Nature Protocols homepage
Advanced search

Protocol

Nature Protocols 3, 1751 - 1765 (2008)
Published online: 23 October 2008 | doi:10.1038/nprot.2008.175

Subject Categories: Genetic analysis | Model organisms | Nucleic acid based molecular biology

Positional cloning by fast-track SNP-mapping in Drosophila melanogaster

Frank Schnorrer1,2,5, Annika Ahlford3,5, Doris Chen1,4,5, Lili Milani3 & Ann-Christine Syvänen3

Positional cloning of chemically induced mutations is the rate-limiting step in forward genetic screens in Drosophila. Single-nucleotide polymorphisms (SNPs) are useful markers to locate a mutated region in the genome. Here, we provide a protocol for high-throughput, high-resolution SNP mapping that enables rapid and cost-effective positional cloning in Drosophila. In stage 1 of the protocol, we use highly multiplexed tag-array mini-sequencing assays to map mutations to an interval of 1–2 Mb. In these assays, SNPs are genotyped by primer extension using fluorescently labeled dideoxy-nucleotides. Fluorescent primers are captured and detected on a microarray. In stage 2, we selectively isolate recombinants within the identified 1–2 Mb interval for fine mapping of mutations to about 50 kb. We have previously demonstrated the applicability of this protocol by mapping 14 muscle morphogenesis mutants within 4 months, which represents a significant acceleration compared with other commonly used mapping strategies that may take years.

Top

Introduction

Over the past few decades, forward genetic screens have proven very useful for identifying the key regulators of complex biological processes in many model organisms such as Caenorhabditis elegans, Drosophila or zebrafish1, 2, 3. Until the introduction of SNPs as markers for gene mapping in model organisms4, 5, 6, positional cloning of chemically induced mutations was a laborious and expensive task. SNPs are ideal as genetic markers because they are mostly phenotypically neutral and occur at high density in the genomes of most organisms. Two early studies created SNP maps for Drosophila and used traditional methods, such as denaturing high-performance liquid chromatography7 or PCR fragment-length polymorphisms and restriction fragment-length polymorphisms5 for genotyping SNP markers for gene mapping at relatively low resolution. Both of these methods require size separation of PCR-amplified DNA fragments by gel electrophoresis, and are therefore not easily applicable to multiplexed genotyping of high-density SNP maps. Furthermore, all of the above-mentioned procedures rely on length polymorphisms of suitable size or on differentiating restriction sites and are therefore only applicable to a small subset of SNPs. More recently, microarray-based methods have become available and now offer the possibility of highly multiplexed genotyping in model organisms, as shown for SNP mapping in zebrafish6.

In this protocol, we describe recently developed and validated resources that integrate a high-density SNP map with an efficient microarray-based mini-sequencing (TAMS) method for multiplexed SNP genotyping and a software tool for SNP scoring. These resources constitute a complete package for accelerated gene mapping in Drosophila 8. This package includes a high-resolution SNP map for common laboratory strains of Drosophila with 2,238 markers at an average SNP density of 50.3 kb, which corresponds to about 6.3 genes (Table 1). The map is publicly available (http://flysnp.imp.ac.at/flysnpdb.php) and is an invaluable resource for the community as it eliminates the need for DNA sequencing for SNP discovery in Drosophila. This resource allows refinement of the region harboring the mutation to a few genes, thus making the identification of mutations by DNA sequencing an easy task.


The TAMS system combines genotyping by highly specific mini-sequencing (single-base primer extension) reactions with the advantages of the microarray format. Compared with standard microarray-based methods, where a single sample is hybridized on each microarray slide6, the 'array-of-arrays' format of TAMS provides much higher sample throughput. The 'array-of-arrays' format allows either 80 samples to be analyzed for up to 196 SNPs or, alternatively, 16 samples to be analyzed for 576 SNPs on a single microscope slide (see Fig. 1b). Genotyping can be performed at a reasonable cost of only 1.6 US cents per SNP genotype. The tag-microarray format of TAMS makes the design of the system flexible because the same arrays can be used for any panel of SNPs. Therefore, the tag-array format renders the TAMS system particularly useful for genotyping in model organisms lacking predesigned assays. The only rate-limiting step in the TAMS assay is a multiplex PCR step, which requires careful design and optimization before the actual genotyping reaction.

Figure 1: TAMS assay.
Figure 1 : TAMS assay.

(a) Components used for the hybridization of TAMS products. One silicon rubber grid with 80 reaction wells and a second with 16 wells are shown in the front. A plexiglas cover for three microarray slides with 80 reaction wells each is shown on the right of an aluminium rack. The plexiglas cover has drilled holes matching the silicon rubber wells. The screws are used to tighten the assembly (see Experimental design for assembling details). (b) An 'array-of-arrays' format is used to analyze multiple SNPs in many samples in parallel. The two different configurations of the system make it flexible with the possibility to easily change the numbers of SNPs and samples to be interrogated (196 SNPs in 80 samples or 576 SNPs in 96 samples). The dimensions for both formats used for array spotting are indicated, e.g., the diameters of the spots and the silicon grid chambers as well as the center-to-center distances between two spots and two subarrays. The spot measures may vary slightly depending on the needle used for spotting. (c) Four color mini-sequencing in solution in which primers hybridize next to each SNP site and are extended with fluorescently labeled terminating nucleotide analogs (ddNTPs). The primers carry 5'-tag sequences that enable them to hybridize to complementary sequences immobilized on a microarray to determine homozygous and heterozygous SNPs. (d) On the left is an example of one subarray scanned at the wavelength visualizing incorporated ddCTP nucleotides labeled with TAMRA fluorophore. Enlargement of the mapping result for C233, an eyes absent allele, is shown on the right. The entire mapping result is depicted in Figure 5. Black bars indicate the determined genetic interval from the stage 1 mapping.

Full size image (95 KB)

Other tag-array formats have also been utilized in genotyping methods on the basis of ligation of allele-specific probes9, 10. In these methods, binding sites for a pair of universal PCR primers for each SNP are introduced through the ligation probes specific for each SNP. These primer-binding sites facilitate amplification of the ligated probes to detectable levels after the genotyping reaction using fluorescently labeled universal PCR primers, and therefore a PCR step is not required before genotyping. Ligation-based approaches are also a promising option for mapping in Drosophila 11. A disadvantage of the allele-specific ligation methods is the need for two long oligonucleotide probes for each SNP and fluorescently labeled PCR primers. Therefore, TAMS assays are less expensive and easier to design and implement for multiplexed genotyping than ligation-based methods. In our recently performed mapping study, the TAMS procedure performed reliably for all genes that we attempted to map8.

Our mapping strategy avoids several of the pitfalls in classical alternative mapping approaches. For example, deficiency-based mapping by lethality scoring may lead to an incorrect interval for the position of the mutation if multiple hits are located on the same chromosome. In deficiency mapping, no result will be obtained when the mutation is located outside of the available deficiencies or if the deficiency break points are not correctly defined. The two currently available deficiency collections with molecularly defined break points have a coverage of 56% and 77%, with an average resolution of 140 and 377 kb, respectively12, 13, 14. Our method has limitations if an inversion is located close to the region of interest on the chromosome, as this inversion will prevent the recovery of relevant recombinants. Furthermore, the reliable success of our '2EP' approach in stage 2 (see below) requires that no other (white +) transgene is in the stock used for mapping.

For effective high-resolution mapping, we apply a two-stage mapping strategy (Fig. 2). In stage 1, SNP genotyping of recombinants is performed by TAMS assays, which involves the incorporation of a fluorescently labeled nucleotide at each SNP site by mini-sequencing, followed by microarray-based detection of the incorporated nucleotides (see Fig. 1)15, 16. The TAMS assay enables highly multiplexed genotyping at a very low cost. We validated TAMS assays for 293 SNPs, evenly distributed across the three major chromosomes of Drosophila, resulting in an average distance of 391.3 kb between two SNPs8. We developed the software tool SNPmapper for fast allele calling and graphical representation of the mapping result. This software is freely available upon request.

Figure 2: Overview of mapping procedure.
Figure 2 : Overview of mapping procedure.

The step numbers correspond to the steps in the procedure section of this protocol. (a) Schematic representation of the two mapping stages. The parental chromosome bearing the mutation is shown in red, the reference chromosomes are shown in blue and light blue (2EP). The mutation is indicated by an asterisk. (b) Crossing scheme for mapping the muscle mutants. In stage 1, tag-array mini-sequencing was used to achieve a 1–2 Mb resolution, which is refined to about 50 kb in stage 2 by the selection of recombination events between two visible markers (2EP). The estimated amount of animals required is indicated. See text for details.

Full size image (88 KB)

In stage 2, we apply high-resolution mapping to refine the region containing the mutation to about 50 kb (Fig. 2). The bottleneck for this high resolution is the generation of a sufficient number of recombinants harboring the recombination break point within the genomic region of interest. To solve this problem, we generated a set of 62 '2EP' stocks with two closely linked EP elements (in cis), which are inserted at a distance of 0.5–2.2 Mb and cover four of the five major chromosome arms at excellent resolution. The recombinants can be easily identified by segregation of the two EP elements that leads to a paler eye color. Consequently, one or two of these 2EP stocks, which cover the mutated region identified in stage 1, can be selected to facilitate fine mapping in stage 2 of the protocol.

Experimental overview

The objective of the protocol presented here is to refine the map position of novel Drosophila mutations from the chromosome level of 20–50 Mb to approx50 kb, which then enables the identification of the mutation by sequencing. Figure 2a gives a schematic overview of the entire mapping procedure, and Figure 2b displays the crossing scheme previously applied by us to map mutations that affect muscle morphogenesis in Drosophila embryos8. In stage 1, the chromosome bearing the mutation of interest is crossed with a polymorphic reference strain. A number of useful polymorphic strains that we have characterized are listed in Table 1. About 100 random recombinant stocks are generated for each gene to be mapped, followed by determination of the phenotype of every recombinant (Figs. 2 and 3). For rapid identification of the phenotype of interest, it is convenient to include a phenotypically relevant green fluorescent protein (GFP) marker on a different chromosome to the mutation and to use fluorescently labeled balancer chromosomes to quickly identify the homozygous animals. Both GFP markers should be easily identifiable in the same animal at the relevant time point. In our example, we used a Mhc-tauGFP reporter, which labels all muscles in the embryo, and scored the phenotype in living embryos homozygous for the recombinant chromosome without removing the chorion (Fig. 3). With this approach, it is feasible to determine the phenotypes of 100 stocks within 3–5 h. DNA from each recombinant stock is then isolated and subjected to genotyping by TAMS at about 60–120 SNP loci, which are equally distributed over the chromosome of interest.

Figure 3: Embryo sample preparation and phenotyping.
Figure 3 : Embryo sample preparation and phenotyping.

(a) Microscope slide with three different samples (e.g., samples 1–3 from block 1). (b) Low-magnification epifluorescent image of embryos containing the balancer (CyO, Kr-GAL4, UAS-GFP) marked by a red arrow (note the GFP expression in Bolwigs organ) and a homozygous recombinant embryo (light blue arrow). All muscles are labeled by Mhc-tauGFP expression. (c) High-magnification ventrolateral view of one segment of a wild-type recombinant. (d) Similar view of a kon-tiki mutant embryo20. Note the round muscles in the kon-tiki embryo in contrast to wild type in which muscles span the entire segment (red arrows). These images were taken without removing the chorion.

Full size image (41 KB)

For each DNA sample, the regions spanning the SNPs are amplified in a few multiplex PCRs, after which the PCR products are pooled and subjected to genotyping by mini-sequencing (Fig. 1c). The mini-sequencing primers are designed to hybridize immediately adjacent to each SNP site. In addition to the Drosophila-specific sequence, each primer carries a unique 5'-'tag-sequence' for capture and identification on a microarray. The multiplex mini-sequencing reactions are performed simultaneously for all SNPs in a single tube, whereby each primer becomes extended by one of four differentially labeled dideoxy-nucleotides that is complementary to the nucleotide at the SNP site. The extended primers are then hybridized to complementary 'anti-tag' oligonucleotides that have been immobilized on a microarray glass slide. The use of generic tag-sequences enables universal, non-SNP-specific array design. We use an 'array of arrays' format to facilitate genotyping of all SNPs for multiple samples on one microarray slide (Fig. 1b). This is enabled by a silicon rubber grid placed on the microarray slide to create an individual reaction chamber for each sample (Fig. 1a). In our predesigned and validated assays, up to 120 SNPs per sample are genotyped in 80 samples in parallel on each slide to yield up to 9,600 genotypes per slide. The genotyping cost is about $150 per slide, including the PCR. One person can process up to six microarrays per day.

The fluorescent signals at four wavelengths, corresponding to the four nucleotides, are measured from the microarray surface at positions defined by the generic 'anti-tags', quantified using the QuantArray software and subsequently processed by the software tool SNPmapper specifically developed for this purpose. This program assigns the genotypes by k-median clustering of the signals and by incorporation of the parental genotype data. The user can specify the method for reduction of background noise as well as the minimal number of called SNP genotypes per sample. The color-coded, graphical output shows the recombinant chromosomes sorted by their phenotype and break points, which facilitates visual identification of the mutated region (Fig. 1d)8. The typical resolution after this first mapping step is about 1 cM, which in most cases corresponds to a physical distance of 1–2 Mb.

In stage 2, 100–200 recombinants are generated between the mutant chromosome of interest and a 2EP chromosome, which covers the region determined in stage 1 (Fig. 2). The rare recombination events between the two EP insertions are identified by the paler eye color of the flies, which results from the separation of the two EP elements. Phenotyping and genotyping of the recombinant stocks in stage 2 leads to a mapping resolution at the limit of the SNP map, which is about 50 kb. For genotyping, we used conventional sequencing in our feasibility study, but we have also precomputed a set of 1,417 TAMS assay primers that can be used for fine-mapping purposes8.

Experimental design

Time management.

Stages 1 and 2 are relatively independent of each other, as none of the recombinant fly stocks generated in stage 1 are required during stage 2. Thus, it is up to the experimenter to first map 20 mutants using stage 1 before moving with the first one into stage 2. For stage one, 6–12 mutants can be mapped by one or two experienced researchers at the same time.

For stage 2, one full-time person is able to process the mapping of 4–8 mutants at the same time. If the SNP detection is done with classical sequencing, as we applied here, mapping of more than four genes at the same time requires a fairly high sequencing capacity.

Selecting reference strains.

It is important that an appropriate polymorphic reference strain is selected. All strains from which we determined SNPs (Table 1) are on average 74.7% polymorphic to each other8 and thus well suited for this procedure. In our case, the mutations were located on an ORiso second chromosome and we selected a CSiso second chromosome as a reference. ORiso2 and CSiso2 show an average SNP frequency of 86.6%.

Optimizing heat shocks.

Before starting the mapping experiments, the timing of the heat shock to eliminate hs-hid chromosomes in Step 6 should be fine-tuned. Timing of the heat shock depends on the size of vials and the hs-hid chromosome used. Too long a heat shock will kill all progeny; too brief a heat pulse may result in a contamination of the stock by surviving hs-hid animals. Generally, 45–60 min is good for vials but, to be safe, absence of the hs-hid chromosome should be visually checked in a few vials.

Designing oligonucleotides.

The design of the oligonucleotides used in TAMS is described in Table 2. A set of validated primers for PCR and mini-sequencing as well as anti-tag sequences are available for 293 SNPs covering the three major chromosomes. Additional oligonucleotides have been precomputed for fine-mapping of 2EP regions. For sequence information of all ready-made primers, see http://flysnp.imp.ac.at/index.php.


Optimizing multiplex PCR.

The most important part of PCR optimization is the primer design. By careful bioinformatics-assisted prediction of interactions, secondary structures and melting temperatures of the primers, tedious experimental testing is significantly reduced (see Table 2). During experimental testing and optimization of the performance of multiplex PCRs, a number of parameters may be varied, for example, reaction times and temperatures as well as the concentrations of reagents. We recommend increasing the concentration of poorly performing primer pairs and using a DNA polymerase enzyme recommended for multiplex reactions. Each primer pair may also be run in an individual reaction first, followed by pooling primer pairs that result in products of good quality. However, for a large number of PCRs, this approach is not practical. In our experience, it is difficult to fully predict the success of a multiplex amplification on the basis of the individual reactions. Therefore, we applied a 'brute-force' approach for establishing the verified TAMS assays, using an excess of primer pairs to later select those that performed best (assay conversion rate of 70%).

Controls for TAMS.

The inclusion of controls in each step of the TAMS procedure enables verification of the success of each of the biochemical reactions. Printed microarrays of poor quality, e.g., with missing spots, donut-shaped spots, too small spots or large fusion spots, can be avoided by the inclusion of two print controls in each production batch. Print control I is a fluorescently labeled anti-tag oligonucleotide that is included at one position in each subarray during the spotting procedure. Print control II is a fluorescently labeled degenerate oligonucleotide, complementary to all of the anti-tag oligonucleotides. Control II can be hybridized for 5 min to a subset of the subarrays in each batch of printed microarray slides at 300 nM concentration in 6.25 times SSC with subsequent washing and scanning. Possible failures of the mini-sequencing reactions can be controlled for by including four 45-mer oligonucleotide templates, each differing at the twenty-second nucleotide position, at a final concentration of 1 nM in the mini-sequencing reaction master-mixture to mimic a 4-allelic SNP. A mini-sequencing primer, complementary to the region adjacent to the variable nucleotide position in the 4-allelic template, is added to the mini-sequencing primer mixture at a final concentration of 10 nM. To control successful hybridization of the mini-sequencing products to the microarrays, a hybridization control can be included. This control consists of a fluorescently labeled oligonucleotide for which the corresponding anti-tag oligonucleotide has been spotted in each subarray. The hybridization control is added to the hybridization mixture at 0.25 nM concentration and is included in the mixture only for every second subarray to also control for leakage between the reaction chambers or wells of the silicon grid. Possible leakage will be observed by a signal from the hybridization control in wells to which it was not added. Additionally, DNA from a wild-type reference strain (as a positive control) and water (as a negative control) for possible PCR contaminations should be included in the PCR and carried through the entire procedure.

Assembling hybridization rack.

To create 16 or 80 separate reaction chambers, place a silicon rubber grid (prepared as described in EQUIPMENT SETUP) over the arrayed slide according to the diamond-pen markings. The side of the grid with the larger diameter of cone-shaped reaction chambers is placed toward the spotted array surface. Place the arrayed slides covered with the silicon grid into the custom-made aluminum reaction rack (Fig. 1a). Three microarray slides can be hybridized simultaneously in each rack. It is important to keep track of the orientation of each slide to later match the scanned signals with the right SNP and sample. Place the polymethyl methacrylate (PMMA or 'plexiglas') cover over the rack and tighten the whole assembly with six screws. This is a critical step in the assembly to avoid leakage without breaking the glass slide.

Top

Materials

Reagents

Equipment

Reagent setup

  • Preparation of 88-mm yeasted apple agar plates for 18-tube blocks
    • For agar plates, autoclave 17.5 g of Bacto Agar and 25 g of sucrose in 750 ml of water, add 250 ml of apple juice and 10 ml of 15% (wt/vol) nipagin (in ethanol). Pour this hot medium into the 88-mm plates and store plates at 4 °C until use. Prepare liquid yeast paste from dry yeast with the appropriate amount of water. Use a custom-made form to stamp the yeast paste onto the apple agar plate to ensure the correct spacing (Fig. 4 and ref. 18).
      Figure 4: Eighteen-tube block handling.
      Figure 4 : Eighteen-tube block handling.

      (a) On the left, custom-made stamp used to stamp yeast paste on an apple agar plate. On the right, a 'yeasted' apple agar plate ready to use. Note the regular format. (b) On the left, empty 18-tube block viewed from the top. On the right, apple agar plate was placed on the 18-tube block and fixed by two rubber strings. The vial positions are indicated. (c) On the left, a filled 18-tube block with flies in each of the tubes. On the right, changed agar plate with embryos at all 18 positions ready for phenotype scoring. Numbers indicate the respective positions, now mirror-imaged compared with b.

      Full size image (39 KB)

  • Preparation of multiplex primer mixes
    • The PCR primers are pooled in panels for the chromosome arm of interest, with each primer at a final concentration of 5 muM concentration. All tagged mini-sequencing primers for the chromosome of interest are pooled at a final concentration of 100 nM.

Equipment setup

  • 18-tube blocks
    • Prepare custom-made 18-tube blocks from 18, single, 10-cm-long, 1.5-cm-wide polystyrene tubes. Glue the tubes together (with a few drops of acetone applied from the top) in a 5–4–5–4 asymmetric geometry. Drill four 0.5–mm-wide holes for oxygen exchange into the closed bottom of each tube. Each tube has assigned positions from 1 in the front left corner to 18 in the back right corner (Fig. 4, and for a detailed construction protocol, see ref. 18).
  • Filling of the 18-tube blocks
    • Anesthetize flies from stock 1 and place them into tube 1, which is then plugged with a foam lid. Fill all tubes analogously, then tap the block softly onto the bench, ideally on some soft surface, to tap the flies down. Remove all 18 lids while tapping down the block to prevent flies from escaping and place an 88-mm apple juice plate with yeast (EQUIPMENT SETUP) on top and fix the plate with two rubber strings. Place the block with the agar plate at the bottom into the incubator (Fig. 4). Change agar plates by tapping the flies down as described above.
  • Preparation of reusable silicon rubber grid
    • Prepare silicon rubber grids to create reaction chambers using an inverted V-bottomed microtiter plate as mold (Fig. 1a). Add the two Elastosil RT 625 components into a 50-ml Falcon tube in a mass ratio of 9:1 (i.e., 46.8 g of A and 5.2 g of B), rotate and turn the tube manually until the components are fully mixed during 5–15 min. Pour the mixture onto an inverted V-bottomed 384-well microtiter plate, leaving about 1–2 mm of the tip of the wells uncovered. Allow the silicon rubber to harden at least overnight at room temperature (20–25 °C). After removing the silicon rubber grid from the plate, use a scalpel to cut the silicon rubber into sections that fit a microscope slide, with the wells matching the printed subarrays. The grid is reusable; wash it in 10% chlorine solution (100% solution containing <5% NaClO), rinse with water and allow it to dry after each use.
  • Preparation of microarrays

ADVERTISEMENT

Top

Procedure

Overview

  1. Stage 1: high-throughput TAMS mapping, generation of recombinantsTiming: 3 fly generations approximately1 month, with only about 4 h hands-on work per mutant to be mappedSelect the mutants to be mapped. One to two full-time persons with fly experience are able to map 6–12 mutants at the same time. Note that in our proof-of-principle experiment, we included a fluorescent marker chromosome labeling embryonic muscles throughout stage 1 (Fig. 2)8. This is not absolutely required: the marker can also be crossed in at the last cross of the mapping procedure before phenotyping as exemplified for stage 2.
  2. Cross each mutant to the reference strain. One vial per mutant is enough.
  3. Collect 10–20 transheterozygous progeny virgins (F 1) bearing the mutant chromosome in trans to the reference chromosome. Cross these virgin females to balancer males in one bottle for each mutant to be mapped. It is convenient to use balancer males that carry a heat-shock-inducible dominant lethal transgene on the other chromosome (in our case Sp, hs-hid) to eliminate this chromosome later.
  4. Collect about 100 recombinant males from the F2 progeny (all males will inherit a recombinant chromosome and can be used) and cross each male individually to 3–4 balancer virgin females in vials.
    Critical step At this step, the virgin females should contain a GFP marker on their balancer chromosome and a heat-shock-inducible dominant lethal transgene on the other chromosome (in our case Sp, hs-hid). We used a balancer chromosome containing a Kr-GAL4, UAS-GFP transgene; all embryos carrying the balancer chromosome can be detected by the expression of GFP in Bolwigs organ at 12–20 h (see ref. 19). The virgin females should also have an additional marker (located on a different chromosome) to label the tissue of interest, ideally a GFP marker that allows rapid scoring of the phenotype of interest. We used an Mhc-tauGFP located on the X-chromosome to label all embryonic muscles (see Fig. 3).
  5. Select individuals with recombinant chromosomes balanced over the balancer chromosome used in Step 4 to establish a stable recombinant stock. This is usually straightforward if a conditional lethal chromosome was included in Steps 3 and 4. If Sp, hs-hid is used, as in our example, selection can be achieved as follows. After about 5 d, discard the adults from each cross and heat-shock the larvae (F3) for about 45 min by placing them in a 37 °C water bath. The resulting adults (F3) establish a stable recombinant stock without any time-consuming virgin collection.
    Critical step Adults need to be discarded before the heat shock, as they may survive the heat shock. In our experience, it is not necessary to wait after discarding the adults, as hs-hid works efficiently in embryos as well as larvae.
  6. Phenotyping of recombinantsTiming: 3–5 h on 2 consecutive days for 96 stocksIt is convenient to process 96 recombinant stocks per mutant at the same time from here onward. Fill the 96 stocks into six 18-tube blocks as described in EQUIPMENT SETUP. A total of 20–50 flies per tube is sufficient. While loading the blocks, take four flies from every stock and place them into a well of a rigid 96-well plate that is placed on ice. After all 96 stocks have been added to the blocks and 96-well plate, put a lid on the plate and store it at -20 °C. This 96-well plate will be used in Step 11 for preparing DNA.
    Critical step From this step onward, it is critical to label all the different recombinant stocks for a gene with a number, e.g., 1–96. It is important to track these numbers together with the position in the blocks and the 96-well plate used later for genotyping.
  7. Place the blocks in the 25 °C incubator for 4–6 h or overnight. The exact duration of incubation depends on the desired stage used for the phenotypic analysis. For muscles, we used embryos about 12–20 h old, thus we left the apple agar plate on the block for 4 h and then incubated the plate for 12 h (overnight) at 25 °C. Alternatively, we left the block in the incubator overnight and then incubated the apple agar plate for another 6 h.
  8. To analyze the embryonic phenotype, pipette three distinct drops of halocarbon oil on a coverslip (about 50–100 mul per drop). Collect embryos under the binocular microscope using a brush. Take embryos from position 1 of the aged agar plate 1 and place them into the first oil drop and distribute them with the brush. Continue by placing embryos from position 2 on the plate into the second drop and so on. Place an 8-mm coverslip on the embryos (Fig. 3a). Each apple agar plate will result in six slides in total. Indicate the appropriate position numbers on the slides.
    Critical step Note that the position on the apple agar plate is mirrored compared with the 18-tube block (Fig. 4). Be careful not to transfer any remaining yeast onto the slide, as it will prevent the embryos from becoming transparent.
  9. Inspect the embryos for their phenotype under a fluorescent microscope. At least two homozygous embryos should be scored (Fig. 3). Embryos can easily be 'rolled' into the correct position with a wooden toothpick placed on the coverslip to ensure that the same region of each embryo is scored.
  10. Score the phenotype (wild-type or mutant) and continue with the other plates. Save the sample name and phenotype in a tab-separated format. This can be accomplished by writing in an Excel sheet and exporting the table as a text file. The column header for the recombinant names should be 'recombinant', for the phenotypes 'phenotype', write 'w' for wild-type and 'm' for mutant (Supplementary Data online). This table will be imported into SNPmapper (see Step 46).
    Critical step Keep all the recombinant stocks until you have obtained the mapping data in case you want to verify a phenotype again.Troubleshooting
  11. DNA isolation for TAMS assayTiming: 4 h for two 96-well platesThaw the 96-well plates with the flies prepared in Step 6. Homogenize the flies in each well in 50 mul of DNA extraction buffer by squishing the flies with a yellow tip and slowly pipetting out the 50 mul. Do this for all the wells individually, taking care to use a new yellow tip each time. It is not necessary to keep the plate on ice.
  12. Incubate for 30 min at 65 °C in a thermocycler. In the meantime, prepare the second plate. It is convenient to process two plates at the same time in Steps 13–21.Pause Point After Step 12, the plates can be stored on ice for 1 h until ready to proceed.
  13. Add 100 mul of freshly mixed LiCl/KOAc (11.5 ml of 5 M KOAc + 28.5 ml of 6 M LiCl) using an eight-channel pipette to each well of both plates. Seal each plate, invert it twice and incubate for 15 min on ice.
  14. Centrifuge for 20 min at 4,650g, 4 °C. We used a Sorvall RC-3B Plus with a H6000 swing-out rotor at 4,000 r.p.m.
  15. Transfer 100 mul of supernatant from each well into a new 96-well plate using an eight-channel pipette. Be careful not to transfer solid debris.
  16. Add 60 mul of isopropanol into each well, seal the plate and invert twice.
  17. Centrifuge for 20 min at 4,650g, 4 °C and discard the supernatant by turning the plate upside down over a sink.
  18. Wash by adding 100 mul of 70% ethanol (vol/vol) into each well, seal and centrifuge for 10 min at 4,650g, 4 °C.
  19. Discard the supernatant by turning the plate upside down and let the plate dry on the bench for at least 1 h.
  20. Resuspend the DNA in 40 mul of water.
  21. Measure the DNA concentration for some of the samples using the Nanodrop spectrophotometer.Pause Point DNA with a concentration >10 ng mul-1 can be stored indefinitely at -20 °C. At lower concentrations, we recommend storage for no longer than 6 months.
  22. Multiplex PCR and cleanup for TAMSTiming: 4.5 h, of which 1.5 h hands-onChoose the TAMS assay set for the chromosome of interest.
  23. Amplify DNA samples and control samples by multiplex PCR according to an optimized protocol with a polymerase suited for multiplex amplification (see details about multiplex PCR optimization and mini-sequencing reaction controls in Experimental design). A typical reaction mix is tabulated below. We recommend using about 10 ng of DNA in each multiplex PCR. One plate of extracted DNA is usually sufficient for all multiplex PCRs of a selected TAMS assay set. The amplification success may be verified by running an aliquot of the PCR product on a 2% (wt/vol) agarose gel for a subset of the samples.
    ReagentVolume per reaction (mul)Final concentration
    10times PCR buffer11times
    25 mM MgCl20.61.5 mM
    2 mM dNTPs0.40.08 mM
    5 muM Primermix0.10.05 muM
    10 U mul-1 SmartTaq DNA polymerase0.10.1 U mul-1
    DNA template2approx10 ng
    H2O to a final volume of 10 mul5.8 

    Critical step In our verified assays, we performed multiplex PCR at a complexity level of up to 24 fragments in the same reaction. However, multiplexing reactions is not easy, as the risk of amplification artifacts tends to increase exponentially as the number of added primer pairs is increased. We recommend aiming at approx10-plex reactions and to carefully optimize the reactions (see Experimental design for guidelines).
  24. Pool the multiplex PCR products for each sample. We recommend running PCRs in microtiter plates (one plate for each different multiplex PCR of the approx96 mutants) followed by pooling of all multiplex reactions for each of the 96 samples into one new plate using a multichannel pipette.
  25. Prepare a master mixture containing the exonuclease (ExoI) and alkaline phosphatase (SAP) reagents for cleanup of the PCR products, as indicated in the table below. Alkaline phosphatase inactivates the remaining dNTPs, and exonuclease I degrades the single-stranded PCR primers, which would result in mis-incorporation of nucleotides during the subsequent mini-sequencing reactions.
    ReagentVolume per reaction (mul)Final concentration

    aThe actual final concentration of MgCl2 will be higher than 7.6 mM, depending on the contribution of MgCl2 from the PCR product mixture.

    PCR products7.1 
    50 mM MgCl21.67.6 mMa
    1 M Tris-HCl pH 9.50.50.05 M
    20 U mul-1 Exonuclease I0.30.57 U mul-1
    1 U mul-1 shrimp alkaline phosphatase1.00.10 U mul-1
  26. Transfer 7.1 mul of the PCR product pool to a new plate, and add 3.4 mul of the cleanup mixture to each sample to a final volume of 10.5 mul.
    Critical step Keep the enzymes and reaction mixtures on ice, as both ExoI and SAP are thermosensitive.
  27. Incubate for 30 min at 37 °C and subsequently inactivate the enzymes by incubation for 15 min at 80 °C.Pause Point The PCR product as well as the product of the cleanup reaction can be left at 4 °C overnight or at -20 °C for indefinite long-term storage.
  28. Cyclic mini-sequencingTiming: 2.5 h, 30 min hands-onPrepare a master mixture with mini-sequencing reagents, as indicated in the table below.
    Critical step The fluorophores are light sensitive. Protect working aliquots from light and keep at 4 °C. Store stock solutions at -20 °C in the dark.

    Preparation of mini-sequencing reaction mixtures

    ReagentVolume per reaction (mul)Final concentration

    aA reaction control can be included in the mini-sequencing reaction (see Experimental design). Include the control templates in the mini-sequencing reaction mixture and the 5'-tagged mini-sequencing control primer in the master mixture of primers.

    bTexas Red-ddATP, TAMRA-ddCTP, R110-ddGTP, Cy5-ddUTP. Cy5-ddUTP should be added at a 1.5- to 2-fold higher concentration than the other ddNTPs to compensate for its lower incorporation efficiency. Depending on the array scanner available, instead of using four nucleotides labeled with different fluorophores in a single mini-sequencing reaction, a single label or two labels may be used in four or two separate reactions, respectively.

    PCR products after cleanup10.5 
    100 nM of each pooled mini-sequencing primera1.510 nM
    100 muM Fluorescently labeled ddNTPsb0.068 (3 times 0.015 + 1 times 0.023)0.10 muM or 0.15 muM
    1% Triton X-1000.300.02%
    25 U/mul-1 KlenThermase0.040.067 U mul-1
    100 nM Reaction control templatesa0.151 nM
    H2O to a final volume of 15 mul2.5 
  29. After cleanup (Step 27), add 4.5 mul of mini-sequencing reaction mixture to the pooled PCR products from each sample to a final volume of 15 mul.
  30. Perform the mini-sequencing reactions using an initial 3-min denaturation step at 95 °C followed by 33 cycles of 20 s at 95 °C and 20 s at 55 °C in a thermocycler.Pause Point The mini-sequencing product can be left protected from light at 4 °C overnight, although we recommend carrying out hybridization (Steps 31–35) on the same day.Troubleshooting
  31. Capture by hybridizationTiming: 3 h, 45 min hands-onAssemble the microarray in the custom-made aluminum reaction rack (Fig. 1a) as described in Experimental design.
  32. Preheat the assembly to 42 °C on a heat block.
  33. Add 7 mul of the hybridization solution to each mini-sequencing reaction mixture from Step 30 leading to a final volume of 22 mul.
  34. Transfer 20 mul of each sample with a multichannel pipette to a separate reaction chamber on the microscope slide.
  35. Hybridize for 2 h at 42 °C in a humid and dark environment, e.g., by placing a wet tissue on the plexiglas lid and covering it with plastic film and aluminum foil.
    Critical step Background fluorescence, caused by drying of samples onto the slide, may occur if the hybridization chamber is not kept humid. It is important to cover the hybridization reaction with aluminum foil to protect the fluorophores from light.
  36. Microarray washingTiming: 30 minPrepare the three washing solutions. Preheat solution II to 42 °C.
  37. After hybridization, take the slides from the reaction rack and rinse immediately with solution I, at room temperature.
  38. Wash the slides twice for 5 min with solution II at 42 °C, and twice for 1 min with solution III, at room temperature, in 50-ml Falcon tubes.
    Critical step While washing, shake the tube gently and avoid exposure to light.
  39. Spin-dry the slides in a Falcon tube without lid for 5 min at 102.5g (rotor radius 113 mm) and store them protected from light.Pause Point The microarrays can be left dry at room temperature overnight, although we recommend scanning them on the same day.
  40. Fluorescence scanning and signal quantificationTiming: 1 hIf the available array scanner permits, balance the signal intensity from each laser channel so that no signals are saturated and the signals from the four fluorophores are as equal as possible. Balancing is feasible if a reaction control with signals from all four fluorophores has been included in the mini-sequencing reaction mixture (see section about controls in Experimental design). Figure 1d shows an example of a scanned subarray.
  41. A quantification program, such as the one supplied with the ScanArray Express instrument or the QuantArray software, is used to quantify the signals from each spot in the scanning images and provides the raw data collected as a text file as output. Subtract the background noise at each wavelength measured either around the spots or at negative control spots. Spotted anti-tag oligonucleotides for which the corresponding tagged primers have not been included in the mini-sequencing reaction mixture serve as control spots.
    Critical step In the quantification process, the position of every spot in each subarray is given a unique SNP and sample identity. The exact format for this identification process will vary depending on the quantification software used. To facilitate the uploading and reading of the quantified raw data in the SNPmapper software, a specific naming convention is, however, recommended. The SNP identity should be SNPid_for/rev=alleles, where 'for' or 'rev' refer to the mini-sequencing orientation (see also Supplementary Data for more details).
  42. Data analysis by SNPmapperTiming: 15 minChoose background removal option by clicking on the corresponding checkbox. Multiple choices are possible. By default, the background measured by the scanner (and provided in the raw data) is subtracted from each signal. Moreover, an additional noise level is determined for each of the four channels on the basis of the signals emitted by the two non-SNP channels of each spot and subtracted. If the output looks noisy, it is also possible to include the removal of single 'outliers' within a user-specified window; e.g., if the window size is 4, 'CS CS OR CS' or 'CS OR CS CS' will be changed to 'CS CS CS CS'.
  43. Select the minimal number of successful TAMS assays per sample (default is 5). Samples where the number of called genotypes is below this threshold are not displayed.
  44. Import files with genotype data of the parental lines and with information about the assayed SNPs (see Supplementary Data for example files).
  45. Load file(s) containing the TAMS raw data (QuantArray format is recommended; see Supplementary Data for example files). Multiple files from different microarray slides (e.g., for chromosome arms 2L and 2R) of the same TAMS experiment and the same recombinant set can be processed.
  46. Load file with the phenotype data (see Step 10). Press Submit button.
  47. Each chromosome arm is shown in a separate window. Images can be saved as png-files (Figs. 1d and 5).
    Figure 5: Stage 1 mapping of C233.
    Figure 5 : Stage 1 mapping of C233.

    SNPmapper output for C233 is shown. Recombinants are sorted according to phenotype, shown in yellow for mutant lines and black for wild-type lines, and genotype with red rectangles for mutagenized ORiso and blue rectangles for reference, CSiso. The determined genomic location is between SNPs 19 and 20 on chromosome arm 2L and is defined by the break points of recombinant numbers 26, 33 and 39 highlighted by red frames.

    Full size image (92 KB)

    Troubleshooting
  48. Stage 2: high-resolution mapping using 2EP stocks, generation of recombinantsTiming: 3 fly generations approximately1 month, with about 10–15 h hands-on work per mutant to be mappedDepending on the result from the first mapping step, select the appropriate 2EP line(s)8 and cross it to your mutant as shown in Figure 2. Set up 1 or 2 bottles of this cross (G0) and flip the cross every 4–5 d to produce enough F1 flies.
    Critical step For the 2EP recombination, it is absolutely critical that no (white +) transgene other than the EP insertions themselves is present in the stocks. Otherwise, the later identification of the recombinants by paler eye color is not possible. Here, we use non-GFP balancers, no muscle marker and an Sp, hs-hid chromosome from which the (white +) was removed. GFP balancer and GFP muscle marker are added in the last cross after the isolation of the recombinants.
  49. Collect 100–200 transheterozygous virgins from the F1 progeny for each mutant and cross them in 6–8 bottles with balancer males; here we use Sp, hs-hid (w-)/CyO males (Fig. 2).
  50. Flip these crosses at least 3 times every 4–5 d to produce enough F2 progeny. In total, you need about 10,000 F2 males per gene, as the recombinant males are present at a frequency of about 1/50 to 1/100.
  51. Collect 100–200 recombinant F2 males, which are selected on the basis of their paler eye color, and cross them individually to 3–4 balancer virgins bearing a labeled balancer and a phenotypically relevant GFP marker for scoring.
    Critical step Identification of the males with paler eye color requires control of the age of the males, as eyes get darker with age. We collected from the bottles every 2 d. In some cases, the very young males need to be aged another day before they can be judged. This step needs to be done with care and requires some experience for recognizing the age of flies and eye color. In our hands, 60–80% of the isolated recombinant candidates were true recombinants; however, this may vary for each 2EP stock8.
  52. Select individuals with recombinant chromosomes balanced over the balancer chromosome used in Step 51 to establish a stable recombinant stock. As described in Step 5, this is done by heat shock of the cross, thus avoiding time-consuming virgin collections.
  53. Phenotyping of recombinantsTiming: 3–5 h on 2 consecutive days for 96 stocksScore the phenotypes of the recombinant stocks exactly as described in Steps 6–10.
  54. DNA isolation for SNP genotypingTiming: 4 h for two 96-well platesPrepare DNA for SNP genotyping as described in Steps 11–20; note that the DNA in Step 20 can be more dilute and resuspended in 50 mul of dH2O.
    Critical step As not all stocks may have been previously genotyped at high resolution (see Table 1), include stocks with all relevant chromosomes in this DNA preparation (e.g., the parental stock in which the mutation was induced, the 2 EP stock and the mutant/balancer stock).
  55. SNP genotyping of recombinants by sequencingTiming: A few days to a few weeks, depending on how many SNPs are detected in parallelWe have precomputed a genome-wide set of high-density TAMS assays, which allow the application of 'local TAMS' assays to reduce cost8. This can be done as described in Steps 22–47. Alternatively, we have previously used PCR and classical sequencing for high-resolution SNP genotyping: first select two or three SNPs close to each EP insertion or just 'outside' the 2EP interval on each side and genotype these SNPs in your parental stock, 2EP and mutant/balancer stock as described below8.
  56. For genotyping, set up PCRs for each of the genomic samples as described below.
    ComponentAmount per reactionFinal concentration
    Genomic DNA0.5 mul 
    Primer 12.5 mul (5 muM)0.5 muM
    Primer 22.5 mul (5 muM)0.5 muM
    10times Taq-PCR buffer2.5 mul1times
    dNTPs0.5 mul (10 mM)0.2 mM
    TaqPol0.25 mul (1.25 units)1.25 units
    H2O16.25 mul 
  57. Place into thermocycler and run the program tabulated below. If the PCRs are performed for the first time, they should be checked by gel electrophoresis for a product.
    StepDenatureAnnealExtend
    13 min, 94 °C  
    2–4130 s, 94 °C30 s, 60 °C2 min 72 °C
    42  5 min, 72 °C
    Troubleshooting
  58. Degrade primers remaining in the PCR product by ExoSAP treatment. Pipette 2.25 mul of PCR product to 0.5 mul of ExoSAP mixture (Invitrogen) and incubate in a thermocycler for 30 min at 37 °C followed by 15 min at 80 °C to inactivate the enzymes.
  59. For sequencing, add 15 mul of H2O to the ExoSAP-treated product, mix and use 6 + 1 mul of primer (5 muM) for a standard sequencing reaction.
  60. Analyze the sequences and call the genotypes; we do this manually with Vector NTI (Invitrogen) or Seqman (DNA-STAR).
  61. After verification of appropriate SNPs, use one SNP at each end of the 2EP interval and genotype these 2 SNPs in all the recombinants. This will identify which of the candidate recombinants are true recombinants. Discard all nonrecombinants.Troubleshooting
  62. Repeat Steps 55–61 with 5–10 equally spaced SNPs in the 2EP interval, but test only about 20 of the >100 recombinants to narrow down the region that is linked to the gene.
  63. Gradually increase resolution by repeating Steps 55–61 with more recombinants and additional SNPs in the relevant region, which is decreasing with each round of genotyping.
  64. Finally, test all recombinants that may have their break point in the region of interest. This process will maximize the resolution, while keeping sequencing costs at a moderate level.
Top

Timing

Estimates of total time required for groups of steps are provided below. Actual hands-on time is indicated in brackets.
Stage 1, rough mapping (1–2 Mb resolution)
Steps 1–5, generation of recombinants: 1 month (4 h)
Steps 6–10, phenotyping of recombinants: 2 d (3–5 h)
Steps 11–21, isolation of genomic DNA: (4 h)
Steps 22–27, multiplex PCR: 4.5 h (1.5 h)
Steps 28–30, mini-sequencing: 2.5 h (0.5 h)
Steps 31–39, hybridization, washing: 3.5 h (1.2 h)
Steps 40–47, quantification and genotype calling: (1.2 h)
Stage 2, fine mapping (50-kb resolution)
Steps 48–52, generation of recombinants: 1 month (10–15 h)
Step 53, phenotyping of recombinants: 2 d (3–5 h)
Step 54, isolation of genomic DNA: (4 h)
Steps 55–64, SNP genotyping of recombinants: variable, few days to few weeks

Top

Troubleshooting

Troubleshooting advice can be found in Table 4.


Top

Anticipated results

We applied our stage 1 protocol to a set of 14 mutations causing muscle morphogenesis defects. We were able to map all of these genes successfully, 7 of them with a resolution <1.5 Mb (see ref. 8). During the mapping of these mutations, we optimized our TAMS protocol, which led to the current version described here. A TAMS result obtained with this optimized version is depicted in Figures 1d and 5 for C233, an eyes absent (eya) allele. In this case, we obtained a resolution of 0.33 Mb after stage 1, which demonstrates the potential of the TAMS procedure. For proof of principle, we subjected two mutations to the stage 2 procedure and were able to map these to an 80-kb interval containing 9 genes and a 91-kb interval containing 12 genes, respectively. In both cases, the intervals contain the correct gene8.

Our mapping strategy is unbiased for any genetic region or the kind of phenotype to be analyzed and only requires that the generation of recombinants is possible. Thus, it can in principle be applied to a wide variety of different biological questions in flies and other organisms.

Note: Supplementary information is available via the HTML version of this article.



Top

References

  1. Jorgensen, E.M. & Mango, S.E. The art and design of genetic screens: Caenorhabditis elegans. Nat. Rev. Genet. 3, 356–369 (2002). | Article | PubMed | ISI | ChemPort |
  2. St Johnston, D. The art and design of genetic screens: Drosophila melanogaster. Nat. Rev. Genet. 3, 176–188 (2002). | Article | PubMed | ISI | ChemPort |
  3. van Eeden, F.J. , Granato, M. , Odenthal, J. & Haffter, P. Developmental mutant screens in the zebrafish. Methods Cell Biol. 60, 21–41 (1999). | PubMed | ISI | ChemPort |
  4. Davis, M.W. et al. Rapid single nucleotide polymorphism mapping in C. elegans. BMC Genomics 6, 118 (2005). | Article | PubMed | ChemPort |
  5. Berger, J. et al. Genetic mapping with SNP markers in Drosophila. Nat. Genet. 29, 475–481 (2001). | Article | PubMed | ISI | ChemPort |
  6. Stickney, H.L. et al. Rapid mapping of zebrafish mutations with SNPs and oligonucleotide microarrays. Genome Res. 12, 1929–1934 (2002). | Article | PubMed | ISI | ChemPort |
  7. Hoskins, R.A. et al. Single nucleotide polymorphism markers for genetic mapping in Drosophila melanogaster. Genome Res. 11, 1100–1113 (2001). | Article | PubMed | ISI | ChemPort |
  8. Chen, D. et al. High-resolution, high-throughput SNP mapping in Drosophila melanogaster. Nat Methods 5, 323–328 (2008). | Article | PubMed | ChemPort |
  9. Baner, J. et al. Parallel gene analysis with allele-specific padlock probes and tag microarrays. Nucleic Acids Res. 31, e103 (2003). | Article | PubMed | ADS | ChemPort |
  10. Shen, R. et al. High-throughput SNP genotyping on universal bead arrays. Mutat. Res. 573, 70–82 (2005). | PubMed | ChemPort |
  11. Macdonald, S.J. , Pastinen, T. , Genissel, A. , Cornforth, T.W. & Long, A.D. A low-cost open-source SNP genotyping platform for association mapping applications. Genome Biol. 6, R105 (2005). | Article | PubMed | ChemPort |
  12. Thibault, S.T. et al. A complementary transposon tool kit for Drosophila melanogaster using P and piggyBac. Nat. Genet. 36, 283–287 (2004). | Article | PubMed | ISI | ChemPort |
  13. Parks, A.L. et al. Systematic generation of high-resolution deletion coverage of the Drosophila melanogaster genome. Nat. Genet. 36, 288–292 (2004). | Article | PubMed | ISI | ChemPort |
  14. Ryder, E. et al. The DrosDel deletion collection: a Drosophila genomewide chromosomal deficiency resource. Genetics 177, 615–629 (2007). | Article | PubMed | ChemPort |
  15. Lindroos, K. , Sigurdsson, S. , Johansson, K. , Ronnblom, L. & Syvanen, A.C. Multiplex SNP genotyping in pooled DNA samples by a four-colour microarray system. Nucleic Acids Res. 30, e70 (2002). | Article | PubMed |
  16. Lovmar, L. , Fredriksson, M. , Liljedahl, U. , Sigurdsson, S. & Syvanen, A.C. Quantitative evaluation by minisequencing and microarrays reveals accurate multiplexed SNP genotyping of whole genome amplified DNA. Nucleic Acids Res. 31, e129 (2003). | Article | PubMed | ChemPort |
  17. Fuger, P. , Behrends, L.B. , Mertel, S. , Sigrist, S.J. & Rasse, T.M. Live imaging of synapse development and measuring protein dynamics using two-color fluorescence recovery after photo-bleaching at Drosophila synapses. Nat. Protoc. 2, 3285–3298 (2007). | Article | PubMed | ChemPort |
  18. Wieschaus, E. & Nüsslein-Volhard, C. In Drosophila, a Practical Approach (ed. Roberts, D.B.) (Oxford Science Publications, Oxford University Press, 1998).
  19. Casso, D. , Ramirez-Weber, F.A. & Kornberg, T.B. GFP-tagged balancer chromosomes for Drosophila melanogaster. Mech. Dev. 88, 229–232 (1999). | Article | PubMed | ISI |