Systematic hybrid LOH: a new method to reduce false positives and negatives during screening of yeast gene deletion libraries



We have developed a new method, systematic hybrid loss of heterozygosity, to facilitate genomic screens utilizing the yeast gene deletion library. Screening is performed using hybrid diploid strains produced through mating the library haploids with strains from a different genetic background, to minimize the contribution of unpredicted recessive genetic factors present in the individual library strains. We utilize a set of strains where each contains a conditional centromere construct on one of the 16 yeast chromosomes that allows the destabilization and selectable loss of that chromosome. After mating a library gene deletion haploid to such a conditional centromere strain, which corresponds to the chromosome carrying the gene deletion, loss of heterozygosity (LOH) at the gene deletion locus can be generated in these otherwise hybrid diploids. The use of hybrid diploid strains permits complementation of any spurious recessive mutations in the library strain, facilitating attribution of the observed phenotype to the documented gene deletion and dramatically reducing false positive results commonly obtained in library screens. The systematic hybrid LOH method can be applied to virtually any screen utilizing the yeast non-essential gene deletion library and is particularly useful for screens requiring the introduction of a genetic assay into the library strains. Copyright © 2006 John Wiley & Sons, Ltd.


The creation of sets of isogenic haploid strains containing deletions of all non-essential genes by the Yeast Genome Deletion project has provided a critical tool for genomic studies in Saccharomyces cerevisiae (Winzeler et al., 1999). The gene deletion library permits simultaneous coverage of the genome in identical experiments and alleviates the need for cloning to determine the mutation causing a given phenotype. Further, the KanMX4–marked gene deletions in the library strains can be easily amplified by PCR and cloned or transferred into other genetic backgrounds by genetic recombination, facilitating phenotype confirmation and subsequent experiments (Hudson et al., 1997). High-throughput studies analysing drug sensitivity (Chang et al., 2002; Giaever et al., 2004; Lum et al., 2004; Parsons et al., 2004), synthetic lethality (Tong et al., 2001, 2004), and other assays have yielded many results not found in traditional mutagenesis screens. The library is also an ideal tool for genomic studies involving plasmid-based reporter assays. However, the presence of additional recessive genetic factors in the individual library strains may obscure the data generated by a genomic screen, by creating false positive results that are not caused by the known gene deletion in each strain, as well as creating false negatives by suppressing the phenotype of the known gene deletion. The presence of genetic abnormalities in individual library strains has been noted (Hughes et al., 2000) but not extensively characterized.

A genomic screen to identify gene deletions that alter the localization patterns of the homologous recombination and repair protein Rad52 was undertaken by transforming a Rad52–YFP plasmid into individual library haploid strains and observing the transformed strains under epifluorescence microscopy. Populations of logarithmically growing cells for each gene deletion strain were scored for the levels of spontaneous relocalization of the Rad52–YFP protein into discrete subnuclear foci, which is a marker for homologous recombination (Lisby et al., 2003). Among approximately one-third of the strains in the gene deletion library screened using this approach, a large number of strains were identified that had significantly elevated levels of spontaneous Rad52–YFP focus formation. On further analysis, we find that most of these ‘hits’ were false positives. In fact, when the 10 strains with the highest levels of spontaneous foci were selected for further analysis, we found that the focus phenotype for nine of the strains analysed was not caused by the marked gene deletion, as the phenotype disappeared when the gene deletions were transferred into another strain background or into the parent strain for the library. Subsequent genetic analysis on several of these gene deletion strains indicated that the Rad52–YFP focus phenotype observed was attributable to additional recessive factors in the individual library haploids.

In response to these results, we developed systematic hybrid LOH to eliminate the influence of additional recessive factors on the phenotype. By taking advantage of conditional centromere constructs in strains of a different genetic background, it is possible to systematically generate hybrid diploids that are heterozygous for every chromosome except the one containing the defined gene deletion. These diploid strains provide maximal complementation of recessive factors while maintaining the homozygosity of the gene deletion for each library strain. This new method suppresses genetic background effects and is widely applicable to almost any genetic screen that utilizes the yeast gene disruption library. Since the method involves a mating step to introduce genetic diversity and complementation of recessive factors in the library strains, it facilitates the introduction of both plasmid-based and genomically-integrated assays into the non-essential gene deletion library strains.

Materials and methods

Strains and media

The individual deletions of non-essential genes made in the BY4742 and BY4739 MATα strains were obtained from the Saccharomyces Gene Deletion Project (Brachmann et al., 1998). The conditional chromosome strains used will be described (Reid et al., in preparation). Briefly, the set consists of 16 strains constructed in the W303 genetic background, each of which carries an identical conditional centromere construct on one of the 16 yeast chromosomes. This construct is composed of a Kluyveromyces lactis URA3 gene linked to a galactose-inducible promoter inserted immediately upstream from the centromere. Additional strains used in these studies are shown in Table 1. Strain BY4742 was used as wild-type control for the library background. Yeast extract–peptone–dextrose (YPD) medium, synthetic complete without leucine (SC–Leu), synthetic complete without tryptophan (SC–Trp), synthetic complete without adenine (SC–Ade), synthetic complete without leucine, methionine, and uracil (SC–Leu-Met-Ura), synthetic complete without leucine containing 5′-fluororotic acid (5-FOA–Leu), synthetic complete with galactose but lacking leucine (SCGal–Leu), and sporulation medium (SPO) were prepared as described (Sherman et al., 1986; Lisby et al., 2001) with 2× the normal concentration of leucine when appropriate. YPD medium was supplemented with the antibiotic G418 at a concentration of 300 µg/ml at indicated steps (YPD + G418).

Table 1. Strains
  • *

    Isogenic to W3749-1A.

  • Isogenic to BY4742.

  • Isogenic to W3646-11D.

BY4742MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0
W3749-1AMATaADE2 bar1::LEU2
W3646-11DMATα ADE2 bar1::LEU2 TRP1 lys2Δ
W2869-8DMATα ADE2 bar1::LEU2 TRP1 lys2Δ srs2Δ::HIS3 RAD52–YFP
W3776-12AMATaADE2 bar1::LEU2 mre11Δ::LEU2 RAD52–YFP
W3789-9AMATaADE2 bar1::LEU2 rad51Δ RAD52–YFP


A CEN-based HIS3-marked plasmid (pWJ1213) for expression of Rad52–YFP from its endogenous promoter was previously described (Feng et al. DNA Repair, in press), and was utilized for all Rad52–YFP experiments in W303 background strains. A LEU2-marked plasmid (pWJ1344) was created to introduce the Rad52–YFP fusion gene into the gene deletion library. A purified fragment of pWJ1213 containing the Rad52–YFP fusion gene was co-transformed into yeast strain W1588-4C (Zhao et al., 1998) for gap-repair via homologous recombination with a PvuII-cut pRS415 vector (Sikorski and Hieter, 1989), to integrate the Rad52–YFP sequence within the multiple cloning site on the pRS415 backbone. The resulting gap-repaired plasmid (pWJ1344) was sequenced for confirmation of the expression cassette. The pWJ1344 plasmid was used for all Rad52–YFP experiments performed using library gene deletion strains.


W303 haploid gene deletion strains and conditional chromosome strains were transformed with the Rad52–YFP containing plasmid, pWJ1344, using standard LiOAc techniques (Schiestl and Gietz, 1989). Library haploids in the initial phase of the experiment were transformed with pWJ1344 using a LiOAc transformation protocol modified to suit simultaneous manipulation of large numbers of strains in the 96-well format. Cells grown overnight at 30 °C in 200 µl YPD are centrifuged and washed in 200 µl distilled H2O. The cells are then centrifuged, washed twice in 200 µl 0.1 M LiOAc and resuspended in 25 µl 0.1 M LiOAc. Next, 30 µl distilled H2O, 120 µl 50% polyethylene glycol, 18 µl 1 M LiOAc, 2.5 µl denatured salmon testis DNA and 4.5 µl plasmid pWJ1344 (approximately 500 ng) are added to the resuspended cells, with vigorous mixing after each step. The cells are incubated overnight at 30 °C, followed by 20 min at 42 °C. Lastly, the cells are centrifuged, resuspended in 20 µl distilled H2O and streak-plated on solid SC–Leu plates. Transformations were successful approximately 80% of the time during the course of the haploid screen.

Transfer of gene deletions to other genetic backgrounds

Gene deletions were transferred from individual library strains into wild-type strains in both the W303 and BY4742 backgrounds using genetic recombination. New gene deletion strains in the library background were constructed using the library wild-type strain BY4742. For the W303 gene deletions, strain W3749-1A was used for all MATa gene deletions and strain W3646-11D was used for MATα gene deletion strains. The gene deletion loci were PCR amplified using primers selected 200–300 bp outside the ORF and the purified PCR products were transformed into the recipient strains using standard LiOAc techniques (Schiestl and Gietz, 1989). Gene replacement was confirmed by PCR analysis and in the case of the W303 strains, by the 4 : 0 segregation of the KanMX marker in sporulated diploids created by mating the W303 mutant strain with the original library mutant.

Preparation of hybrid diploids for systematic hybrid LOH

Sixteen strains, each containing a conditional centromere construct on one of the 16 yeast chromosomes, were transformed with the Rad52–YFP plasmid pWJ1344, using standard methods. Haploid library strains arrayed in the 96-well format were grown overnight in fresh liquid YPD + G418, while conditional centromere strains were grown overnight in SC–Leu medium to maintain the LEU2-marked plasmid. Each library strain was mated to the appropriate strain containing a conditional centromere on the chromosome that contains the gene deletion in the respective library strain. Library and conditional centromere strains were mated overnight in liquid YPD. Each subsequent dilution step described below involved resuspension in the consecutive media and transfer of 5 µl into 150 µl of medium. The mating cultures were diluted and grown overnight in SC–Leu-Met-Ura to select for diploid strains containing markers in each strain as well as on the plasmid. To prevent the survival of and further propagation of surviving haploid conditional centromere strains in the liquid cultures, they were then rediluted and grown overnight in YPD + G418, and further diluted and grown overnight in SC–Leu-Met-Ura to ensure diploidy. Strains were then transferred to SCGal–Leu medium overnight to destabilize the centromere of the conditional chromosome, followed by transfer to 5-FOA–Leu for 2 days to select for cells that have lost the conditional chromosome. Cultures were then plated onto SC–Leu solid medium. All strains were grown at 30 °C until preparation for microscopy. In 5% of the population, 5-FOA resistance is acquired through gene conversion at the centromere rather than by loss of the conditional chromosome leaving the cell heterozygous at the gene deletion locus (see Results and discussion). By collectively assaying multiple isolates from this stage, we eliminate the possibility of working with a single cell that arose from such a gene conversion event. Thus, strains were grown as populations and not streak purified following the acquisition of 5-FOA resistance.


Examination of Rad52–YFP focus levels by microscopy was performed as previously described (Lisby et al., 2003). Cells were grown overnight in SC–Leu medium at 23 °C. For each gene deletion strain, 200–300 cells were scored for Rad52–YFP focus formation.

Tetrad analysis

The systematic hybrid LOH method was designed for screening the gene deletion library and is based on the conditional loss of one chromosome in a diploid strain, to eliminate complementation of the gene deletion on the homologous chromosome. To determine the efficacy of this method, gene deletion mutants were subjected to tetrad analysis, to investigate the frequency of selectable chromosome loss and of endoduplication of the monosomic chromosome. Two or more gene deletions on each arm of all chromosomes except chromosome III were selected for this analysis. Homozygous hybrid diploids were prepared as described above, then restreaked onto SC + 5-FOA plates and grown for 2 days. The diploid strains were then replica-plated onto SPO plates and dissected after 4 days. Twenty-four tetrads were dissected for each gene deletion and after 3 days growth on YPD plates were replica-plated to YPD + G418 plates to score for the KanMX marker at the gene deletion locus. Tetrads exhibiting 2 : 2 segregation for the KanMX marker on chromosomes XV and IV were replica-plated onto SC–Ade and SC–Trp, respectively (see Results and discussion).

Results and discussion

Problems with screening library haploids

The gene deletion library was utilized to screen the genome to identify genes that affect the relocalization of the central yeast recombination and repair protein Rad52 from a diffuse nuclear pattern into discrete subnuclear foci. The formation of Rad52 foci has been demonstrated to reflect the establishment of active repair complexes in response to DNA damage (Lisby et al., 2001, 2003). In mitotically growing wild-type strains, Rad52 foci are observed in ∼5% of all cells, but higher levels of spontaneous foci are observed in strains containing deletions of genes involved in various aspects of DNA metabolism (Lisby et al., 2001) and lower levels are seen in strains containing mutations in genes that are necessary for recombination foci (Lisby et al., 2004). The characterization of spontaneous Rad52–YFP focus levels in gene deletion strains may identify new genes involved in DNA dynamics and the maintenance of genomic integrity.

A plasmid containing a Rad52–YFP fusion gene was transformed into individual strains of the disruption library. Mitotically growing cells for each Rad52–YFP transformed strain were observed under epifluorescence microscopy and the fraction of cells exhibiting Rad52–YFP foci was systematically recorded. Over 1800 gene deletion strains were screened by this protocol, representing 32% of the non-essential ORFs in the yeast genome.

To independently confirm the phenotypes of these mutants, as well as the efficacy of the screen, 10 mutants with the highest levels of spontaneous Rad52 foci were selected for further analysis before any further screening was conducted. To separate the effect of the deleted gene of interest from any unlinked mutations in the library background (BY4742), the deletion locus from each of the 10 strains was transferred into the W303 genetic background, by first isolating a gene disruption fragment by PCR and then transforming W3749-1A or W3646-11D (see Materials and methods). Nine of the these disruptions lost the elevated focus phenotype in the W303 background, exhibiting focus levels equivalent to those seen in the wild-type strain. Only one mutant, ctf4Δ::KanMX, exhibited elevated Rad52–YFP focus levels in the W303 background consistent with what was observed for that gene deletion in the library background. To assure that the phenotypic differences in these gene deletions were not the result of general differences between the library and W303 backgrounds, several deletions were retransformed into the library parent strain BY4742 and analysed for spontaneous Rad52–YFP focus levels. The focus levels observed in these strains were the same wild-type levels observed in the W303 background (Table 2), indicating that the increased focus phenotype was not due to the disruption per se.

Table 2. Rad52–YFP focus levels for gene deletions in different genetic backgrounds
MutantLibrary haploid (%)W303 haploid (%)New BY4742 haploid (%)W303 X Library diploid* (%)
  • *

    Homozygous diploids produced by mating W303 and library haploids.

  • nd, not determined.


To investigate the difference between the library gene deletion strains and the new gene deletion strains, further genetic analyses were performed. Library haploids containing the gene deletions spt8Δ::KanMX, arp6Δ::KanMX, hap4Δ::KanMX and ylr047cΔ::KanMX were mated to a wild-type W303 strain, sporulated and subjected to tetrad analysis and microscopy to score Rad52–YFP focus levels for the meiotic products. If the gene disruption were solely responsible for the increased focus phenotype, we expect 2 : 2 segregation in such a cross, with co-segregation of the KanMX disruption and increased Rad52–YFP focus levels. If a second gene were also segregating, we would see 2 : 2, 1 : 3 and 0 : 4 tetrads. As seen in Table 3, no clear segregation pattern was seen, indicating that the increased focus phenotype in these strains is not governed by one or two loci (Table 3). Next, we found that the genetic factors contributing to the phenotype in these strains were recessive, as wild-type levels of foci were observed in hybrid diploids homozygous for all gene deletions except ctf4Δ::KanMX (Table 2). These results indicate that the elevated levels of Rad52–YFP foci observed in the library haploids were the result of multiple recessive factors in the deletion library strains independent of the documented gene deletion.

Table 3. Rad52–YFP focus levels in spore colonies from dissected heterozygous hybrid diploids
TetradWild-type (%)Wild-type (%)spt8Δ (%)spt8Δ (%)
  1. Library haploid strains were mated to W303 wild-type strain W3749-1A, sporulated and dissected. Spore colonies were genotyped, transformed with pWJ1344, and prepared for microscopy as described.

 Wild-type (%)Wild-type (%)arp6Δ (%)arp6Δ (%)
 Wild-type (%)Wild-type (%)hap4Δ (%)hap4Δ (%)
 Wild-type (%)Wild-type (%)ylr047cΔ (%)ylr047cΔ (%)

The documented gene deletion in each library strain not only is insufficient to cause the Rad52 focus phenotype, but also may not directly influence the phenotype. Therefore, we suspected that the recessive factors that cause the focus phenotype in these strains, and by extension false positive results seen in other genomic screens, may have arisen during the transformation events that produced the gene deletion or introduced the plasmid containing the Rad52–YFP construct. Furthermore, it is possible that the individual strains may have accumulated suppressor mutations during the multiple stages of growth associated with propagating thousands of strains with different growth rates as a set. Finally, expression profile analysis revealed significant levels of aneuploidy among strains in the deletion library (Hughes et al., 2000), another factor that can obscure the contribution of a gene deletion to an observed phenotype. Effective analysis of the gene deletion strains requires the complementation of these additional recessive factors without complementation of the documented gene deletion for each strain.

Screening library gene deletions as hybrid diploids

We devised a method to systematically generate hybrid diploids that retain homozygosity at the gene deletion locus. This systematic hybrid LOH method allows us to assay the function of the gene deletion in each strain in the non-essential library, separate from additional recessive factors that could influence the Rad52–YFP focus phenotype. A set of 16 strains in the W303 genetic background, each of which contains a conditional centromere on one of the 16 yeast chromosomes, is used. The conditional centromere construct consists of a K. lactis URA3 gene situated immediately upstream of a galactose-inducible centromere. In addition, a plasmid or any other genetic assay can be established in the conditional chromosome-containing strains to simplify the introduction of the assay into the gene deletion library by mating.

As illustrated in Figure 1, the library deletion strains are mated with a W303 strain that bears the Rad52–YFP assay plasmid and both a conditional centromere and counterselectable K. lactis URA3 marker on the chromosome corresponding to the one that contains the gene deletion in the library strain. These hybrid diploids are then grown on medium containing galactose as the carbon source, which induces transcription through the conditional centromere, disrupting its function in mitotic chromosome segregation. Growth in medium containing 5-FOA selects for cells in which the chromosome has been lost during subsequent divisions. The diploid cells are initially monosomic for the chromosome containing the gene deletion, but over subsequent generations, for most chromosomes, euploid cells overtake the culture following duplication of the monosomic chromosome. This is evident in the reproducible generation of four viable haploid spores from these diploids during meiosis (see below).

Figure 1.

Method summary. The method for the systematic creation of hybrid diploid strains that are homozygous for the desired allele. The haploid containing the gene deletion is mated with the conditional centromere strain corresponding to the chromosome containing the deletion. Growth in galactose destabilizes the centromere on the conditional chromosome, permitting loss during non-disjunction. Growth in 5-FOA selects for cells that have lost the chromosome. Our studies suggest that, in most populations, endoduplication of the monosomic chromosome follows rapidly

Chromosome loss and endoduplication

After selection, three cell types can give rise to a 5-FOA resistance phenotype: cells that are 2n–1 following chromosome loss; cells that are 2n due to endoduplication of the monosomic chromosome following loss of its homologue; and cells in which the URA3 marker is lost through mutation or gene conversion while the W303 chromosome is still maintained. The frequency of these cell types in populations of 5-FOA-resistant diploids was evaluated using tetrad analysis. Diploids produced according to the methods detailed above were sporulated and the resulting tetrads dissected on YPD plates. The spore colonies were then replica-plated to YPD plates containing G418. Sporulated diploids that successfully lost the conditional chromosome but never duplicated the monosomic chromosome produce only two viable G418 resistant spores (2n–1), while diploids that have endoduplicated the Δ::KanMX-marked chromosome from the library strain produce four viable G418-resistant spores (2n). Diploids that produce four viable spores where G418 resistance segregates 2 : 2 have acquired 5-FOA resistance through a mechanism other than loss of the conditional chromosome and remain heterozygous at the gene deletion locus.

To measure the frequency of the three kinds of events described above, tetrad analysis was performed on diploids generated from 66 independent library mutants, including deletions on both arms of each chromosome. However, chromosome III was not analysed because chromosome loss eliminates mating type heterozygosity, which is required for sporulation. The majority of the tetrads analysed reflect homozygous diploid cells that have endoduplicated the disruption chromosome (83%), while 12% of the tetrads remain 2n–1 for the gene deletion chromosome (Figure 2). The two chromosomes with the most 2n–1 diploids were chromosome I (100% aneuploid) and VI (43% aneuploid), consistent with a higher tolerance of monosomy for smaller chromosomes. Approximately 5% of all tetrads analysed revealed diploids that remained heterozygous at the gene deletion locus. This last class is best explained by a gene conversion or mutation in the URA3 gene (see below).

Figure 2.

Analysis of loss of conditional chromosome and endoduplication of the monosomic chromosome containing the KanMX marked gene deletion. Hybrid diploids were produced using the method described here, sporulated, dissected and replicated to YPD + G418 plates to follow the gene deletion marker. Tetrads scored as 2 : 0 reflect chromosome loss and stability with a 2n − 1 chromosome count, while tetrads scored as 4 : 0 reflect chromosome loss and subsequent endoduplication. In tetrads where the gene deletion marker segregates 2 : 2, gene conversion may have eliminated the centromeric URA3 locus without loss of the chromosome

It is important to note that the analysis described above is dependent upon sporulation. Therefore, differences in sporulation efficiency between aneuploid and euploid cells could bias the results in favour of the euploid class. For example, a relative delay or defect in sporulation among 2n–1 cells would result in their under-representation in the tetrad analysis. If that were the case, the diploid population would contain more 2n–1 cells and fewer endoduplicated 2n and/or gene deletion heterozygotes than our analysis estimated.

The class of events that remains heterozygous for the gene deletion after 5-FOA selection likely reflects loss of the URA3 marker at the conditional centromere locus through gene conversion. For two chromosomes, IV and XV, each contains an additional recessive marker on the conditional chromosome. We analysed the segregation pattern for those markers in tetrads that exhibited 2 : 2 segregation of the KanMX-marked gene deletion locus. All 28 tetrads that showed 2 : 2 segregations of the KanMX-marked deletion locus on chromosome XV also exhibited 2 : 2 segregation at the distal ADE2 locus, indicating that the conditional chromosome was not lost in these cells. Interestingly, for the eight tetrads segregating 2 : 2 for gene deletions on chromosome IV, four show loss of heterozygosity at the centromere-linked TRP1 locus (4 : 0), suggesting that co-conversion of the centromeric URA3 marker and TRP1 locus occurred in these cells. Other possible explanations for 2 : 2 segregation of the KanMX marker include disomy for that chromosome in the library haploid strain, an inactivating mutation in the URA3 gene, or a misidentification of the chromosomal location of the marked gene deletion.

Hybrid diploids eliminate false positives and negatives from the haploid screen

To show that the systematic hybrid LOH method eliminates background effects when screening the gene deletion library, we next analysed the 10 original deletion mutants described above (Table 2). As shown in Table 4, only ctf4Δ::KanMX exhibited spontaneous foci in excess of wild-type levels, mimicking the result observed in W303 haploids. Additionally, three gene deletions known from previous studies to increase levels of spontaneous Rad52 foci (rad51Δ, srs2Δ and mre11Δ) maintained elevated focus levels in the hybrid diploid strains (Table 4).

Table 4. Rad52–YFP focus levels for gene deletions in different genetic backgrounds
MutantLibrary haploid (%)W303 haploid (%)Heterozygous hybrid diploid* (%)Homozygous hybrid diploid** (%)
  • *

    Heterozygous diploid diploids produced by mating library haploids with corresponding conditional centromere strains as described.

  • **

    Homozygous diploids refer to hybrid diploids after destabilization of and selection against the conditional chromosome.


Next, the Rad52–YFP focus screen was repeated, using hybrid diploids for the same set of 1800 strains initially screened as haploids via transformation. The first seven unique gene deletions uncovered in the diploid screen were transferred into wild-type W303 haploid strains and all seven maintained the phenotype observed in the hybrid diploids (Table 5). We also found that the distribution in the systematic hybrid LOH screen is very different from that observed in the haploid screen (Figure 3); 157 haploid strains and 39 diploid strains exhibited Rad52–YFP foci in more than 20% of the cells observed. Significantly, only nine gene deletions exhibited this phenotype in both the haploid and hybrid diploid screens, suggesting that the additional genetic factors in the library strains generate false negative as well as false positive results. Thus, the strategy of systematically screening the gene deletion library as hybrid diploids after LOH of the gene deletion chromosome successfully eliminates additional recessive factors in the individual library strain that both produce and suppress the elevated Rad52–YFP focus phenotype.

Figure 3.

Number of gene deletions strains with > four-fold increase in Rad52–YFP focus levels found among same set of 1785 strains screened as haploids and hybrid diploids. Among 157 hits from the haploid screen and 39 from the hybrid diploid screen, only nine gene deletion strains exhibited the phenotype in both sets

Table 5. Mutants obtained in diploid screen have phenotypes confirmed in W303 background
MutantHomozygous hybrid diploid (%)W303 haploid (%)


W303-derived conditional centromere strains can easily and systematically be used in conjunction with the non-essential gene deletion library to produce hybrid diploids that are homozygous for the library gene disruption. The hybrid diploids minimize the contribution to the observed Rad52–YFP focus phenotype of other recessive genetic factors from the library strain. While the possibility remains for genetic factors from the library background, such as dominant alleles or mutations on the same chromosome as the documented deletion, to manifest effects on the focus phenotype in the hybrid diploid, systematic hybrid LOH will sufficiently reduce the number of false positives as well as false negatives obtained in a screen to an easily manageable level. The excessive number of false positive results identified when screening for a Rad52–YFP focus phenotype is likely dependent upon the sensitivity of this assay, which scores an active phenotype in live cells that can result from defects in diverse cellular processes. The level of false results may not be typical of any screen using the gene deletion library. However, the hybrid LOH method should reduce false positive and negative results in any genomic screen, providing the greatest value when applied to assays with similarly high sensitivity to additional genetic factors. Beyond the broad utility of the mating method in separating the gene deletion in each library strain from background factors and suppressor mutations, it is also particularly useful for genomic screens that require the introduction of additional genetic assays. As shown for the Rad52–YFP focus formation screen, the method facilitates the introduction of a plasmid-based reporter into the gene deletion strains, obviating the requirement for 4800 individual transformations. Furthermore, more complicated constructs, such as multiply-marked chromosomes, can be incorporated in the conditional centromere strains for introduction into the library strains for their assay in hybrid diploids.


We thank members of the Rothstein laboratory and Jean Gautier and Lorraine Symington for helpful discussions in the development of the method. We also acknowledge the valuable assistance of Sergey Kalachikov of the Columbia Genome Center in preparations for experiments using the gene deletion library. This work is sponsored by NIH Grant No. GM50237 (to R.R.) and grants from the Danish Natural Science Research Council (to M.L.) and the Alfred Benzon Foundation (to M.L.).