SEARCH

SEARCH BY CITATION

Keywords:

  • polymorphism discovery;
  • Arabidopsis;
  • haplotype;
  • SNP;
  • indel;
  • microsatellite

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

We have adapted the mutation detection technology used in Targeting Induced Local Lesions in Genomes (TILLING) to the discovery of polymorphisms in natural populations. The genomic DNA of a queried individual is mixed with a reference DNA and used to amplify a target 1-kbp region of DNA with asymmetrically labeled fluorescent primers. After heating and annealing, heteroduplexes are nicked at mismatched sites by the endonuclease CEL I and cut strands are visualized using Li-cor gel analyzers. Putative polymorphisms detected in one fluorescence channel can be verified by appearance of the opposite cut strand in the other channel. We demonstrated the efficiency of this technology, called Ecotilling, by the discovery in 150+ individuals of 55 haplotypes in five genes, ranging from sequences differing by a single nucleotide polymorphism to those representing complex haplotypes. The discovered polymorphisms were confirmed by sequencing and included base-pair changes, small insertions and deletions, and variation in microsatellite repeat number. Ecotilling allows the rapid detection of variation in many individuals and is cost effective because only one individual for each haplotype needs to be sequenced. The technology is applicable to any organism including those that are heterozygous and polyploid.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Understanding and manipulating genetic variation is paramount to elucidating gene function, identifying human disease genes, breeding, and conserving natural diversity. These tasks should be greatly facilitated by the information that genome sequencing provides on all genetic loci of a species: comparison of homologous loci can reveal functional variants of genes, and nucleotide polymorphisms can enable gene mapping and discovery by association or linkage disequilibrium analysis (Ardlie et al., 2002; Nielsen and Zaykin, 2001; Rafalski, 2002). The many available genome sequences, however, were extracted from one or few individuals of each species. Thus, although sequence information is powerful, it provides only a starting point to understand how genomes vary in populations. To realize this potential, sequence information on multiple individuals in a population must be obtained. However, sequencing even a subset of the genes in numerous individuals is expensive, and so several alternatives to sequencing have been proposed to discover allelic polymorphisms (Kristensen et al., 2001; Kwok, 2001; Tsuchihashi and Dracopoli, 2002). Methods that rely on differential mobility in chromatography or electrophoresis have high-throughput potential as they can be applied to many individuals of a population (Li et al., 2002; Xiao and Oefner, 2001). However, they usually can only identify the existence but not the type of polymorphism. Methods based on hybridization to microarrays can potentially discover and type many of the polymorphisms between two genomes, but applying this approach to many individuals is costly, and fewer than 50% of potentially detectable polymorphisms can be identified (Borevitz et al., 2003; Tillib and Mirzabekov, 2001). Thus, there is a need for cost-effective tools that can sample variation accurately on a large scale.

Targeting Induced Local Lesions in Genomes (TILLING) is a low-cost, high-throughput reverse genetic method that combines random chemical mutagenesis with PCR-based screening of gene regions of interest (Colbert et al., 2001; McCallum et al., 2000a,b). A 1-kbp region of a gene is amplified with fluorescent primers using as template the pooled DNA of mutagenized individuals. Nucleotide changes are identified by enzymatic digestion of heteroduplexes with the mismatch clearage endonuclease CEL I and detection of cleaved fluorescent products on electrophoretic gels (Kulinski et al., 2000; Oleykowski et al., 1998). One Li-cor gel analyzer run at 96 lanes per gel can probe an approximately 1-kbp region in pools assaying over 2000 individuals per day. Single nucleotide polymorphism (SNP) discovery for cataloging variation is essentially the same task as detection of induced point mutations, and so the question arises as to whether the high-throughput method used for TILLING is practical for SNP discovery and genotyping.

We developed a method for detecting multiple types of natural polymorphisms in natural populations, a strategy that we call ‘Ecotilling’. Using up to 192 accessions of Arabidopsis thaliana, we show that Ecotilling can be used to effectively discover natural polymorphisms in a large set of individuals. Further, given the positional information provided by CEL I cutting, it is easy to identify polymorphisms and haplotypes, and by grouping the identical DNA haplotypes existing in the population, DNA sequencing can be limited to the fraction of individuals representing unique haplotypes. Ecotilling is a general strategy for DNA polymorphism discovery in any organism.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

In the course of TILLING genes in Arabidopsis, we observed that occasional contamination by other accessions (ecotypes) in our mutagenized Columbia-0 (Col) populations was readily detected as single or multiple bands appearing repeatedly in the same lane position (representing one contaminated DNA pool of eight individuals) in multiple screened genes (Figure 1). The ends of the amplified DNA are labeled differentially with either the IRD700 dye, on the left end, or IRD800 dye, on the right end. The two fluorescent dyes are detected in different channels, and two images are generated for each electrophoretic run. Candidate polymorphic sites identified in the IRD700 channel can be confirmed in the IRD800 channel, which shows the cleavage product labeled on the right end. This observation suggested that, in addition to allowing efficient detection of induced mutations, the TILLING method should be ideal for the detection of natural polymorphisms. We developed a system for such an application as described in Figure 2.

image

Figure 1. Discovery of variant haplotypes contaminating a TILLING screen.

The selected lanes are from the gel images for the IRD700 and IRD800 channels of a Li-cor analyzer and were generated during a TILLING screen. Each lane displays the 1-kbp amplification product of the queried gene produced from a DNA pool of eight mutagenized individuals. After melting and annealing of PCR products, nucleotide mismatches in heteroduplexes are cut by the endonuclease CEL I. The cut DNAs are visible as bands of faster mobility than the full-length product. Mutations are rare, and two are unlikely to be found in the same pool or lane. Thus, the appearance of multiple cut products in the same lane (marked by arrows) was best explained if contamination by different accessions had occurred among the mutagenized lines. This possibility was confirmed by the repeated discovery of polymorphisms in the same pools in other screened genes (data not shown). We concluded that in the contaminated lanes one of the eight individuals is either a hybrid from a fortuitous out-cross, or grew from a contaminant seed contributing extraneous DNA to the pool of Col-0 ecotype DNA. Thus, the darker bands marked by symbols on the right of the gels are the product of mismatches in heteroduplexes between the contaminant and Col DNA (see Experimental procedures) digested with endonuclease CEL I. Mobility of the left- and right-labeled cut strands can be compared in the two channels and the inferred molecular weights add to the full-length product.

Download figure to PowerPoint

image

Figure 2. The Ecotilling process involves preparation of DNAs from individuals of a population, in this case geographical accessions (ecotypes) of Arabidopsis.

Each DNA is combined with reference DNA from the standard accession, Col-0, and arrayed in plates. Fluorescently labeled primers are used to amplify an approximately 1-kbp product, which after melting and annealing to form heteroduplexes is nicked at mismatches by CEL I. Cut strands are detected on Li-cor analyzers. After this screen, only the variant haplotypes are sequenced.

Download figure to PowerPoint

Figure 3 shows a Li-cor image for a 1-kbp window from the DNA methyltransferase 2 (DNMT2; also DMT11, At5g25480) locus in 96 ecotypes. The DNA of each ecotype was mixed with an equal amount of DNA from the standard Col ecotype and amplified with PCR primers specific for DNMT2. Each lane displays a top band (full-length product), a bottom band (primer), and background bands (lighter intensity) common to all lanes. If the DNA sequence of the ecotype in the 1-kbp window is identical to that of Col, no other bands should be produced. If one or more sequence differences are present, novel bands are produced with the mobility of the corresponding cut fragments. On the left of the picture are marked the displayed polymorphisms that identify several haplotypes, both frequent and rare, of DNMT2. Note that by lining up these bands with the background bands, it is easy to confirm that bands found at the same horizontal position do indeed co-migrate.

image

Figure 3. Example of a 96-ecotype gel image (IRD700 fluorescent dye channel), revealing 10 polymorphic sites within 1 kbp of the DNMT2 gene.

Exons (boxes) and introns (lines) are drawn on the left. Arrow points at the bottom indicate markers at each 8th lane.

Download figure to PowerPoint

Each nucleotide polymorphism was first recorded by its gel mobility, which approximates position within a few nucleotides. A PERL program, ‘e-squint’, facilitated the task of logging the observed polymorphisms in a database. Thus, each haplotype can be archived as tabular mobility data. This information alone can represent diversity in the sampled population. Sequence data, however, are even more useful. Another program, ‘e-pick’ generated sequencing request files from the e-squint data, facilitating the arraying of representative template DNAs and appropriate primers for sequencing. A sample representing each haplotype was sequenced. We analyzed sequences on both strands and determined that all the polymorphisms that were present could be detected by CEL I. An effective and cost-saving measure would be to select the left or right sequencing primer by its proximity to the polymorphism. We performed a multiple alignment of sequencing chromatograms using sequencher software to automatically identify the base change, which in each case confirmed the mobility of the gel band (Figure 3). Thus, this gel image required a total of 10 sequencing runs to identify the polymorphisms in a 1-kbp segment from 96 ecotypes. In each case, sequencing confirmed the CEL I assay results.

We subjected four other genes to the same analysis to test the robustness of the approach. As shown in Figure 4, we detected a total of 45 haplotypes that differed from those of Col at 1–31 sites. All CEL I-detected signals that were sequenced were verified as true polymorphisms. In addition, we detected by sequencing some polymorphisms that had not been detected by the CEL I assay. Not unexpectedly, we found that most of these polymorphisms were present in the terminal 80 nucleotides of each amplified fragment. The corresponding CEL I-generated products migrate in a noisy gel area in both IRD channels and can be obscured (Greene et al., 2003). We sequenced 96 randomly chosen haplotypes – encompassing four genes – of the Col type, i.e. from accessions in which no polymorphisms were detected. We found three haplotypes that differed from the Col standard haplotype. Two had polymorphisms near the ends of the TILLED fragment, and the corresponding CEL I products would have been in the noisy region of the gel. A third divergent haplotype was detectable in the original CEL I gel, but had been overlooked in the original examination. Thus, errors are possible, but are relatively infrequent. An error could also result if two haplotypes differ from each other and from the standard Col by having a different nucleotide substitution at the same position. Both would yield the same Ecotilling gel pattern in relationship to the Col standard and would be classified as a single haplotype. We sequenced the phytochrome-interacting factor 2-4 (PIF2-4) haplotype in 15 ecotypes, and the domains rearranged DNA methyltransferase 1-2 (DRM1-2) haplotype in 35 ecotypes finding no differences at the polymorphic base site. Therefore, in the present cases, all haplotypes were the same, but this type of error will no doubt occur at low frequency. Another source of error, failure to amplify the queried allele, was tested by amplifying ecotypic DNA without adding the Col standard. Depending on the gene, amplification failures ranged from 0 to 4% and may be caused by multiple polymorphisms in the regions complementary to the primers.

image

Figure 4. Haplotype diversity at four tested loci.

The exons (boxes) and introns (angled lines) of the gene model on the left represent the sampled region of each gene. The haplotypes are displayed as vertical lines with polymorphisms as ticks (SNPs), triangles (deletions), and inverted triangles (insertions). The number next to the triangle indicates the size of the indel. Starred ticks represent polymorphisms that were not sequenced. A red line connecting a polymorphism to the gene model indicates a non-synonymous change or indel in the predicted protein sequence. The red-highlighted deletion symbol in haplotype DRM1-10 indicates a frameshift. The frequency of each haplotype is indicated at the bottom.

Download figure to PowerPoint

The genes examined provide examples of the pattern of variation occurring in Arabidopsis. For example, the DRM1 gene (At5g15380) has a predicted null allele in two ecotypes, indicating that, at least in certain genetic backgrounds, it may be dispensable in a natural environment. DRM1 is probably redundant with a closely linked gene, DRM2 (At5g14620; Cao and Jacobsen, 2002). A different pattern was seen with the AtWERNERexo gene (At4g13870), which encodes the exonuclease domain of a Werner syndrome protein homolog (Hartung et al., 2000). Variation was largely limited to intronic regions, indicating that although variant haplotypes are present in nature, most do not encode variant proteins. Several haplotypes found at a putative GATA-type transcription factor locus (called ‘C7’) (At4g36620) had non-synonymous changes. The most divergent haplotype was found at the PIF2 locus (At5g24500), which encodes a putative protein of unknown function. PIF2-2, the haplotype of ecotype Lezoux-0, displayed 31 changes (3 deletions, 3 insertions, and 25 SNPs) out of about 900 sampled residues (3.4%). We sequenced the DNA regions flanking the TILLED fragment of PIF2-2, which spans the predicted exons and intron. No changes were detected in a 400-bp region directly upstream of PIF2, while in a 650-bp region, directly downstream 3-bp changes were present in the proximal 150 nucleotides and none in the distal 500. Thus, haplotypic variation at the PIF2 locus was limited to the approximate 1-kbp core of the 2.2 region queried.

By examining multiple haplotypes of each gene, we were able to test the ability of the CEL I assay to detect different types of polymorphisms, including insertion or deletions (indels). In addition, we were surprised to see that the assay could identify polymorphisms caused by microsatellite repeat variation. In the example in Figure 5, a difference in one trinuncleotide repeat between the PIF2-1 haplotype of Col and the PIF2-9 haplotype of Monte Tosso-0 produced multiple bands predicted to be three nucleotides apart. We interpret this pattern as preferential D-looping of one type of triplet (for example, GAA instead of AAG and AGA) and cutting by CEL I at different D-looped sites along the microsatellite region. Interestingly, cutting at multiple possible D-loops did not dilute the signal and prevent detection. In addition, a 9-bp deletion was also detected by the formation of multiple bands. Deletions as large as 21 bp were detected. Lastly, although the CEL I assay can effectively display multiple polymorphisms between two haplotypes, it is less accurate when the number of polymorphisms is very high, as in the case of the PIF2-2 haplotype. This haplotype consists of more than 30 polymorphisms: as a result, too many mismatches decreased resolution and signal intensity in the CEL I analysis, allowing detection of only a subset. This subset, however, was sufficient to define the variant haplotype whose sequence was later determined.

image

Figure 5. Detection of different types of DNA polymorphism.

Selected lane regions of a larger Li-Cor image display the CEL I digestion products of heteroduplexes between the reference PIF2-1 allele from Col-0 and PIF2-8 from Mechtshausen-0 (A), PIF2-9 from Monte Tosso-0 (B), and PIF2-10 from Merzhausen-0 (C). The displayed products are a SNP (A), a microsatellite unit repeat polymorphism (B), and a deletion polymorphism (C). The double band observed for the SNP may result from length heterogeneity of the PCR product. The IRD700 image was flipped vertically to facilitate comparison; however, because of the assay (see Figure 1), different strands of the same product are imaged in each IRD channel.

Download figure to PowerPoint

We found that accurate E-squinting (identifying and attributing gel bands to each sample) was relatively laborious because the lanes in a 96-sample gel are very close and sometime hard to distinguish. Lowering the number of samples per gel by loading a 96-well gel with 50–70 samples, i.e. leaving some lanes blank, made for much easier scoring.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

A cheap and fast natural polymorphism discovery and genotyping method is likely to be of great importance for determining the spectrum of variation and for genetic mapping based on linkage association analysis. We show that the TILLING method we have developed for discovery of induced mutations is applicable, with few modifications to the discovery of polymorphisms. For Ecotilling, instead of using as templates pools of DNA from mutagenized lines, we used the DNA of A. thaliana ecotypes, each mixed with the reference Col ecotype DNA. We amplified a 1-kbp segment from each investigated gene using a window selected for mutation discovery, usually according to criteria that maximize incorporation of conserved exons and codons suitable for ethylmethanesulfonate-induced mutagenesis (Till et al., 2003a,b). These criteria are useful for discovery of polymorphisms that may have functional consequences for the encoded protein. The choice of segment, however, could be any chromosomal region suited for amplification. We found that the method could be used to survey natural variation in many individuals accurately and cheaply: we discovered 55 haplotypes in five loci sampled over roughly an 800-bp window (1 kbp − 2 × 80 bp, the terminal noisy regions).

Our comparison of the many accessions of A. thaliana confirms that this is an attractive model organism for studying nucleotide diversity. The different wild accessions constitute a rich source of diversity, in both phenotype and nucleotide sequence. Nucleotide diversity (Nei and Li, 1979) ranges between 0.003 and 0.01 at random sites (Breyne et al., 1999; Hardtke et al., 1996; Miyashita et al., 1999; Spiegelman et al., 2000) and at selected genes (Aguade, 2001; Caicedo et al., 1999; Kawabe and Miyashita, 1999; Kawabe et al., 1997, 2000; Miyashita, 2001), but can be higher in certain regions (e.g. RPP5 gene cluster; Noel et al., 1999). We found different patterns of allelic variation, depending on the gene investigated confirming that a survey of diverse ecotypes may help the functional examination of any gene of interest. In the case of DRM1, a null allele was identified among the ecotypes, suggesting that this gene is dispensable. Another unexpected finding is the highly variant PIF2-2 haplotype, whose unusual divergence may have been caused by extraordinary retention of an ancient allele or by introgression of a foreign chromosomal region through an interspecific cross.

Ecotilling information will be very valuable for functional analysis and will be provided together with induced mutants by the Arabidopsis TILLING Project (ATP). We have selected 50 standard ecotypes (http://walnut.usc.edu/finalPlate.htm), whose survey will be incorporated as a user requested service at the ATP site (http://tilling.fhcrc.org:9366). Given the generality of the method, we expect that its application should be widespread. For example, Ecotilling was effective at genotyping rice accessions, both indica and japonica (B. Till et al., unpublished results). Errors and failures are low (a few percent): those caused by the operator can be minimized by training while those caused by failure of amplification of the queried genotype can either be tolerated, or, if accuracy is paramount, identified by additional measures. For example, reference and query can be amplified separately, labeling a single terminus (the right or left, respectively). The resulting products are then mixed, melted, and annealed to perform heteroduplex analysis. Another measure is the use of a reference DNA in which an induced mutation is expected to produce a ‘control’ signal in every lane.

Under the experimental conditions used, the endonuclease CEL I cuts with partial efficiency, allowing the detection of multiple mismatches in a DNA duplex. As a result, interrogating an unknown homologous DNA by heteroduplexing to a fragment of known sequence should reveal the number and position of polymorphic sites. Both nucleotide changes (all types of base mismatches) and small indels (1–30 bp) were identified. Surprisingly, we were able to observe changes in triplet repeats as a ladder of fragments cut at multiple sites, perhaps because the multiple displacement loop polymorphism that is formed in a heteroduplex may be cut more efficiently by CEL I than in SNP heteroduplex. The ability of this assay to resolve indels without compromising detection of polymorphisms 3′ of the indel is particularly advantageous in heterozygotes where direct sequencing would cause reading frameshifts. Detection is on gels with nearly base pair resolution, and background patterns are uniform across lanes; bands that are of identical size can be matched, thus discovering and genotyping NPs in a single step. Furthermore, once the DNAs from the accession are arrayed and the primers are available, a gel image can be produced in less than 8 h (involving a PCR reaction, cutting with CEL I and electrophoresis in a Li-cor analyzer) and the gel pattern is immediately interpretable, allowing an early assessment of diversity at the sampled locus. Depending on the number of analyzers available, large numbers of accessions can be rapidly tested. For example, we discovered 10 haplotypes among 192 ecotypes for a 1-kbp stretch of DNMT2. Last, TILLING of larger fragments (up to 1.6 kbp) is also possible, increasing the size of the region over which polymorphisms can be detected reliably in a single Ecotilling run, further decreasing the cost of discovery (B. Till and C. Burtner, unpublished observation).

The ability of the Ecotilling approach to resolve multiple input sequences should be applicable to heterozygous and polyploid organisms, where single or multiple polymorphisms at two or more alleles may be present. Such an ability has been thoroughly demonstrated by the routine detection to date of 855 homozygous and 1842 heterozygous mutants in pools of eight Arabidopsis individuals (Greene et al., 2003; Till et al., 2003b). Two CEL I reactions are necessary to detect whether an individual is heterozygous at a given gene: one with and one without the reference DNA. The non-doped CEL I reaction will display the heterozygous sites, whereas all nucleotide sites that are negative in the non-doped reactions are deemed homozygous. In the case of an autotetraploid individual, such as frequently found in plants and cancer cells, a non-doped reaction will display all existing heterozygosities regardless of their genotypic state, such as simplex (aAAA) or duplex (aaAA). In addition, non-doped reactions can reveal the existence of duplicated loci as indicated by an Ecotilling analysis of MutS homolog gene in Arabidopsis (L. Hensel and A. Odden, personal communication).

The ability to identify a polymorphism in 1/16 molecules using this method (Greene et al., 2003; Till et al., 2003b) should allow pooling of samples in mapping studies. For example, seeking a rare recombinant (V-A) between a locus V that can be scored visually and a linked locus A marked by an SNP in the testcross V-a//v-A × v-a//v-a would entail eight-fold pools of DNA extracted from Vv individuals arrayed at 768 per 96-well plate. Amplified A sequences subjected to CEL I analysis without doping would allow the identification of rare V-A//v-a individuals among the predominant V-a//v-a type.

In a number of circumstances, Ecotilling compares favorably to the most common alternative – full sequencing. For polymorphism discovery in populations of homozygous organisms, the frequency of haplotypes will determine which of the two is most advantageous, and larger ratios of haplotypes to sampled individuals favor sequencing. With many fewer haplotypes than individuals, Ecotilling greatly reduces the number of sequences that needs to be determined. Sequencing is then carried out on individuals selected to represent each haplotype. The exact haplotypic frequency at which one system becomes more advantageous depends on variables, including the cost of sequencing, the choice of sequencing one or both strands, the degree of automation, the sampled individual number, and the cost of labor. For example, in the context of ATP (http://tilling.fhcrc.org:9366/), Ecotilling is several fold cheaper than full sequencing. There are scenarios in which Ecotilling should be the method of choice. The first case is when haplotyping is sufficient and sequence determination is unnecessary. This is the case, for example, of forensic investigations when a rare haplotype must be matched to one of many candidates, or of conservation studies when one must sample diversity within a species. The second scenario, as discussed above, is when heterozygous indels occur at the studied locus. The third scenario is when heterozygous polyploids are studied.

Other methods for natural polymorphism discovery are still in developmental stage, such as mass spectrometric detection (Krebs et al., 2003). Microarrays are promising tools (Cutler et al., 2001; Larsen et al., 2001) but have strengths and weaknesses different from those of Ecotilling. Recently, Borevitz et al. (2003) demonstrated the discovery of polymorphisms between the Landsberg erecta (Ler) and Col ecotypes of Arabidopsis using an oligonucleotide array designed for expression analysis in which each feature represents a short 3′ region of each gene. Although fewer than 50% of the potentially detectable polymorphisms were identified, the method yielded many confirmed polymorphisms with a genome-wide coverage. It would be unsuited for discovery and comparison of haplotypes at specific loci, unless many more oligonucleotide features per gene were deposited on the array. Such arrays might be available at competitive prices in the future, but at present their cost is likely to be prohibitive. These considerations suggest that microarrays and Ecotilling are complementary tools: the first well suited for global natural polymorphism discovery among a few genotypes, the latter best at surveying diversity at specific loci among many genotypes.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Plant lines and DNAs

Ecotypes were obtained from the Arabidopsis Biological Resources Center (http://www.arabidopsis.com), DNA was extracted from 200 mg of tissue obtained from a single plant using the Q-Biogene FastDNA kit as previously described by Till et al. (2003a), and quantified on a 1.5% agarose gel using lambda DNA as a concentration reference (Invitrogen, Carlsbad, CA, USA). DNA from all samples was normalized to a final concentration of 3 ng µl−1. Seed was collected from each DNA source plant to establish single plant descent lines. Following extraction, DNA samples were arrayed in a 96-well format. Pooled DNAs were diluted to 0.015 ng µl−1 for screening and mixed with an equal amount of Col-0 DNA.

Screening for polymorphisms

The genes for Ecotilling were chosen from the ATP queue of requests (Till et al., 2003b). The primers and product size were as follows: DNMT2, left primer: taagcagcagcagaagaagaagcatgg, right primer: accaggcagaaataacgtggtcttgaa, fragment length 935; DRM1 (At5g15380), left primer: agaaagccgaacacggaccagctc, right primer: actcgtccataagccgctcaatcg, fragment length 1001; C7 (At4g36620), left primer: ccacacacaacacacacttctctttctcc, right primer: gtcggcgacgttaagcctccaag, fragment length 809; PIF2 (At5g24500), left primer: caaccgacgatgacgatgcttctg, right primer: ccttcggctgacattgctgctttc, fragment length 1002; and AtWRNexo (At4g13870), left primer: agaatctcattgctgcaagctttgttct, right primer: gccactgacagcatcaggaaggtc, fragment length 1034.

PCR was performed in a 10-µl volume with 0.075 ng genomic DNA mixed 1 : 1 with Col wild-type DNA. PCR and CEL I reactions were carried out as described by Colbert et al. (2001) and Till et al. (2003a).

The UNIX programs e-squint and e-pick were used for entry of ecotypic screen data and for the arraying of sequencing templates. They were modified from the analogous squint and pick, which are used for the entry of mutant screen data and the array of mutant sequencing templates (Till et al., 2003a).

DNA sequencing

For DNA sequencing, amplifications were performed with 0.005 ng genomic DNA in a 20 µl final volume following the manufacturer's suggestions for the Ex-Taq polymerase (Takara, Otsu, Shiga, Japan) as described by Till et al. (2003a). We sequenced both strands, with the exception of DNMT2 (see Results). Samples were run on ABI 3100 capillary sequencing machines operated by the Fred Hutchinson Cancer Research Center's genomics facility. Sequence trace analysis was performed using sequencher™ 4.1 software (Gene Codes). Detected polymorphisms were confirmed by comparing to the cut strand mobility information gathered in the CEL I screens to the polymorphism identified by sequencing.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

We thank Jorja Henikoff for e-squint and e-pick programs, past members of the high-throughput TILLING team, including Terri Bryson, Nina Accornero, Trent Colbert, Rebecca Lechalk, and Mike Steine. We thank Linda Hensel for sharing unpublished results. We also thank Jerry Davison for discussion. This work was funded by the National Science Foundation Plant Genome Research Program.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  • Aguade, M. (2001) Nucleotide sequence variation at two genes of the phenylpropanoid pathway, the FAH1 and F3H genes, in Arabidopsis thaliana. Mol. Biol. Evol. 18, 19.
  • Ardlie, K.G., Kruglyak, L. and Seielstad, M. (2002) Patterns of linkage disequilibrium in the human genome. Nat. Rev. Genet. 3, 299309.
  • Borevitz, J.O., Liang, D., Plouffe, D., Chang, H.-S., Zhu, T., Weigel, D., Berry, C.C., Winzeler, E. and Chory, J. (2003) Large-scale identification of single-feature polymorphisms in complex genomes. Genome Res. 13, 513523.
  • Breyne, P., Rombaut, D., Van Gysel, A., Van Montagu, M. and Gerats, T. (1999) AFLP analysis of genetic diversity within and between Arabidopsis thaliana ecotypes. Mol. Gen. Genet. 261, 627634.
  • Caicedo, A.L., Schaal, B.A. and Kunkel, B.N. (1999) Diversity and molecular evolution of the RPS2 resistance gene in Arabidopsis thaliana. Proc. Natl. Acad. Sci. USA, 96, 302306.
  • Cao, X. and Jacobsen, S.E. (2002) Role of the Arabidopsis DRM methyltransferases in de novo DNA methylation and gene silencing. Curr. Biol. 12, 11381144.
  • Colbert, T., Till, B.J., Tompa, R. et al. (2001) High-throughput screening for induced point mutations. Plant Physiol. 126, 480484.
  • Cutler, D.J., Zwick, M.E., Carrasquillo, M.M. et al. (2001) High-throughput variation detection and genotyping using microarrays. Genome Res. 11, 19131925.
  • Greene, E.A., Codomo, C.A., Taylor, N.E. et al. (2003) Spectrum of chemically induced mutations from a large-scale reverse-genetic screen in Arabidopsis. Genetics, 164, 731740.
  • Hardtke, C.S., Muller, J. and Berleth, T. (1996) Genetic similarity among Arabidopsis thaliana ecotypes estimated by DNA sequence comparison. Plant Mol. Biol. 32, 915922.
  • Hartung, F., Plchova, H. and Puchta, H. (2000) Molecular characterisation of RecQ homologues in Arabidopsis thaliana. Nucl. Acids Res. 28, 42754282.
  • Kawabe, A. and Miyashita, N.T. (1999) DNA variation in the basic chitinase locus (ChiB) region of the wild plant Arabidopsis thaliana. Genetics, 153, 14451453.
  • Kawabe, A., Innan, H., Terauchi, R. and Miyashita, N.T. (1997) Nucleotide polymorphism in the acidic chitinase locus (ChiA) region of the wild plant Arabidopsis thaliana. Mol. Biol. Evol. 14, 13031315.
  • Kawabe, A., Yamane, K. and Miyashita, N.T. (2000) DNA polymorphism at the cytosolic phosphoglucose isomerase (PgiC) locus of the wild plant Arabidopsis thaliana. Genetics, 156, 13391347.
  • Krebs, S., Medugorac, I., Seichter, D. and Forster, M. (2003) RNaseCut: a MALDI mass spectrometry-based method for SNP discovery. Nucl. Acids Res. 31, e37.
  • Kristensen, V.N., Kelefiotis, D., Kristensen, T. and Borresen-Dale, A.L. (2001) High-throughput methods for detection of genetic variation. Biotechniques, 30, 318322, 324, 326 passim.
  • Kulinski, J., Besack, D., Oleykowski, C.A., Godwin, A.K. and Yeung, A.T. (2000) CEL I enzymatic mutation detection assay. Biotechniques, 29, 4446.
  • Kwok, P.Y. (2001) Methods for genotyping single nucleotide polymorphisms. Annu. Rev. Genomics Hum. Genet. 2, 235258.
  • Larsen, L.A., Christiansen, M., Vuust, J. and Andersen, P.S. (2001) Recent developments in high-throughput mutation screening. Pharmacogenomics, 2, 387399.
  • Li, Q., Liu, Z., Monroe, H. and Culiat, C.T. (2002) Integrated platform for detection of DNA sequence variants using capillary array electrophoresis. Electrophoresis, 23, 14991511.
  • McCallum, C.M., Comai, L., Greene, E.A. and Henikoff, S. (2000a) Targeted screening for induced mutations. Nat. Biotechnol. 18, 455457.
  • McCallum, C.M., Comai, L., Greene, E.A. and Henikoff, S. (2000b) Targeting induced local lesions in genomes (TILLING) for plant functional genomics. Plant Physiol. 123, 439442.
  • Miyashita, N.T. (2001) DNA variation in the 5′ upstream region of the Adh locus of the wild plants Arabidopsis thaliana and Arabis gemmifera. Mol. Biol. Evol. 18, 164171.
  • Miyashita, N.T., Kawabe, A. and Innan, H. (1999) DNA variation in the wild plant Arabidopsis thaliana revealed by amplified fragment length polymorphism analysis. Genetics, 152, 17231731.
  • Nei, M. and Li, W.H. (1979) Mathematical model for studying genetic variation in terms of restriction endonucleases. Proc. Natl. Acad. Sci. USA, 76, 52695273.
  • Nielsen, D.M. and Zaykin, D. (2001) Association mapping: where we've been, where we're going. Expert Rev. Mol. Diagn. 1, 334342.
  • Noel, L., Moores, T.L., Van Der Biezen, E.A. et al. (1999) Pronounced intraspecific haplotype divergence at the RPP5 complex disease resistance locus of Arabidopsis. Plant Cell, 11, 20992112.
  • Oleykowski, C.A., Bronson-Mullins, C.R., Godwin, A.K. and Yeung, A.T. (1998) Mutation detection using a novel plant endonuclease. Nucl. Acids Res. 26, 45974602.
  • Rafalski, A. (2002) Applications of single nucleotide polymorphisms in crop genetics. Curr. Opin. Plant Biol. 5, 94100.
  • Spiegelman, J.I., Mindrinos, M.N. and Oefner, P.J. (2000) High-accuracy DNA sequence variation screening by DHPLC. Biotechniques, 29, 10841090, 1092.
  • Till, B.J., Colbert, T., Tompa, R. et al. (2003a) High-throughput TILLING for functional genomics. In Plant Functional Genomics: Methods and Protocols (Grotewald, E., ed.). Totowa: Humana Press, pp. 205220.
  • Till, B.J., Reynolds, S.H., Greene, E.A. et al. (2003b) Large-scale discovery of induced point mutations with high throughput TILLING. Genome Res. 13, 524530.
  • Tillib, S.V. and Mirzabekov, A.D. (2001) Advances in the analysis of DNA sequence variations using oligonucleotide microchip technology. Curr. Opin. Biotechnol. 12, 5358.
  • Tsuchihashi, Z. and Dracopoli, N.C. (2002) Progress in high throughput SNP genotyping methods. Pharmacogenomics J. 2, 103110.
  • Xiao, W. and Oefner, P.J. (2001) Denaturing high-performance liquid chromatography: a review. Hum. Mutat. 17, 439474.