• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. Aug 6, 2002; 99(16): 10575–10580.
Published online Jul 29, 2002. doi:  10.1073/pnas.162136299
PMCID: PMC124979
Genetics

High-resolution SNP mapping by denaturing HPLC

Abstract

With the availability of complete genome sequences, new rapid and reliable strategies for positional cloning become possible. Single-nucleotide polymorphisms (SNPs) permit the mapping of mutations at a resolution not amenable to classical genetics. Here we describe a SNP mapping procedure that relies on resolving polymorphisms by denaturing HPLC without the necessity of determining the nature of the SNPs. With the example of mapping mutations to the Drosophila nicastrin locus, we discuss the benefits of this method, evaluate the frequency of closely linked and potentially misleading second site mutations, and demonstrate the use of denaturing high-performance liquid chromatography to identify mutations in the candidate genes and to fine-map chromosomal breakpoints. Furthermore, we show that recombination events are not uniformly dispersed over the investigated region but rather occur at hot spots.

Keywords: SNP detection, recombination mapping, DHPLC, nicastrin, recombination hot spots

Genome-wide genetic screens in model organisms like Drosophila facilitate the identification of genetic loci involved in biological processes (1). However, the isolation of the affected gene is a tedious process, particularly because the most commonly used mutagen, EMS, primarily induces point mutations. Standard procedures to localize point mutations involve (i) identification of the affected chromosome (arm); (ii) mapping by noncomplementation with chromosomal deficiencies; and (iii) recombination mapping relative to visible markers. In this way, it is generally possible to map mutations to a region of a few hundred kilobase pairs (kb). Limiting factors are the resolution of the meiotic map, the accuracy of determined deficiency breakpoints, and misleading effects caused by second site lethal mutations.

To meiotically map mutations to the level of a single gene, sequence polymorphisms such as single-nucleotide polymorphisms (SNPs), nucleotide insertions, or deletions are exploited as genetic markers. These are widespread, even in regions devoid of phenotypic markers, usually genetically inert, and dominant. An all-molecular approach to recombination mapping is aided by genome-wide SNP maps that have been established for some Drosophila strains (2, 3). However, the resolution of these maps (114–1,000 kb) is in many cases not sufficient to localize the candidate gene and also is not much higher than the resolution attained by classical techniques. Moreover, in contrast to more recently introduced model organisms like Arabidopsis thaliana and Caenorhabditis elegans, for which clearly defined lines exist, the genetic backgrounds of Drosophila strains are often heterogeneous and difficult to trace, such that SNP maps cannot readily be applied to other strains. Furthermore, the complete reliance on sequence polymorphisms requires PCR amplification of the genomic DNA, which becomes a rate-limiting factor (4). A mapping strategy combining visible and molecular markers for recombination mapping can reduce efforts and costs (5).

When a genetic locus has been mapped at low resolution by any of the described classical methods, it is necessary to establish a high-resolution SNP map between mutant and tester strains. For this purpose, sequence polymorphisms have to be identified in the target region, which normally involves the amplification of genomic DNA, DNA sequence determination, identification of polymorphisms, and the subsequent development of a detection assay (4). As the exact nature of the polymorphism is irrelevant for mapping purposes, a method permitting the reliable detection of the mere difference in DNA sequence without determining the actual nature of the difference could save the sequencing and assay development steps.

Possibly the most advanced method for mutation detection without sequencing is denaturing HPLC, which permits the resolution of heteroduplex DNA in PCR fragments of up to ≈1,000 bp in length (6–8). The underlying principle of the technique is a slightly altered melting behavior of heteroduplexes versus homoduplexes, leading to a difference in retention time on ion-pair reversed-phase HPLC columns. On a chromatogram, DNA homoduplexes generally elute in one peak, whereas DNA heteroduplexes produce one or more additional peaks. The sequence difference is thus translated into an altered chromatogram that eliminates the need to obtain information about the change at the DNA sequence level.

Here we demonstrate a recombination mapping procedure combining visual and SNP markers achieving low and high resolution, respectively. SNP mapping was successfully carried out on a denaturing high-performance liquid chromatography (DHPLC)-based map with no information on the nature of the polymorphism. The strategy thus both minimizes the rate-limiting PCR step and drastically reduces the effort associated with SNP detection. This latter aspect is of general interest, because it also applies to any utilization of SNPs, e.g., for mapping of human disease loci. Furthermore, we fine mapped the breakpoints of recombinant chromosomes and found evidence for meiotic hot spots of recombination. The SNP map was also used to determine the breakpoints of a deficiency deriving from an unrelated strain background. Finally, we applied DHPLC for screening of candidate genes and found a surprisingly high rate of second-site mutations.

Materials and Methods

Drosophila Stocks and Generation of Recombinants.

The FRT82B strain containing eyFLP was obtained from B. Dickson (Institute of Molecular Pathology, Vienna) (9) and originates from the Rubin lab (10). Recombinants were generated by the crossing schemes depicted in Figs. Figs.22 and and33 between the mutant FRT82B strain and either a strain carrying the w+ marked EP3455 inserted at 96B1 (11) or a strain harboring a phenotypically neutral y+ marked enhancer promoter (EP) transposable element inserted at 96B20 (S. Breuer, personal communication). The EPy+ element was constructed by D. Nellen (Institut für Molekularbiologie, Universität Zürich, Zürich) and is inserted in a y w strain background. y w flies have been kept in our lab for years and were provided by W. Gelbart (Department of Molecular and Cellular Biology, Harvard University) (12). The rab1 mutation was found in a collection of P-element insertions (13) (F. Rintelen, personal communication) and derives from a y w PlacW strain. It is unclear whether the high incidence of polymorphisms at the investigated region in y w versus y w PlacW flies is because of the accumulation of spontaneous mutations or a different origin of the third chromosome. The rab1737/14 allele, which proved to be cell lethal, was recombined onto a FRT82B w+ chromosome. Crosses were set up at 25°C.

Fig 2.
The crossing scheme to recover recombinants with a distal, y+ marked, and phenotypically neutral EP element. (a) A recombination event between the mutation (asterisk) on the red FRT82B chromosome and the black EPy+ parental chromosome ...
Fig 3.
(a) Crossing scheme (related to Fig. Fig.22b) and SNP-mapping experiment to establish a left border for the critical region by recombination with a w+-marked EP element. Recombinants will be w and y. (b) A polymorphism ...

Screening and Classical Mapping.

y w; FRT82B males were fed with 33 mM EMS according to standard protocols and crossed to females of the genotype y w eyFLP; FRT82B w+ cl3R3.7/TM6B y+. The F1 progeny were scored for eyes and heads of abnormal size. The underlying principle of the clonal expansion of mutant head tissue is the exclusive action of the eyFLP recombinase in head progenitor cells and the unmutagenized FRT82B chromosome carrying the recessive cell lethal mutation cl3R3.7. Two cell populations are generated after eyFLP-induced recombination at the FRT sites: one homozygous for the newly induced mutation and the other homozygous for the cell lethal mutation, which will be eliminated. By using the FRT82B w+ y+ marker chromosome (10), the loci were meiotically mapped relative to the visible markers w+ (at cytological position 90E) and y+ (96E).

DNA Isolation.

Genomic DNA was prepared by homogenizing a single recombinant fly in 40 μl of TE (10 mM Tris/1 mM EDTA, pH 8.0), containing 25 mM NaCl, 0.2% Triton X-100, and 200 μg/μl of proteinase K, incubating the suspension for 30 min at 37°C, and finally inactivating proteinase K for 5 min at 95°C.

PCR.

We searched the Drosophila genome (release 2) (14) for intergenic regions or introns from cytological position 96A13 to 96C2. Primer pairs were designed to amplify fragments at melting temperatures of ≈55°C. PCR reactions contained 1 μl of DNA, 1 μl primer each (10 pmol/μl), 0.2 mM each dNTP, 2 mM MgCl2, 5 μl of 10 × buffer, and 1.25 units of Taq (Sigma). After being melted for 3 min at 95°C, they were cycled 32 times at 94°C for 20 sec, 55°C for 30 sec, 72°C for 1 min, and finally extended for 2 min at 72°C. To allow for efficient heteroduplex formation, the reactions were subsequently heated to 95°C for 30 sec and then slowly cooled to room temperature. However, we obtained comparable results when omitting this final step.

DHPLC.

Analysis was carried out on a Transgenomic (Omaha, NE) HPLC apparatus. For the calibration step, we first analyzed the predicted melting profile of the fragments by the supplied wavemaker (Transgenomic) software to determine the optimal temperatures and gradient conditions to be tested. We then directly loaded PCR reactions derived from homozygous and heterozygous flies and analyzed them at the determined temperatures, injecting 5 μl per run. Calibration was successful when the chromatogram of a heterozygous fragment deviated from the control homozygous fragment. For SNP mapping, amplicons derived from recombinants, homo-, and heterozygous control flies were run at the established conditions.

Polymorphism Density and DHPLC Accuracy.

To test the accuracy of DHPLC, we sequenced 29 heterozygous fragments, 18 of which showed a polymorphism by DHPLC. The exact nature of the polymorphism was determined by “double-peak” analysis. In only one case, sequencing revealed a SNP that was not detected by DHPLC. Therefore, in our hands, DHPLC accuracy is 95% (18/19), which is comparable to numbers achieved in more elaborate studies (15). By amplifying a smaller segment surrounding the unrecognized SNP, the polymorphism could be detected by DHPLC as well (Fig. (Fig.11d, SNP3). In the ≈23.4-kb region tested, the average polymorphism rate between the two parental strains is 1 per 377 bp. More than 80% of the polymorphisms are SNPs, and the others are small insertions or deletions. Interestingly, introns were more than twice as polymorphic as intergenic regions (Table (Table1).1).

Fig 1.
The chromosomal organization of cytological region 96A18 to 96B20 and the outcome of the SNP-mapping experiments. (a) The order of the annotated genes from proximal to distal and the insertion sites of the markers used to generate recombinants are indicated. ...
Table 1.
EPy+ and FRT82B strains are sufficiently polymorphic to aid SNP mapping

Statistical Analysis.

If the recombination breakpoints in the investigated interval between nct and the EPy+ at 96B20 were randomly dispersed, one would expect a Poisson distribution of recombination events. The average recombination ratio in the investigated region of 267.9 kb and 1.01 cM (33 y+ flies among 3,263 “pinheads”) physical and genetic distance, respectively, is 3.775 cM/Mb. The recombination rate per investigated interval in cM per Mb is given in Fig. Fig.5.5. Those values were categorized into arbitrary classes, and the expected number of events was determined according to the Poisson formula P(k,n) = nken/k! (which yields the probability for k events when the mean is n) (Table (Table2).2). χ2 = Σ(yixi)2/xi for those values amounts to 66,744, indicating an extreme deviation from the null hypothesis that the observed distribution is a Poisson distribution (3° of freedom).

Fig 5.
The recombination rate and the distribution of breakpoints between CG7012/nct and the EP element at 96B20 determined for 32 recombinants. The genetic distance between those two markers is 1.0 cM and the physical distance is 0.268 Mb. The map is drawn ...
Table 2.
Actual versus expected frequencies of recombination rates classified into seven arbitrary intervals

Results

A Genetic Locus with a Growth Phenotype Resembling Ras Mutations.

Our gene identification strategy is exemplified by mapping the recessive genetic locus S5A, identified in a F1 mosaic screen for genes affecting cellular growth. Exploiting the eyFLP technique (9), animals were generated that were largely homozygous for a mutagenized FRT82B chromosome in the head but heterozygous for the mutation in the body (Materials and Methods). Mutations affecting cell growth but not differentiation produce flies with smaller (pinheads) or larger heads. The Drosophila Ras gene influences both growth and differentiation, and Ras mutations produce a very small head with scarred eye tissue in this assay (16, 17). The S5A locus comprises four lethal alleles, two of which are Ras-like (Fig. (Fig.22d). A more detailed analysis of the screen and the clonal phenotype of the locus will be described elsewhere.

Low-Resolution Mapping.

Mutations in S5A associated with the pinhead phenotype were placed 2.3 cM proximal to cytological position 96E by recombination mapping with a marker chromosome (Materials and Methods). The lethality associated with the four alleles was not complemented by deficiency Df(3R)96B, which reportedly uncovers 96A21 to 96C2. This cytological interval comprises approximately 440 kb of genomic DNA, thus illustrating that resolution achieved by classical techniques can be competitive to whole-genome SNP maps. We chose two markers situated within the candidate region EPy+ and EP3455 (see below) to generate recombinants for fine-mapping.

Establishing an SNP Map in the 96 A–C Region.

Assuming that noncoding regions exhibit a higher rate of polymorphism than coding DNA, we designed primer pairs to amplify PCR fragments from intergenic regions or from introns greater than 1 kb (Materials and Methods). The fragments chosen for amplification had an approximate size of 800 bp and were spaced at intervals of approximately 17 kb. In a control experiment, we determined by sequencing that 95% of the polymorphisms present in these fragments had been resolved by DHPLC (Materials and Methods and Table Table1).1). We then tested whether PCR products derived from control homozygotes and from flies heterozygous for the parental mutagenized chromosome, FRT82B, and the marker chromosomes EPy+ or EP3455 (Figs. (Figs.113) could be discriminated by DHPLC. Of 40 products amplified from FRT82B/EPy+ heterozygotes, 27 (67.5%) exhibited an altered chromatogram when compared with homozygotic amplicons. The SNP map established for the FRT82B/EPy+ strain combination spanned 594 kb (from 96A14 to 96C2) and had an average resolution of approximately 22 kb (Fig. (Fig.11a). The other heterozygote (FRT82B/EP3455) was not polymorphic enough to permit the construction of a high-resolution map but was nevertheless useful, because a single SNP can provide information about the relative position of marker and mutation (see below).

Mapping of Mutations.

Recombinants between S5A and the markers EPy+ and EP3455 chosen for fine mapping are rare but are efficiently recovered by the crossing schemes depicted in Figs. Figs.22 and and3.3. Two recombinant products can be isolated from the cross with EPy+, which is located distally to the locus (Figs. (Figs.11a and and22a). One product loses both the mutation and the marker and can be selected on the basis of the fact that it is viable over another allele of the same locus (Fig. (Fig.22b). The other product carries both the mutation and the marker and can be detected by the small head assay (Fig. (Fig.22c; see below). Such recombinants can be recovered only when the marker is distal to the gene, thus reducing the critical region to maximally 373 kb (96A21 to 96B20; Fig. Fig.11a).

Recombinants of the first crossing scheme were directly assessed because the SNP map had been calibrated for this strain combination (FRT82B and EPy+). Two predications can be made from the SNP data obtained from individual recombinants. First, the position of the crossing over can be mapped (Fig. (Fig.11b). Second, the relative position of the mutation and the marker can be deduced. A recombinant fly will be homozygous (no SNP detected) on one side of the crossing over and heterozygous (SNP detected) on the other, depending on whether the marker is situated distally or proximally to the gene (compare Figs. Figs.22b and and33b and Fig. 6, which is published as supporting information on the PNAS web site, www.pnas.org). Thirty-two recombinant chromosomes that lost the mutation and the EPy+ marker (Fig. (Fig.22a) were recovered and analyzed by DHPLC (Materials and Methods). The most proximal recombination event (i.e., the recombination closest to the gene) was localized between two SNPs at 96B4–B5 (Fig. (Fig.11 a and b), and the critical region was thus reduced by 325 kb. Furthermore, they exhibited a homozygous profile distally and a heterozygous profile proximally, indicating that the EPy+ transposon is located distally to the locus (Figs. (Figs.11b and 6).

The applied crossing scheme depends on the following prerequisites: First, the mutation must be homozygous lethal or exhibit an easily scorable phenotype. Second, at least two alleles are required, because lethal second site mutations could be mapped instead. Therefore, recombinant chromosomes of single or nonlethal alleles should be isolated by the alternative scheme selecting for the pinhead and marker phenotypes. However, this approach poses the problem of introducing the third strain background of a FRT82B chromosome carrying a cell lethal rab1 mutation that permits the expansion of the mutant tissue (Fig. (Fig.22c). For this third chromosome, the SNP map has not been calibrated. However, because SNPs are codominant, the calibrated sequence difference between FRT82B and EPy+ must unambiguously discriminate between rab1/FRT82B or rab1/EPy+ heterozygotes, respectively. Thus, we recalibrated the local SNP map for the third strain background by assessing the profiles of the two new heterozygous combinations. Now the breakpoints in the recombinants could be directly mapped, because they are characterized by a switch from one heterozygous profile to the other (Fig. (Fig.11c). The most proximal recombinant recovered from this analysis was located only 4 kb apart from the S5A locus (Fig. (Fig.11 a and c).

Although the density of polymorphisms in FRT82B/EP3455 heterozygotes is low, the relative position of the EP3455 marker and the mutation could be determined. We found a polymorphism between the two parental strains directly distal to the transposon insertion site (SNP7, Figs. Figs.11a and and3).3). A recombinant between the mutant and EP3455 exhibited a heterozygous profile, indicating that EP3455 is situated proximally to the gene (Fig. (Fig.33b and Fig. 6). Therefore, the insertion point sets the left border of the candidate region, which is thus reduced by another 64 kb and contains only 10 genes in a 45-kb interval (Fig. (Fig.11a).

Candidate Gene Approach.

Three of the 10 genes in the critical region, CG7012, CG13638, and CG10951, were considered primary candidates on the basis of the function of homologs in other species. As EMS, the mutagen used, primarily induces point mutations (18), we crossed the mutant chromosomes back to the parental FRT82B strain, amplified the coding regions from these heterozygotes, and searched for sequence changes by DHPLC. CG7012/nct, the Drosophila homolog of nicastrin (19), exhibited a DHPLC-resolvable polymorphism in all four alleles. By subsequent sequence and genetic analysis, it was shown that these polymorphisms are responsible for the phenotype (Fig. (Fig.44a and data not shown). On two of the four chromosomes, we also found mutations in CG10951, coding for a nimA-like kinase (niki) with a C terminus homologous to the RCC1 Ran guanine exchange factor (Fig. (Fig.44b). The lethality of the niki heteroallelic combination could not be rescued by either a tub-niki or a niki genomic rescue construct, suggesting that the lethality is not associated with the niki mutations but is because of the closely linked nct mutations on the same chromosome. The nct phenotype is unaffected by the second site mutations in niki (data not shown).

Fig 4.
Mutations of two nicastrin alleles and of CG10951/niki. (a) The domain organization of nicastrin and the DHPLC profiles of FRT82B control homozygotes and nct6F2/FRT82 and nct5P3/FRT82 heterozygotes, respectively. SP, signal peptide (hatched); DYIGS, conserved ...

These data show that chemically induced mutations can be efficiently scanned by DHPLC. Furthermore, they suggest that closely linked second site mutations are more common than previously appreciated.

Mapping the Breakpoints of a Deficiency.

During this study, we realized that one of the breakpoints of deficiency Df(3R)96B has been mapped inaccurately. To determine the breakpoints at high resolution, we made use of our local high-density SNP map. However, in the unrelated Df(3R)96B/TM6B strain background, this map is not applicable. We therefore crossed the deficiency chromosome into the parental backgrounds of EPy+ and FRT82B, respectively. Because SNPs are codominant, the calibrated sequence difference must appear in either of the two heterozygotes. The hemizygous region uncovered by the deficiency, on the other hand, retains a homozygous profile in both combinations (Fig. (Fig.11d). In this way, we could place one chromosomal breakpoint between tok and tld at 96A22 and the other between CG6031 and CG13646 or between 96B8 and B11. The right border is shifted toward 96B10, because the deficiency complements a lethal P-element insertion in CG6238/slingshot (Fig. (Fig.11d).

Gene-Poor Regions at 96B Are Recombination Hot Spots.

We were interested in whether recombination events are randomly dispersed, because the presence of hot spots would limit the resolution of SNP mapping. The sites of recombination were determined in all 32 recombinants, revealing that 22 events are located in two major intervals. These cover a region of 76 kb, whereas the other events occurred in the remaining 192 kb (Fig. (Fig.5).5). At the given resolution, the difference between meiotically inert and active intervals is up to 20-fold. The skewed distribution of crossing-over events deviates significantly from a Poisson distribution expected for a random distribution of recombination (see Materials and Methods, statistical analysis). Plotting the recombination rate against the number of genes in the intervals (Fig. (Fig.5)5) reveals a negative correlation of recombination events and gene density. Presently, the scarcity of data and the relatively low resolution of the SNP map do not permit the distinction of whether it is the low gene-density per se or another feature within that gene-poor region that makes those sites more recombinogenic.

Discussion

The mapping strategy described here combines recombination mapping using visible marker for low-resolution and SNP markers for high resolution to locate a mutation to a genomic region of a few kilobase pairs. This “mixed method” is akin to the principle described by Martin et al. (5), who argue that the rate-limiting SNP detection reactions are reduced at least by a factor of four. Aside from the SNP assays, the difference between the two strategies is that Martin et al. describe a one-step process resulting in medium resolution, whereas our method is divided into a low- and high-resolution step. The two-step strategy has also been applied by Berger et al. (2), who followed an all-molecular sequence-based approach. Our protocol guarantees maximal resolution with a minimal number of recombinants, because choosing closely linked markers provides a recombination event in the region of interest. The SNP map is then calibrated for the same strain combination, so that recombinants can be directly analyzed. This procedure allows saving one generation compared with the strategy of crossing recombinant chromosomes to standard SNP chromosomes (3).

Here we show that “SNP mapping without knowing the SNP” by merely resolving sequence differences between two strains is an exact and very efficient mapping strategy. Three advantages of this method can be put forward. First, calibrating SNPs by DHPLC is very accurate. Ninety-five percent of the SNPs determined by sequencing in a 23.4-kb region were detected by DHPLC. Second, the method is semiautomatic. Third, and most important, once a SNP has been established, there is no need to further develop SNP detection assays. SNP detection by DHPLC has been demonstrated previously (8, 20), and DHPLC has been used for SNP mapping recently (3, 21), but in these cases it was introduced as an assay system to detect an already established sequence-verified SNP.

Screening candidate genes in the critical region by DHPLC [also shown by Cargill et al. (22)] for EMS-induced point mutations led to the surprising discovery that two of the four nct alleles carry a second site mutation in a closely linked gene, which we termed niki. These mutations are not responsible for the growth phenotype. The mutation rate for high doses of EMS (50 mM) has been calculated by Bentley et al. (23) to be 1/209 kb. The chromosomes investigated in this study were mutagenized with 33 mM EMS. Therefore niki, a gene with a 2-kb coding region, has a less than 1% likelihood of being affected and a less than 1 in 10,000 chance of being hit twice. Pastink et al. (18) found two double substitutions among 52 EMS-induced mutations in the vermilion locus. Similarly, with the related mutagen MMS, we induced two point mutations within a single yeast gene (24). It is thus conceivable that a chemically induced mutation favors the occurrence of second site mutations in the vicinity. Such closely linked mutations may pose serious problems when one is trying to map single alleles, because it cannot be taken for granted that even in small candidate regions an identified mutation is causative for the phenotype. Generating recombinants by selecting for the phenotype (Fig. (Fig.22c) rather than against the associated lethality (Fig. (Fig.22b) can reduce, but not eliminate, this risk.

A potential limitation of gene mapping by recombination is the nonrandomness of recombination events. In the course of mapping mutations to the nct locus, we found that the distribution of crossover sites between nct and EPy+ in the 32 recombinants was nonrandom. Two recombination hot spots (responsible for 0.44 and 0.19 cM, respectively) were identified. These hot spots fall within regions of low gene density. Comparatively few recombination events were identified in gene-rich regions. Collecting similar high-resolution data throughout the genome could help in addressing several interesting issues: Does the observed negative correlation of gene density and meiotic activity apply as a general rule, or is it valid only for the 96B region? Is the recombination rate within such an interval uniform, or can it be localized to a single site? In other words, is it the low gene density per se that renders a region meiotically active, or can the apparent hot spot be attributed to another feature of the chromosome, for example a promoter? In yeast, meiotic hot spots generally coincide with promoters of transcriptionally active genes (25), whereas human hot spots may reside in promoters, inter-, and intragenic regions (26). In the 96B region, the CG13647 promoter could be the primary site of recombination, because it contains SNP16, which is located in the center of 14 recombination events (Figs. (Figs.11a and and5).5). The high incidence of polymorphisms (an average of 1 in 377) and the high recombination rate (0.44 cM between SNP15 and SNP17) should allow the generation of an ultrahigh-density SNP map to test this prediction.

Supplementary Material

Supporting Information:

Acknowledgments

We are grateful to Felix Rintelen for the rab1 mutation, to Sebastian Breuer and Denise Nellen for the EP line, to Rita Bopp for selecting recombinants, to Barbara Flückiger for assisting the DHPLC, to Olaf Nairz and Roman Kaelin for statistical advice, and to Michael Spörri and Giancarlo Tomio for sequencing. We also acknowledge Peter Gallant and Sean Oldham for critically reading the manuscript. K.N. was supported by fellowships from the Human Frontiers Science Program, European Molecular Biology Organization, and a European Union-sponsored Training and Mobility of Researchers grant (to E.H.).

Abbreviations

  • EMS, ethyl methanesulphonate
  • DHPLC, denaturing high-performance liquid chromatography
  • SNP, single-nucleotide polymorphism
  • EP, enhancer promoter element

Notes

This paper was submitted directly (Track II) to the PNAS office.

References

1. St Johnston D. (2002) Nat. Rev. Genet. 3, 176-188. [PubMed]
2. Berger J., Suzuki, T., Senti, K. A., Stubbs, J., Schaffner, G. & Dickson, B. J. (2001) Nat. Genet. 29, 475-481. [PubMed]
3. Hoskins R. A., Phan, A. C., Naeemuddin, M., Mapa, F. A., Ruddy, D. A., Ryan, J. J., Young, L. M., Wells, T., Kopczynski, C. & Ellis, M. C. (2001) Genome Res. 11, 1100-1113. [PMC free article] [PubMed]
4. Syvanen A. C. (2001) Nat. Rev. Genet. 2, 930-942. [PubMed]
5. Martin S. G., Dobi, K. C. & Johnston, D. S. (2001) Genome Biol. 2, 1-12. [PMC free article] [PubMed]
6. Underhill P. A., Jin, L., Zemans, R., Oefner, P. J. & Cavalli-Sforza, L. L. (1996) Proc. Natl. Acad. Sci. USA 93, 196-200. [PMC free article] [PubMed]
7. Oefner P. J. (2000) J. Chromatogr. B Biomed. Sci. Appl. 739, 345-355. [PubMed]
8. Wolford J. K., Blunt, D., Ballecer, C. & Prochazka, M. (2000) Hum. Genet. 107, 483-487. [PubMed]
9. Newsome T. P., Asling, B. & Dickson, B. J. (2000) Development (Cambridge, U.K.) 127, 851-860. [PubMed]
10. Xu T. & Rubin, G. M. (1993) Development (Cambridge, U.K.) 117, 1223-1237. [PubMed]
11. Rorth P., Szabo, K., Bailey, A., Laverty, T., Rehm, J., Rubin, G. M., Weigmann, K., Milan, M., Benes, V., Ansorge, W. & Cohen, S. M. (1998) Development (Cambridge, U.K.) 125, 1049-1057. [PubMed]
12. Smith D., Wohlgemuth, J., Calvi, B. R., Franklin, I. & Gelbart, W. M. (1993) Genetics 135, 1063-1076. [PMC free article] [PubMed]
13. Deak P., Omar, M. M., Saunders, R. D., Pal, M., Komonyi, O., Szidonya, J., Maroy, P., Zhang, Y., Ashburner, M., Benos, P., et al. (1997) Genetics 147, 1697-1722. [PMC free article] [PubMed]
14. Adams M. D., Celniker, S. E., Holt, R. A., Evans, C. A., Gocayne, J. D., Amanatides, P. G., Scherer, S. E., Li, P. W., Hoskins, R. A., Galle, R. F., et al. (2000) Science 287, 2185-2195. [PubMed]
15. Spiegelman J. I., Mindrinos, M. N. & Oefner, P. J. (2000) Biotechniques 29, 1084-1092. [PubMed]
16. Prober D. A. & Edgar, B. A. (2000) Cell 100, 435-446. [PubMed]
17. Halfar K., Rommel, C., Stocker, H. & Hafen, E. (2001) Development (Cambridge, U.K.) 128, 1687-1696. [PubMed]
18. Pastink A., Heemskerk, E., Nivard, M. J., van Vliet, C. J. & Vogel, E. W. (1991) Mol. Gen. Genet. 229, 213-218. [PubMed]
19. Yu G., Nishimura, M., Arawaka, S., Levitan, D., Zhang, L., Tandon, A., Song, Y. Q., Rogaeva, E., Chen, F., Kawarai, T., et al. (2000) Nature (London) 407, 48-54. [PubMed]
20. Giordano M., Oefner, P. J., Underhill, P. A., Cavalli Sforza, L. L., Tosi, R. & Richiardi, P. M. (1999) Genomics 56, 247-253. [PubMed]
21. Spiegelman J. I., Mindrinos, M. N., Fankhauser, C., Richards, D., Lutes, J., Chory, J. & Oefner, P. J. (2000) Plant Cell 12, 2485-2498. [PMC free article] [PubMed]
22. Cargill M., Altshuler, D., Ireland, J., Sklar, P., Ardlie, K., Patil, N., Shaw, N., Lane, C. R., Lim, E. P., Kalyanaraman, N., et al. (1999) Nat. Genet. 22, 231-238. [PubMed]
23. Bentley A., MacLennan, B., Calvo, J. & Dearolf, C. R. (2000) Genetics 156, 1169-1173. [PMC free article] [PubMed]
24. Nairz K. & Klein, F. (1997) Genes Dev. 11, 2272-2290. [PMC free article] [PubMed]
25. Nicolas A. (1998) Proc. Natl. Acad. Sci. USA 95, 87-89. [PMC free article] [PubMed]
26. Jeffreys A. J., Kauppi, L. & Neumann, R. (2001) Nat. Genet. 29, 217-222. [PubMed]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...