  μl of forward and reverse HRM primers (10 pmol μl -1 ) and SDW to 10 μl. Using an Eppendorf 96-well Master-Cycler 5331, reaction volumes were subjected to 95°C (5 min), followed by 30 cycles of 95°C (10 sec), 55°C annealing (10 sec) and 72°C extension (30 sec). Included within each 96-well analysis plate were samples in triplicate, also triplicate wild-type DNA samples and two sets of negative controls, one without DNA added and one without HRM primer added.

Lightscanner™ analysis of HRM products
PCR plates containing HRM amplicons were placed into a Lightscanner™ (Idaho Technology, Salt Lake City, USA) and the temperature raised to 60°C for 5 mins to ensure all samples were equilibrated. The melt temperature range was set at 63.3°C-95.4°C with a ramp setting at 0.1°C and a second hold at each step. Exposure was set at "Auto", background correction to exponential, curve shift to 0.020, standards to 'Auto Group', and sensitivity at normal +2.8. HRM analysis was then performed on the dissociation of double-stranded DNA PCR products, which had been saturated with the low PCR-toxic dye, EvaGreen from the initial HRM PCR. The LightScanner™ Data Analysis software (Version 2.0, Idaho Technology) was used to analyse the data and produce normalised disassociation curves and difference plots.