INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS WEB SITE REFERENCES REFERENCES

Forward genetic screens in model organisms remain a crucial tool for uncovering new biological information (Matthews and Kopczynski 2001; Sternberg 2001). These approaches require extensive recombination mapping of a mutation to discover the identity of a gene. Traditional methods in model systems have typically relied on the use of visible phenotypic markers for linkage mapping of mutations. However, single nucleotide polymorphism (SNP) markers are currently favored because of their relative abundance and because they can eliminate confounding interaction with the mutant phenotype, although in some cases outcrossing introduces genetic modifiers.

To date, the only strategy for SNP-based cloning in the nematode Caenorhabditis elegans (C. elegans Sequencing Consortium 1998) is the snip-SNP approach (Wicks et al. 2001). Here we present an alternative tripartite approach for rapid SNP-based mapping in the worm. We first established a set of finely spaced genome-spanning SNP markers and then combined this resource with a tiered mapping strategy that progressively narrows the region containing the gene of interest. Finally, we used a high-throughput SNP assay that allowed reliable and rapid genotyping with low marker development costs. This strategy afforded rapid gene cloning in C. elegans and can be tailored for use in other model organisms with a sequenced genome.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS WEB SITE REFERENCES REFERENCES

Our initial goal was to create a set of reliable SNP markers at a density of one marker every 100 kb across the C. elegans genome. To achieve this density, a large set of predicted SNPs scattered throughout the C. elegans genome was required. We chose to identify polymorphisms between the Hawaiian CB4856 strain of C. elegans and the commonly used Bristol N2 laboratory strain, because the CB4856 strain is known to have the most even distribution of SNPs across the chromosomes (Koch et al. 2000; Wicks et al. 2001). A small insert (1.7 kb ± 0.5 kb) library was constructed from CB4856 genomic DNA. Double-end sequencing of cloned inserts from this library produced 16,941 high quality sequencing reads, which represented 7.3% sequence coverage of the genome. An informatics software pipeline (Vysotskaia et al. 2001) was then used to predict likely polymorphisms between CB4856 sequencing reads and the canonical N2 genomic sequence (WS version 48). The pipeline identified a total of 10,711 predicted polymorphisms; 9602 of these were unique from those previously reported. These unique polymorphisms include 6902 substitutions (SNPs), 1885 deletions (one or more bases removed), and 815 insertions (one or more bases added). We estimate the overall rate of polymorphism between the two strains to be one substitution/insertion/deletion per 840 bases. Transitions accounted for 57% (3906 SNPs) of substitution SNPs, whereas 43% (2996 SNPs) were transversions. These observations agreed with previous findings from a smaller dataset (Wicks et al. 2001). Validated SNP data are available from NCBI dbSNP (accession nos.: 4388625-4389689) and GenBank dbSTS (accession nos.: G73810-G74874).

We combined the publicly available C. elegans CB4856 SNP information (http://genome.wustl.edu/gsc/C_elegans/SNP/index.html) with our data to obtain a total of 17,189 predicted polymorphisms throughout the C. elegans genome and then systematically chose one substitution SNP spaced approximately every 100 kb. Oligonucleotide primers were designed flanking these predicted SNPs for PCR amplification. The presence of a SNP was confirmed by sequencing the PCR product and/or by a fluorescence polarization-template directed incorporation (FP-TDI) SNP genotyping assay (see below). The latter method proved to be faster than sequencing and had a comparable failure rate of ~10%, which includes mispredictions, primer failure, and assay failure. To date, a set of 1099 markers have been confirmed and formatted for a genotyping assay; 427 (39%) of these 1099 confirmed SNPs were derived from our putative new 9602 SNPs. Our substitution SNP marker set has an average spacing of 91 kb ± 56 kb across the genome (Fig. 1A). The most telomeric SNPs on average are located ~72.5 kb from the telomeres of each chromosome, ranging from 16.6 kb (left end of chromosome V) to 178.3 kb (right end of chromosome III).

View larger version (32K): [in this window] [in a new window]   Figure 1   Distribution of single nuclear polymorphism (SNP) markers across the Caenorhabditis elegans genome. (A) Marker distribution along the physical map of chromosome. The location of the Tier 1 SNP markers are shown with open squares, and the location of Tier 2 markers are shown by open circles. The location of each of the validated 1099 markers is shown along the chromosomal axis. The largest gap is 449 kb (chromosome I), and the average validated marker spacing is 91 kb ± 56 kb. (B) Recombination rates across the six chromosomes. The physical (Mb) location of all predicted polymorphisms (y-axis) is plotted versus the extrapolated genetic (cM) location (x-axis) of the associated cosmid across the genome (Genetic Map Position from Wormbase; Stein et al. 2001). This illustrates the rate of change in the recombination rate across the chromosome and shows the "gene cluster" effect in the center of the autosomes (Barnes et al. 1995). This information is used during mapping, because the local recombination rate affects the number of putative recombinants that must be genotyped to obtain a 100-kb interval.

By use of the comprehensive SNP marker set, we implemented a mapping strategy that uses iterative phases to progressively refine a genomic region of interest (ROI) containing a mutant gene. In C. elegans forward genetic screens, mapping often begins by genotyping phenotypically mutant F2 offspring from a cross of a homozygous mutant animal (N2 background) to a wild-type animal (CB4856 background) (see Methods). High-throughput genotyping of 30-60 offspring, using a preselected marker set of 30 SNPs (Table 1 

View larger version (59K): [in this window] [in a new window]

View larger version (25K): [in this window] [in a new window]   Figure 2   Tiered mapping strategy. Shown is a schematized mapping workflow. (A) Tier 1 mapping localizes the gene of interest to a chromosome by assessing linkage to one centrally located SNP. (B) Tier 2 mapping localizes the gene to a subregion of the chromosome. This region can vary in size between 1 and 6.7 Mb. In the dpy-18 example shown, the gene falls between two SNP markers that are 2 Mb apart on chromosome III. This resolution is routinely achievable by genotyping 30-60 recombinants. (C) Tier 3 mapping begins by identification of informative animals with recombination breakpoints within the region defined by Tier 2 and then fine mapping with Tier 3 markers to narrow the region of interest. For dpy-18, Tier 3 mapping localizes the gene to a region as small as 97 kb and thus narrowed the candidate region to ~0.1% of the worm genome. The number of recombinants required to achieve this mapping resolution will depend on the local recombination rate (Fig. 1B; Genetic Map Position from Wormbase; Stein et al. 2001) along the chromosome in the vicinity of the mutated gene. (D and E) Tier 4 mapping and/or mutation detection. Further refinement of the candidate interval occurs by validation of additional SNP markers and genotyping of informative recombinants. For dpy-18, the location of the predicted substitution SNPs located within the 97-kb region of chromosome III are shown. During this fine-mapping process, candidate gene approaches such as cosmid rescue or RNA interference (RNAi) can also be used to help identify the mutation.

Our mapping strategy required high-throughput genotyping for maximum speed and efficiency. Of the several available methods for SNP analysis (Kwok 2001), we selected the FP-TDI assay for its consistent and interpretable results, low assay-setup cost, and automated detection. We performed the FP-TDI assay (Chen et al. 1999) using commercial reagents (AcycloPrime-FP SNP Detection Kit, Perkin Elmer Life Sciences, Inc.) in 384-well format with a standard liquid handling robot (Tecan Genesis, Tecan). Figure 3 shows examples of data obtained using this assay. The separation of dyes that is achieved using control N2 and CB4856 genomic DNA is illustrated in Figure 3A and B and shows the consistent clarity of base-calling data obtained from six randomly chosen chromosome II SNPs on F2 recombinants (crude worm lysates).

View larger version (51K): [in this window] [in a new window]   Figure 3   Sample fluorescence polarization-template directed incorporation (FP-TDI) mapping data. (A) High quality discrimination data from the FP-TDI assay of 48 random SNP markers on control genomic DNA from N2 and CB4856 strains. SNPs are detected through analysis of a single base pair extension from a sequencing primer that hybridizes just adjacent to the polymorphic nucleotide (Chen et al. 1999). Clusters are readily identified and unambiguous. (B) FP-TDI data from eight recombinant samples (crude worm lysates), each tested with six SNP markers. In this case, the FP-TDI kit used distinguishes A from T, the most prevalent SNP change in C. elegans. (C) FP-TDI analysis with the four Tier 2 SNP markers on chromosome III using 35 samples from a mapping cross to localize dpy-18. We localized the recessive gene to a 2.0-Mb subregion of chromosome III. SNP data are converted to table format for interpretation, and the informative recombinants that defined the interval are labeled Left 1, Left 2, and Right 1. (D) The 97-kb interval defined by the Tier 3 markers CE3-194 and CE3-195 is shown. There are 16 predicted genes in this region (AceDB version WS-48); for Tier 4 mapping there are nine predicted substitution SNPs in this interval. RNAi of the predicted gene sequence Y47D3B.10 is known to give a Dumpy (Dpy) visible phenotype (Hill et al. 2000), indicating that this gene is a strong candidate for dpy-18. We confirmed by directed sequencing that dpy-18(e364) did contain a mutation G to A in exon 3 altering the TGG(Trp) codon to TAG(Stop).

This strategy has been applied extensively for identification of novel genes. To date, we have mapped more than 50 loci and cloned >30 genes from several forward genetic screens. We can routinely identify a gene of interest within a 2- to 4-mo time frame. To illustrate this process, Fig. 3C shows results from Tiers 1 and 2 analysis of just 35 DNA samples in the mapping of the cloned gene dpy-18 (Hill et al. 2000) to a 2.0-Mb region of chromosome III. This simple analysis narrows the ROI to only 2% of the genome. Figure 3D shows the 97-kb interval bound by the two validated FP-TDI SNP markers flanking dpy-18. This panel shows the resolution attainable with the set of markers currently available. In addition, information from the RNAi screen of chromosome III (Gonczy et al. 2000) indicates the sequence Y47D3B.10 as being a good candidate gene for dpy-18. We sequenced this e364 allele and confirmed the published mutation in the third exon (Hill et al. 2000), which introduces a premature stop codon into the coding sequence.

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS WEB SITE REFERENCES REFERENCES

We have presented a tripartite, comprehensive strategy for systematic and high-throughput gene identification in C. elegans. This strategy required the development of finely spaced, genome-wide SNP markers and combined an iterative mapping approach with the high-throughput FP-TDI SNP marker assay. We optimized the FP-TDI assay for automated reaction setup and nucleotide analog detection. The FP-TDI assay is highly reliable and allows greater flexibility in selecting which SNPs are assayed, as well as how many samples are genotyped. Our strategy effectively speeds mutation detection and gene cloning in C. elegans, especially when combined with tools for candidate gene analysis such as cosmid rescue and RNAi. Many aspects of our approach are transportable to other model systems and could allow for rapid and systematic gene identification in these systems.

	METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS WEB SITE REFERENCES REFERENCES

Library Construction and Sequencing

Random, genome-wide DNA sequences from the Hawaiian C. elegans strain CB4856 were obtained by constructing a small insert genomic library for shotgun sequencing. Library construction was described previously (Vysotskaia et al. 2001). Double-end sequencing of clones was performed on ABI 3700 (Perkin Elmer) DNA sequencers.

SNP Prediction

The CB4856 sequence traces were aligned against Bristol N2 genomic sequence (C. elegans Sequencing Consortium 1998) using a custom script that takes into account the quality of the neighboring sequence as well as that of the potential polymorphic base (Vysotskaia et al. 2001). Polymorphism information can be found in NCBI dbSNP (accession nos. 4388625-4389689) and GenBank dbSTS (accession nos.: G73810-G74874).

SNP Confirmation

We modified primer3 (http://www-genome.wi.mit.edu/genome_software/other/primer3.html) and designed primers for PCR amplicons ranging between 150 and 300 bases that contain the selected putative SNPs. An initial set of ~100 of the predicted polymorphisms between the Bristol N2 and CB4856 strains were confirmed by sequencing the PCR amplicon from each strain. Sequencing was performed using standard protocols, and products were resolved using capillary electrophoresis on ABI 3700 (Perkin Elmer) instruments.

All 1099 SNPs were also confirmed by FP-TDI (Chen et al. 1999). We used the SNP-kit (AcycloPrime-FP SNP Detection Kit, Perkin Elmer Life Sciences, Inc.) and modified the volumes for compatibility with 384-well PCR. Reactions were set up on the Tecan Genesis 150 robot (Tecan). Briefly, a 200- to 300-bp region of the genome containing the SNP was amplified using standard PCR (6 µL reaction volume). Excess primers and dNTPs were removed by addition of a 6-µL cocktail of shrimp alkaline phosphatase (Roche) and E. coli Exonuclease I (USB) reaction (12 µL final reaction volume). The single base extension reaction (6 µL reaction volume added to above) was performed using the SNP kit components (acyclo dideoxynucleotide triphosphate [ddNTP] terminators are used instead of fluorescently labeled ddNTPs) and a 30-mer oligonucleotide. Addition of the oligonucleotide, complementary to the sequence on one DNA strand immediately 5' of the polymorphic base, allows incorporation of one of the two acyclo terminators in the kit depending on the sequences within the amplified PCR product. Allelic discrimination occurs through measuring the change in fluorescence polarization of the dyes associated with the incorporated nucleotide.

SNP markers, the sequence of all required primers, and standard assay conditions for FP-TDI have been deposited in GenBank and dbSNP and are also available on the Exelixis web site (http://www.exelixis.com/discovery/elegans).

Crossing Strategy

Mapping a recessive mutation created in the Bristol N2 background commences by crossing homozygous mutant (N2-background) animals with wild-type CB4856 animals. The F1 progeny are then segregated away from other progeny and allowed to self-fertilize. The resulting F2 animals are then picked onto 6-cm Petri plates and phenotyped for the mutation. Alternatively, after picking of the F2 animals onto plates, the F2 can be allowed to self-fertilize and lay a brood of F3 animals, and these are then phenotyped. Only F2 animals that are homozygous for the recessive mutation (or homozygous without it) are potentially informative and are genotyped.

To map a dominant mutation, essentially the same procedure is followed except the F1 animals are backcrossed to CB4856 animals, and F2 showing the mapping phenotype (mutant/CB4856) are singled from the resulting outcross progeny.

DNA Sample Preparation

DNA samples for PCR were prepared as described previously (Williams et al. 1992). This procedure usually yields DNA at a concentration of 100 ng/µL. A portion of the population not used for the DNA lysate can be saved for reconfirmation of a phenotype.

	WEB SITE REFERENCES

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS WEB SITE REFERENCES REFERENCES

http://genome.wustl.edu; Washington University, School of Medicine, Genome Sequencing Center.

http://www.exelixis.com; Exelixis, Inc. home page.

http://www-genome.wi.mit.edu; Whitehead Institute Center for Genome Research.